UN; Urban Drainage In Specific Climates

8/3/2019 UN; Urban Drainage In Specific Climates

http://slidepdf.com/reader/full/un-urban-drainage-in-specific-climates 1/158

INTERNATIONAL HYDROLOGICAL PROGRAMME

URBAN DRAINAGE IN

SPECIFIC CLIMATES

Chief EditorCedo Maksimovic

Volume III

Urban drainage in arid andsemi-arid climates

Editor of Volume IIIM. Nouh

IHP-V 1 Technical Documents in Hydrology 1 No. 40, Vol.lllUNESCO, Paris, 2001

(SC-200 1 wS/43)

Ei IRTCUD



The designations employed and the presentation of materialthroughout the publication do not imply the expression of any

opinion whatsoever on the part of UNESCO concerning the legalstatus of any country, territory, city or of its authorities, orconcerning the delimitation of its frontiers or boundaries.

The ideas and opinions expressed in this book are those of theindividual authors and do not necessarily represent the views of

UNESCO.



PREFACE

This volume, dealing with urban drainage in arid and semi-arid climates, is part of a three-volumeseries on Urban Drainage in Specific Climates within the framework of Theme 7: Integrated urbanwater management Project 7.3 lntegrated urban drainage modelling in different climates: tropical,

arid and semi-arid and co/d of the Fifth Phase of UNESCO’s International Hydrological Programme.The other two volumes of the series address urban drainage aspect in the Humid Tropics and in

Cold Climates. The Editor of this volume is Prof. Mamdouh Nouh of Sharjah University, United ArabEmirates, having led a team of 13 leading international experts in the preparation of this volume.

The series of the three volumes has been produced under the co-ordinating role of IRTCUD(International Research and Training Centre on Urban Drainage), an organisation established underthe auspices of UNESCO. Prof. Cedo Maksimovic of Imperial College, London, and Director ofIRTCUD, served as Chief Editor of the series.

Chapter 1 is an introductory chapter common to the three volumes, giving a general overviewof urban drainage principles and practice.

Chapter 2 discusses the specific climate features affecting the hydrologic characteristics of

arid and semi-arid areas. It points to the low, highly variable, rainfall, high evaporation rates, and thesparseness of vegetation and to the main features of arid zone landscapes.

Chapter 3 emphasizes the effect of such characteristics on urban drainage, showing thatmany problems originate from the application of traditional methods of design developed in otherclimatic zones.

Chapter 4 describes the various factors affecting runoff prediction and design of urban

facilities. Common methods of estimating design rainfall and runoff are also reviewed anddiscussed.

Chapter 5 addresses important issues related to urban stormwater pollution. Due to the dry

conditions and reduced vegetation in arid areas, higher pollutant concentrations and loads duringstormwater events occur in these areas. The importance of monitoring of stormwater quality whereaccurate estimates of pollutant loads are required is underlined.

Chapter 6 discusses the state-of-the art of the traditional methods of urban drainage in aridand semiarid regions, both for small and large catchments.‘

Chapter 7 proposes sustainable solutions for urban drainage pr-oblems in arid and semiaridregions, using, for instance, techniques of rainwater harvesting in the urbanised areas and adoptingthe methods of water spreading over the infiltrating surface of catchments.

Chapter 8 reviews practices of maintenance and management of urban stormwater drainage

appropriate for arid and semiarid regions.

.111



Chapter 9 describes the case studies dealing with: (i) the effects of dust storms onstormwater quality, (ii) stormwater hydrograph prediction in arid catchments using the Storm WaterManagement Model (SWMM), and (iii) effect of urbanisation on hydrograph components and onrunoff water quality.

This is the first time, to our knowledge, that the nature and problems of urban drainage inarid and semi-arid climates have been treated systematically, extensively and in depth in a singlevolume, thus our deep appreciation for this ground breaking effort goes to Prof. Nouh and his team

and to the Chief Editor of the series, Prof. Maksimovic. We trust that this work will benefit thenumerous countries in all continents that have arid and semi-arid regions. W e welcome commentsand feedback on this vaiume from users all over the world that may enable us to produce animproved edition later on.

The Secretariat of the International Hydrological Programme

Division of Water Sciences, UNESCO

1, rue Miollis75732 Paris Cedex 15, France

http://www.unesco.org/water/

iv





CONTRIBUTORS

Chapter 1: Theoretical BackgroundCedo Maksimovic, London, UK.

Chapter 2: General Characteristics of Arid and Semiarid Regions

L. Mays, Arizona, USA.

Chapter 3: Problems of Urban Drainage in Arid and Semiarid Regions

M. Nouh, Sharjah, United Arab Emirates

J. Simons, California, USA

Chapter 4: Storm Hydrology of Urban Drainage

Thomas A. McMahon, Francis H.S. Chiew, Tony H.F. Wong, Hugh P.

Duncan, Melbourne, Victoria, Australia

Chapter 5: Urban Stormwater PollutionFrancis H.S. Chiew, Hugh P. Duncan, Tony H.F. Wong, Thomas A.

McMahon, Melbourne, Victoria, Australia

Chapter 6: Traditional Methods of Urban Drainage in Arid and Semiarid Regions

M. F. Hamouda, Safat, Kuwait.

N. Al-Awadi, Muscat, Oman

Chapter 7: Sustainable Solutions for Urban Drainage Problems in Arid and Semi-arid Regions

M. Nouh, Sharjah, United Arab EmiratesK. Al-Shamsy, Alexandria, Egypt.

Chapter 8: Urban Drainage Maintenance and Management Issues in Arid and Semi-arid Regions

Tony H. F. Wong, Francis H.S. Chiew, McMahon, T.A., Duncan, H.P.,

Melbourne, Victoria, Australia

Chapter 9: Case Studies

M. Nouh, Sharjah, United Arab Emirates

A. Al-Rumhy, Jeddah, Saudi Arabia

V



CONTENTS

Chapter 1 General overview of urban drainage principles and practice ..................1.1 Introduction .....................................................1.2

1.3General characteristics of urban drainage and sustainability concept ........Urban drainage system as a part of an integrated river basin water ........management - principles of design and operation

1.4 Basic principles of rainfall-runoff and pollution modelling and outlook for theirapplication in particular climates .....................................1.4.1 Water quality aspects .......................................1.4.2 Quality aspects. ...........................................

1.5 Common UD models and needs for their improvements and update.........1.6 GIS and informatic support. ........................................

1.7 Concluding remarks and acknowledgement. ...........................Bibliography ...........................................................

Chapter 2 General characteristics of arid and semi-arid regions ........................2.1 Physical features .................................................

2.1.1 What is aridity?. ...........................................2.1.2 Geomorphology ...........................................2.1.3 Soil characteristics .........................................2.1.4 Aeolian systems ...........................................

2.2 Climate ...................................................... ......

2.2.1 Causes of aridity ...........................................2.2.2 Climate areas.

............................................2.2.3 What are deserts? ........................... :.' .............2.3 Hydrology .......................................................

2.3.1 Rainfall ..................................................2.3.2 Infiltration .................................................2.3.3 Runoff and flooding .........................................2.3.4 Erosion and sediment transport. ..............................

Conclusions ...........................................................Bibliography ...........................................................

Chapter 3 Problems of urban drainage in arid and semi-arid regions ....................3.1 Design particularities ..............................................

3.1 .l Rainfall ..................................................3.1.2 Evaporation. ..................... : ........................3.1.3 Infiltration ................................................3.1.4 Sedimentation .............................................3.15 Water quality ..............................................

3.2 Maintenance, operation, and management. ............................3.3 Data acquisition and processing .....................................3.4 Application of common urban drainage models .........................3.5 Interaction with other urban water systems .............................Conclusions ...........................................................Bibliography ...........................................................

Chapter 4 Storm hydrology of urban drainage. ......................................4.1 Effect of urbanisation on runoff. .....................................4.2 Estimation of design rainfall .........................................

4.2.1 Design event ..............................................4.2.2 Rainfall intensity-frequency-duration relationships .................

vii

1

1

1

4

7

710

15161619

2121212223242424262627272829323333

3535354141434447

5051525555

585859

5960



4.2.3 Temporal distribution of rainfall intensity ......................... 61

4.3 Estimation of design runoff. ........................................ 62

4.3.1 Rational method for estimating design peak discharge ............. 62

4.3.2 Methods for estimating complete runoff hydrograph ................ 63

4.3.2.1 Determination of rainfall excess hyetograph ............... 63

4.3.2.2 Time-area method for estimating runoff hydrograph ......... 64

4.3.2.3 Routing of stormwater runoff. .......................... 66

4.3.3 Event and short time-step stormwater runoff models ............... 66

4.4 Estimation of daily and longer time-step runoff. ......................... 67

4.4.1 Annualrunoff .............................................. 68

4.4.2 Daily and monthly runoff. .................................... 69

4.4.2.1 Simple conceptual model to characterise daily runoff. ....... 70

4.4.2.2 Conceptual rainfall-runoff model for estimating daily runoff .... 70

4.4.2.3 Data and model calibration and verification ................ 72

Conclusions ........................................................... 73

Bibliography ........................................................... 73

Chapter 5 Urban stormwater pollution. ............................................5.1 Stormwater pollution ..............................................

5.2 Urban stormwater quality process ....................................

5.3 Event mean concentrations of water quality parameters ...................

5.4 Estimation of stormwater pollution load ................................

5.4.1 Water quality monitoring .....................................

5.4.2 Estimation of daily and long-term pollution load ...................

5.4.3 Modelling of event and sub-daily time step pollution load ............

5.5 Other issues in the estimation of stormwater pollution loads ................

5.5.1 GIS and pollutant mass loading ...............................

5.5.2 Link-mode modelling of pollution generation and transport in largecatchments

5.5.3 Contaminants associated with different sediment sizes .............


Chapter 6 Traditional methods of urban drainage in arid and semi-arid regions6.1 Small catchments ................................................

6.1.1 Amount of storm runoff ......................................

6.1.2 Design of storm sewers .....................................

6.1.2.1 Determination of storm water flow rates ...................

6.1.2.2 Hydraulic grade calculations ...........................

6.1.3 French drains .............................................

6.1.4 Drainage system changes ...................................

6.1.5 Infiltration systems .........................................

6.2 Largecatchments ........................... .....................

6.2.1 Estimation of flow ..........................................

6.2.2 Urban drainage models ......................................6.2.3 Computer programs .......................................

6.2.4 Urban storm drainage system .................................

6.2.4.1 Gutters ............................................

6.2.4.2 Inlets. .............................................

6.2.4.3 Catch basins .......................................

6.2.4.4 Grated inlets. .......................................

6.2.4.5 Manholes ..........................................

6.2.5 System design. ...........................................

6.2.5.1 Storm sewer design. .................................

6.2.5.2 Gutter design .......................................

6.2.5.3 Street inlet design ...................................

6.2.5.4 Detention and retention storage facilities ..................

6.2.5.5 Stormwater culverts. .................................6.2.5.6 Infiltration ponds. ....................................

6.3 Evaluation of urban drainage methods. ...............................

7575

767881

81

818284

8485

8687

87

88

88

8990

90

919292929393

949496

96

9797

9797999999

100100101102103

.Vlll



6.4 Traditional vs modern methods. .................................6.4.1 Collection of stormwater ................................. : : : :6.4.2 Storage and reuse of stormwater. .............................


Chapter 7 Sustainable solutions for urban drainage problems in arid and semiarid regions7.1 Water harvesting .................................................

7.1 .l Design of rainwater harvesting system ..........................7.1.2 Maintenance of rainwater harvesting systems ....................7.1.3 Costs of rainwater harvesting systems ..........................7.1.4 Rainwater harvesting systems in arid and semi-arid areas ..........7.1.5 Types of rainwater harvesting collecting systems ..................7.1.6 Efficiency of rainwater harvesting systems .......................7.1.7 Modelling of rainwater collection systems ........................

7.2 Infiltration potential and natural drainage ..............................7.2.1 Comparison between different direct techniques ..................7.2.2 Managerial aspects .........................................

Conclusions ...........................................................

-Bibliography ...........................................................

Chapter 8 Urban drainage maintenance and management issues in arid and semiaridregions8.1 General ........................................................

8.2 Stormwater detention and retention systems ...........................8.2.1 Flood retarding basins ......................................8.2.2 Stormwater quality improvement facilities ........................

8.2.2.1 On-site detention basins ..............................8.2.2.2 On-site retention systems .............................

8.3 Channelmanagement...........................................~ ..


Chapter 9 Casestudies ..........................................................9.1 Effect of duststorms on stormwater quality .............................

9.1 .l Average TSP level .........................................9.1.2 Spatial variation of duststorm .................................9.1.3 Temporal variation of duststorm ...............................9.1.4 Catchment size .................... .-. ......................

9.2 Application of SWMM model ........................................9.2.1 Effect of catchment size .....................................9.2.2 Effect of infiltration rates .....................................9.2.3 Effect of rainfall characteristics ................................

9.2.3.1 Spatial variation of rainfall .............................9.2.3.2 Temporal variation of rainfall ...........................

9.2.3.3 Dry period between two successive rainstorms. ............

9.2.3.4 Total depth of rainfall .................................9.2.3.5 Duration of rainfall ...................................

9.2.4 Effect of duststorms. .......................................9.3 Effect of urbanization ..............................................Conclusions ...........................................................Bibliography ...........................................................

103103104104104

106106107107108109112114

116118118119119120

121

121122122124128128130131131

132135135135136137138139140141

141142

143144145146147147149

ix



Chapter 1

General overview of urban drainage principles and practice

1 .I Introduction

This- chapter is meant to serve as a common introduction to UNESCO’s three volumespublication dealing with urban drainage in three particular climate zones:

l ASA - Atid and Semi Aridl HT - Humid tropical and subtropical andl CC - Cold climate zones

It’s structure reflects the need to underline the similarities of urban drainage problemsin particular climate zones and to address the need for breakthroughs in both research andapplication of the adequate tool in these region. The chapter provides an introduction to thecontemporary state of the art in analysis, modelling, design and management of urbandrainage systems indicating that the particular aspects are covered in separate volumes foreach zone . The principles of “fitting” the urban drainage solutions into integrated wtchmentmanagement plans is introduced. Two principal components of the integrated flood mitigationsolution such as: structural and non-structural measures applied in two parts of a river basin(catchment) i.e.: urban and suburban zone and the rest of the catchment - rural and naturalareas are presented. The differences o f the situations in developing and developed countrieshave been highlighted.

The concept of natural drainage within the broader framework of sustainable solutionsis re-iterated and its major components will be presented by distinguishing between therehabilitation of aged systems and the construction of new ones.

The scope dependent nature of storm system modelling is presented in the form “anappropriate tool for each task”. The major types of modelling concept (quality, quantity,interactions, integrated) are briefly analysed, by placing an emphasis on data needs and datareliability as well as on the need for development of a new generation of modules that will beable to cope with particular aspects of specific climates.

1.2 General characteristics of urban drainage and sustainability concept

Water in urban areas; and urban storm drainage as a part of the urban infrastructure, aretopics which are gaining in importance in recent years. Cities now house 50% of the worldpopulation, consume 75% of its resources, yet occupy only 2% of the land surface. By themiddle of the next century, it is confidently predicted that 70% of the global population will livein urban areas. The number of megacities (> 10 million inhabitants) will increase to over 20,80% of which are in developing countries (Niemcynowicz, 1996). Properly designed andoperated urban drainage systems with its interactions with other urban water systems arecrucial element of healthy and safe urban environment.

The concept of sustainable development is provoking a profound rethinking in ourapproach to urban water management (ASCELJNESCO-IHP, 1998). Sustainable



development is that which “meats the needs and aspirations of the present generation withoutcompromising the ability of future generations to meet their own needs* (WCED, 1987). So,sustainable solutions have a “now” and a “then” component, and improvements thoughnecessary in the present must not be carried out at the expense of future needs andsituations. An alternative definition (IUCN-UNEP-WWF, 1991) asserts that sustainable

development is that which “improves the quality of human life while living within the carryingcapacity of supporting ecosystems”. Here, the emphasis is placed on mankind’s demand forand impact upon earth resources and the environment. Finally, Agenda 21 behoves us to“think global, but act local”. Public participation becomes important and demands individualresponsibility. Sustainable services must be environmentally friendly, socially acceptable andfinancially viable into the next millennium (Butler 8 Maksimovic 1999). The sustainabilityconcept calls for overall rethinking and this implies paying attention to particular situations inthe local area. Learning about natural and man made processes that affect the runoff qualityand quantity is of prime importance. This publication is thus expected to point out the mostimportant issues that affect the way that we analyse, design, build and operate our stormdrainage sys tems in a nature friendly fashion. Our current knowledge about the physicalprocesses involved is far from satisfactory, even in temperate climates where the most ofresearch has been carried out in the past. Knowledge about processes affecting urbanstorm drainage systems in particular climates covered in this publication (arid and semi-arid,

humid tropical and subtropical and cold) is far from satisfactory. However the publication isaimed at providing an up to date look at solutions to flooding and water quality problems.The concept of sustainability calls for amenity and resources recycling to be taken intoaccount as well. The authors are aware of the fact that many issues raised here requirefurther studies, research and development and that the issues raised will provoke furtherrefinements.

In densely populated developed countries (UK, Germany, some parts of the USA,Japan, etc.), urban drainage consumes a high proportion o f the investments into urbaninfrastructure. The reasons for this are the obvious need for an integrated approach to urbanwater management, and raised public awareness of the pollution caused by urban effluents,which affect both the urban areas themselves and the receiving water bodies. The situation indeveloping counties is also changing rapidly in the sense that all parties involved in planning,design, management and maintenance as well as funding ( World Bank, aid agencies etc.)

are becoming aware that storm drainage can not be ignored. On the contrary, it has to beincorporated into integrated urban infrastructure projects with their mutual interactionsencompassing not only the conventional problem of flood mitigation but also health hazardreduction (water quality concerns) and problems of urban amenities and resourcesmanagement (Figure I. 1).

Although cities are in contact with water from various origins (ground water, streamsflowing through or near the city etc.), the major concern o f urban drainage systems is wateroriginating in the city area itself, i.e. water from local rainfall (urban storm runoff) and itsinteraction with the water originating from the rest of the river basin.

The change of the role of urban storm drainage (USD) and developments ofinformation processing technology have imposed a need for new tools and products to beused in the problem solving procedure. Methods for flood protection by local storms and forassessment of the effects of pollution transported by storms on receiving waters have been

significantly improved during the past two decades with the introduction of computer basedsimulation, design, optimisation, real time control and management. The achievements ofmodem infonatics (i.e., a higher level of information processing) have made a significantimpact on all aspects of problem solving. However, despite significant development achieved,there is still a big gap to be bridged since a compact and reliable package that adequatelypredicts dynamics and spatial distribution of urban floods and that incorporates source controlmeasures does not seem to exist in the world.



Figure 1 I Stormwater quality, quantity and amenity and resources management of equalimportance

In modem societies, the status of urban drainage as a part of the integratedinfrastructure system varies from one country to another, depending primarily on the level ofdevelopment and the society awareness of the importance o f this problem. In general, theimportance of the system increases with the level of development, but there are alsoexceptions. The awareness of the wet-weather pollution potential has rapidly increased inrecent years. The systems, which used to have a simple function of collecting storm waterand conveying it to the nearest point of disposal as soon as possible, have gradually evolvedand are being replaced by the integrated systems which are gaining in importance. Their rolehas changed and now in addition to covering urban flood protection, pollution control andmanagement they are starting to cater for improvement of the quality of life by bringing waterfeatures - creating urban amenity in the city. Additionally, storm water is considered to be aprecious resource, which can be retained near the source to be reused, recharged to theunderground for aquifer replenishment or to create habitat for the return of wildlife todesignated urban areas etc.

Conventional urban drainage systems are separate such as shown in Figure 1.2 orcombined in which case both waste and stormwater share the same pipe. During dry

weather, water is directed to treatment plant (if existing) and during wet weather, part of themixed water in combined sewers diverts to receiving stream via Combined Sewer Overflows(CSO). If the city is served by a wastewater treatment plant, CSOs may be one of the majorpoint sources of receiving water pollution. In practice, separate systems rarely remain fullyseparate; there is always some storm water in foul system and waste water in storm systems.In most cases they behave like two combined systems with various degrees of waste waterdilution. Treatment plant suffers from intermittent overload during storm periods. Increasedenvironmental concern has lead to development of the concept in which, at least indeveloped countries, conventional storm drainage systems are gradually being replaced bythe systems based on runoff quantity and quantity control. The system consists of severaltechniques that aim at controlling the problem as near to the source as possible - thus theterm “source control”. They all attempt to mimic the natural processes involved. Thetechniques include storage, treatment and infiltration, by a “water management treatment

train” (Figure 1.3) that results in significant reduction o f peak and volume of runoff, improvedwater quality and a possibility of using storm water as a resource and as an element of urbanamenity. However, the means of implementing the element and principles o f this technologyin urban drainage in particular climates is an art still to be mastered despite significantachievement in some countries - for example Sweden, Stahre (1999) in cold climate, city ofCuritiba, Brazil, in tropical, several cases in Israel etc in arid climate conditions as resentedby Simon (1996). However, in order to reach greater sustainability in both conventional andinnovative urban drainage systems, better understanding of the physical processes,interactions between the systems and environment in particulate climatic conditions isneeded. This publication is supposed to wver part of missing information and to address theproblems that need further investigations.

3



Figure j .2 A conventional separate foul and surface water drainage

b Conveyance

Site qmtrol

Lg$3.l control i $. -

&fdtmtionv

Figure 1.3. Surface drainage management train - likely sustainable solution (ClRIA555)

1.3 Urban drainage system as a part of an integrated river basin water

management - principles of design and operation

It is well known that the river basin has been considered an entity that determines both therange and reach of human activities with respect to water in both ancient and modemsocieties. The catchment is used as a unit for planning and management of not only water,but also of other resources, as well as human and economic activities. In the case of urbandrainage of a particular city, the relevance of a catchment is greater for smaller catchmentsand decreases as the size of the catchment increases, in the sense that the relative effect ofthe quantity and quality of runof? water generated by that particular drainage systemdiminishes with the size of the catchment and with the distance from the point of storm waterdisposal. However, the integrated effect of all stomt drainage systems contributing to thebalance of surface water and to the flux of suspended sediment and other pollutants has to betaken into account at the level of river basin or sub-basin upstream of the point under

consideration, especially in densely populated areas. The interaction of storm drainage

4



systems with downstream municipalities and water users is strong in those cases when thedrainage peak flow uses up the capacity of the river channel, so that no capacity is left fordownstream runoff. In these cases, the downstream-upstream relationships and links have tobe analysed in order to either share the existing capacity or to share the costs o f itsenlargement. Small river basins in densely populated areas are therefore more sensitive tothis problem and shall be analysed in the following discussion. On the other hand, the rivers

carrying water from large catchments serve as receiving waters for both solid and dissolvedpollutants, and the effect of urban storm water disposal has to be analysed from the point ofview of its pollution and contribution to the silting of downstream water, including reservoirs.

Alterations to the natural water balance within the catchment area can have bothpositive and adverse effects on upstream and downstream water users. In that respect,integrated planning and design of urban drainage systems requires that both effects areanalysed and an unbiased assessment is made in all phases of the planning andmanagement process. Figure 4. outlines an approach which integrates catchment wide,metropolitan/municipal as well as local area planning and management considerations.

T MANAGEMENT

WATER Nc’PPL.\

Figure 1.4. Urban Storm Water Master Plan as a part of the Catchment Management Plan

The general goal of integrated water management is a sustainable utilisation of waterresources respecting the social, economic and environmental interests. Considering the closeinterrelationship between the society and economy, the first two groups are usuallyaggregated into socioeconomic issues. It also includes institutional issues. It should berecognised that the goals and objectives of integrated water management are formulated atvarious spatial scales, involving all three components, According to Butler and Maksimovic(1999) the institutional aspects cover the following:

l Development of improved informatic support tools for planning, design andoperational management based on improved quantii and quality of data.

l Incorporation of more (relevant) components and stakeholders into the decrsion-making process (e.g. sustainability, public attitudes).

l Development of methodologies to evaluate the uncertainty and risk associated withfuture water management strategies.

l Decision on how to consult and educate the public concerning the importance o furban water issues.

l Devising suitable organisational/institutional structures to incorporate theintegrated, holistic system management we advocate.

l Enacting appropriate supporting water legislation and standards.

The fundamental qualities of integrated water management are its holistic nature, whichrecognises the system complexity and inter-connectivii of its elements, demonstrated byexchange of information, energy and matter, and the style of planning actions. The holistic

5



approach also equally involves local/municipal and regional authorities, engineers andnatural scientists, environmentalists and decision makers, politicians of all patties, governingand in opposition, as well as the people affected (Geiger, 1994 and Geiger & Becker, 1997).Sustainable water management ensures that no matter is accumulated or energy is lost, byrecovery and reuse techniques. This approach requires novel, environmentally soundtechnologies. In the urban drainage field it calls for a wider application of source control. Inthe context of urban and industrial water resources, the most pertinent water uses are watersupply (safe, reliable and equitable), drainage and flood protection (affordable), sanitationwith maximum reuse, recreation (protecting public health), aesthetic and cultural values, andecosystem health. Solutions applied at urban catchment level have to be analysed in termsof it upstream and downstream interactions. The conditions may vary in various climateconditions and these will be analysed in the main chapters of this publication .

Contributions of urban storm drainage projects to the conflicts and uncertainty in waterresources plans at a river basin level, can be analysed by taking into consideration the waysin which the existing urban structures, their features, and the newly planned drainageelements affec t both water balance and quality in a particular urban area. In this respect,the major difference between urban and rural (or natural) part of a river basin, is the reducedinfiltration potential o f urban areas and the fast response in generation of surface runoff. Amutual interaction of urban runoff and flows in adjacent steams is shown in Fig. 5. Water

running from the upstream parts of catchments flows either through the city’s regulatedstream, or through its system of urban drainage infrastructure. The major difference inapproaches to integrated solutions is indicated by the ratio of the urban peak flow to the flowin the receiving stream, at the downstream end of the urban area. The forms of urban floodingcaused by other man made and natural disasters such as storm surges that usually coincidewith heavy rainfall , dike break (Iwasa, Inoue, 1987) have also to be taken into account.

We take the Danube, as a large river flowing through the large cities of Vienna,Bratislava, Budapest, Belgrade, etc., as an example. In the most extreme events of heavystorms over these cities, the local runoff contributes only a very small proportion of the flow inthe river, and one can claim that the management of urban storm drainage system in thesecities will not significantly affect the flow in tie Danube. This effect is diminishing as onetravels form the spring to the river mouth. In that respect, and with reference to water quantity,these systems do not affect the peak flow in the receiving water and can be designed

independently. However, small streams near large cities (i.e. small value of the factor Y -equation 1) are strongly affec ted by the urban runoff peak flows, and implementation ofsource contro l could be strongly recommended. In some cases it is the only solution. In manycases, the peak flow generated by urban runoff is comparable to the conveyance capacity ofthe receiving stream. In these cases, the management of urban drainage has a significanteffect on the receiving water and its downstream reaches. Consequently, the solution for theparticular storm drainage problems has to be developed at the catchment level, and in anintegrated way. However, growing wncems over the quality of surface runoff require that theinteraction of particular city’s pollution load is addressed in conjunction with other pollutioncontributions from both upstream and downstream urban areas. The difference in the capacityof the main receiving water calls for classification of the concepts of storm drainage solutions,depending on the ratio of the peak flows, likely to occur at the point o f disposal (end of thepipe), to the average discharge in the receiving stream (shown in Figure 1.5.)

w=

QR,,-QR

QR(1.1)

where :QR = inflow at the upstream end of the urbanised area; Qh = outlet flow at the

downstream end of the urbanised area.



Al

Figure 1.5. Classification of urban sub-catchments and interaction between urban runoff andthe adjacent river.

In the cases when the receiving urban stream reaches its full capacity under the effectof a given return period flood wave. Source control solutions will be strongly dependent on thevalue of the factor !Pand of the urban catchment location (part of the city at high elevationsand storm water drained by gravity, or part of the city at low elevations which can be floodedby receiving water). A high value o f the factor Y, means the high capacity of the receivingwater, thus the storm runoff from the local urban area might not affect water level in the river.Thus the implementation of source control may be less beneficial to the for the part of thecity located on the left river bank situated in the higher elevations (Figure 1.2.) than to theportion of the city located on the opposite side of the river. In the latter case, theimplementation of source control measures could result in significant savings in bothconstruction and operation management costs. Both the cities of Budapest and Belgradeare good examples in which the portions of the city on opposite sides of the receiving waters(Danube in the case of Budapest and both Sava and Danube rivers in the case of Belgrade),have completely different flooding vulnerabilities and different source control suitability. Thusthe portions on different river banks would benefit differently if source control measureswould have been applied.

As a conclusion to this section one can say that there are no universal rules of thumbfor implementation of source control techniques. The most appropriate solutions have to befirstly sought through the resolution of conflicts between land and water users at catchmentlevel and than at the level of municipal storm water plan. Both structural and non-structuralmeasures have to be analysed in terms of the suitability to an application of source controland benefits that can be reached. Stormwater quality issues, which were almost ignored inthe past, have to be addressed properly in terms of their spatial and temporal distribution andeffects on receiving water bodies. A possible approach in the evaluation of suitability byGIS support is given in the paper of Macropoulos et al ( 1998) and will be discussed later inthe section 05).

1.4 Basic principles of rainfall-runoff and pollution modelling and

outlook for their application in particular climates

1.4.1. Water quantity aspects

Modelling in urban drainage serves various purposes such as overall assessment o f thecatchment response as a part of strategic and master planning to detailed network andancillary elements design, assessment of pollution, operational management, real timecontrol and analysis of interactions among subsystems. The type of model applied depends

on the goal of modelling, spatial coverage, data and technology availability but most oftenon the knowledge, skills and experience of the modeller. Once familiarised with a certain

7



model, the user tends to apply it even in the cases in which that particular model is notappropriate. In principle the simple lumped models (black box and similar) in which thewhole catchment is treated as an entii, can provide reliable results and good fit withmeasurements obtained on the same point from which the data have been used forcalibration. One cannot expect to get realistic results for the points within the catchment (network) unless the measurement is performed on that point and “new model” obtained by

calibration against that data sets. The only reliable approach is to obtain more reliable data ona catchments physical characteristics and then develop and apply physically based model inwhich uncertainty is reduced by replacing the role of physically meaningless parameters withthese characteristics. The general principles of conceptual and physically based models ofboth water quality and quantity have been known for several decades Maksimovic,Radojkovic (1966) , Yen (1966) O’Laughlin et al (1996).

Detailed description of modelling principles in temperate climates in which thesemodels have been developed, and from which the data were wllected for model calibration, isbeyond the scope of the present chapter. More details on the attempt to use these models inspecific climates will be given in the separate chapters.

The conceptual models are based on assumptions such as constant runoff coefficient,SCS curve numbers, rational formula, time-area, unit hydrograph etc. originating yearsbefore the computers reached the level of development that allowed their broader application

in daily practice. Although they were developed for application in natural and ruralcatchments, they continued to be used in urban areas, where the conditions are significantlymore complex, spatial variability of soil and impervious areas require much finer spatialresolution, and man made object require detailed specification of infrastructure system andtheir interaction with the flow pattern. When property calibrated against measurements,these models can produce seemingly logical results especially if one is modelling the wholeurban area as a single catchment and model calibration performed against data in one point.This could be useful for example in design of centralised storage facilities, inflow to treatmentplants and similar cases when the response of the whole catchment is considered. However,for detailed runoff modelling of complex features such as trunk systems with broad sub-catchment areas, street drainage systems with detailed property drainage components andsub-catchments, models of this nature generate results of high level of uncertainty.

Physically based models in which a more detailed presentation of the catchment

characteristics are made and distributed modelling is applied should theoretically be lesssensitive to subjective assessment of model parameters. In the simplest terms, the wholecatchment is divided (delineated) in smaller sub-catchments which, depending on thepurpose of modelling, can vary in shape and size arbitrarily as to accommodate, the mostrealistic model presentation of flow pattern (Figure 1.6). The temperate climate approachconsiders the following element of modelling :

l Rainfall as an input: single storm, series, historical rainfall, etc.0 Interception (surface depression)l lnfilJration (steady, unsteady, unsaturated soil, simple solution or Richard’s equation

based solutions)0 Surface runoffl Gutter flow

l Flow in ancillary structures0 Pipe flow

When it comes to runoff modelling in specific climate conditions it is evident that thisapproach needs further upgrades as to accommodate features like:

l Different forms of precipitation (snow fall in CC)l Different forms of interception (HT)l Different forms of runoff formation (snow melt in CC)l Effec t of different cultural, planning, building and other effect on interception (ASA

and other)l Strong interaction of surface runoff with sediment transport (ASA)l Lack of proper infrastructure and interactions with solid waste and waste water (

developing countries, low income habitats)l Interaction with ground water (infiltration, exfiltration - all)

8



l Interaction with source control featuresl Interactions with real time and other control structures.

It is evident that these features require separate attention, although not necessarily acompletely new model. Most of them can be accommodated into reliable, well conceptualisedphysically based model. This does require more knowledge, reliable data, properinterpretation.

Some of the shortcomings of the models o f this nature as presented in the paper ofMaksimovic et al (1999) are:

l Concepts dating back many decades. The development of contemporaryinformation technology’s computing power has not always been mirrored byimprovements in the models,

l The outdated concepts are often hidden behind powerful graphics and presentationglamour

l Modelling of urban water interactions are almost nonexistent and integratedmodelling is in its infancy

l

Many models lack modularity, transparency and transportability (automatic “scalingup and down”),l Data quality and completeness is usually not property addressed by software

developers, its users often lack the knowledge of basic assumptions on whichmodels are built,

l Data acquisition and processing are not compliant with model structure andcomplexity, or models are not capable o f producing proper results from availabledata base (DRIPS Syndrome - Data Rich Information Poor Systems

l Complete d igital data on the urban infrastructure and on the spatial distribution ofbasic urban environment features (land use, DEM- digital elevation models etc.) israrely found at an appropriate horizontal and vertical resolution,

l Thorough testing against high quality data sets is often exercised neither bydevelopers nor by users,

l High level of independent, international verification of new products is rarelyperformed. In-house verification tends not to reveal the weak points of the products,a proper educational component is often missing.

Additionally, O’Loughlin et al (1996) claim: ” Despite this availability of information, tools andguides, and the success of rainfall-runoff models in providing generally acceptable basis fordesign of infrastructure works, there are limitations to the modelling of rainfall-runoffprocesses . They identify four major reasons for this:

l Insufficient datal Variability of rainfall inputsl Insufficient temporal detaill Model incompatibility

Concerning the level of detail they point out that engineers have long been skilled atidealising or conceptualising systems, to produce manageable models involving typically 10 to100 elements to represent a complex urban drainage network. Now that there is a capacity towork with more detail, it is necessary to look at appropriate levels for various tasks and therelationships between models of various scales.

It has already been mentioned that for studies concerning general response (in theterms of both quality and quantity) of the catchment or sub-catchment of a considerable size,it may suffice to apply a lumped approach in which spatial variability of catchementcharacteristics as well as of precipitation is ignored. Providing that reliable measurements atthe end of catchments are available, the results of input-output correlations are used instead.Some models of this nature will be discussed in the particular chapters of separate volumes.

9



1, Ckuchmcnt with dminage system i direct surface runoff TO-GALRAiNF.4LL2 impervious surfact: AAAAbbbbb

ret;ltion3 pervious surface

retentioa4 ierception

.infiluation6 gurtcr flow

2:

7 base flow I-I8pipeflow I

b) WAIIER BALANCE

~~ ) AN EXAMPLE OFa) DELLNUTION PRINCIPLE DELINEATEDURBAN AREA

Figure 1.6. Summary of physically based approach requiring a reliable catchment delineation

1.4.2. Quality aspects

Storm water runoff becomes polluted when it washes off concentrated and diffused pollutionsources spread across the catchment. An example of the average concentrations found instorm runoff is presented in Table 1 l . (Source: Xanthopoulos and Hahn 1993 and Cordery1977).

In addiion to soil ero&on caused by raindrop impacts and shear stress action, twomajor sources contribute to storm water pollution in temperate dimate zones:

a. diffused sources (Figure 1.7) originating primarily from atmospheric fallout andvehicle emission, additionally spread by the vehicles and wind and

10



b. concentrated sources originating mostly from human activities - bad housekeeping(industrial wastes, chemicals spread in urban areas - gardening for example)exposed to and widespread by wash-off by storm runoff.

Both of the processes generate soluble and suspended material. Throughout theprocess of transport, depending on hydraulic conditions, settling and re-suspension takes

place on the surface and in pipes, as well as biological and chemical reactions. Theseprocesses are often considered to be more intense in the initial phase of the storm (first flusheffect), however, due to temporal and spatial variability of rainfall and flowing water, first flusheffects are more pronounced in pipes rather than on surfaces Deletic (1998), where highconcentrations of pollutants can be expected throughout the runoff process. The success ofrunoff quality modelling exercise is strongly dependent on the quality of model (its reliability torealistically reproduce processes taking place in nature), and the reliability of data againstwhich the model has been calibrated

Table 1 .I. An example of average concentrations of pollutants in storm runoff

,,_ uantity Mean Concentration

Conductivity (uS/cm) 108-470BOD (mg/l) 7.3- 15

TOC (mg/l) 26-28.3NH4(mg/l) 1.92-2.75

Pb Ms/l) 160-525

iWW 320-2000

Wg4 35-57

PH 6.47-6.78

COD (mg/l) 47-146

DOC(mg/I) 3.1.-5.1

P(mdl) 3.1.-5.1

W-w/U 1.6-2.95

Cd(W) 2.8 - 6.4

CN&l~l) 23-184

Coliforms (/I OOml 2.2 - 5.6 (10*6)

Similar to quantity modelling, storm runoff quality modelling can be undertaken at

various levels of complexity, starting again with simplest input - output relationships. Moreadvanced models deal with spatial distribution of diffused pollution sources and analysis ofunsteady process of incipient of solid particles motion, bringing them to suspension,

transport along the paved areas, deposition in grassed areas (Deletic 1999) transportthrough the pipes and disposal either into receiving water body or into treatment plant.

In order to enable the comparison of modelling approaches between the models being

used in temperate climates with those in development or in need to represent the conditionsin other climate conditions, the basic principles of quality modelling are briefly summarised.Most of the models in current practice model the runoff quality by correlating theconcentration of pollutants to the concentration of particles of suspended solids which aremodelled in the phase of build-up and wash-off. The most common approach in build-upmodelling is based on the assumption of an exponential relationship between the amount of

solids available on the surface, M, and the duration of antecedent dry weather period, fdV.

This equation was adopted in the model of Deletic at al (1977) - Figure 1. 8:

(1.2)

where M [g/m*] is the amount of solids available on the surface, T [day] is the time elapsedfrom the start of the first rainfall in the series, t dV [day3 is the duration of antecedent dry

weather period, and t’ [day] is the virtual time, M, [g/m ] is the maximum amount of solids

expected at the surface, and k [day-‘] the accumulation constant.The virtual time is calculated by assuming that deposition is zero at t’days before the

start of the antecedent rainfall, as indicated in Figure 1.8.

11



A spatial distribution of solids is modelled, based on records from the literature, adifferent approach to prior models, which all assume that sediment is distributed evenly overthe modelled surface.

It should be noticed here that this approach build-up modelling could be successfullyused in ASA climates where most of the solids accumulated are either atmospheric depositor are transported by wind. However, in cold climates where a great deal of pdlution isexperienced in the snowmelt period from de-icing activities, which are not uniformlydistributed over the entire catchment, alternative methods have to be applied (for exampleGIS supported spatial distribution of salt used in de-icing). In this respect a critical evaluationof other models used in both quantii and quality modelling in particular climate conditionsshould be made as for their suitability for application in specific climates.

snowmelt

industri 18%nd landfil.

human activities

Figure 1.7. Diffused pollution sources in urban area

The reliable modelling of suspended solids wash-off has to be combined with surfaceand pipe flows to which the solids entrainment module has to be attached. The approach

applied in Deleti} et al (1997) will be used as an illustration. In this approach, the solids wash-off one dimensional mode l contains the following sub-blocks: 1. overland flow; 2. solidsentrainment; 3. suspended solids transport by overland flow. Overland f/ow is modelled usingthe kinematic wave equation, which has been used before for the modelling of surface runoff.So/ids entrainment is assessed by a new method, developed by the first author, whichconsiders independently rainfall and overland flow effects on amount of material lifted fromthe surface.

The rainfall effect is assessed by means of the kinetic energy of rain drops, while theeffect of flow is expressed by shear stress. One calibration coefficient is needed for thismethod. The general principles of modelling will be described in more detail enabling thus thecomparison to be made between the commonly applied approach and the one that could beused in presenting the specific aspects wash-off in ASA, CC and HT climates. Physicallybased modelling deals with mass and momentum conservation principles which are that

12



simplified or adjusted for the specific features of the particular catchments ‘characteristics,boundary conditions internal and external local climate induced boundary conditions.

For a unit width of the road surface (Fig. 1.9.a) the continuity equation, Eq. 1.3, and thefull momentum equation, Eq. 1.3, can be written as:

ah+aq ,a 77" (1.3)

(1.4)

rain rain rainMC-------------------------------

Figure 1.8: The concept used in modelling of solids build-up at the surface

-1

Figure 9: a) Road surface flow; b) Gutter flow

where, h [m] is the water depth, q [m3/s/m] is the unit ovetiand flow, ie [mk] is the effective

rainfall intensity, Ss [-] is the surface slope (the natural slope of the stree t surface), tb [Pa] isthe bed shear stress, ti [Pa] is the additional shear stress due to rainfall drops, x is the spatialcoordinate, and t is the time from the start of rain. It should be noted here that the sourceterm on the right hand side of the equation 1 is based only on the contribution from directrainfall. In CC conditions for example this term has to be modified as to include the effects ofsnow melt and freezing, which have to incorporate the temporal variations theirthermodynamic properties. Similarly in HT and ASA conditions it might be necessary toinclude the evaporation term which has not been included here.

The initial and boundaty conditions are given below,

W,O) = O;qW) = O;q(O,t) = qup 0) (1.5)

where qup i’s the unit overland flow at the end of the upstream section. The effective rainfallintensity ie was calculated by Linsley’s equation,

13



i, =i(l-edP’Yd) (1.6)

t

where i [m/s] is rainfall intensity; P = Iidt [mm] is the total amount of precipitation up to time

0

f, and yd [mm] is the retention coefficient and is dependent on surface type.The bed shear stress, tb was defined as:

Tb =;c,,(;)2

The friction coefficient, Ct,, depends on the flow type :

I3 Re<Re, laminar flow

CTb = R;,

-Re > Re,, turbulent flowRec3

(‘1.8)

where Re=q/v is the Reynolds number and v is the water kinematic viscosity. C,, CA and CJare constants that depend on surface type, and Ret, is the critical Reynolds number betweenlaminar and turbulent flow.

The effect of rain drops was modelled by an additional shear stress, IL‘~which isdifficult to define separately. Therefore, the total shear stress, written as,

(1.9)

which incorporates both phenomena was used.The local and convective terms (marked as Term 1 and Term 2 in Eq. 1.4) as well as

the pressure gradient term (marked as Term 3) are much smaller then the remaining two andare usually neglected and only gravity and friction terms (Term 4 and Term 5) were keptwithin the dynamic equation. The resulting equation is well known as the kinematic waveequation, and has been used for the moddling of both overland and gutter flow .Consequently, for modelling of gutter flow the following equations can employed (Fig. 1.9.b).

aA aq~+-g=qW +qr+i,L,

Q=0-375H8/3

n tgcp(1.11)

where Q [m3/s] is the gutter flow; qw [m2/s] is the unit inflow from the sidewalk; qr [m2/s] is the

unit inflow from the road surface; ie [mls] is the effect ive rainfall intensity Lg [m] is the gutter

width, A [m2] is the cross section area, H [m] is the water depth by the curb, n [m-1’3 ] is theManning roughness coefficient, Sg [-] is the longitudinal slope of the gutter, p p] is the

transverse angle of the gutter. Eq. 1 ll is known as luard’s (1946) formula which differsslightly from Manning’s expression, but gives be tter results for the shallow flow in a triangularcross section channel . In specific climate conditions the right hand side of the equation 1.10can be modified as to include additional terms the contribute to water balance.

It was assumed that there is no flow at the beginning of a rainfall event (the initialconditions). The inflow from the upstream reach was used as an upstream boundarycondition. Furthermore, the solids entrainment, pollution transport by overland flow and gutterflow are modelled by making use of kinetic energy of rainfall drop impact, carrying capacity ofsurface runoff and principle of turbulent transport and diffusion in open channel flow (Deleticet al. 1997). Although these principles are universal thus applicable in other climateconditions, the appropriate modifications have to be made in transport and diffusion

equations in order to incorporate their specific conditions, mainly in the source and sink terms14



of mass conservation and transport equations. Some of these principles are discussed in themain body of the text, however it should be noted that they are to be further investigated,tested and checked against reliable data. In this respect this publication is to be seen as asource of information on both current practice and need for further investigations in order torealistically reflect the conditions in particular climates. Additionally, it is noted that the aboveconsiderations have only dealt with suspended solids.

1.5 Common UD models and needs for their improvements and update

Physically based models are based on the analysis o f processes on the surface and innetworks, and is performed by taking into account detailed features on the surface

(topography, soil characteristics, land use, connectivity between elements etc.) and of thenetworks and ancillary structures. This section will mention just a few (more detailedpresentations are given in the other chapters ) of the existing models available either freelyor commercially:

l SWMM (US EPA’s Storm Water Management Model) - Huber (1995). This is oneof the first models developed, with a high degree of physically based principlesincorporated. Its initial versions (still in frequent use in its original main frameversion) have served as a basis for development of the other models which havetaken advantage of later development of personal computer technology.

l Hydroworks (HR Wallinford -Wallingford Software)The latest versions of the package are user orientated and can be used formatching with data sources and in composition of reports.

l MOUSE (DHI -Danish Hydraulics Institute-1990) Broadly used internationally. Thedevelopers have made an effort to incorporate some of the developments of PCtechnology (for example data base management in network simplification). Adiscrepancy is noticed between versatile pipe flow model and surface runoff onewhich would benefit from upgrading and proper matching with GIS and surfaceflooding routines.

l

Hystem E&an (ITWH - Fuchs and Scheffer ( 1990)l Bemus (IRTCUD, Djordjevic et al 1998)

However the models seem to have reached a level in which most of the modeldevelopers seem to have lost enthusiasm for further upgrade and improvement of models’capability in dealing with complexity of urban environment. Adding powerful graphic andwlourful images does not contribute to the reliability of modelling as long as the upgrade ofthe physical background is not improved. In addition to the above specific particularfeatures of particular climate regions, the following aspects need to dealt with in either modeldevelopment or customising for application in particular climate conditions:

l Capturing, filtering, compaction and processing of high spatial resolution data(primarily obtained by remote sensing. These data would enable a better

representation of terrain and land use) and its matching with GIS tools, the use ofwhich could enhance the analytical power of the models.l Analysis of the effect of maintenance and management practices (de-icing, sewer

flushing, gullyspot cleaning, street sweeping and of the other storm runoff andquality relevant activities) on water quality

l Analysisof the effects of source control practicesl Surface flooding (interactions of surcharged underground network with superficially

flooded areas, flood risk analysis)

These new incentives seem to be needed for a significant breakthrough to be made.This publication aims to provide some material which could serve as a guideline fordevelopment of new generation of models or improvement of the existing ones.

15



1.6GIS and informatic support

Geographical Information Systems are know to deal with acquisition, processing andimplementation of data of a spatial nature (Boroughs 1988). Despite significant progressbeing made in this technology and its application in various water and environmental

engineering fields, their application in urban drainage is still relatively limited. Significantprogress has been made in the use of GIS based data in creation of data bases linked urbanwater infrastructure system simulation models (for example AquaBase - Kuby 1998). For thecreation of initial data sets (GIS layers) various sources of data can be used (Figure 1 lO).The systems are extremely powerful in providing input data to models after the elementarymanipulations with layers presenting physical features of the catchment (such as elevationmodel and land use) and superficial and underground network have been performed. Startingin the late eighties, with some of the first papers on GIS application in urban drainage - Elgyet al (1993) the research group of the present author has developed a methodology forhandling arbitrary data sources and automatic creation of input files for storm drainagemodelling. An example of data preparation for creation of input files for catchment delineation(Maksimovic (1995) is given in Figure 1 l 1. The results o f application of catchmentdelineation is presented in Figure 1.12. Figures 1.13 present the results of application of GISfunctionalities in the analysis - assessment of the suitability of a catchment of implementationof source control techniques and Figure 1 I3 depicts the results of the application of thisanalysis in the survey of the applicability of source control in the same catchment(Macropoulos et al 1998 and Macropoulos et al 1999). The works of Prodanovic (1999) andDjordjevic et al (1998) provide further development towards GIS - assisted physically-basedflood modelling in urban areas based on the dual drainage concept.

There is a huge unexploited potential of GIS application in particular climates. In theindividual chapters, authors present current techniques in data analysis and modelling. Mostof the specific features of the urban catchment in particular climates are of a spatial naturewhich renders them particularly applicable to quantification by GIS (e.g. suitable forapplication of GIS. It can be used in quantification of both physical features (such as soilpropensity characteristics, soil erosion, pollutant potential distribution, snow cover, asphalttemperature, solar radiation exposure). These and other GIS applications are yet to beresearched and made a part of the daily routine.

1.7. Concluding remarks and acknowledgement

The material presented in the present three volumes is result of the team work of numerousspecialists gathered around the UNESCO IHP V programme under the theme 7: IntegratedWater Management in Urban Areas within the Theme 7.3. Urban Drainage in specificclimates. The series of the three volumes has been produced under the co-ordination role ofthe regional IRTCUD (International Research and Training Centre on Urban Drainage) unitsfor particular dimate regions : humid tropical in Brazil, cold in Norway and arid and semi aridin Sharjah. The production of the present volumes would be impossible without UNESCO’sendorsement and coordination roles of the key w-editors: Prof. Carlos Eduardo Morelli Tucci(for HT volume), Dr. Sveinnung Saegrov, Mrs. Jadranka Milina (MSc) and Prof. Sveinn T.Thorolfsson (for CC volume) and Prof. Mamdouh Nouh (for ASA volume). Thanks are due tothe contributing authors of the chapters in individual volumes. Their names are listed in therelevant vdumes.

It is sincerely hoped that that publication of these three volumes will encourage furtherresearch and development in those regions in which there is still much to be learned aboutthe governing physical processes and in which the most appropriate sustainable solutionscan be found to the problems of urban flooding, storm water quality management, amenitydevelopment, provision, enhancement and resources recycling.

16



Q Video images

5 Satellii images

l GPS Data

s=l Digital data from

c Photogrammetry total stations

%Dynamic positioning 8 bathimetry data

Figure 1.10. Sources of data for GIS applications

Pre processing

of primary data

and creation of

secondary files

for

subcatchment

delineation

Figure 1.11. Pre processing and post-processing of data for catchment delineation

Fig. 1.12. GI.S supported catchment delineation (Maksimovic et al 1994)

17



Fig. 1.13. Suitability of the Klka catchment for application of infiltration techniques

-30% to -24%

-24% to -18%

-18% to -12%

-12% to -6%

-6%tol%

l%to7%

7%to13%

13% to19%

19% to25%

25% to31%

31% to37%

37% to 43%

43% to 49%

49% to 55%

55% to61%

61% to67%

67% to 73%

II 73% to79%

Fig.l- 14. Reduction in maximum water level for 10 years return period rainfall

18



Bibliography

BOROUGHS, P. A (1986) ‘Principles of Geographical lnforrnation Systems for LandResources Assessment: Oxford Science Publications, Oxford

BUTLER, D.; C. MAKSIMOVIC (1999) ‘Urban Water Management; Challenges for the nextMillennium’ . Progress in Environmental Sciences, Vol. 1 No. 3 p. 213-235

CIRIA (1999) ‘Sustainable urban drainage systems; Design manual for. Scotland andNorthern Ireland’ Report No. C521, Construction Industry Research and InformationAssociation, London

DELETIC, A; C. MAKSIMOVIC; M. NETIC (1997) ‘Modelling of Storm wash& ofSuspended Solids from Impervious Areas’. Journal of Hydraulic Research , Vol. 35,No. 1, p. 99117.

DELETIC, A. (1998) ‘The First Flush Load of Urban Surface Runoff. Water Research, Vol.32 No. 8, p. 2482-2470.

DELETIC, A. (1998) ‘Sediment Behaviour in Grass Filter Strips’. Water Science andTechnology, Vol. 39, No. 9, p. 19138.

DHI (1999) ‘MOUSE Modelling of Urban Sewer Systems on Microcomputers; Users Guideand Technical Reference’. Edition 3.0.

DJORDJEVIC, S.; C. MAKSIMOVIC ; D. PRODANOVIC (1998) ‘An Approach to Simulationof Dual Drainage’. Water Science and Technology, Vol. 39, No. 9, p. 95-194.

ELGY. J., C. MAKSIMOVlC; D. PRODANOVlC (1993A) ‘Matching Standard GIS Packageswith Urban Storm Drainage Simulations Software’. Sixth International Conference onUrban Storm Drainage, Niagara Falls, Canada

ELGY. J., C. MAKSIMOVIC; D. PRODANOVIC (19938) ‘ Using Geographical InformationSystems for Urban Drainage Hydrology’. Proc. of the International Conference onApplication of Geographical Information Systems in Hydrdogy and Water Resources,HydroGlS 93. K. Kovar and H. P. Nachnebel (ed) Vienna, Austria

FUCHS. L.; C. SCHEFFER (1991) ‘HYSTEM-EXTRAN, Version 4.3, Model Description andDocumentation, Microcomputer in Urban Drainage’. Inst. fuer Technishe andWissenschaftlische Hydrologie,Hanover.

GEIGER, W. (1998) ‘Principles of Integrated Water Management for the Revival of Old

Industrial Areas’. Springer Vet-lag, ASI Series, p. 57- 108 , Editors: A. G. Buekens andV. V. Dragalov. Geiger, W. F. and M. Becker (1999) ‘Revisiting the Past? NewApproaches to Urban Drainage in the Emscher Area’, Proc. of EngineeringFoundation Conference: Sustainable Urban Water Resources in the 21*. Century,Sept. 7-12, Malmo, Sweden, Publ. ASCE, Editors: Ch. Rawney, P. Stahre, L. A.Roesner, pp 123 -135 Haestad Method (1997): Computer Applications in HydraulicEngineering. 166 p.

HUBER, W. (1995) ‘EPA Storm Water Management Model - SWMM’. In V.P. Singh (ed.)Computer Mode/s in Watersheshead Hydrology, Water Resources Publication, pp.783 898

IWASA,Y. ; K. INOUE (1987) ‘Comparative Study of Flood-Runoff Analysis in View ofDisaster Research’. Proc. US-Asia Conference on Engineering for Mitigation NaturalHazard Damage

IZZARD, C. F. (1948) ‘Hydraulics o f Runoff form Developed Surfaces’, Proc. of HighwayResearch Board p. 129150.

KUBY, R. (1998) ‘Common Ground for GIS Support to Water Supply, Sewerage andDrainage (AquaBase)‘Training course: GIS in Urban Water, Internal publication ofImperial Cdlege and CUW-UK,

MAKSIMOVIC, C. and M. RADOJKOVIC (1986) ‘Urban Drainage Modelling’, Proc. Int. SympUDM’88 Dubrovnik, Yugoslavia, Pergamon Press 540 p.

MAKSIMOVlC, C., D. PRODANOVlC, J. ELGY; L. FUCHS (1994) ‘GIS (or GIM) in WaterProjects-Tools or Toys’. Proc. International Conference of Hydroinformatics Deft,Bal kema

MACROPOULOS, C., C. MAKSIMOVIC; D. BUTLER (1999) ‘Effectiveness of Storm Drainagein urban Flood Risk Reduction under Source Control Implementation’. FifthInternational Mime Symposium: System Operational Effectiveness l-3 December (in

Press)MACROPOULOS, C.,D. Butler and C. MAKSIMOVIC (1998) ‘GIS supported evaluation of

19



source control applicability in urban areas’. Water Science and Technology, Vd,. 39 No.3, p.243252

MAKSIMOVIC, C., L. FUCHS, D. PRODANOVIC and J. ELGY (1995) ‘Full Scale Applicationof Standard GIS Packages with Urban Storm Drainage Simulation Software’. KurierAbwasserung Germany, Book 1, p. 3-13

MAKSIMOVlC, C., D. BUTLEr and N. GRAHAM (1999) ‘Emerging Technologies in theWater Industry’. In Water Industry Systems; Modelling and Optimization Applications,

Ed. D. Savic and G. Walters, Research Studies Press, p. 3984.NIEMCZYNOWICZ, J. (1998) ‘Challenges and interactions in water future’. Environmental

Research Forum-3 4, l-10, Transtec Publications, Switzerland.O’LOUGHLIN, G., W. HUBER and B. CHOCAt, (1998) ‘Rainfall-Runoff Processes and

Modelling’, Journal of Hydraulic Research, Vol. 34., No 8, p. 733-752.OSTROWSKI, M.W. and W. JAMES, (1998) ‘Requirements for group decision support

systems for urban stormwater management’. Fourth Int. Conference on Developmentsin Urban Drainage; UDM’98, London

PRODANOVIC D.( 1999) ‘Unapredjenje metoda primene hidroinformatike u analizi oticanja saurbanih pow&a’ (in Serbian) (Improvements of application methods in analysis ofrunoff from urban areas), PhD Thesis presented at the University of Belgrade,Yugoslavia, November 1999, p 198.

SIMON, A. (1998) ‘Estimation of the Utilisation of Possibilities of Overland Flow in UrbanAreas in Israel’s Coastal Plain’. In lntegfated Water Management in UrbanEnvironment Edited by J. Nierrczynowicz, Trensttxh Publications, p. 221- 232.

STAHRE, P. (1999) ‘Ten Years of Expe&nces of Sustainable Stormwater Management inthe Cii of Malmo Sweden’. Proc of s’” International Conference on Urban StormDrainage, Sydney, Australia, p. 1087 - 1097.

XANTHOPUOLOS, C. and H. HAHN (1993) ‘Anthropogenic Pollutants Wash-off from StreetSurface’. Proc. Of the Sixth Int. Conference on Urban Storm Drainage, Niagara Falls,Canada

YEN B. C., (1986) ‘Rainfall-Runoff Process on Urban Catchment and its Modelling’. Invitedlecture at Int. Symp. UDM’86 Dubrovnik, Yugoslavia , Pergamon Press (Ed.Maksimovic & Radojkovic)

20



Chapter 2

General characteristics of arid and semi-arid regions

2.1 Physical features

2.1.1 What is aridity?

And means dry, or parched, and the primary determinant of aridity in most areas is thelack of rainfall (Slatyer and Mabbutt, 1964). Aridity is defined as a lack of moisture which isbasically a climatic phenomenon based upon the average dimatic conditions over a region(Agnew and Anderson, 1992); therefore, arid regions have been identified by dimatologicalmapping. Of the many classifications based on climate Meigs(1953) developed a set of maps forUNESCO that received wide international acceptance and were recognized by the World

Meteorological Organization. Meigs divided xeric environments into extremely arid, arid, andsemiarid. Meigs defined the arid areas as those in which the rainfall is not adequate for regularcrop production, and semiarid areas as those in which the rainfall is sufficient for short-seasoncrops and where grass is an important element of the natural vegetation.

To avoid confusion the term desert is based upon land surface characteristics and can beconsidered as areas of low or absent vegetation cover with an exposed ground surface (Goudie,1985). Agnew and Anderson (1992) considered the arid realm to encompass arid and semi-aridenvironments from desert through to steppe landscapes. They used the term desert to conveyhyper-arid conditions where rainfalls are particularly low and vegetation is sparse. Figure 2.1shows the arid regions.of the world based on UNESCO (1977). Note that the four areas shownare hyper-arid, arid, semi-arid and sub-humid.

Figure 2.1. Arid regions of the world (UNESCO, 1977)

21



2.1.2 Geomorphology

From a geomorphological viewpoint no single process dominates arid environments. Aridlands vary from tectonically active mountainous regions in North and South America to thegeologically stable shield areas in Africa and Australia. Table 2.1 lists the major and regions ofthe continents and their main geomorphology and vegetation types (after Shmida, 1985 aspresented in Dick-Peddie, 1991). The boundaries o f these regions conform closely to the arid

and hyper and homodimates of Meigs(l953). Dick-Peddie (1991) present summary discussionsof each continent along with the extent and vegetation of the semiarid, arid, hyperarid, andriparian habitats within each region.

Table 2. 1. Major arid regions of the continents and their geomorphology and vegetation types

Regions Geomorphology type Vegetation types

North AmericaGreat Basin

SonoraChihuahuaBaja California

South AmericaPatagoniaPeru-ChileMonte

AsiaArab

Syrian

TharIranTurkestan

Takla MakanGobi

AfkicaSahara

Sahel

Somali

NamibKalahari

Australia

Basin and mountains

Plateau, crystalline mountainsLimestone plateauDiverse, coastal, sand

Plains, dissected low plateau Perennial grasslandCoastal, badlands, Salinas Thorny succulent savannaBasins, mountains, salinas Thorny succulent savanna

Huge sands, limestone plateau

Limestone plateau, ergs

Sans, Salinas, gravelsSalinas, ergs, interplainsLoess, sandy plains

Interplains, ergsBasins, mountains, sands

Huge sands, ergs, diverse

Diverse plains, plateau

Coastal sands, old relief

Coastal sands, diverseOld diverse plateaux, mountainsHuge plains, salines, dd relief,sands

Artemisia-Atriplex steppes andpygmy open woodland

Thorny succulent savannaOpen shurbland and thicketsThorny succulent savanna anddwarf shurbs

Chenopod-Zygophyllum desert,dwarf shrubs, Artemisia andPoaceae on sandsArtemisia - chenopod steppes

and desert dwarf shrubsThorny-rattanoid savannaChenopod desert dwarf shurbsArtemisia-Stipa steppes andpygmy open woodlandChenopod desert dwarf shurbsChenopod - Tamarix desertshrubland

Chenopod-Zygophyllum desert,dwarf shrubs, Artemisia andPoaceae on sandsThorny-rattanoid savanna

Thorny savanna and desertdwarf shrubsSucculent desert dwarf shrubsThorny-succulent savanaChenopod shrublands andsclerophyl evergreen lowwoodland

Source: Modified from Shmida (1985) as presented in Dick-Peddi (1991)

22



Agnew and Anderson (1992) define the following features of arid zone landscapes basedupon Goudie (1985) Heathcote (1983) and Thomas (1989):

l Alluvial fans which are fan-shaped deposits found at the foot of the slope, gradingfrom gravels and boulders at the apex to sand and silt at the foot of the fan, called abajada when coalesced.

l Dunes which are aeolian deposits of sand grains (unconsolidated mineral partides)forming various shapes and sizes depending upon the supply and characteristics of

the material and the wind system.l Bedrock fields including pediment, a piano-concave erosion surface sloping from the

foot of an upland area; and hamada, a bare rock surface with little or no vegetation orsurficial material.

l Desert flats with slight slopes possibly containing sand dunes, termed playa when thesurface is flat and periodically inundated by surface runoff.

l Desert mountains are the most wmmon feature of and lands.l Badlands well dissected, unconsolidated or poorly cemented deposits with sparse

vegetation.

2.1.3 Soil characteristics

Soil characteristics are influenced primarily by low rainfalls, high evaporation rates, andlow amounts o f vegetation. The soils, therefore, have low organic matter, an accumulation ofsalts at the surface, little development of day minerals, a low cation-exchange capacity, a dark orreddish wlor due to desert varnish, and little horizon development due to the lack of percolatingwater (Fuller, 1974). Even though there are vast areas covered by thin, infertile soils, there arehowever, arid lands where soils are highly productive having a very high potential for agriculture.Dregne (1976) presents the following:

l Entisols cover 41.5% of and lands (immature soils ranging from barren sands to veryproductive alluvial deposits.

l Aridisols cover 35.9% of arid lands ( red-brown desert soils, dry and generally onlysuitable for grazing without irrigation)

l Vertisols cover 4.1% of and lands (moderately deep swelling day which is difficult to

cultivate)

Agnew and Anderson (1992) report the fdlowing:l Mollisols cover 11.9% of arid lands (one of the world’s most important agricultural

soils)l Alfisols cover 6.6% of arid lands (high base saturation, reasonably high day contents,

agriculturally productive.

As a result of the climate in arid lands, soil formation is dominated by physicaldisintegration with only slight chemical weathering. Elgabaly (1980) defines three main types ofsoils found in arid lands:

l Saline soils - Characterized by the presence o f excess neutral salts (pH less than 8.5) that

accumulate on the surface in the form of a loose crust depending upon the depth and salinityof the groundwater table.l Saline-alkaline soils - Charactenzed by the presence of excess soluble salts (pH

approximately equal to 8.5). The structure is more compact at a certain depth and darker inwlor.

0 Sodic soils - Characterized by the presence of low soluble salts (pH greater than 8.5).Surface cdor is usually darker and day accumulates in the B horizon and a columnarstructure eventually develops.

The formation or origin of salt-affected soils is connected with: a.) climate, as saline soils arean element of arid lands; b.) relief, as saline soils are more common in low lands such as deltasand floodplains; and c.) geomorphdogy and hydrology as saline soils are related to the depth ofthe water table.

23



2.1.4 Aeolian systems

Aeolian systems are those in which wind plays a dominant role in sculpturing thelandscape. According to White, et al.(1992) actiie aeolian systems occur in regions which arecircumscribed by the 150 mm rainfall isohyet. By this criteria then they occupy about 20% of theEarths land surface, evenly divided between the hot deserts and the cold deserts of the middleand polar latiiudes. The creation of dune forms by aeolian deposition is the most distinctive

feature of aeolian systems. Aeolian deposition creates distinctive landforms that are shaped onlarge spreads of sand. Dunefields only occupy a limited proportion of the area of individualdeserts, commonly 20% -30%. Sands tend to collect in desert lowlands where the winds are lesssevere and erosive than in the adjacent uplands

2.2 Climate

2.2.1 Causes of aridity

One of several processes can lead to aridity, however Hill (1966) believes that the majorcause of aridity is explained through the global atmospheric circulation patterns. Thompson(1975) lists four main processes that explain aridity as presented by Agnew and Anderson (1992):

l Hiah uressureAir that is heated at the equator rises, moves polewards and descendsat the tropical latitudes around 20 to 30 degrees latitude. This descending air iscompressed and warmed, thus leading to dry and stable atmospheric conditionscovering large areas such as the Sahara Desert (see Figures 2.2 and 2.3).

Figure 2.2. General atmospheric circulation during January (Agnew and Anderson, 1992)

l Wind direction-Winds blowing over continental interiors have a reduced opportunity toabsorb moisture and will be fairly stable with lower humidities. These typically dry,northeasterly winds (in the northern hemisphere) are seasonally constant andcontribute to the aridity of South West Asia and the Middle East.

24



1------...-----..fn Hfgh rem#. tqpnslOlEd!~ - --

P warlnwh-d

Figure 2.3. General atmospheric circulation during July (Agnew and Anderson, 1992)

l Touooraphy -When air is forced upward by a mountain range (Figure 2.4) it will cooladiabatically (A to B) at the saturated adiabatic rate once the dew point is reached (Bto C) with possible precipitation. On the leeward side of the mountain the same airdescends (C to D) warming at the dry adiabatic rate and hence the descending air iswarmer at corresponding altiiudes compared to the ascending air. Hence a warmer,drier wind blows over the lands to the leeward side, providing the ascent is sufficientto reach the dew point temperature.

l Cold ocean currents -Onshore winds blowing across a cold ocean current dose tothe shore will be rapidly cooled in the lower layers (up to 500 m). Mist and fog mayresult as found along the coasts of Oman, Peru, and Namibia, but the warm air aloftcreates an inversion preventing the ascent of air and hence there is little or noprecipitation. As this air moves inland it is warmed and hence its humidity reduces.

The majority of semiarid and arid regions are located between latitudes 25 and 35degrees (see Figure 2.1) where high pressures cause warm air to descend, resulting in drystable air masses. Aridity caused by orographic aridity is common in North and South America,where high mountain ranges extend perpendicularly to the prevailing air mass movements. Asdescribed above these air masses are cooled as they are forced up mountains, reducing theirwater holding capacity. Most of the moisture is precipitated at the high elevations of thewindward slopes. The relatively dry air masses warm as they descend on the leeward side of themountain ranges, increasing their water-holding capacity and reducing the chance of anyprecipitation. This orographic aridity is referred to as the rain shadow effec t (Dick-Peddie, 1991).The positioning in a continent where distance from oceans lessens the chance of encounteringmoisture-laden air masses is the cause of the semiarid and arid conditions of central Asia. Coldocean currents cause the coastal arid regions of Chile and Peru and the interior part of northernArgentina, where cold ocean currents in dose proximity to the coast supply dry air that comes onshore, but as the mass is forced up the mountain sides there is no moisture to be lost as the airmass cools.

25



2.2.2 Climate areas

Logan (1968) distinguished the four areas as subtropical, continental interior,rainshadow. and wol coastal arid lands, with the following definitions ( Agnew and Anderson,1992):

Figure 2.4. Rainshadow effect leading to aridity (Agnew and Anderson, 1992)

l SubtroDical areas (e.g. Sahara, Arabia, Sonora, Australia, and Kalahari) are

characterized by anticydonic weather producing clear skies with high groundtemperatures and a marked nocturnal cooling. The dimate has hot summers andmild winter so the seasonal contrasts are evident with rare winter temperatures downto freezing. Convective rainfalls only develop when moist air invades the region.

l Continental interior areas (e.g. arid areas of Asia and western USA) have largeseasonal temperature ranges from very cold winters to very hot summers. Snow fallcan occur however its effectiveness may be reduced by abalation as it lies on theground through winter. Rainfall in the summers is unreliable and can occur as violentdownpours.

l Rainshadow areas ( leeward sides of mountain ranges such as the Sierra Nevada,the Great Dividing Range in Australia and the Andes in South America) occur whereconditions are diverse but are not as extreme as the continental interior areas.

l Cool coastal areas (e.g. Namib, Atacama, and the Pacific coast of Mexico)-have

reasonably constant conditions with a cwl humid environment. When temperatureinversions are weakened by moist air aloft, thunderstorms can develop.

2.2.3 What are deserts?

Desert implies aridity, however desert is a less precise term. There is no worldwideagreement as to what constitutes arid land and as to what gradations occur within the concept ofarid (Dick-Peddie, 1991). Shmida (1985) equated extremely arid environments with extremedeserts, and environments with deserts or true deserts, and semiarid environments withsemideserts. Mares, et al. (1985) equated arid and semiarid environments with semideserts,which results in equating extremely and environments with deserts (true deserts). Other authorsuse isohyets of annual precipitation to place limits on the various xeric zones. According to Dick-

26



(1953). In most instances the world’s deserts trend to be located in Meigs arid and extremely aridhomoclimates.

2.3 Hydrology

2.3.1 Rainfall

Precipitation includes rainfall, snowfall, and other processes by which water falls to theland surface, such as hail and sleet (Chow, et al., 1988). The formation of precipitation requiresthe lifting o f an air mass in the atmosphere so that it cools and some of its moisture condenses.In arid environments the processes leading to aridity tend to prevent the cooling throughmaintaining air stability, creation of inversions, or through the warming o f the atmosphereresulting in lowering the humidity. The influence of these processes depends upon theatmospheric conditions. When rainfalls do occur they can be intense and localized downpours asmoist air breaks through. Precipitation variability for the world is shown in Figure 2.5 .

Slatyer and Mabbutt (1954) point out that the primary feature of precipitation in arid areasis the high variability of the small amount received, fo r which it is not uncommon for the standarddeviation of the mean annual rainfall to exceed the mean value. They also point out that in mostarid regions precipitation characteristics fo llow somewhat similar patterns, reflecting a high orderof variability in time and space of individual storms, of seasonal rainfall, and of annual and cyclical

totals.

I----.. -__- - ---- --- . __-- .._-

Figure 2.5. Variations in annual rainfalls(after Rumney, 1988, as presented in Agney and Anderson, 1992)

Schick (1988) discusses the immense temporal variability of rainfall and the very highintensities in hyperarid areas of the world. In typical cloudbursts in the extreme desert areas thetransition between total dryness and full-blast rain is near instantaneous, with the first fewminutes of the rainfall having intensities in excess of 1 mm/min. The excessive intensities inhyperarid seem to be associated with relatively high temperatures, and are therefore the result ofconvective processes. The convective storms tend to form at preferred distances from eachother, as opposed to being randomly scattered in space. Sharon (1981) found that theconvective storms in the extremely arid Namib Desert had preferred distances of around 40 - 50km and 80 - 100 km. The rain front are often very sharply defined both in direction of cellmovement as well as laterally. Sharon (1972) reported cases where the velocity of rain cells in

27



the extreme desert was found to vary from near zero ( a stationary cloudburst) to several tens ofkilometers per hour. The lateral boundaries o f maving cells tend to be sharp. There are alsowidespread rainfalls that cover vast desert areas with lower intensity but relatively high quantityrains.

Goodrich, et al. (1990) studied the impacts of rainfall sampling on runoff computations inarid and semi-and areas of the Southwestern U. S. This study concluded that the appropriaterainfall-sampling interval for arid land watersheds depends on many factors including the

temporal pattern of the rainfall intensity, watershed response time, and infiltration characteristics.The study recommended that either breakpoint rainfall data of data sampled at uniform timeincrements be used for watersheds with equilibrium times smaller than about 15 minutes and thata maximum interval of 5 minutes be used for more slowly responding basins. Using a physicallybased rainfall-runoff modeling (KINEROS, Woolhiser, et al, 1990), they found that the outflowhydrographs were more sensitive to the rainfall input than to the model parameters. Feneira(1990) used Opus, an agricultural ecosystem model with an infiltration-based hydrology option, tosimulate field responses in arid and semi-arid areas to rainfall inputs of various time intervals.The results using synthetic rainfall data from the statistical analysis of rainfall data fromwatersheds in the Southwestern U. S. showed a strong sensitivity of runoff predictions to the timeinterval of input rainfall data.

2.3.2 Infiltration

In arid lands the physical process if soil formation is active, resulting in heterogeneoussoil types, having properties that do not differ greatly from the parent material, and having soilprofiles that retain their heterogenous characteristics (Elgabaly, 1980). Soils in arid lands maycontain hardened or cemented horizons known as pans and classified according to the cementingagent such as gypserious, calcareous, iron, and so on. The extent to which the horizons affectinfiltration and salinization depend upon their thickness and depth of formation as they constitutean obstacle to water and root penetration. In salt mediums salt crusts can form at the surfaceunder specific conditions. Large quantities of soluble salts cause coagulation of clay particles.Enrichment of soil in sodium salts modifies the soil structure as a result of the dispersion andswelling properties of sodic clays generally formed and the soils become impermeable. Changesin soil structure due to the action of different salts has an important influence on the behavior ofsoil under irrigation and drainage. Salt affected and sodic soils have a very loose surface

structure making it susceptible to wind erosion and water erosion.

There are numerous methods available to the hydrologist to compute infiltration (Chow,et al., 1988), varying from the constant infiltration rates to the Green-Ampt method. Which modelto use should depend upon the use for the hydrologic modeling and the availability of data forcalibration of the model. The Green-Ampt infiltration model is a physically based that is arecommended method for semi-arid and arid lands. The Green-Ampt equation for cumulativeinfiltration, F(t), as a function of time t, is expressed as

F (t) = Kt + YAB In ( 1 + F(t)/YAO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)

and the infiltration rate, f(t), is given as

f(t) = K ((wAe/F(t)) + 1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ._._......._.._..........

where the Green-Ampt parameters are defined as

(2)

K = hydraulic conductivity (saturated;\v = soil suction head; andA8 = change in moisture content from the initial content, 8i , to the porosity, n.

28



Table 2. 2. Green-Ampt Infiltration Parameters for Various Soil Classes

Soil Class Porosity EfectivePorosity

Wetting front Hydraulicsoil suction conductivity

head9 Wcm) K(cm/h)

Sand 0.437 0.417 4.59 11.78(0.374-0.0-500) (0.354-0.480) (097-25.36)

Loamy sand 0.437 0.401 6.13 2.99(0.363-0.506) (0.329-0.473) (1.35-27.94)

Sandy loam 0.453 0.412 11.01 1.09(0.351-0.555) (0.283-0.541) (2.67-45.47)

Loam 0.463 0.434 8.89 0.34(0.375-0.551) (0.334-0.534) (1.33-59.38)

Silt loam 0.501 0.486 16.68 0.65

((0.420-0.582) (0.394-0.578) (2.92-95.39)Sandy clay loam 0.398 0.330 21.85 0.15

(0.332-0.464) (0.235-0.425) (4.42-108.0)

Clay loam 0.464 0.309 20.88 0.10(0.409-0.519 (0.279-0.501) (4.79-91 .lO)

Silty clay loam 0.471 0.432 27.30 0.10(0.418-0.524) (0.347-0.517) (5.67-131.50)

Sandy clay 0.430 0.321 23.90 0.06

(0.370-0.490) (0.207-0.435) (4.08-140.2)Silty clay 0.479 0.423 29.22 0.05

(0.425-0.533) (0.334-0.512) (8.13-139.4)Clay 0.475 0.385 31.63 0.03

(0.427-0.523) (0.269-0.501) (6.39-156.5)

The numbers in parentheses below each parameter are one standard deviation around the parameter value given (Rawis,Brakensiek, and Miller, 7983)

The determination of the effect of impervious area is particularly important (Dawdy,1990). Impervious area increases volume of runoff and increases the velocity of the water, bothof which tend to increase peak flows. However the effect of increased impervious area depends

upon its location in the basin and the “connectedness” of the impervious area to the channel(Dawdy, 1990). Runoff from the impervious areas not directly connected to the channel systemmust flow over pervious areas, and thus contribute less runoff.

2.3.3 Runoff and flooding

In arid and semiarid regions flash floods are caused by high intensity, short durationstorms with a high degree of spatial variability. Runoff hydrographs from these storms typically

exhibit very short rise times, even for large catchments (Goodrich, et al. 1990).

Overland flow

When the rainfall rate exceeds the infiltration capacity and sufficient water ponds on thesurface to overcome surface tension effects and fill small depressions, Hortonian overland flowbegins (Woolh iser, et al., 1990). When viewed from a micro-scale over-land flow is a three-dimensional process, however, at a larger scale it can be viewed as a one-dimensional flowprocess in which the flux is proportional to some power o f the storage per unit area, as

Q=ahm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (3)

where Q is the discharge per unit width, h is the storage of water per unit area (or depth if the

surface is a plane), and m are parameters related to slope , surface roughness, and whether theflow is laminar or turbulent. The continuity equation for flow is expressed as

29



ah/i% +iTQ/& = q(x,t)

. . . . . . . . . . . . . . . . . (4)

where t is time and, x is the spatial coordinate, and q(x,t) is the lateral inflow rate. Substitutingequation into equation the resulting kinematic wave equation for one-dimensional flow on a planesurface (Hortonian overland flow) is

ah I at + amh”-‘ah I & = q(x, t)

. (5)

The kinematic assumption requires only that discharge be some unique function of the amount ofwater stored per unit of area; it does not require sheet flow ( Woolhiser, e t a al., 1990). Thekinematic wave formulation is an excellent approximation for most overland flow conditions(Woolhiser and Liggett, 1967 and Moms and Woolhiser, 1980).

Lane, et al. (1978) studied the partial area response (variable source area response) onsmall semiarid watersheds. This refers to the response of a watershed when only a portion of thetotal drainage area is contributing runoff at the watershed outlet or point of interest. Thegeneration of overland flow on portions of small semiarid watersheds was analyzed using three

methods: an average loss rate procedure, a lumped-linear model, and a distributed-nonlinearmodel (kinematic wave). The results showed that significant errors in estimating surface runoffand erosion ra tes are possible if a watershed is assumed to contribute runoff uniformly over theentire area, when only a portion of the watershed may be contributing.

Floods and channel routing

From a geomorphology viewpoint the response of fluvial systems to flood dischargesdepends in large part on the amount of time that has passed since the major climatic perturbationswitched the mode of operation of hillslopes (midslopes and footslopes) from net aggradation tonet degradation ( Bull, 1988). During the early stages of hillslope stripping, the amount ofavailable sediment was so large that intense rainfall-runoff events caused debris flows andaccelerated valley floor aggradation. During later stages an opposite result occurs when majorrainfall events accelerated the removal of remaining sediment on the hillslopes, thereby causingstill larger increases of stream power relative to resisting power. Resulting degradation cutsthrough the valley fill to bedrock.

Unsteady free surface flow in channels can also be represented by the kinematic waveapproximation to unsteady, gradually varied flow, given as

dA/dt+iTQ/&=q,(x,t)

. . . . . . . . . . . . . . . . . . . . (6)where A is the cross-sectional area, Q is the channel discharge, and q, (x,t) is the net lateralinflow per unit length of channel. Using the kinematic assumption, Q can be expressed as aunique function of A so that Eq. 6 is

8A / iTit+ (dQ / dA)(iTA / 8x) = q, (x, t)

. . . . . . . . . . . . . . . . . (7)

The kinematic assumption is embodied in the relationship between channel discharge and crosssectional area, is

Q =aRm’A . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)

Where R is the hydraulic radius. If the Manning’s equation is used, CL 1 O S”2 n and m = 5/3.

30



Table 2. 3. Recommended Manning’s roughness coefficients for overland Flow (Woolhiser, et al.,1990)

Cover or treatment

Concrets or aspaltBare sandGraveled surfaceBare clay loam (eroded)Fallow - no residueChisel plow

Moldboard plow (fall)ColterRange (nature)Grass (bluegrass sod)Short prairierass

Residue rate Value Rangerecommended

tons/acre0.011 0.010 - 0.0130.01 0.010 - 0.0160.02 0.012 - 0.030.02 0.012 - 0.0330.05 0.006 - 0.16

<l/4 0.07 0.006 - 0.17l/4 - 1 0.18 0.07 - 0.34

l-3 0.30 0.19 -0.47>3 0.40 0.34 -0.46

<l/4 0.08 0.008 - 0.41l/4- 1 0.16 0.10 -0.25

l-3 0.25 0.14 -0.53

>3 0.30<l/4 0.04 0.03 - 0.07l/4- 1 0.07 0.01 -0.13

1-3 0.30 0.16 - 0.470.10 0.05 -0.130.13 0.01 - 0.320.10 0.02 - 0.240.45 0.39 - 0.630.15 0.10 -0.20

Transmission losses

Many semiarid and arid watersheds have alluvial channels that abstract large quantitiesof streamflow referred to as transmission losses (Lane, 1982, 1990; Renard, 1970). Theselosses are important in the determination of runoff because water is lost as the flood wave travelsdownstream. The transmission losses are an important part of the water balance because theysupport riparian vegetation and recharge local aquifers and regional groundwater (Renard, 1970).Procedures to estimate transmission losses range from inflow-rate loss equations, to simpleregression equations, to storage -routing as a cascade of leaky reservoirs, and to kinematic wavemodels incorporating infiltration (Smith, 1972; Wwlhiser, et al., 1990). Stream channels alsotransport water across alluvial fans from mountain fronts to lower portions of the watersheds.These channels are unstable and are variable in time and space; however they retain theirephemeral character and thus transmission losses can exhibit their influence on flood peaks,water yield, and groundwater recharge just as for ephemeral stream channel networks (Lane,1990). From the viewpoint of flood routing and transmission losses, the main differences

between ephemeral stream channel networks forming the drainage patterns in watersheds andephemeral channel segments transversing alluvial fans are due to the nature of their structureand linkage (Lane, 1990). In watersheds the channel systems tend to be dendritic with the mainchannels collecting tributary inflow in the downstream direction. On alluvial fans the channelsegments tend to be singular or bifurcating in the downstream direction. On alluvial fans thereusually is no tributary inflows; however, channels can split or diverge resulting in tributaryoutflows in the downstream direction. As pointed out by Lane(l990), in spite of the differencesmany of the same flow processe s occur in watersheds and on alluvial fans; therefore methodsthat have been developed to consider streamflow and transmission losses in individual streamchannel segments can be applied to both ephemeral streams in watersheds and to ephemeralstream segments transversing alluvial fans.

31



2.3.4 Erosion and sediment transport

Sheet and rill erosion

Water is the most widespread cause of erosion, which can be classified into sheeterosion and channel erosion. Sheet erosion is the detachment of land surface material byraindrop impact and thawing o f frozen grounds and its subsequent removal by overland flow

(Shen and Julian, 1993). Transport capacities of thin overland flow or sheet flow, increases withfield slope and flow discharge per unit width. As the sheet flow concentrates and the unitdischarge increases, the increased sediment transport capacity scours microchannels, alsoreferred to as rills. The till erosion is the removal of soil by concentrated sheet flow. Surfaceerosion begins when raindrops impact the ground and detach soil particle by splash (Shen and

Julien, 1993)

Upland erosion

Bennett (1974) presented the following mass balance equation similar to that forkinematic water flow to describe the sediment dynamics at any point along a surface flow path,given as (Wollihiser, et al., 1990)

. . . . . . . . . . . . . . . . (9)

in which C, = sediment concentrationA= cross-sectional area of flowe = rate of erosion of the soil bedqs = rate of lateral sediment inflow for channels

The rate of erosion of the soil bed, e, can be composed of two components: 1) soil erosion by thesplash of rainfall on bare soil, gs, and 2) hydraulic erosion (or deposition),g,,. Hydraulic erosion isdue to the interplay between the shear force of water on the loose soil bed and the tendency ofsoil particles to settle under the force of gravity. The total rate of erosion of the soil bed is thenexpressed as

e=gs+gh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10)

Channel erosion

Channel erosion includes both bed and bank erosion, which can be very significant inalluvial channels. Sediment transport capacity is generally proportional to the water dischargeand channel slope. The sediment transport capacity vanes inversely with the bed sediment size.Lane (1955) proposed the following qualitative 3equilibrium relationship between hydraulic andsediment parameters,

QSozQQsd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11)

In which Q is the channel discharge, So is the downstream slope of the channel, Qs is the bedsediment discharge, and d is the bed sediment size.

Sediment transport simulation for channels can be nearly the same as for upland areas.The difference is that splash erosion is neglected for channel erosion and the term qs becomesimportant in representing lateral inflows (Woolhiser, et al., 1990).

Wind erosion

Wind erosion can be important in semiarid and arid areas. The rate of wind erosiondepends upon the particle size distribution, wind velocity, soil moisture, surface roughness, andvegetative cover. Chepil and Woodruff (1954) proposed an empirical equation for estimating therate of wind erosion. This equation provides a rough estimate of wind erosion rates.

32



Conclusions

The physical and climate features as well as the distinguished hydrologic characteristics of aridareas are reviewed and discussed. Alluvial fans, dunes, bedrock fields, desert flats andmountains, and badlands; well dissected - unconsolidated or poorly cemented deposits withsparse vegetation, are the main features of and zone landscapes. Low rainfall, high evaporation

rates, and low amounts of vegetation influence the soil characteristics. The soils have loworganic matter, an accumulation of salts at the surface, little development of clay minerals, a low-exchange capacity, a dark or reddish wlor due to desert varnish, and little horizon developmentdue to the lack of percolating water. Their formation is dominated by physical disintegration withonly slight chemical weathering. Main types of soils include saline soils, saline-alkaline soils, andsodic soils are found in these areas. The specific climate features affect the hydrologiccharacteristics. The main feature of precipitation is the high variability in time and space of thesmall amount received. The soils have a very loose surface structure making it susceptible towind and water erosion. Infiltration is of relatively high rates and variable in space. Flash floodsare caused by high intensity, short duration storms with high degree of spatial variability. Runoffhydrographs from these storms typically exhibit very short rise times, even for large catchments.Transmission losses are important in the determination of runoff. Due to the particular propertiesof flash floods and soils, both sheet and channel erosions as well as wind erosion are important

processes in the arid areas, resulting in significant amounts of suspended sediment to beproduced and transported with the floods to downstream facilities.

Bibliography

A Woolhiser GNEW, C. AND E. ANDERSON (1992) ‘Water Resources in the And Realm”,Routledge, London.

BENNETT, J. P. (1974) “Concepts Of Mathematical Modeling Of Sediment Yield”, WaterResources Research, AGU, Vol. 10, No. 2, pp. 485492.

BULL, W. B. (1988) “Floods, Degradation And Aggradation”, in Flood Geomorphology, ed. by V.R. Baker, R. C. Kochel, P. C. Patton, John Wiley & Sons, New York.

CHEPIL, W. S. AND N. P. WOO DRUFF (1954) “Estimates Of Wind Erodibility Of Field Surfaces”,

J. Soil Water Cons. Vol. 9, No. 6, pp. 257-265.CHOW, V. T., D. R. MAIDMENT, AND L. W. MAYS (1988) “Applied Hydrdogy”, McGraw-Hill,New York.

DAWDY, D. R. (1990) “The Sorry S tate Of The Flood Hydrology In The And Southwest”, inHydraulics/Hydrology of Arid Lands, ed. by R. H. French, American Society of CivilEngineers, New York.

DICK-PEDDIE, W. A. (1991) “Semiarid And Arid Lands: A Worldwide Scope, Semiarid Lands AndDeserts”, Soil Resource and Reclamation, ed ited by J. Skujins, Marcel Dekker, Inc., NewYork.

DREGNE, H. E. (1976) “Soils of Arid Regions”, E lsevier, Oxford, 1976.ELGABALY, M. M. (1980) “Problems Of Soils And Salinity”, in Water Management for And Lands

in Developing Countries, A. K. Biswas, et al., editors, Pergamon Press, Oxford.FERREIRA, V. (1990) “Temporal Characteristics Of Arid Land Rainfall Events”, in

Hydraulics/Hydrology of Arid Lands, ed. by R. H. French, American Society of CivilEngineers, New York.FULLER, W. H. (1974) “Desert Soils”, in Desert Biology, Vol. 2, ed. by G. W. Brown, Academic

Press, London.GOODRICH,. D. C., D. A. WOOLHISER, AND C. L. UNKRICH (1990) “Rainfall-Sampling Impacts

On Runoff’, in Hydraulics/Hydrology of Arid Lands, ed. by R. H. French, AmericanSociety of Civil Engineers, New York.

GOUDIE, A. (ed.) (1985) “Encyclopaedic Dictionary O f Physical Geography”, Blackwell, Oxford.HEATHCOTE, R. L. (1983) “The And Lands: Their Use and Abuse”, Longman, London.HILLS, E. S. (1966) “Arid Lands”, Methuen, London.LANE, L. J. (1982) “Distributed Model For Small Semiarid Watersheds”, Journal o f the Hydraulics

Division, ASCE, Vol.108, No. HYlO, pp. 1114-l 131.

33



LANE, L. J. (1990) “Transmission Losses, Flood Peaks, And Groundwater Recharge”, inHydraulics/Hydrology of And Lands, ed. by R. H. French, American Society of CivilEngineers, New York.

LANE, L. J., M. H. DISKIN, D. E. WALLACE, AND R. M. DIXON (1978) “Partial Area Responseon Small Semiarid Watersheds”, Water Resources Bulletin, AWRA, Vol. 14, No. 5, pp.1143-l 158.

LOGAN, R. F., Causes (1968) “Climates, And Distribution Of Deserts”, in Desert Biology, edited

by G. W. Brown, Academic Press, London, pp 21-50.MEIGS, P. (1953), “World Distribution Of Arid And Semi-Arid Homoclimates”, Rev. Res. On And

Zone Hydrol., UNESCO, Paris, pp. 203-210.MICHAUD, J., AND S. SOROOSHIAN (1994), “Comparison Of Simple Versus Complex

Distributed Runoff Models On A Mid-Sized Semiarid Watershed”, Water ResourcesResearch, AGU, Vol. 30, No. 3, pp. 593605.

MORRIS, E. M. AND D. A. WOOLHISER (1980), “Unsteady One-Dimensional Flow Over APlane: Partial Equilibrium And Recession Hydrographs”, Water Resources Research,AGU, Vol. 16, No 2, pp. 355360.

RAWLS, W. J., D. L. BRAKENSIEK, AND N. MILLER (1983) “Green-An@ Infiltration ParametersFrom Soils Data”, Journal of the Hydraulics Division, ASCE, Vol. 109, No. 1, pp. 62-70.

RENARD, K. G. (1970), “The Hydrology Of Semiarid Rangeland Watersheds”, ARS 41-162, U. S.Department of Agriculture, Agricultural Research Service, Washington, D.C..

RUMNEY,, G . R. (1968) “Climatology and the World’s Climates”, Macmillan, New York.SHARON, D. (1972) “The Spottiness Of Rainfall In A Desert Area”, J. Hydrology, Vol. 17, p 161-175.

SHARON, D. (1981) “The Distribution In Space Of Local Rainfall In The Namib Desert”, J.Climatology, Vol.1, p. 69-75.

SHEN, H, W. AND P. Y. JULIEN (1993) “Erosion And Sediment Transport”, in Handbook ofHydrology, edited by D. R. Maidment, McGraw-Hill, New York.

SCHICK, A. P. (1979) “Fluvial Processes And Settlement In Arid Environments”, GeoJournal,Vol. 3, p. 351360.

SCHICK, A. P. (1988) “Hydologic Aspects O f Floods In Extreme Arid Environments”, in FloodGeomorphology, ed. by V. R. Baker, R. C. Kochel, P.,C. Patton, John Wiley & Sons, NewYork.

SHMIDA, A. (1985) “Biogeography Of The Desert Flora”, in Ecosystems of the World, Vol. 12A,Hot Deserts and Arid Shrublands, edited by M. Evenari, I. Noy-Meir, and D. W. Goodall,Elsevier Science Publishers, Amsterdam, pp. 23-77.

SLATYER, R. 0. AND J. A. MABBUTT (1964) “Hydrology Of And And Semiarid Regions”,Section 24 in Handbook of Applied Hydrology, edited by V. T. Chow, McGraw-Hill BookCompany, New York.

THOMAS, D. S. (ed.) (1989) “And Zone Geomorphology”, Belhaven Press, London.THOMPSON, R. D. (1975), “The Climatology Of The Arid World”, Paper No. 35, Department of

Geography, Reading University.UNESCO (1977) “Map Of The World Distribution Of Arid Regions”, MAB Technical Note 7, Parts.WHITE, I. D., D. N. MOTTERSHEAD, AND S. J. HARRISON (1992) “ Environmental Systems:

An Introductory Text”, 2nd edition, Chapman & Hall, London.WOOLHISER, D. A. (1975) “Simulation Of Unsteady Overland Flow’, in Unsteady Flow in Open

Channels, Vol. II, edited by K. Mahmood and V. Yevjevich, Water ResourcesPublications, Fort Collins.

WOOLHISER, D. A. AND J. A. LIGGETT (1967) “Unsteady, One-Dimensional Flow Over APlane-The Rising Hydrograph”, Water Resources Research, AGU, Vol. 3, No. 3, pp. 753-771.

WOOLHISER, D. A., R. E. SMITH, AND D. C. GOODRICH, KINEROS (1990) “A KinematicRunoff And Erosion Model: Documentation And Users Manual”, U. S. Department ofAgriculture, Agricultural Research Service, ARS-77.

34



Chapter 3

Problems of urban drainage in arid and semi-arid regions

The previous chapter reviews the particular characteristics of arid and semi-arid climates.This chapter emphases on the effect of such characteristics on urban drainage. It emphasison specific problems in urban facilities’ design by traditional methods.

3.1. Design particularities

Proper design of urban facilities requires reliable prediction of water quantity and quality.

Such prediction needs - among others - information on rainfall, runoff, evaporation,infiltration, sedimentation, and water quality.

3.1 .I. Rainfall

For the purpose of design of urban drainage facilities, a rainfall event should be identified andthen used in generating design hydrographs. The rainfall event is either an actual rainfallevent or an artificial one. Generally, the former is used when runoff volume is of interest whilethe latter is used when the peak flow is of prime consideration. The artificial rainfall event iscommonly termed as “a design storm” and is determined based on depth-duration-frequency“DDF” analyses of historical data. It is used in the design of the majority of urban drainagefacilities.

The traditional method of design is to develop DDF curves and then use them to

identify the depth of a design storm for a selected duration and a selected return period. Thisdepth is then modified to account for the areal effect on point rainfall. The modified depth isused together with an assumed time distribution of rainfall to determine the design“hyetograph”, which is the prime input to a variety of runoff prediction methods.In arid and semi-arid dimates, problems in the application o f the above traditional method o f

design could include:. In developing DDF curves for historical rainfall records, independent rainstorm events

should be available, and thus criterion for the independence should be established. Oneapproach is to perform autocorrelation analyses for various rainstorm durations todetermine the time lag between rainfall periods such that there is no significant statisticalcorrelation between them. This approach could not be feasible in most of the world aridareas due to the normal shortage of necessary reliable data.Another approach ( Restrepo-Posada and Eagleson, 1982) is to identify the minimum time interval between two

successive rainstorms during which average rainfall is or close to zero (i.e. dry period).According to this approach, tv~ successive rainstorms statistically belong to each other ifthey are separated by a dry period less than the identified minimum dry period. The twosuccessive rainstorms are statistically independent if they are separated by a dry periodequal to or longer than the identified minimum time interval. Based on the assumptionthat the Poisson process describes the completely random arrivals of storm events, theyhave established that the intervals (i.e. rainless dry periods) between Poisson arrivals aredistributed exponentially. In applying this approach, trial dry period values are assumed,for each individual set of record, and the coefficient of variation of the dry perioddistribution are accordingly computed. The value of the dry period corresponding to acoefficient of variation equal to unity can therefore be determined and considered as theminimum dry period (i.e. rainless period) necessary to separate any two successiverainstorm events into statistically-independent rainstorm events. A typical example is

shown in Fig. 3.1. Analyses of rainstorm events in the southwest region of Saudi Arabia(Nouh, 1987b) have indicated that this minimum dry period necessary is in the order of 5

35



days. Same analyses of rainstorm events in other arid catchments produced similarresults as shown in Table 3.1.

Figure 3.

0.9.9 1 1 I 1 I 1

0 1 2 3 4 6 6

TRIAL VALUE OF WI~IMUU TM% BEfWEEM STORMS tMVSlRIAL VALUE OF WI~IMUU TM% BEfWEEM STORMS tMVSl

,l . Minimum dry period between successive rainstorms (Abha, Saudi Arabia)l . Minimum dry period between successive rainstorms (Abha, Saudi Arabia)

Table 3.1. Minimum dry period between two successive rainstorms

Country Number of Investigated Minimum DryFtainstorms Period (Days)

Saudi Arabia 298 5Qatar 118 4Oman 217 4UAE 189 5Kuwait 112 5Bahrain 96 4Yemen 56 4

. The traditional DDF analysis is based on the assumption that rainfall depths within acertain duration have variate values which wme from a single probability distribution

function, and accordingly the records of rainfall depths are normally fitted to a selectedsingle variate probability distribution function. This assumption may not be valid in somearid catchments where rainfall depths vary significantly with season and may begenerated by more than one mechanisms o f dimate (Nouh and El-laithy, 1988b). Atypical example of this situation is in the southwest region of Saudi Arabia, where rainfalldepth is affected by two systems o f dimate; the monsoon system during summer and thecyclonic system during winter. Such seasonal rainfall variation is shown in Table 3.2(Nouh, 1987a) where seasonal maxima were expressed as percentages of thecorresponding annual maxima. In this table, M5 refers to the annual maximum rainfalldepth with a return period of 5 years and for durations from 10 minutes to 120 minutes.Similarly, MlO, M20, . . , Ml00 are the same for return periods of 10, 20, . . _ 100 years,respectively. The considered season for summer is from May to October, and that forwinter is from November to April.

The results of fling rainfall records to various probability distribution functtons in sucha case (Nouh, 1987b) have indicated that the rainstorm depth is best described by a mix

36



of two probability distribution functions (two extreme value type 1 distribution functions forthe maximum depth, and TV lognormal probability distribution function for the averagedepth). The mix of the functions f (x) is described as:

f(x)=pf,(x)+(l-p)fz(x), . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)

where p is the proportion units that have f, (x) variate values which come from aprobability distribution function for summer rains, and the remainder “(l-p)” have variatevalues which wme from another probability distribution function f 2 (x) for winter rains.

Table 3.2. Percentage of seasonal to annual rainfall maxima for diierent return periods

M5 I Ml0 M20

percentage percentage percentage

Winter 1 Summer Winter ~Summer

20304050

I 6070a090

192938

46

5664

7286

22344354

I 64768897

M30 M!50 Ml00

percentage percentage percentage

Winter 1 Summer Winter I Summer Winter 1Summer

I21 16 2333 26 3443 30 44

53 36 54

66 53 6681 58 79

86 70 8796 82 98

The procedure of the fitting in such a case. is to split the rainstorm reconds within aduration into two sets: the first set includes the summer rains while the other indudes thewinter rains. A best-fitted probability distribution function fo r each set of records isidentified. Then, an optimization methodology can be adopted to determine a proportionunits “p.” The methodology changes automatically the value of “p” until1 the standarderror, resulting from comparing the observed rainfall depths with the estimated ones, isminimized.

. The DDF curves provide a way to estimate the average rainstorm depth for a specified

duration and a specified return period over a catchment. The effect of rainfall spatial andtemporal variabilities over small size catchments is normally ignored in the traditionalmethod of urban drainage design. However, many investigators have found that theproduced runoff from a rainstorm varies with both rainstorm depth and spatial (Schiling,1983) as well as temporal (Akan and Yen, 1984) variabilities of the rainstorm over thecatchment. In arid areas, such variabilities are significant even over small sizecatchments. Neglecting these variablitites can be one important source of errors inrunoff prediction.

The depth of rainfall should, therefore, be reduced to account for such high variabilities ofrainfall over catchments. The reduction factors are found tc vary with both catchment sizeand rainfall return period (Nouh, 1991). The following table shows typical results obtainedfrom data analyses in the eastern and western zones of the southwest region of Saudi

Arabia (Nouh, 1988~~).

Table 3.3. Areal reduction factors (%) in western and eastern areas in the southwest regionfor Saudi Arabia

-prid:)M

2 5 10 m so la0

95 93 *3 9s 92 93

Sl 75 1s 7s 7s 748s85UWM 83

SI s8 s8 1 so 48

I8484847473 763 a36 839 934 4s43

4943494949 47

bW,pIiLld:~ O#lW9priod:~

2 J IO 30 50 IQ) 2 J 10 38 m 108%95959595 9s 9999999Bu 97

479747969491 41 91 91 90

8s 9393939393 90I 9s 9s 9s 95 9s 93

mmb’lomm 6882828ZPQ n

878787nr 8s 9393939393 91

7070703010 70 mma*la m91 91 91 91 91 91 Y 94 u 91 94 9(

37



. Traditionally, the return period of the design rainstorm is normally assumed to ,be thesame as the return period of the produced runoff. This assumption is proved to be invalidin arid catchments (Nouh, 1990a) where rainstorms as well as various hydrologicabstractions (i.e. evaporation, infiltration, depression storage, . ..etc.) have distinguishedcharacteristics. These characteristics result normally in having return periods of runoff

longer than the corresponding return periods of rainstorms. The ratios between the returnperiods of runoff simulated by the SWMM model and those of the corresponding designrainstorm increase with the increase in the return period, and vary with both catchmentand rainstorm characteristics. The following figures show the trend of such variations(Nouh, 1990a). In these figures, A is catchment size, f= is the constant intItrationcapacity, SRV and TRV are measures of the spatial and temporal variabilities of rainfallover the catchments, respectiiely, and defined as:SRV = A measure of spatial rainfall variation, taken as the ratio of the time-averagerainfall intensity of a hyetograph at the center of a rainstorm to the time-average rainfallintensity of the rainstorm over the catchment,TRV = A measure of temporal rainfall variation, taken as the average ratio of time topeak intensity of hyetographs to the total duration of the hyetographs over the catchment,D is total rainfall depth (mm), and P is the average daily variation of temperature over

catchments (Co ).

Figure 3.2. Return period ratios for simulated runoff peaks (top) and for simulatedrunoff volumes (bottom) using different shapes of hyetograph.

. The decision on the time distribution of rainfall depth could be another source o f runoffprediction in arid dimates. Akan and Yen (1984), among others, have shown thatdifferent shapes of hyetographs of same rainfall depth produce different runoffhydrographs. Figures 3.4 & 3.5 show different hyetographs simulated from the same

rainfall depth, and the corresponding runoff hydrographs.

38



sRvpo.2,TRv=o.2P=&l’C,f 3321 r/hr

1.0. - -a*---. l ____ L . . ..a w

2 5 10 50 loo 5 10 50 100

RmmIl PBLIOD O? TRIARGW IJESICII llAnwmM, y-ta

Figure 3.3. Effec t of catchment, climate and rainfall characteristics on the return periodratios for the simulated peaks (left) and volumes (right)

Analyses of large number of rainstorms in arid climates in the Arabian Gulf Sates (1990b)

have indicated that the hetograph in these climates can be reasonably described by thefollowing linear relationships:

i U)Hpv&t-a),(tl-a)i (t) HP e

for 0 S t I a (2)for a 5 t s D .’ .’ .’ .’ .’ . . .’ . . .’ . (3)

where HP is the peak intensity of hetograph, and k and a are parameters to be determinedfrom the following Table 3.4.

For rainstorm return period longer than 5 years, the parameters k and a may be modifiedas km and am , respectively, as:

km = k ( 1 + Ak T + Bk T* + Ck T3 ) ; for5ITI 100 . . . (4)

am=a( l-AaT+BaT2+CaT3) ; for5 ~TI 100 . . . . (5)

where A, B, and C are coefficients to be determined from the following Table 3. 5.

For rainstorm duration more than one hour, further modification may be adopted as:

km (d) = km ( 1 + 0.268 e-“d ) ;forlSdS12 . . . . . . . . . . . . . (6)

am(d)=am(1-0.109e-"d) ;forlSdS12 . . . . . . . . . . . . . . . (7)

where km (d) and am (d) are the modified parameters km and am , respectively, for a duration ofd hours.

39



r Iaama

1 Kmif~rbcbul

t Iorahf iold

a 8Uff

4 lifaldm

6 IC8-0f8 ll#riDbCore*r)

0.4

0.3

YEs: 0.2

ii

0.1

0

30 lILIpsri.

Figure 3.4. Shapes of design hyetographs

60

3 trBposo~d8l 0 CoDSo*ito -

0 30 90

Figure 3.5. Simulated runoff hydrographs from different hyetographs

40



Table 3.4. Average values o f k and a ( d < 1 hr and T < 5 yrs)

Location

Saudi Arabia,Southwest

Saudi Arabia,CentralSaudi Arabia,NortheastKuwaitBahrainOmanUAEQatar

Location

Saudi Arabia,SouthwestSaudi Arabia,CentralSaudi Arabia,NortheastKuwaitBahrainOmanUAEQatar

No.

Summer Rain&

914

175

981248a8

aa

10

12

1513121012

ms wi

k No.12 809

nter Rainsto

a11

10 637 11

9 514 15

8 33 179 26 1511 29 IO8 32 810 27 9

Table 3. 5. Coefficients for modifying hyetograph parameters.

1.35, 0.67

1.10,0.50

0.85, 0.50

0.90, 0.501.70, 0.802.70, 1.002.50, 1.20

Summer Rainstorms , Winter Rainstorms

(X 5) (Xb7) (X $5)

3.47, 1.15 4.77, 2.13 0.93, 0.55 6.33, 3.17

1.05, 0.90 2.15, 1.00 0.50, 0.50 4.50, 2.50

0.65, 0.95 1.50, 0.90 1.55, 0.80 1.50, 2.00

0.70, 0.90 2.00, 0.90 1.70, 1.00 1 oo, 2.002.50, 0.90 3.80, 1.40 2.56, 1.00 1.75, 0.501.55, 0.90 2.70, 1.00 2.50, 1.00 1.50, 0.801.50, 1.00 2.50, 1.00 1.80, 1.0 1.50, 0.80

ns

k9

a

a

;9a12

2.50, 0.90

6.50, 1.00

6.00, 1.001.50, 0.601.50, 0.701.50, 0.70

3.1. 2. Evaporation

Because urban drainage deals generally with short duration processes, the evaporationprocess is normally neglected in the traditional procedure of design. However, in aridclimates, the evaporation amount is significant and represent an important rainfall abstractioneven during rainstorms duration shorter than 30 minutes (Nouh, 1982, 1987c, 1988a). Thus,neglecting this evaporation amount may result in considerable over-estimation of runoff.

3. 1.3. Infiltration

Infiltration is a major rainfall abstraction in and dimates. Its rate is a function of soilpermeability and initial soil-moisture content. The estimation of such rate at a time in most

cases is guess-work unless site specific inflltrometer results are available. Even at a site atwhich infiltration tests have been performed, a change in ion concentration due to majorrainfall or runoff events or due to some form of surface pollution may alter the soilpermeability, and accordingly infiltration parameters drastically. Nouh (1996c) hasinvestigated the effects of changes in the infiltration parameters on runoff prediction in aridcatchments. The investigation was based on comparing observed hydrographs withhydrographs simulated by calibrated D& model. In the simulation, three routing methods areconsidered. These are the characteristics, the explicit and the implicit methods. The results,which are summarized in Fig. 3.6, show that infiltration parameters have significant effects onrunoff predicted using any of the routing methods. In the figure, EVC is a soil moistureparameter; BMSN is the available soil water at field capacity; and RR is the proportion ofrainfall that infiltrates into the soil.

For the purpose o f urban drainage design, infiltration is accounted for by one of thefollowing methods:

41



l Use of the infiltration curves that have been proposed by ASCE and WPCF for sandysoils, residential areas, and industrial-commercial areas (ASCE, 1970). These curveshave the major drawback of not accounting for soil water storage, which is animportant consideration in arid catchments.

-la 1.2

-m- IiEzzrE-

I

Figure 3.6. Effect of changes in the infiltration parameters of the D& model on (A) sum ofsquares of differences between observed and predicted flows; (B) magnitude of peak

discharge; and (C) magnitude of runoff volume.

l Use of equations describing the infittration process ( e. g. Horton, Green and Ampt,Phillip,. . .etc). A review of these equations can be found elsewhere (Viessman et al.,realistically to determine.

4 Use of the SCS procedure (USDA, 1975). In this method, the initial soil abstraction isassumed to be equal to 0.20 times the catchment storage.to be invalid in and climates (Nouh et al., 1988b).

This assumption is provedIt has been found (Nouh et al.,

1988b) that the initial abstraction is within the range of 0.25 and 0.40 times the andcatchment storage. Due to the high spatial variation in soil parameters, even with theuse of a calibrated value fo r the initial abstraction, significant prediction error ‘PE(defined as the ratio of the difference between observed and predicted peak dischargeto the observed peak discharge) have been identified (Nouh, 1988c). Table 3.6 show

42



typical results in some arid catchments in Saudi Arabia. It can be seen that PEincreases as catchment size and/or as return period of predicted flow increases.

In addition to the above difficutties in determining the infiltration in arid catchments, there arethe problem of recovery and the problem o f high temporal and spatial variabilities of infiltrationcapacity. The former problem arises when rainfall ceases and there is a recovery of

infiltration capacity with time. The extent of recovery at any point in time depends on thedryness of the period. Although this condition is recognized in the SWMM continuoussimulation model, the results of the simulation model in extremely and catchments have beenfound to be less than satisfactory (Nouh et al., 1988b). The latter problem of the highvariability of infiltration may be overcome by subdividing the total catchment into components,each has an approximately uniform soil and cover properties, and then deal with eachsubdivision independently. The results of this approach suggest further development inreasonably determining the infiltration in arid catchments.

Table 3.6. Prediction error “PE” (%) for different flow return periods

tion

Size of drainagebasin: km2

Return period af prediction flood flow : years

f 2 3 5

80 21.5 23.0 26.8

120 23.0 26.7 30.21200 23.0 27.5 31.72671 31.4 33.6 35.74713 35.0 36.0 36.5

33000 37.0 41.5 42.5

Meanpredictionserror : %

26.9229.9731.0336.6336.8341.78

1

3. 1.4. Sedimentation

Climate in arid areas affect the concentration of sediments in runoff. The high temperatureand low humidity and their large daily variations destruct the soils from pervious surfaces andprovide loose materials to be carried with flash stonnwater to the drainage system. Inaddition, dust, dirt, sediments carried with sandstorms and pollutants of various kinds, settled

from the atmosphere and generated by urban activities, accumulate on the impervioussurfaces during the long rainless period and are eventually washed off by the flash runoffduring the rainstorms. These mechanisms result in large amounts of sediments to betransported to the various elements of drainage systems.

The transported sediments create one or more of the following two major problems in thedesign of urban facilities:

l Decrease of accuracy in routing the flow through the various drainage systemcomponents (Nouh, 1996e). The main reason is that the common routing modelsassume clear-water flow. This assumption is not realistic due to the large amount ofsuspended sediments carried by the flow. The flows in fact could be described assediment-laden flows or hyperconcentrated flows, which have characteristics differentfrom those of dear-water flows (Nouh, 1989, 199od). Generally, dynamic viscosity as

well as friction slope change with the change of suspended sediment concentrations.Recalling the relative dynamic viscosity as the ratio between the dynamic viscosity inflow with sediment (i.e. hyperconcentrated or sediment-laden flow) and the dynamicviscosity in dear-water flow, Fig. 3.7 shows such changes with mean suspendedsediment concentration.

l Partial or full blockage of sewers due to settlement of large amounts of sedimentsduring the rapid recession of hydrographs, and then the consolidation of suchsediments during the long dry period “rainless” between successive rainstorms. Suchblockage leads to formation of street lakes, which cause many environmental hazards(Fig. 3.8).

43



hEAN CONCENTRATlON OF SUSPENDED SfMMprr T”. ml

Figure 3.7. Variation of viscosity and friction slope with suspended sediment concentrations

Figure 3.8 Formation of Street Lakes Due to Drainage Sewer Blockage

3. 1. 5. Water quality

Stormwater quality in arid climates is mainly affected by suspended sediments transportedwith flash floods. The sediments may be originated from soil erosion during rainfall eventsand/or from duststorms that normally exist in and climates. The amount of soil erosiondepends on local conditions; such as degree o f urbanization, type of land use, densities ofautomobile traffic and animal populations, and rainstorm characteristics. The effect of

44



various factors on soil erosions and accompanied transported pollutants can be foundelsewhere (Foster et al., 1974; Wischmeier et al., 1978; McElroy et al. 1976). Duststormsalso play important role in the transportation of suspended sediments and pollutants instormwater runoff. Recent investigations (Nouh, 1998a) have shown that the increase in theconcentrations of total suspended particulate matter “TSP” in the air increases theconcentrations of suspended sediment “c” in sewer flows (Figs.3.9 & 3.10). In addition, both

the spatial and tempora l variations, measured by the coefficient of variation and thecoefficient of kurtosis, of a duststorm have significant effect on the concentrations “c” (seeFigs. 3.11 & 3.12). Another investigations (Nouh, 1998b) have shown that both the duststormmean concentration of TSP and its spatial and temporal variabilities have considerable effecton the concentrations of pollutants in the stormwater runoff (see Figs 3.13, 3.14, 3.15).

1

0 500 1000 1500 2000 2500 3000

CONCENTRATION OF TSP, u&urn

Figure 3.9. Variation of mean TSP s

concentrations with mean C in sewerflows

0 02 0.4 0.6 0.8 1

COEFlm OF VARL%TlON OF DUSEXORM

Figure 3. 11. Effect of spatial variationof duststorm on C in sewer flows

0 loo0 2ooo 3ooo 4ooo

TSP CONCENTRATION, ug’cu.m

Figure 3. 10. Variation of maximum TSP

with maximum C in sewer flows.

-0.50.5 0 0.5.5 1 1.5.5 2 2.5.5

COEFFICIENT OF KURTOSSOF DUSTSTORMOEFFICIENT OF KURTOSSOF DUSTSTORM

Figure 3. 12. Effect of temporal variationof duststorm on C in sewer flows.

45



21 +‘N02;@03 +)lC +TP j

10-

1.6

0.8

0 . . ..i....i....i.*..i....i..*.

500 1000 1500 2000 2500 3000 3500

TSP CONCENTRATION (ug/cu.m)

Figure 3. 13. Influence of mean concentration of TSP (mg/m” ) on meanconcentration of pollutants (mg/l) in stormwater runoff.

-NCY+03 j-DC ; -cq j

-ca$ +“‘$I +C@O j

Xd*iCQ &Pb*ld +Z&IO

0.5 1.5 2.5 3.5 4.5 5.5 f

COEF”ClENTOF “ARRIATION OF D”S,-STORM

Figure 3.14. Influence of duststorm Figure 3.15. Influence of duststormcoefficient of variation on concentration coefficient o f kurtosis on concentration of

of pollutants (mg/l) in stormwater runoff pollutants (mg/l) in stormwater runoff.

46



A recent study (Nouh, 1997) has indicated that the concentrations of pollutants in stormwaterrunoff can be predicted from a regression type relationships as:

Cd Ad + a,dqBld + a2dsp2d . . . . . . . . . . . . . . . . . . . . , . . . . (8)

Pd= adC,bd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)

where C is the mean concentration of suspended sediments (gpl); q is the stormwater runoffdepth over catchment (mm); S is the total suspended particulate matter concentrationaveraged over the catchment (@ m’); h, a, 6, a and b are regression parameters to bedetermined. The subscript d refers to a division range of suspended sediment diameters.

The regression parameters together with their statistical inferences are given inTables 3.7 to 3.9. The tables indicate that the suspended sediment concentrations in thestormwater runoff are affected by the characteristics of both flash floods and duststorms overcatchments. Although the effect of flash floods is more signiffcant than that of duststorms, theinfluence of duststorms on suspended sediments and pollutants transport can not be ignored.

This observation is absolutely valid especially in developing areas.

3. 2. Maintenance, operation, and management

Generally, maintenance is required from time to time to protect the integrity of drainagefacilities. The maintenance intends to resolve problems mainly generated by erosion,sedimentation, and accumulation of leaves and trash in a drainage system. Due to thedistinguished climate characteristics in and areas, maintenance in and catchments should bemade more frequent than that in similar catchments in non-arid areas. Normally,maintenance in non-arid catchments is made once a year or after a major rainstorm. In andcatchments, the maintenance should be made after each rainstorm (I.e. 3 to 4 times yearly).

Extreme fluctuations in temperature, high nocturnal and low diurnal humidity, and

violent solar action that normally exist in arid areas cause chemical changes in certain soilelements, frequently resulting in the breakdown of rocks and disintegration of soils. Thebroken rocks and the disintegrated soils furnish the flash intense rainfall with large amounts ofsolids to be carried downstream. These solids size varies from very fine sediment to largeboulder (Fig. 3.16).

Figure 3. 16. Sediments of varying size transport to drainage facilities

47



Table 3.7. Regression p arameters for mean concentration of suspended sediments.

d (mm)d > 0.20

h (lower, upper) al (lower, upper)

23.3( 19.6,27.4) -1.43(-l .49, -1.38)

p1 (lower, upper)

0.53(0.49,0.58)

a2 (lower, upper) f32 lower, upper) R’ Ir

3.62 (3.11,3.94) 0.033(-0.008, 0.041) 0.76 0.1030.20 2 d > 0.060.06 2 d > 0.020.02 2 d > 0.002

0.002 2 d

37.6(3 1.4,43.8) -1.41(-1.43, -1.32) 0.67(0.61,0.72) 4.26 (3.91,4.72) 0.043 (0.037,0.050) 0.81 0.1015 1.3(48.6,55.2) -1.53(-1.62, -1.39) 0.73(0.61,0.82) 4.93 (4.28, 5.61) 0.053 (0.042, 0.061) 0.83 0.092

67.4(63.7,74.8) -1.71(-1.89, -1.59) 0.92(0.88, 1.03) 5.27 (4.87, 5.63) 0.078 (0.071, 0.087) 0.88 0.087

73.6c69.1.77.3) -1.74(-1.91, -1.62) 0.93cO.8 .0.98) 6.47 (6.11.6.93) 0.082 (0.069.0.093) 0.87 0.083

Table 3. 8. Regression parameters for heavy metals concentrations.

d Copper “Cu” Lead “Pb” Nickel “Ni” Zinc “Zn” Iron “Fe”

(mm> a b r a b r a b r a b r a b r

d > 0.20 0.032 0.046 0.89 0.007 0.051 0.62 0.056 0.009 0.39 0.033 0.106 0.71 0.089 0.117 0.780.20 2 d > 0.06 0.042 0.057 0.88 0.009 0.063 0.63 0.062 0.031 0.57 0.029 0.098 0.73 0.097 0.103 0.810.06 2 d > 0.02 0.776 0.093 0.89 0.005 0.096 0.69 0.091 0.053 0.59 0.045 0.102 0.78 0.107 0.125 0.81

0.02 r d > 0.002 0.916 0.153 0.96 0.006 0.162 0.86 0.072. 0.096 0.64 0.061 0.120 0.86 0.126 0.216 0.870.002 2 d 0.103 0.196 0.93 0.011 0.123 0.77 0.083 0.104 0.69 0.097 0.155 0.85 0.133 0.238 0.91

Table 3.9. Accuracy o f performance of the regression equations.

1 dL

(mm>

d > 0.20 1.08 0.24 18 1.12 0.19 110.20 2 d > 0.06 1.06 0.22 16 1.13 0.13 15

0.06 2 d > 0.02 0.92 0.20 11 1.08 0.12 12

0.02 r d > 0.002 0.96 0.16 9 0.96 0.10 9

0.002 2 d 0.98 0.11 6 0.97 0.15 7

C Copper “Cu”Rat Dev Abs Rat Dev Abs

.Lead “Pb” Nickel “Ni” Zinc “Zn” Iron “Fe”

Rat Dev Abs Rat Dev Abs Rat Dev Abs Rat Dev Abs

1.26 0.33 26 1.26 0.66 59 1.18 0.29 23 1.14 0.24 191.17 0.29 30 1.36 0.51 53 1.12 0.22 22 1.08 0.23 171.12 0.23 34 1.28 0.47 47 0.92 0.23 19 1.04 0.21 150.87 0.22 19 1.94 0.45 42 0.96 0.21 18 0.96 0.19 150.88 0.27 25 1.96 0.46 38 0.94 0.19 20 0.98 0.16 13

40



In addition, because hyetographs as well as resulting hydrographs are characterized by sharprise, slide of embankments does normally exist with large amount of sediment transport to thedrainage facilities. Rainfall has normally significant impact on the movement of slidematerials, through surface erosion and mass washing events - sediment slumping andsliding. In these cases, two types of solids can be produced and transported to the drainagefacilities: the suspended fraction of silt, and the coarser materials that are rolled down along

the slope. The removal of fine-grained soils from the slopes allows the rainwater to entertissues and fractures, while the movement of huge boulders on the slopes increases the voidsin the soils. The water filling of the generated tissues and fractures in addition to the rise ingroundwater levels during the heavy rainfalls result in high pore water pressure that reducethe soil shear strength and its slope stability, leading to serious landslides. Fig. 3.17 showstypical examples o f such landslide. Details of these problems are reported elsewhere (Nouhand El-Laithv. 1988b 1.

Figure 3. 17. Typical Landslides in And Climates

In addition to the above, and as previously mentioned, the arid climate is normallycharacterized by existence of duststorms, which cause considerable amount of solids to besuspended in atmosphere and to settle on land surfaces. The amount of these suspendedsolids in atmosphere can be in the order of 10 times more than that in areas of insignificantduststorms. As rainfall moves through the atmosphere it washes out the suspended solids

and carries them to the land surface. Upon reaching the ground it will dislodge some particles

49



(mostly soil on pervious surfaces, and wide variety of settled solids and debris on impervioussurfaces) and dissolve other materials. The produced stormwater runoff carries the particlesdislodged by initial precipitation impact, other particles dislodged by the movement of therunoff itself, and a variety of dislodged materials to the drainage system. The result is thetransport of considerable amounts o f sediments to the drainage system. The long dry periodbetween two successive rainstorms assist in accumulating large quantity of sediments settledfrom atmosphere, leaves and trash, . .miscellaneous rubbish on the land surface to bewashed by the flash floods to the drainage facilities, causing serious operation managementproblems.

Normal practice in arid climates include:. Embankment and slopes protection against sliding may be by placing granular materials

and/or planting grass cover on the slopes.. Retardation of sheet erosion by using grass cover plantation and/or riprap placement on

the surface.. lnstallment of concrete sediment racks in the detention basin to avoid the transport of

debris and boulders. The area of the rack should be large enough to hold up quite alarge mass of material without impeding the flow of water.

. lnstallment of trash racks to hold rubbish, papers, leaves, etc.

Routine maintenance in such climates should include after each rainstorm checking theembankrnents and repair the damages, checking the concrete and metal components of thedrainage system and make necessary restoration, clean both the concrete sediment racksand trash racks, and clean the settled sediments and the rubbish materials from the streets.

3. 3 Data acquisition and processing

Acquisition of reliable data in and climates requires sophisticated sampling program due tothe following factors:

1. The random nature of stormwater runoff2. The short duration and high rise of both hyetographs and hydrographs

3. The high spatial variation of catchment parameters4. The high spatial and temporal variations o f rainfall and stormwater runoff5. The high concentration of sediments in the stormwater6. The long dry period between two successive rainstorms

The random nature of stormwater runoff and the high spatial variations of rainfall, runoff, andcatchment parameters need large amount of reliable data in order to properly describe thedistribution of water quality and quantity. The high temporal variation does not allow reliabledata to be sampled manually. The high concentration of sediments in the stormwaterimposes limitations on the use of some sampling equipment and on the implementation ofcommon s tormwater runoff models in stonnwater simulation. The long dry period betweentwo successive rainstorms allows settlement of large amounts of sediments and accumulationof rubbish materials on the catchment surface, which impose further limitations on data

sampled by common methods. Considering the above factors, successful sampling can bedone as in the following:

Water aualitv measurements:Due to the short duration hyetogmohs, especially in small size catchments, peak loading ofpollutants borne by stormwater takes place before personnel are able to occupy sites andstart manual sampling. So, automatic sampling can only provide reliable data. A series offlow measurements and samples are collected at the same time of the measurements duringthe runoff event monitored. The time between two successive samples is as short as 5minutes. Laboratory analyses of the collected samples can then be made. Based on themeasurements and the laboratory analyses, the runoff hydrograph and a curve of pollutionloading as a function of time may be plotted. The area under the curve gives a good estimateof the total pollution load for the event. This method is expensive, but provide reliableinformation on either the total loading or the mean loading (concentration) of pollution.

50



Flow measurements:Many methods are generally used to measure stormwater flows. These include Currentmeter, Flumes, Weirs, Float velocity, and Tracer dilution. Analyses of the majority of thesemethods are given elsewhere (Grant, 1979). In arid climates, because of the previouslyexplained distinguished characteristics, the use of a flume is found to be the most suited one(Nouh, 1988b). However, it requires continues cleaning, especially from sediment andrubbish settlement.

Rainfall measurements:There are several rainfall measurement devices availabte for runoff study. Measurementtechniques include sight gages, weighing and counting tips of a balanced duel cell bucket. Inarid climates, reliable rainfall measurements may be taken by using automatic gages. It isrecommended to have a rain gage every 10 km* of catchment. For catchments of largetopographical variations, the rainfall gage should cover an area of catchment as small as 5km’ (Nouh, 1988b).

3.4 Application of common urban drainage models

As it has been indicated earlier that stormwater flows in arid climates are normallyaccompanied by transport of large amount of suspended sediments. The flow properties inthese situations are different from the properties of clear-water (i.e. clear water free ofsediment) which are commonly assumed to be in most urban drainage models. Theanticipated predicted results are hydrographs of .low (may be unacceptable) accuracy. It hasbeen found (Nouh, 1987d) that the increase in suspended sediment concentrations in thestormwater runoff decreases the channel friction slope “Sr”, which is basically used in therunoff prediction by common urban drainage, models. The decrease in the friction slope “ASt ” due to a concentration of suspended sediment in the flow is expressed relative to thefriction slope ‘Sr y (evaluated using a calibrated Manning’s coefficient for clear water) andplotted against the sediment concentration in Fig. 3.18. It can be seen that the decrease inthe friction slope can be as much as 20% of its clear water value if the concentration ofsuspended sediment in the flow is as high as 40 gpl.

CWTRATI(w 2’. gpl .

Figure 3.18. Percent decrease in friction slope with increase in suspended sedimentconcentrations

The deficiency of urban drainage models in predicting stormwater runoff is demonstratedthrough the use of SWMM in two urban catchments of diierent size in the southwest region ofSaudi Arabia ( Nouh, 19874 ). The catchments are of similar properties and have almost thesame rainfall characteristics. However, the smaller size “RABHA” is of area equal to 1430 x1O4m* , whereas the larger size “AI-DAWAL” is of area equal to 3718 x 10” m* . The SWMMwas used to predict peak flow, time to peak flow, and volume of runoff hydrograph. Theprediction was made twice; the first prediction was made according to the common procedure

51



using E+ evaluated as if the water free of sediments (i.e. dear-water flow), whereas thesecond prediction was made with S modified according to the concentrations of suspendedsediments in the flows (i.e. using Fig. 3.18). The two predictions were compared with thecorresponding measured quantiiies and the results are shown in Fig. 3.19. Similar resultswere obtained by using ILLUDAS and UCUR models (Shaheen et al. 1998).

D

5

Figure 3. 19. Measured and predicted hydrographs in two arid catchments

It is apparent that the consideration of suspended sediment concentrations in the stormwaterrunoff is a key issue in the process runoff prediction by common models.

3.5 Interaction with other urban water systems

Generally, stormwater runoff interacts with other urban water systems. If properly collected, itcan be a measure for environment protection and a source for fresh water to cover significantpart of the water supply shortages that are expected to be faced in many countries in theyears to come. Fig. 3.20 Shows the world water shortages in 2025 (Seckler, 1999).

This role of stormwater runoff is dearly noticeable in arid climates, where renewablefresh water supply is insufficient to meet urban demand. Considering the Arabian GulfStates, as countries of typical arid climates, severe water shortages would be faced. This willrequire that almost all-fresh water resources be allocated for domestic consumption or othermunicipal and industrial purposes. While Kuwait and Qatar face this problem today, within a25year period countries such as Bahrain, Saudi Arabia and United Arab Emirates will havefrom their current renewable natural water reserves (without desalination) just about the limitof what is needed for survival, with no fresh water available for agriculture. It is estimated thatthe available quantity of water from natural renewable sources, excluding desalination in theyear 2025 will range from 10 to 20 m3 /person/year. The actual domestic water consumptionrequired in a modem home with indoor plumbing, hot and cold running water and minimalhousehold equipment such as a washing machine ranges from 35 to 70 m3 /person/year. Theexpected population growth together w ith water supply and demand in some of the ArabianGulf States are shown in Table 3.10.

52



Water Scarcity in 2025

Group 7 - phyrlul wat8r scarcRy

cl

chup 2 - economk water rcu ctly

group 3 - modemte noed forwatar devrtopmaat

n$!j$! Not estimated 1WMl Indicator of Relative Water Scarcity

Figure 3. 20. World water shortages in 2025 (after Seckler, 1999)

53





the total wastewater treatment and disposal cost or about US$ 0.10 out of a total of US$0.35/m3 . Recycling of wastewater can have the multiple benefit of protecting the environmentand serving as a major source of water and nutrients for the soil. Some of the Arabian GulfStates (UAE included) have embarked on successful wastewater recycling programs.

On the &her hand, the rare flash floods that are generated from intensive rainstormsin these satates, have significant economic and environmental impacts. The floods, if

properly managed, can be significant fresh water source instead of being a source of publicinconvenience and serious environmental impacts. Fig. 3.21. Shows construction damagesdue to flash floods in the southwest region of Saudi Arabia (Nouh et al. 1988b). Recentinvestigations (Zuhair et al. 1999) have shown that proper harvesting of flash floods in theplaces of severe water shortages in the Arabian Gulf States can secure a significant freshwater quantii as in the following. Kuwait: 12% of water demand for agriculture; Oman(Muscat): 27% of water demand for industry; UAE (Al-Ain): 16% of water demand foragriculture; and Saudi Arabia (Abha): 11% of water demand for industry and landscapeirrigation. These results confirm the strong interaction of stormwater runoff with other urbanwater systems.

Conclusions

This chapter discusses specific problems arise from the design of urban facilities by traditionalmethods in arid and semi-arid areas. The problems are originated from the fact that thedistinguished climatic characteristics in arid and semi-arid areas are different from those inother climates, where the traditional methods of design are developed and practiced.

In arid and semi-arid areas, rainfall amounts are small with high seasonal variations.So, utilizing a mix of two probability distribution functions to model rainfall depth could bemore appropriate than using a probability distribution function to model the same. In addition,the rainfall has high spatial and temporal variations, even over catchments as small as 5 km* ,so accounting for these variability are seriously needed. This chapter proposes a reductionof the rainfall amount by a factor, which has been found to depend on catchment size and onrainfall return period. On the other hand, the return period of simulated runoff hydrographshas been found to be significantly diierent from that of design rainstorm. This chapter

proposes a methodology to realistically identify the design hyetograph and to estimate therunoff return period from a given return period of rainstorm.

Evaporation and infiltration are found to be important hydrologic abstractions evenduring the short duration of runoff. Evaporation is subject to high ddly as well as seasonalvariations. The infittration is subject to high spatial and temporal variations. Realisticconsideration of evaporation and infiltration in the urban runoff prediction by tradiionalmethods is not available at present, so this chapter suggests further developments to bemade to reasonably determine the infiltration in and and semi-arid catchments.

Sedimentation is another important factor affecting runoff hydrograph and its quality.Transported suspended sediments, generated from soil erosion and from particles depositedfrom heavy duststorms, affect the physical properties of water and thus results in deweasingthe accuracy of available flow routing models. The transported sediment results in sewerblockage, and requires careful drainage system maintenance program. This chapter a

methodology is presented to predict the concentration of suspended sediments in sewer flow,and also to predict the effect of such concentrations on the chemical constiiuents in the flow.Due to the above limitations of the existing tradiional methods, further research in the

prediction of runoff in and and semi-arid climates is recommended.

Bibliography

AMERICAN SOCIETY OF ClVlL ENGINEERS AND WATER POLLUTION CONTROLFEDERATION (1970) “Design and Construction of Saniiry and Storm Sewers,”Manual of Practice, No. 9, Water Pollution Control Federation, New York.

Al&N, A. 0. AND YEN, 8. C. (1980) “Effect Of Time Distribution Of Rainfall On OverlandRunoff,” Proceedings, 3ti International Conference on Urban Storm Drainage,

IAHRIIAWPRC, Goteborg, June 1984, pp. 193-202.

55



FOSTER, G. R., WISCHMEIER, W. H. (1974) “Evaluating Irregular S lopes For Soil LossPrediction,” Transactions of ASCE, Vol. 17, No. 2, pp. 305309.

GRANT, D. M. (1979) “Open Channel Flow Measurements Handbook,” Instrumentationspecialties Co., Lincoln, Nebr.

RESTREPO-POSADA, P. J. AND EAGLESON, P. S. (1982) ‘Identification Of IndependentRainstorms,” Journal of Hydrology, Vol. 55, pp. 303-319.

MCELROY, A. D. ET AL. (1976) “Loading Functions For Assessment Of Water PollutionFrom Nonpoint Sources, y EPA-600/2-76-151, U.S.EPA, Washington D. C.NOUH, M., (1982), ” On The Uses Of Minicomputers For Identification Of Nonlinear Surface

Runoff Systems,” Int. J. of Control and Computers, Int. Assoc. of Sc. & Tech. forDevelopment, Publisher: ACTA Press, Vol. 10, No. 1, pp. 1521.

NOUH, M. (1987a), “Analysis Of Rainfall In The Southwest Region Of Saudi Arabia,”Proceedings, Institution of Civil Engineers, Part 2, Vol. 83, pp. 339349.

NOUH, M. (1987b), “Point Rainfall Distribution In The Southwest Of Saudi Arabia,” Journal ofEngineering Sciences, Vol. 13, No. 1, pp. 85106.

NOUH, M. (1987c), ” Effect Of Rainfall Runoff Model Assumptions On Optimal Storm SewerSystem Design,” Arabian Journal for Science and Engineering, Publisher: John WI@& Sons Ltd, Volume 12, Number 1, pp. 19-35.

NOUH, M. (1987d), “Effect Of Very Large Sediment Concentrations On Highway Sewer

Flows,” the 4th International Conference on Urban Storm Drainage and the XXIICongress of the International Association for Hydraulic Research, The InternationalAssociation for Hydraulic Research and the lntemational Association of WaterPollution Research and Control, Laussane, !%tzerland, pp. 343348.

NOUH, M. (1988a), ” Estimate Of Floods In Saudi Arabia Derived From Regional Equations,”Journal of Engineering SSciences, Publisher: John Wiley & Sons, Volume 14,Number 1, pp. l-26.

NOUH, M. AND EL-LAITHY, A. (1988b), “Construction Damages Due to Floods in WadiAddilah,” Find Technicd Report Prepared on the Saudi Arabian National WaterResearch Project Number AR-562 for the King Abdulaziz City for Science andTechnology, Riyadh, Saudi Arabia, 212~.

NOUH, M. (1988c), ” On The Prediction Of Flood Frequency In Saudi Arabia,” Proceeding ofthe lnstiiution of Civil Engineers of UK, Part 2, Publisher: Thomas Telford Ltd, Volume85, pp. 121-144.

NOUH, M. (1989) “The Von-Karman @efficient In Sediment Laden Flow,” Journal ofHydraulic Research, The International Association for Hydraulic Research, Volume27, Number 4, pp. 477499.

NOUH, M. (1990a), ” Relationships For Return Periods Between Design Storms AndSimulated Runoff In Urban And Catchments,” the International Conference on UrbanStorm Drainage, The International Association for Hydraulic Research/TheInternational Association of Water Poflution Research and Control, Suita, Osaka,Japan, July, pp. 1137-l 142.

NOUH, M. (1996b), “Design Hyetographs For Arabian Gulf States,” Seventh Congress of theAsian -and Pacific Regional Diiision,The International Association for HydraulicResearch, Beijing, China, November, pp. 317-326.

NOUH, M. (199(k), ” Calibration And Verification Of The DR3 Model In Arid Watersheds,” the

lnternationd Symposium on the Hydrological Basis for Water ResourcesManagement, The International Association of Hydrological Sciences, Beijing, China,October 1999, Paper B19, IAHS Publication No. 199, pp. 67-77.

NOUH, M. (199Od), “Reliability Of Experimental And Numerical Methods Of Hyper-Concentrated, Sediment-Laden, And Clear-Water Flood Flow Routing,” theInternational Conference on Physical Modeling of Transport and Dispersion, TheInternational Association for Hydraulic Research/ The American Society of CivilEngineers, Massachusetts institute of Technology, Cambridge, Massachusetts, USA,August, pp. 14B.l-148.6.

NOUH, M. (199Oe), “Hyperconcentrated, Sediment-Laden, And Clear-Water Flow RoutingUsing Numerical Methods,” Int. Conference on River Flood Hydraulics, IAHR/Wallingford Hydraulic Research, Wallingford, England, September, pp. 359-368.

NOUH, M. (199Of)“Relationships For Return Periods Betwen Design Storms And SimulatedRunoff In Urban Arid Catchments,” the lnternational Conference on Urban Storm

Drainage, The International Association for Hydraulic Research/The International

56



Association of Water Pollution Research and Control, Suita, Osaka, Japan, July, pp.1137-1142.

NOUH, M. (1991) “Urban Drainage In Arid Climates,” Invited Paper, Proceedings(Supplements Volume), The International Conference on Urban Drainage and NewTechnologies. UNESCO/ IAHR, Dubrovnik, Former Yugoslavia, June, pp. 8893.

NOUH, M. (1998a), “Effect Of Duststorms On Sediment Concentrations In Sewer Flows In An

Arid Catchment,” 7th International Conference on Urban Storm Drainage,IAHR/IAWPRC, Hannover Germany 9-l 3 September, pp. 212-224.

NOUH, M. (19986) “Influence Of Duststorms On Stormwater Runoff Quality,” 3rdInternational Conference on Environmemal Pollution, European Center for PollutionResearch / UNESCO, Budapest, 1519 April, Vol. 1, pp. -357.

NOUH, M. (1997) “Regression Analyses Of Heavy Metal Concentrations In Urban FlashFloods,” 3rd International Conference on Flow Regimes from InternationalExperimental and Network Data, IAHSIUNESCOMIHO, Slovenia l-4 October.

SCHILING, W. (1983)” Effect Of Spatial Rainfall Distribution On Sewer Flows,” Proceedings,Seminar on Rainfall as the Basis for Urban Runoff Design and Analysis,Copenhagen, August, pp. 177-188.

SECKLER, D. (1999) Water For Foods In 2925: The Major Issues,” Newsflow, No. 2/99,IWMI.

SHAHEEN, S., NOUH, M., AL-NASRIRY, A. (1998),“Prediction Of Stormwater Runoff In AridCatchments,” Journal of Water Resources Engineering, Vd. 13, No. 4, pp. 18-32.

VIESSMAN, W., LEWIS, G., AND KNAPP, J. (1989) ‘Introduction To Hydrology,” Harper &Row, Publishers, New York.

USDA, SOIL CONSERVATION SERVlCE (1975) “Urban Hydrology For Small Watershed,”Technical Release 55, Washington D. C.

WISCHMEIER, W. H., AND SMITH, D. D. (1978) “Predicting Rainfall Erosion Losses: AGuide To Conservation Planning,” USDA Agricultural Handbook No. 537.

ZUHAIR, A., NOUH, M., EL-SAYED, M. (1999) “Flood Harvesting In Selected Arab States, ”Final Report No. M31/99, Institute of Water Resources, 217~.

57



Chapter 4

Storm hydrology of urban drainage

This chapter deals with the various factors affecting runoff prediction and design of urban facilities.Common methods of estimating design rainfall and runoff are reviewed and discussed.

4.1 Effects of urbanisation on runoff

Urbanisation causes changes in catchment hydrdogy due to the increase in the impervious area andthe reduction in catchment storages as waterways become channeled and piped. The effects ofurbanisation on runoff can be summarised in terms of the following changes to the characteristics of

runoff hydrographs from a developing catchment (see Figure 4.1)l increased peak discharges and runoff volumes

l decreased time of concentration

l increased frequency and severity of flooding

l urban waterways altered from an ephemeral system to a perennial system.Stormwater runoff volumes in urban areas are considerably higher than runoff volumes in non-

urban areas because impervious surfaces such as paved roads, parking lots and roofs prevent rainfallfrom infiltrating into the ground. The runoff coefficient of directly connected impervious surfaces isabout 0.9 (some of the rainfall is loss to surface pores and depressions) while the runoff coefficient innon-urban areas varies from less than 0.1 in arid areas to greater than 0 .3 in temperate and wetareas. The effect of urbanisation on runoff volumes is therefore greater in arid areas than in non-aridareas because the increase in runoff coefficient as a result of urbanisation is greater in and areas.

The example in Table 4.1 shows that urbanisation can increase the annual runoff in arid and semi-andareas by more than four times.

higher peak discharge:\

Figure 4.1 Effect of urbanisation on the characteristics of runoff hydrograph(adapted from Schueler, 1987)

it

bigger runoff I’ \

volumes / : less gradual Pat urban-development

Time

Peak flow rates in urban areas are also much higher than in non-urban areas because of higherrunoff vdumes and shorter times of concentration. The higher flow velocities result from the increasein the amount of impervious surfaces which are smoother than surfaces in rural catchments, and from

58



natural streams being straightened and lined with concrete to move stormwater away from built-upareas. As urbanisation results in a decrease in the time of concentration, the catchment response torainfall becomes more sensitive to the high rainfall intensities and short duration events. The effect ofcatchment urbanisation on peak discharges can thus be expected to be more pronounced for regionswith IFD relationships that are highly skewed (see Section 4.2).

Table 4.1 Example calculation of post-urban and pm-urbanmean annual runoff volumes in a semi-arid area

Mean annual rainfall = 300 mm

Impervious area runoff coeffident = 0.9Pervious area runoff coefficient = 0.1

Urban catchment

Fraction impervious = 0.4

Impervious area runoff= 0.4 x 0.9 x 300 = 108 mm

Pervious area runoff=0.6x0.1x300=18mm

Total runoff = 126 mm

Rural catchment

Fraction impervious = 0.0

Impervious area runoff=Omm

Pervious area runoff= 1.0x0.1x300=30mm

Total runoff = 30 mm

The peak flow rates for the more frequent storms (more than once a year) in urban areas can bemore than 100 times higher than in non-urban areas (see Figure 4.2). In the extreme events (1 in 50

or 1 in 100 years), the peak flow rates in urban and non-urban areas are more similar because inextreme events the saturated pervious surfaces behave more like the impervious surfaces. It istherefore wmmon for flash floods to occur in urban catchments, particularly in areas where thedrainage systems are not designed adequately.

The changes in catchment hydrology as a result of urbanisation have a direct impact on the aquaticecosystem, most notably the loss of aquatic habitats and biodiversity due to poor stormwater qualityand increased frequency and severity of habitat disturbances. Vegetation removal in urban areas canalso accelerate streambank erosion.

4.2 Estimation of design rainfall

4.2.1 Design event

The concept of a design event is used in designing urban stormwater projects, like the sizing of pipesin drainage systems. For economic reasons, some risk of failure is allowed for in the design. Theconsiderations that are taken into account in selecting an appropriate design standard for hydraulicstructures may include the operation of the structure during above-design events, the consequencesof above-design events (in terms of public safety, disruption to services, flood damages andenvironmental impacts) and the frequency of above-design occurrences.

Implicit in the adoption of a probabilistic approach to selecting a design standard is theconcept of designing a structure for a given risk of “failure”, or perhaps more appropriately, risk ofabove-design operating conditions. The terms “recurrence interval” and “return period” are commonly

59



used in the water engineering profession but often in a manner that can be misleading. A commonmisconception is the implication that these probabilistic events are exceeded at regular intervals asdefined by the “return period” or “recurrence interval” This can be a problem when disseminating

information to the public and decision makers.

Denver, USAI rainfall of 380 mm)

1 10

Average recurrence interval (years)

100

Figure 4.2 Ratio of flood peaks in two urban catchments relative to pm-urban flood peaks

Two more acceptable probability terms are ‘average recurrence interval” (ARI) and “annualexceedance probability” (AEP). The term “average” is added to the former to re flect that therecurrence interval of a particular sized rainfall or runoff event is, in the long term, equal to the averageinterval specified. The term AEP is perhaps the more technically correct term in that it represents the

probability of a rainfall or runoff event exceeding the design value at least once in a year.Estimates of design runoff are required to design hydraulic structures, but because runoff dataare not available in most areas, a model is usually used to convert a design rainfall to design runoff. Itis also difficult to obtain long runoff records that retlect present catchment conditions because o f thecontinuous development in urban areas.

The design rainfall is characterised by the AEP, average rainfall intensity and the rainduration. The time/temporal distribution of rainfall over the event (hyetograph) is also needed togenerate storm hydrographs. The rain intensity can also vary spatially, but this variation is usuallyignored in urban catchments because they are generally small catchments.

It is worth noting that rainfall depths are smaller in arid and semi-and areas than elsewhere.Not only does rain seldom fall, not much falls when it does. Although intense storms that cause flashflooding are wmmon in urban catchments, the volumes are generally smaller in arid and semi-andareas. Therefore, channel capacities and detention basins required for urban catchments in arid and

semi-arid areas are considerably smaller than those required elsewhere and stormwater quality in aridand semi-arid areas can often be treated with less than half the storage needed for other areas.

4.2.2 Rainfall intensity-frequency-duration relationships

The rainfall intensity corresponding to the selected storm exceedance probability at a site is neededfor design. This information is usually provided in the form of a rainfall intensity-frequencyduration(IFD) chart. The IFD chart for Alice Springs, Australia in Figure 4.3 illustrates the characteristicrelationship of decreasing probabilistic average rainfall intensity for increasing storm duration for any

60



given storm AEP. These charts are readily available for most areas, through the analyses of rainstormdata from many locations by regional or national meteorological, engineering or hydrological agencies.

0.1 ' I I I

0.1 1 IO 100

Storm duration (hrs)Figure 4.3 Rainfall intensity-frequencyduration chart forAlice Springs, Australia (mean annual rainfall of 280 mm)

Accurate IFD relationships can be obtained by analysing local rainfall data. Where local dataare available, and/or where IFD charts are not available, a rainfall frequency analysis can be used todetermine the IFD relationships. In the analysis, the annual maximum rainfall depths over durations ofinterest (typically from fwe minutes to 72 hours) are first extracted from the rainfall records. Theannual rainfall data series obtained for each duration is then fitted to a statistical distribution todetermine the AEP of the various rainfall depths. Common statistical distributions used for rainfallfrequency analysis include the log-normal, log-Pearson III and Generalised Extreme Values.However, the use of mix of two probability distribution functions may better suit arid areas of significant

seasonal rainfall (see chapter 3).

4.2.3 Temporal distribution of rainfall intensity

The variation of rainfall intensity over a storm event (hyetograph) is required to estimate a runoffhydrograph. Estimates of runoff hydrographs are needed for some applications, for example thedesign of stormwater detention storage facilities.

The temporal pattern of storm events can vary considerably, often without any consistenttrend because of the inherent stochastic nature of the meteorological factors influencing stormmechanisms. Nevertheless, there is a need to derive a probabilistic storm temporal pattern toestimate a design hydrograph. Consistent with the probabilistic approach to estimating stormwaterrunoff, the probabilistic storm temporal pattern represents the average temporal pattern to be used inconjunction with the design average rainfall intensity to determine the design runoff hydrograph.

The average temporal pattern of rainfall can best be determined by analysing lots of localrainstorm data. Nevertheless, like the IFD charts, recommended probabilistic temporal patterns ofdesign storms are sometimes available, through the analyses of many rainburst data for variouslocations by national meteorological agencies.

As indicated in chapter 3, the temporal as well as the spatial rainfall variations in andcatchments are high, and have significant impact on the produced runoff. These variations also affectto great extent the accuracy of design models (see chapter 9).

61



4.3 Estimation of design runov

There are a number of methods for estimating design runoff from urban catchments, all of whichinvolve the conversion of rainfall over the catchment to runoff. Table 4.2 provides a generaldescription of the most popular methods in increasing order of complexity. The methods can becategorised into three groups.

. Methods that estimate only the peak discharge from the design rainfall intensity, like the rationalmethod. These methods are usually used to analyse individual pipe capacities.

. Methods that convert rainfall excess hyetograph to a full catchment runoff hydrograph. Examplesinclude unit hydrograph, time-area, storage routing and kinematic wave methods.

l Conceptual models that simulate the catchment physical processes in detail. Examples includeSWMM and HSPF. These models can be used as an event model to estimate design runoffhydrographs from single storm events, or as a continuous model to produce continuoushydrographs spanning years of flows. These models are usually used to simulate flow behaviourthrough large drainage networks.

Table 4.2 Selected hydrological models for urban drainage design and analysis(adapted from O’Loughlin and Robinson, 1998).

ModelEmpiricalequationsRationalmethod

UnithydrographsTime area

StorageroutingKinematicwavePhysicalmodels

computers optrona

1 ncluding those involving storages 1 equired)Hydrograph 1Design or analysis of all sizes of systems, 1Medium to complex

including those involving storages(computers required)

Hydrograph Design or analysis o f all sizes of systems, Medium to complexincluding those involving storages (computers required)

Continuous Detailed analysis of large drainage Complexhydrograph netirks and scientific research. (computer required)

Requires lots of data.

4.3.1 Rational method for estimating design peak discharge

The rational method is the most commonly used method for estimating peak stormwater discharge.The underlying principle of the rational method is that under a uniform rainfall intensity, the maximumdischarge from a drainage area will occur when the entire area is contributing to runoff. The entire

area starts contributing to runoff when rainwater from the hydrologically most remote point reaches thedesign location. This time is called the time of concentration.

The rational method assumes that the peak discharge is proportional to the rainfall intensity

QP = CIA (4.1)

l This chapterpresents methods for estimating “ditlkzd” runoff 6om urban catchments. Sewerage overflowsand discharges in combined stormwater-sewage systems are not considered.

62



where Q ,, is the peak or design discharge, C is a dimensionless runoff coefficient, I is the rainfallintensity, and A is the drainage area.

The runoff coefficient depends on the land use and the size of the storm. In urban areas, themost important land use parameter is the amount of impervious area. The runoff coefficient forimpervious surfaces is typically about 0.9, and the runoff coefficient for pervious surfaces vary fromless than 0.1 in and areas to more than 0.3 in temperate and wet areas. For urban areas, the runoff

coefficient in equation (4.1) can therefore be viewed as the areal weighted average of the runoffcoefficients in impervious and pervious surfaces. The runoff coefficient also varies with the size of thestorm. For frequent events with low rainfall, there is little runoff contribution from the pervioussurfaces, and the runoff coefficient is essentially the fraction imperviousness of the catchment. Therunoff coefficient becomes higher for higher intensity storms (low AEPs) due to the reduced effect ofinftltration in the pervious surfaces.

The time of concentration is an important parameter in the rational method, because indetermining the design peak discharge, the rainfall intensity for the duration equal to the time ofconcentration is used. In urban catchments, the time of concentration is estimated as the sum of flowtravel times for the longest flow path. The main flow paths considered include roof-to-gutter conduits,overland flow, gutter flow along roads and pipe and/or open channel flow. The travel time for eachflow path can be calculated using standard hydraulic methods. As the flow velocity is dependent onthe geometric and hydraulic properties of the various flow paths, it may be necessary to adopt an

iterative process to estimate the time of concentration. For example, the design rainfall intensitydepends on the time of concentration, which in turn depends on the adopted rainfall intensity becausethe travel time decreases with increasing rainfall/flow.

In urban catchments, sometimes a higher peak flow can occur at a duration less than the timeof concentration. This is called a partial area effect and it occurs because impervious areas contributea large proportion of the flow due to their high runoff coefficient and rapid response. Here, the rationalmethod should be used to estimate the peak discharge from the partial area (lower subarea) as wellas the combined catchment, and the larger of the two estimates used as the design peak discharge.

In the rational method, the assumption of uniform rainfall intensity will lead to anunderestimation of the peak discharge. On the other hand, neglecting the storage effects in thecatchment leads to an overestimation of the peak discharge. The rational method is therefore moreaccurate for small catchments with little channel storage and for short duration storms (short time ofconcentration) because short duration storms tend to be more uniform in time than long duration

storms. In and catchments, where the spatial and temporal variations of rainfall are high, the accuracyof the rational method is unacceptable in catchment size larger than 2 km* (Shaheen et al., 1998).

4.3.2 Methods for estimating complete runoff hydrograph

4.3.2.1 Determination of rainfall excess hyetograph

Rainfall excess is defined as the component of rainfall that becomes surface runoff. The computationof rainfall excess involves the estimation of losses from rainfall. The losses may include interception,surface depression storage and infiltration. Losses due to evapotranspiration are negligible becauseof the relatively short duration of rainstorms.

Methods used to calculate rainfall losses include (see Figure 4.4)

1st. loss calculated as a constant fraction of rainfall (runoff coefficient method)2nd. constant loss rate

3rd. initial loss

4th. initial loss followed by a continuing loss occurring at a constant rate

5th. initial loss fdlowed by a continuing loss occurring at a constant fraction of rainfall

6th. infiltration curve or equation representing infiltration capacity, with losses decreasing with time.The choice of method depends on the study problem, data availability and likely runoff processes.

The sixth method is generally used only for large design projects and where there is sufficient data. Itis usually used with detailed event models (see Section 4.3.3).

63



(i) constant fi-action (ii) consta.nt loss

(iii) initial (iv) initial loss followed

by constant continuing loss

(v) initial loss followed by

constant fraction continuing loss(vi) infiltration curve

Figure 4.4 Loss models used to estimate rainfall excess

In urban catchments, it is useful to separate the impervious and pervious parts. Forimpervious surfaces, there is usually a small amount of initial loss (about 1 mm) due to water filling thesurface depressions and pores. The losses can therefore be estimated using method (iii) or method(iv) or (v) with very little continuing loss. Losses from the pervious surfaces are usually determinedusing methods (iv) or (v), with the initial and continuing loss parameters estimated by analysing rainfalland runoff data in the catchment. Since there is no reason for expecting loss rate values to conform toa particular distribution, the median o f the derived values from several storms is probably the most

appropriate for design. Where data are not available on the catchment of interest, loss rates can beestimated using rainfall and runoff data from nearby similar catchments. Like the IFD charts and thetemporal distribution of rainfall, recommended design loss rates are sometimes available, and thesecan be used where local data are not available to compute the losses.

4.3.2.2 Time-area method for estimating runoff hydrograph

Many of the hydrograph methods listed in Table 4.2 are based on variance and modification of theprinciples of the time-area method of computing stormwater runotf. Like the rational method, the timearea method neglects the effect of catchment storage. However, it accounts for the temporaldistribution of rainfall. For steady uniform rainfall conditions, both the rational method and time-area

64



method will give identical peak discharges. However, by accounting fo r the temporal d istribution ofrainfall, the peak discharge determined using the time-area method will be higher than that determinedusing the rational method.

The time-area method is based on combining the rainfall excess hyetograph with the time-areadiagram of the catchment. The time-area diagram represents the relationship between subareas ofthe catchment and its corresponding stormwater travel time from the centroid of the subarea to the

catchment outlet. The method is illustrated in Figure 4.5 and described below.

1. Draw isochrones (lines of equal travel time of runoff to the catchment outlet) for the catchment.The time interval of the isochrones should be consistent with the rainfall time interval.

2. Derive the cumulative catchment area to travel time relationship.

3. Derive the area versus travel time relationship (time-area d iagram). It appears in this illustrationthat the time-area diagram can be derived directly from step (i), but step (ii) is usually necessaryfor urban catchments because of the complicated drainage network.

4. Combine the rainfall excess hyetograph and the time-area diagram, with the start time of therainfall excess hyetograph aligned to the vertical axis of the time-area diagram. At each timestep, runoff is calculated as the sum of the products of corresponding ordinates of the rainfallexcess hyetograph and the time-area diagram.

Derivation of the time-area diagram

2 3 4 5 6

Time interval

Combination of rainfall excess hyetograph and

time-area diagram to calculate discharge hydrograph

123456Time interval

Time I ~!&!w--- ~~_12

)&AIlhAz+ 12A1

3 1 ,A3+ 12A2+4 1 ,Aa+ 12A3+5 i I,As+ IzA4+6 I IA& IA-s+7 I k!As+8 I9 IIO j11 1

14 IhAz+ WII3A3+ u2+ Is.41

I3A4+ w3+ IsA + ~AI

4As+ 5&+ kA3+ kA2

S&+ &As+ b&+ k44&v+ kAs+ k&

b&b+ bAskA6

i012345 6 7 8 9 10 11

Time interval

Figure 4.5 Time-area method for estimating runoff hydrograph

65



4.3.2.3 Routing of stormwater runoff

In large urban catchments, the conveyance of stormwater runoff through the drainage network wouldalter the shape of the runoff hydrograph due to the effect of storage in the system. Routing modelsattempt to simulate the translation and attenuation of the hydrograph as it moves through the drainagesystem. Various computer softwares for routing stormwater runoff are available (see O’Loughlin andRobinson, 1998).

In applying routing models, the catchment is divided into subcatchments. Runoff hydrographfor each subcatchment is computed and routed through the drainage network which “links” thehydrographs from the various subcatchments. A storage or hydrologic routing method is usually used.The storage routing method computes the outflow hydrograph at the downstream end of eachindividual drainage “IinK (pipe or open channel) by solving the continuity equation and the storagefunction.

The continuity equation can be written as

I - Q = dS/dt (4.2)

and expressed in finite difference form as

0.5 (II+ It+r) - 0.5 (Qi + Qt+,) = (St+, - St)/ q t (4.3)

where I is the inflow rate to the drainage link, Q is the outflow discharge from the drainage link, S isthe volume of temporary storage in the drainage link, and q lt is the time interval in the finite differenceapproximation of the continuity equation.

Equation (4.3) can be rearranged such that all the known variables are placed on the left hand side

0.5 (It+ It+,) - 0.5 Qt + SJot = 0.5 Qt+, + St+,/Ot (4.4)

To solve the routing problem, the continuity equation is combined with the storage equation whichexpresses the relationship between the storage in the drainage link and the discharge

s = fn (Q) (4.5)

The storage function represents the combined effect of the various hydraulic and geometricfactors affecting the discharge characteristics of the drainage system. Common relationships usedinclude power relationships like S = aQb or relationships based on uniform flow equations likeManning’s equation.

For a given relationship, a table relating the outtlow discharge Q and (0.5 Q + S/at) and (0.5Q - S/Dt) can be established. The term (0.5 Q, - S&It) on the left hand side of equation (4.4) can bedetermined from the tabulated relationship as Qt is known. This and the inflow terms give the righthand side of equation (4.4), from which Q +l can be determined from the tabulated relationshipbetween Q and (0.5 Q + Wit).

4.3.3 Event and short time-step stormwater runoff models

Two of the more commonly used water quantity and water quality models that simulate the physicalcatchment processes in detail are SWMM (Huber and Dickinson, 1988) and HSPF (Johansen et al.,1984). SWMM was initially designed as a single event model but can now also be used to simulatecontinuous runoff hydrog-raphs. HSPF is usually used for continuous simulation, typically with hourlytime steps.

SWMM is by far the most popular urban stormwater model. Its reputation for being a difficultmodel to use is probably compensated by the extensive body of literature describing its applications tovarious problems. Nevertheless, modelling with SWMM requires a significant time investment, and

66



models like SWMM are generally useful only for reasonably large design pro jects and where there aresufficient data to estimate/calibrate the many model parameters.

SWMM represents the modelling area as several subcatchments with the runoff estimated forthe subcatchments routed through the drainage network (gutters, pipes, sewers and channels). Therouting routines in SWMM allow for the simulation of complex drainage networks, indudirg overflowsand pressurised flows.

The catchment surface is modelled as a nonlinear reservoir with a finite capacity (calledmaximum depression storage) (see Figure 4.6). Surface runoff occurs only when the depth of water inthe reservoir exceeds the maximum storage capacity. Each subcatchment is divided into three parts,pervious surface, impervious surface with a maximum storage capacity and impervious surface withzero storage capacity. The input into the reservoir is precipitation excess and the output is surfacerunoff.

Continuity equation Manning’s equation

Add= AI-Q

dt n r

A is subcatchment W is width of overland flow

area n is Manning’s roughnesd is depth of coefficient

dP is maximum storage capacityS is subcatchment slope

Figure 4.6 Nonlinear reservoir representation of subcatchment in SWMM

The precipitation excess is calculated as rainfall less infiltration and evapotranspiration. Inevent modelling, the evapotranspiration is negligible and can be ignored. infiltration occurs only fromthe pervious surface and is modelled using either the Horton or Green-Ampt equation. The equationsare used to calculate the infiltration capacity (reducing the infiltration capacity with time), with all therainfall exceeding the infiltration capacity at that time step becoming rainfall excess. SWMM alsoallows for the routing of infiltrated water through upper and lower subsurface zones that can contributeto delayed subsurface runoff. This may be important only in continuous runoff simulation over longperiOdS.

The catchment (nonlinear reservoir) process is modelled using a continuity equation andManning’s equation, with the combined equation solved numerically to estimate surface runoff (see

Figure 4.6)..

4.4 Estimation of daily and longer time-step runoff

At the outset in any modelling exercise it is important that the objectives of the exercise be clearlyenunciated. This is important so that a model of appropriate detail can be prescribed and datarequirements specified. If an annual runoff estimate can provide the decision-maker with an outputwith an error level appropriate to that decision, using a daily model would be inappropriate. There willbe occasions of course where a more sophisticated model (and data) than necessary are available.

67





can be large scatter in the data. As runoff from small events is generated only from the effectiveimpervious surfaces, the slope in the rainfall-runoff plot gives an estimate of the fraction of catchmentwith effective impervious area, and the intercept of the rainfall axis is an estimate of the initial loss.

6- .

g” l/i . .

40

!z

30

2 20

;-I

. .

g2- l m

.

10 i.+

cz . .

0 n 0

Data from 180 ha urban catchment Data fhm 200 ha urban catchmentin Dhahran, Saudi Arabia in Melbourne, Australia

(mean annual rainfall of 80 mm) (mean amma rainfkll of 650 mm)

Figure 4.8 Event rainfall-runoff plot to estimate fraction o f catchment with effective impervious area

0 20 40 0 20 40 60

Storm rainfall (mm)

Plots show all storm events. Points above the line indicate events where

surface runoff is also generated kom the pervious area.

Slope is an estimate of the diction of catchment with effective impervious area.

Intercept o f the rainfkll axis is an estimate of impervious area initial loss

The fraction of catchment with efiedive impervious area can also be estimated from aerialphotographs and knowledge about the drainage system. The aerial photograph can be used toestimate the fraction o f catchment with directly conneded impervious area (impervious surfaces thatare directly connected to the drainage system). This is often similar to the fraction of catchment witha// types of impervious surfaces, except in catchments where roofs are not directly connected to the

drainage system. The fraction of catchment with effective impervious area is typically about 80 to 90%of the fraction of catchment with direct/y conneded impervious area (see Boyd et at., 1993) becausenot all impervious surfaces that appear to be directly connected are directty connected (for example,blockages in gutters can result in water flowing onto pervious surfaces).

The impervious area runoff coefficient, rd, in equation (4.6) is typically about 0.8 to 0.95,depending on the initial loss and the number of raindays. The pervious area runoff coefficient, r,, isdependent on the climate and physical catchment characteristics. The pervious area runoff coefftcientis 1-r in drier areas, and in arid and semi-arid areas, it is usually less than 0.1.

4.4.2 Daily and monthly runoff

Daily estimates o f runoff are often required to investigate shorter term impacts, to determine seasonalcharacteristics of urban runoff, to study water quality management options, and as inputs to water

quality models. Runoff from effective impervious surfaces can be easily modelled because all therainfall becomes runoff after an initial I- is satisfied.Runoff from the pervious area (‘pervious’ refers to all surfaces that are not effective

impervious area) can be simulated using conceptual models that mimic the catchment processes.These models consist of one or more storages and equations that describe the movement of waterbetween the storages. There are hundreds o f models available in the literature for simulating runofffrom rainfall and potential evapotranspiration data for rural catchments, and most of them can also beadapted to estimate pervious area runoff in urban catchments. There is usually little differencebeMzn these conceptual models when used to estimate runoff in urban areas, partly because mostof them can estimate runoff satisfactorily once they are calibrated, but mainly because pervious areascontribute only a small proportion of the total runoff from urban catchments.

69



4.4.2.1 Simple conceptual model to characterise daily runoff

In the absence of data, the simple,wnceptual model in Figure 4.9 can be used to characteriti dailyrunoff. The catchment is divided into an effective impervious area and a pervious area. All the dailyrainfall in the effecGve impervious area becomes runoff once the daily initial loss is satisfied. In the

pervious area, surface runoff is generated when saturation occurs, and baseflow is simulated as alinear recession of the soil store. Evapotranspiration is dependent on the amount o f water in the soilstore, but cannot exceed the atmospherically controlled rate of areal potential evapotranspiration (seeFigure 4.9).

I

outdoor

use T= (m

Rain

gq x 10, PET)

surface runoff _

Figure 4.9 Structure of a simple conceptual daily rainfall-runoff model for urban catchments(parameters are highlighted in bold italics)

With no model calibration, the reliability of the runoff volumes estimated using this simplemodel would be no better than the annual volumes estimated using equation (4.6). However, unlike

equation (4.6) the model can provide an indication of the daily flow characteristics, as illustrated inFigure 4.11. The impervious area rainfall threshold is typically about 1 or 1.5 mm (‘thres’ in Figure4.9) while the pervious area storage capacity and baseflow factor generally range between 50 and200 mm and 0.01 to 0.1 respectively (the parameters ‘Scap’ and ‘k’ in Figure 4.9).

4.4.2.2 Conceptual rainfall-runoff model for estimating daily runoff

Where there is runoff data for model calibration, better estimates of daily runoff can be obtained bysimulating the pervious area catchment processes in more detail. An example of such a model is thedaily urban runoff model developed by the Cooperative Research Centre for Catchment Hydrology inAustralia (CRCCH). The model retains the simplicity of the earlier model, but provides a betterconceptual representation of the processes. The model structure is shown in Figure 4.10 with the nineparameters (one for the impervious area and eight for the pervious area) highlighted in bold italics

(see also Chiew and McMahon, 1999).Like the earlier model, all the rainfall in the effective impervious area becomes runoff once the

daily initial loss is satisfied. The remaining area is modelled as two separate parts with differentstorage capacities (related to effective soil depth). The first has a smaller storage capacity andrepresents parts of the catchment that saturates easily. The second represents the remainder of thecatchment with a greater soil storage capacity.

Rainfall on the pervious area is first subjected to an infiltration function that determines theinfiltration capacity. Rainfall that exceeds the infiltration capacity becomes infiltration excess runoff,while the remaining moisture fills the soil moisture stores. Surface runoff occurs when the storagecapacities are exceeded (when saturation occurs). Water from the soil stores recharges agroundwater store when the storage exceeds a certain amount (‘field capacity’). The recharge is

70



calculated as a parameter (which mimics the hydraulic conductivity) times the amount the storageexceeds the ‘field capacity’. The infiltration excess runoff, infiltration, saturation excess runoff andgroundwater recharge from the two pervious parts of the catchment are estimated separately.

Rainfall +

Outdoor water use

fhres

saturation excess runoff

imuervmus

Need to specify the fraction imperviousness, and thetwo fi-actions of the remaining area (AZ and AZ) andtheir storage capacities (Slcap and S2cap).

Infiltration capacity = coeflx exp (-sq x storage / storage capacity).Rainfall exceeding the capacity becomes infiltration excess runoff.

Evapotranspiration = min (S&cap x 10, PET),where S and Scap are the area1 weighted values of the two storages.

Evapotranspiration is satisfied first fkom the larger store.

Figure 4.10 Structure of the CRCCH daily urban runoff model(parameters are highlighted in bold italics)

Baseflow from the groundwater store is simulated as a linear recession o f the groundwaterstore. Evapotranspiration is calculated as a linear function of the weighted soil wetness (see Figure4.10) but cannot exceed the potential rate. The evapotranspiration demand is satisfied first from thelarger store, therefore allowing for some redistribution of water between the two stores.

The model thus distinguishes the runoff contributions from surface (impervious area runoff,infiltration excess runoff and saturation excess runoff) and subsurface (baseflow) flows. This isimportant for water quality modelling because the water quality characteristics in the surface andsubsurface flows can be very diierent.

The plot in Figure 4.11 shows an example comparison of the daily runoff estimated by the tworainfall-runoff models described here for an urban catchment with the recorded daily flowcharacteristics. As expected, the more complex model (Figure 4.10) which was calibrated performedbetter. Nevertheless, the simple model (Figure 4.9) without any calibration, can also simulatesatisfactorily the relative daily flow characteristics, although in some cases, the actual runoff estimatescan be poor. Thus, where adequate rainfall-runoff data are available, a more complex model could be

71



used to estimate daily runoff via model calibration, but where there is little data, the use of a simpleuncalibrated model may be able to provide an indication of the daily flow characteristics.

- Recorded runoff

- Runoff estimated using simple made1in Figure 4.9 (no calibration)

---- Runoff estimated using model in

Figure 4.10 (model was calibrated)

Percentage of time daily runoff is exceeded I

Figure 4.11 Flow duration plots comparing daily runoff simulated by two conceptual models with the recordedrunoff (using data for a 450 ha urban catchment in Canberra, Australia, mean annual rainfall of 550 mm)

4.4.2.3 Data and model calibration and verification

The input data required to run the conceptual daily rainfall-runoff models are rainfall and arealpotential evapotranspiration. The availability of reliable continuous daily rainfall data is importantbecause the model simulations are most sensitive to the input rainfall data. The catchment-averagerainfall data can be obtained from one or several rain gauges within or close to the catchment.

The modds presented above are lumped conceptual models, vvith single ‘lumped’ parametersused to represent the entire catchment. There are also no routing algorithms to route the flows to thecatchment outlet. Nevertheless, the use of these models is usually sufficient because they are usedon small urban catchments (generally less than 10 km*, and seldom more than 100 km*), and becausealthough they run on a daily time step, they are often used to provide estimates of runoffcharacteristics over longer time periods (weekly, monthly or seasonal). In any case, where accurateestimates of runoff over short time-steps are needed and where data are available, the more complexevent and short time-step models are used (see Section 4.3.3).

Generally, evapotranspiration has little influence on the water balance on a daily time scale.In such cases, the inter-annual variability of potential evapotranspiration is small compared to thevariability in rainfall. Therefore, the use of mean monthly potential evapotranspiration is sufficient forthe modelling exercise. The mean monthly areal potential evapotranspiration can be estimated as afactor of pan evaporation data or estimated from climate data using Penman-Monteith equations orMorton’s wet environment evapotranspiration algorithms (see Chiew and McMahon, 1991). However,in case of arid and semiarid climates, evaporation is significant and has substantial effects on theproduced runoff (Nouh, 1991; Shaheen et al., 1998) and thus for such climate further research toachieve acceptable accuracy of the prediction models is needed. Results of case studies andaccuracy of the most popular SWMM model in arid climates are presented and discussed in chapter 9.

72



The availability of runoff data for model calibration is probably the most importantconsideration for using the models because the models can only be as good as the data used tocalibrate them. In the model calibration, it is also important to test the ability of the optimizedparameters in estimating runoff for an independent period where the data are not used for the

calibration; A split sampling method is commonly used where the available data is divided into twoperiods. The model is first calibrated on the first half of the data, and tested against the remainingdata. Where there is little data for model calibration and verification, a cross-verification method can

be used to increase the amount of independent data to assess the model. Here, the data set isdivided into several periods. Data from each period is left out in turn and the model is calibratedagainst data from the remaining periods. The optimised parameter values are then used to estimaterunoff for the period that was left out and compared against the recorded runoff.

Conclusions

Urbanisation has significant impacts on stormwater runoff. As urbanization increases, peakdischarges and runoff volume increase but time of concentration decreases. Due to the small

amounts of runoff volumes in arid areas, required detention basins are normally smaller than thoseneeded for other areas. However, the basins should be designed based on careful consideration ofrare flash floods. Common statistical distributions used for rainfall frequency analysis include thelognormal, log-Pearson III and Generalized Extreme Values. However, the use of mix of twoprobability distribution functions may better suit arid areas of significant seasonal rainfall. Temporalrainfall variations have significant impact on the produced runoff in arid areas, and thus there is a needto derive a realistic probabilistic storm temporal pattern to estimate a design hydrograph in theseareas. Available common methods used by professional engineers for design of urban facilities maynot be unsuitable for application in arid catchments due to the distinguished characteristics of soilsand rainfall in these catchments (see chapter 3) and also due to the normal scarcity of data neededfor the calibration and verification of the methods.

Bibliography

BOYD, M.J., BUFILL, M.C. AND KNEE, R.M. (1993) “Pervious And Impervious Runoff In Urban

Catchments”, Hydrological Sciences, Vol. 38, pp. 463-478.CHIEW, F.H.S. AND MCMAHON, T.A. (1991) “The Applicability Of Morton’s And Penman’s

Evapotranspiration Estimates In Rainfall-Runoff Modelling,” Water Resources Bulletin, Vol. 27,pp. 61 I-620.

CHIEW, F.H.S. AND MCMAHON, T.A. (1999) “ Modelling Runoff And Diffuse Pollution Loads InUrban Areas,” Water Science & Technology, Vol. 39, pp. 241-248.

GRAYSON, R.B. AND CHIEW, F.H.S. (1994) “An Approach To Model Selection,” Proceedings of theJoint 25th Congress of the International Association of Hydrogeolog ists and the 22ndInternational Hydrology and Wate r Resources Symposium of the Institution of Engineers,

Australia, Adelaide, November 1994, National Conference Publication (Institution of EngineersAustralia), 94/10(l): pp. 507-512.HUBER, W.C. AND DICKINSON, R.E. (1988)” Storm Water Management Model, Version 4: User’s

Manual,” University of Florida, Gainesville, U.S.A.JOHANSEN, R.C., IMHOFF, J.C., KITTLE, J.L. AND DONIGAN, A.S. (1984)” Hydrocomp Simulation

Program - Fortran (HSPF): User’s Manual Release 8.0,“. USEPA, Athens, Georgia, U.S.A.MORTON, F.I. (1983),“0pera tional Estimates Of Actual Evapotranspiration And Their Significance To

The Science And Practice Of Hydrology,” Journal of Hydrology, Vol. 66, pp. l-76.NOUH, M. (1991) “Urban Drainage In Arid Climates,” Invited Paper, Proceedings (Supplements

Volume), The International Conference on Urban Drainage and New Technologies, UNESCO/IAHR, Dubrovnik, Former Yugoslavia, pp. 86-93.

O!LOUGHLIN, G.G. AND ROBINSON, D.K. (1998)“Urban Stormwater Management,” Book Eight ofAustralian Rainfall and Runoff, Institution of Engineers, Australia.

73



SCHUELER, T.R. (1987), “Controlling Urban Runoff: A Practical Manual for Planning and DesigningUrban BMPs,” Washingto Metropolitan Water Resources Planning Report, 185 pp.

SHAHEEN, S., NOUH, M., AL-NASRIRY, A. (1998), “Prediction Of Stormwater Runoff In AridCatchments,” Journal of Water Resources Engineering, Vol. 13, No. 4, pp. 18-32.

74



Chapter 5

Urban stormwater pollution

5.1 Stormwater pollution

A pollutant can be defined as a material present in a concentration greater than that which naturally

occurs in the water, air or soil. Stormwater pollution loads in urban areas are much higher than inunimpaired areas because urban activities generate more pollutants, and higher runoff volumes inurban areas can transport more of the available pollutants to the drainage system.

Stonnwater pollution comes from point and non-point sources. Point sources are those wherethe polluted water is discharged at a single location, such as a factory or a sewage treatment plant.Non-point or diffuse sources are those where pollution is generated over a large area and flows intothe drainage system at more than one point. This chapter is concerned with only non-point sourcepollution. Non-point source pollution in urban runoff can come from atmospheric deposition,automobiles and roads, residential and industrial activities, construction activities and soil erosion (seeTable 5.1 and Makepeace et al., 1995).

The different types of pollutants in urban stonnwater are briefly described below.

sediments in urban waterways can smother aquatic habitat and reduce channel capacity.

Sediments also reduce the aesthetic appeal of waterways and increase the need for filtration inwater supplies. Suspended solids (SS) is used to describe sediments suspended in water.Turbidity, a water quality parameter that quantifies the cloudiness of water, is often related tosuspended solids. Nutrients and heavy metals can easily adsorb to sediment and utilise it as themedium for transportation in urban runoff. The deposition of sediments can result in the release ofthese nutrients and toxins at a later time when the ambient condition is favourable for their release.It is for this reason that many management methods target the removal of sediments with theexpectation that a significant amount of organic and inorganic pollutant till also be removed.

Nutrients (phosphorus (P) and nitrogen (N)) are essential to living organisms, but excessive levelscan upset the natural balance of the ecosystem. Excess nutrients promote the growth of onespecies of aquatic plant to the exclusion of others (such as blue-green algae in the process ofeutrophication). The thick mats of algae formed at the water surface reduce light penetration andoxygen exchange between the water and the atmosphere, chpking the waterways. The buildup of

toxins in excessive algal growth can cause the closure of fisheries, water farming industries andpublic beaches.

Oxygen demanding materials include biodegradable organic debris, such as decomposing foodand garden wastes, that contribute to oxygen depletion in stormwater. Chemical oxygen demand(COD) and biological oxygen demand (DOD) are measures of the oxygen used when thesematerials react with chemical and biological substances in the water. Low oxygen levels in thewater can stress the aquatic community and facilitate chemical reactions that lead to sedimentdesorption of nutrients and metals.

Heavy metals include lead (Pb), zinc (Zn), copper (Cu), chromium (Cr), cadmium (Cd), nickel (Ni)and other inorganic substances. Heavy metals are toxic to animals, birds and humans. The toxic

75



effects can either be chronic (gradual buildup causing long-term illness and eventual death) oracute (high concentration causing sudden illness or death).

Toxic organic wastes come from garden and household chemicals (herbicides and pesticides),industrial chemicals and landfill-waste leachates. Organic pollutants can accumulate in anecosystem, causing long-term toxicity.

Pathogenic micro-organisms include bacteria, viruses and protozoa found in soil, decayingvegetation and animal wastes. The Escherichia wli (E. wli) bacterium is widely used as anindicator of faecal pollution levels in stonnwater. Bacteria and pathogens excreted in human andanimal faeces may initiate outbreaks of infectious hepatitis and gastrointestinal diseases.

Hydrocahons ome from oil and grease used for combustion and lubrication, and from surfactantsin detergents. Oil spills in urban areas are wmmon and can cause short-term toxicity problemswhile surfactants can damage biological membranes of aquatic plants and animals.

Litter includes plastics, paper, bottles and other rubbish discarded by people. Liier is aestheticallyunpleasant, smells and attracts vermin. Some litter, such as broken glasses and syringes, canpose health risks. Liier and plant refuse such as dead leaves are commonly termed grosspollutants due to their large unit size.

5.2 Urban stormwater quality process

The process of stormwater contamination is often viewed as occurring in two phases - buildup andwashoff (see Chiew et al., 1997a). Buildup is the accumulation of pollutants on catchment surfacesover dry periods and washoff is the removal of surface pollutants during storms. However, it may bemore helpful to view the stormwater quality process as a process of dynamic equilibrium betweendeposition and removal at a point, and between contributing and non-contributing areas (see Duncan,1995). The pollutant sources include atmospheric deposition, vehides, road and pavement wear,residential and industrial activities, construction materials, vegetation and nearby dry soil (see Table5.1). Removal processes include wind, vehicle induced eddies, decomposition, street sweeping, andof course washoff by rain (James and Shivalingaiah, 1988).

Figure 5.1 shows the typical pattern of pollutant load on impervious catchment surfaces over

time. The level of surface pollutant load depends on the rate of deposition and the length of the dryperiod, and any removal by redistribution, decomposition, street sweeping or washoff. Because thesystem is in dynamic equilibrium, a larger departure from equilibrium will generate a larger restoringeffect. The cleaner a surface is made, the faster it gets dirty again, by redistribution of material fromsurrounding areas (Novotny et al., 1985). This explains why the surface pollutant load increasesrapidly after a storm, but the rate of increase in the surface load reduces with increasing length of dryperiod. The accumulated load is also typically large compared with the washoff load in any singleevent, and therefore the surface pollutant load tends to remain largely the same over time (see Chiewet al., 1997b).

Washoff is the removal of soluble and particulate pollutants by rainfall and runoff. Duringstorms, turbulence created by falling raindrops and flowing water loosens particles, which becomesuspended in water and are carried to the drainage system. Pollutants washed out from theatmosphere by rainfall (wet deposition) can add to the load carried in the flow.

In an urban drainage system, water flow is concentrated in gutters, pipes and channels andthe pollutants can be transported as bottom sediments, suspended particles and dissolved material.The rate of pollutant transport depends on the water velocity and depth, and the degree of turbulence.Dissolved pollutants can become adsorbed onto particle surfaces, and fine particles can flocculate toform larger particles. Most of the pollutants in sediments are associated with smaller particles due totheir greater surface area relative to the larger pat-tides. Pollutants attached to fine particles are easilycarried because a small flow velocity is sufficient to mobilise them and keep them in suspension.

Pollutant concentration often peaks before the peak in stormwater runoff, a process known as‘first flush’ (see Figure 5.2). The poll&graph varies between pollutants, with peak levels of dissolvedpollutants and micro-organisms occurring before peaks in particulate pollutants because less energy isrequired to detach them from catchment surfaces and keep them in suspension. The first flush effect

76



is more detectable in smaller catchments and on impervious areas. In larger catchments, the flowsfrom individual subcatchments arrive at different times, leading to a longer lasting but less detectablefirst flush.

Table 5.1 Common sources of pollutants in urban runoff

Pollutant Source Sediments Nutrients Oxygen Heavydemanding metals

I

Road andpavement wearVehicle wearVehicle fuels andfluidsFuel combustionSoil erosion

Human and animalwastePesticides and

Paint and solventsIndustrial activitiesHouseholdchemicals

I I I

Main sources of metals in urban stormwater

Petrol additivesLubrication oil I I I I

Pesticides and fertilisersDve and oaint

I II I

\ Paper I I I

In arid and semi-arid areas, the surface pollutant load is likely to be high. The equilibrium loadis high under dry conditions and reduced vegetation, and a long inter-event period means that the

actual load closely approaches the equilibrium load. Where rainfall is dominated by thunderstormactivities, rainfall and runoff energies are also likely to be high. Both these factors lead to typicallyhigher pollutant concentrations and loads during stormwater events in arid and semi-arid areascompared to elsewhere. Sediment loads transported during storm events are also generally muchhigher in arid and semi-arid areas because sparse native vegetation in arid areas offers littleprotection against soil erosion. However, the total annual pollutant loads from urban catchments inarid and semi-arid areas are generally less than those from temperate and humid areas, becauseannual runoff volumes are lower and storm events are less frequent in arid and semi-arid areas.

77



I Time

Figure 5.1 Typical pattern of surface pollutant load on impervious catchment surface over time

Time

Figure 5.2 Hypothetical pollutant and flow hydrographs showing first flush effect of pollutant remtrJal

5.3 Event mean concentrations of water quality parameters

Most of the pollutant load from urban areas is generated during big storm events. This is particularlyso in and and semi-arid areas where the storms are fewer and more intense than elsewhere. There isalso little subsurface flow in and and semi-and areas. As such, pollutant concentration during stormevents is an important parameter for estimating urban pollution loads. The event mean concentration

(EMC) is commonly used to describe the storm pollutant concentration, and it is defined as thepollutant load washed off by a storm event d ivided by the event runoff volume.

The event mean concentration can be estimated by monitoring pollutant concentration anddischarge over a storm event. The EMCs can vary between storm events, and EMCs of differentstorms are usually averaged to provide a representative EMC value for a catchment. The EMCdepends on many physical and climatic factors of the catchment, and can vary by more than an orderof magnitude between catchments. Therefore a good event monitoring program is essential whereaccurate estimates of pollutant loads are required.

The EMC is generally higher in drier areas. For example, the plots in Figure 5.3 showstatistically significant inverse relationships betin EMC and mean annual rainfall for severalimportant water quality parameters. The data for Figure 5.3 come from over 500 worldwide urban

78



stormwater quality data sets reported in the English language literature analysed by Duncan (1999)and include the data from the U.S.A. Nationwide Urban Runoff Program (see Athayde et al., 1983).

The same study found very few significant differences between the EMCs from the sub-categories of urban land use - residential, industrial and commercial. The EMCs vary considerablybetween catchments, but the standard land use zonings explain only a very small part of the variation(see Figure 5.4).

The mean k one standard deviation EMC values of the 21 water quality parameters reportedby Duncan are tabulated in Table 5.2. The table also gives the EMC values for areas with a meanannual rainfall of less than 550 mm for eight water quality parameters where there are data from atleast ten catchments. The data for the arid and semi-arid areas come mainly from Saudi Arabia,California and inland western U.S.A . It should be noted that all the data (except for pH) are analysedin the log domain.

W) linear relationship is statistically significant at a = 0.0 1

(S) linear relationship is statistically significant at a = 0.05

OX-3 linear relationshiD is not statisticallv significant at a =

Sumended solids Total Nitrogen CHS)

1’ ’ 0.1’0 500 1000 1500 2000 0 500 loo0 ‘500 2000

Chemical Oxygen Demand (HS) Total Organic Carbon (HS)

‘oc@l . . , 10001 1

3 10 ’ I 1 ’ I0 500 1000 1500 2000 0 500 1000 1500 2000

9Total Zinc (S) Total Copper (S)

. I 10 I I

=:ng ..01 1 0.001 I I

0 500 1000 1500 2000 0 500 1000 1500 2000

Total Chromium CNS) Total Iron (NS)

’ 0.1’ I500 1000 1500 2000 0 500 1000 1500 2000

Mean annual rainfall (mm)

Figure 5.3 Event mean concentration versus mean annual rainfall for several water quality parameters

79



Residential

Industrial

Commercial 1 : I1 10 100 1000

Residential

Industrial

Commercial

0.01 0.1 1 10

Residential i i ~i ~

Industrial

Commercial

0.001 0.01 0.1 1

Concentration (mgk)

Figure 5.4 EMC values (mean one standard deviation) for different urban land uses showmg thelarge variability and difficulty in distinguishing between the different land uses

(from analyses of worldwide data by Duncan, 1999)

Table 5.2 Mean f one standard deviation EMC values fr& analyses of worldwide data by Duncan(1999) - analyses carried out in the log domain

I Water aualitv Darameter I Unit I Number of data points and mean EMC (and lower and upper

Total Organic CarbonPHT~dkiitv-. -.v..Total Lead (Pb)

Total Zinc (Zn)Total Copper (Cu)Total C ’ . ‘- ‘.

mg/L 23 24 (13-44) I IpH 48 6.9 (6.2 - 7.6)

NTIJ 16 61_.-lil

-0 (14-260)mg/L 0.14 (0.040- 0.52) 13 0.22 (0.08 - 0.58)

mg/L 156 0.24 (0.089- 0.65) 10 0.33 (0.20 - 0.57)mg/L 140 0.050 (0.017 - 0.15) IO 0.064 (0.033 - 0.12)- ,. r- ‘.3 (1.3- 15)amlum (l;a) 1 UQlL 1 31 1 4

I rla/L 1641 2!

1Faecal Streptococci 1#/lOOmL I 19 19000(1900-190000) I I I

80



5.4 Estimation of stormwater pollution load

The total pollutant load washed off urban surfaces depends on the runoff volume and theconcentration of a given contaminant. Even in the absence of runoff data, the storm runoff volumecan be estimated reasonably reliably as a proportion of the rainfall. However, as discussed in Section

5.3, the pollutant concentration can vary considerably from one catchment to another. Because ofthis, the monitoring of stormwater quality is essential where accurate estimates of pollutant loads arerequired.

5.4.1 Water quality monitoring

Monitoring should be carried out over storm events because most pollutants are transported duringstorms. Pollutant load is the product of runoff vdume and pollutant concentration, but for a givencatchment the runoff volume varies over a much larger range than the pollutant concentration. Duringa single storm, most of the pollutant load is also transported during the higher discharges, even whenfirst flush occurs. The most effective event monitoring is with automatic samplers. These samplersare triggered by storm events (usually by water depths) and collect many water quality samples duringthe event that can then be tested in the laboratory. The total pollutant washoff load as well as the load

characteristics throughout the storm (loadograph) can be estimated from the discharge hydrographand the pollutograph (hydrograph of pollutant concentration) (see Figure 5.5).

Many authorities take ‘grab samples from urban waterways periodically (usually monthly) andtest them for various water quality parameters. This data is useful in providing an indication ofpollutant concentrations during dry weather (from subsurface flow), but may be of little use inestimating the total pollutant loads from baseflow and stormwater flow combined. Data from grabsampling can also be used to study long-term water quality trends.

32 600-

&

iaJz2 300-.-a

OL

0.40

0.30

0.20

0.10

0.00

Figure 5.5 Hydrograph, pollutograph and loadograph for one storm event in 1994 in a 150 ha urbancatchment in Melbourne, Australia (mean annual rainfall of 650 mm)

5.4.2 Estimation of daily and long-term pollution load

The average annual pollution load can be estimated as (see Chiew and McMahon, 1999)

81



Pollution Load = EMC x Runoff (5.1)

As discussed above, the EMC can vary considerably between catchments, and it should be measuredwhere accurate estimates of pollution loads are required. Nevertheless, in the absence of data, theEMC values in Table 5.2 can be used together with the estimated runoff to provide a guide to theprobable range of diffuse pollution load generated from a catchment.

For most studies, there is only sufficient data to estimate daily diffuse pollution load as

Pollution Load = (surface runoff from impervious and pervious area) x EMC+ subsurface flow x dry weather concentration (5.2)

The different components of runoff can be estimated using the conceptual daily rainfall-runoffmodels discussed in Chapter 4. Table 5.2 provides the typical range of EMC values for various waterquality parameters, but accurate EMC values can only be obtained from water quality monitoringduring storm events. The dry weather concentration of most urban pollutants is usually much lowerthan the EMC, and it can be determined from several dry weather baseflow samples. In and andsemi-and areas, the modelling of pollution loads from baseflow may be supertluous because there islittle baseflow. The baseflow water quality is also generally much better than the surface runoff waterquality.

There is little reason to use a more complex model than the linear one described by equation(5.2) to estimate daily pollution loads because it is difficult to define a clear relationship between runoffand EMC (see Section 54.3). Nevertheless, despite the la& of data, it is not uncommon to see manystudies estimate the daily pollution load as a power function of runoff,

Pollution Load = a Runof? (5.3)

where a and b are parameters found by optimisation or by relating the pollutant load and runoff data.

54.3 Modelling of event and sub-daily time step pollution load

Two of the more commonly used stormwater quality models are SWMM (Huber and Dickinson, 1988)and HSPF (Johansen et al., 1984). SWMM was initially designed as a single event stonnwaterquantity and quality model, but now permits continuous simulation of stormwater hydrograph andpollutograpMoadograph. HSPF is usually used for continuous simulation, typically with hourly timesteps.The hydrologic components of these and other event models have been discussed in Chapter 4.

Water quality models either conceptually simulate the stormwater pollution process orempirically relate the pollutant load to the runoff. In the conceptual representation, the modelsconsider the pollutant buildup and washoff processe s separately. For example, SWMM estimates thesurface pollutant load either as a linear, power, exponential or Michaelis-Menton function of the lengthof dry days (see Figure 5.6). The model then estimates pollutant washoff using an exponentialwashoff algorithm that reduces the available surface pollutant load exponentially over the storm event(see Figure 5.6). The input into the water quality component of these models is the stormwaterhydrograph and the output is the poll&graph and loadograph, which can be routed through thedrainage network to the catchment outlet.

The model parameters depend on the catchment surface characteristics and other factors andcan be determined by calibrating the model against locally monitored water quality data. However,given the scarcity of event water quality data and the number of parameters used, the detailedmodelling of pollutant buildup and washoff processe s in these models is rarely justified. Therefore,these models are generally used only for research purposes and in planning studies for comparingrelative simulations for alternative design strategies.

The pollutant washoff is governed by two main processes, the shear stress generated by flowand the energy input of rainfall. Recent studies (Vaze and Chiew, 1999) have shown that eventpollutant loads can be estimated satisfactorily as power functions of the rainfall intensity or the runoffrate,

82



Event Load = a 2 (very short-duration rainfall intensity or runoff rate) b (5.4)i=l

with the parameters, a and b, determined by calibrating the equation against locally monitored eventpollutant load data. The implication of this is that washoff is dominated by short periods o f high

intensity rainfall and/or runoff, which are not adequately represented by data averaged over longerdurations. The rainfall energy is the larger component in overland flow, but runoff energy is likely todominate in channel flow.

start of storm (POtart of storm (PO

(a, b and limit are model parameters)a, b and limit are model parameters)

Time (t)ime (t)

Figure 5.6 Buildup a lgorithms and exponential washoff decay algorithm used in SWMM

In most applications, if there are sufficient data to calibrate the models on a particularcatchment, the optimised parameter values (in Figure 5.6 and equation (5.4)) can be used to estimatethe pollutant washoff loads and washoff characteristics satisfactorily for future events on thatcatchment. However, because of the large variability between catchments, it is unlikely thatparameters calibrated against data from one catchment can be confidently used to estimate pollutantloads from another catchment.

In the model calibration, it is important to test the ability of the optimised parameters toestimate the pollutant loads for independent data that are not used for the calibration. A split samplingmethod is commonly used where the availabte data are divided into two. The model is calibratedagainst the first half of the events, and then tested against the remaining events. However, as it israre to have data from many events, a cross-verification method can also be used to increase theamount of data for independently testing the calibrated model. Here, each event is left out in turn, withthe model calibrated using data from all the other events. The optimised parameter values are thenused to estimate the pollutant load for the event that was left out, which is then compared against therecorded load.

83



Many modelling studies use an empirical power function to relate the pollutant load to the totalstormwater runoff volume

Event Pollutant Load = a (Event Runoff Volume) b (5.5)

The coefficients, a and b, are usually determined by plotting the event load and runoff on a log-log

scale. The correlations reported in the literature for the relationship in equation (5.5) are almostalways spurious and are very much higher than the actual correlations in the basic data. This isbecause runoff is used both to estimate the load (load is either estimated from the dischargehydrograph and the poll&graph or as total runoff times EMC) and as the explanatory variable inequation (5.5). Therefore, to reftect the true correlation in the basic data, the relationship betweenEMC and event runoff, rather than the relationship between event load and runoff, should be used(see Figure 5.7).

Data from 180 ha urban catdxnent Data Corn 670 ha urban catchment

in Dhahran, Saudi Arabia in Sydney, Australia

(mean annual rainfU of 80 mm) (mean annual rainfall of 1200 mm)

-__3 RL = 0.01

g 400 .

3 n I n

E200' l .

8 c .

0

3000

0 2 4 6 I 0 2 4 6 8 10

R* = 0.74.

0 2 4 6

Storm runoff (mm)

R* = 0.05.

.I I.

.= n. n

P

loo00

8000

4000

2000

0

I . I

R* = 0.26

:>m

I

0 2 4 6 8 10

Storm runoff (mm)

Figure 5.7 Pollutant EMC versus runoff and load versus runoff showing spurious correlation in loadversus runoff relationship

5.5 Other issues in the estimation of stormwater pollution loads

5.5.1 GIS and pollutant mass loading

Geographic information system (GIS) is increasingly being used with pollution load models to estimateand present relative load contributions from different catchments to receiving waters. The pollutionloads are either estimated directly using a pollutant mass loading (kg/ha) for each urban land use oras the product of runoff and EMC, with the runoff estimated as a proportion of rainfall, and an EMC(mg/L) value associated with each land use (e.g., Wong et al., 1997).

The pollutant mass loading and EMC values are usually obtained from the literature with somesampling on different land use locations within the catchments to narrow the uncertainty in the data.

84



This method can at best provide only approximate estimates of the long-term pollutant loads fromurban catchments.

Nevertheless, GIS is a powerful technology because it can facilitate the handling of largevolumes of data and can carry out complex spatial operations and link spatial and descriptiveinformation. The GIS can also present graphical results of the spatial estimates of pollution loads in auseful way that will allow catchment managers to consider alternative management practices or to

identify potential problem areas to be monitored or treated.

5.5.2 Link-node modelling of pollution generation and transport in large

catchments

It is often n ecessary to integrate the pollutant loads estimated from individual catchments (using themodels presented in Section 5.4) to assess their impact on the receiving waters. The addition of long-term pollutant loads from individual catchments is usually facilitated in the water quality models or canbe done using spreadsheets or GIS.

However, increasingly more in-house models are being developed to simulate the fate of thepollutants as they are transported through the drainage network and natural urban waterways to thereceiving waters. The in-transit processes simulated by these link-node models (see Figure 5.8) mayinclude deposition and scouring/resuspension of sediments, biochemical transformation of pollutants,

and the pollutant adsorption to and release from sediments. The link-node models can also simulatethe removal of pollutants by structural treatment methods (in particular, ponds and wetlands), and arecommonly used to study altemative water quality management strategies. The empirical and/orprocess based algorithms used to simulate the in-transit p ecesses are usually intuitive and realistic,but there is rarely field data to test these algorithms as used in the model.

H Nodes represent pollutant export from subcatchment and inputs fromupper subcatchments. Nodes can also be used to represent stormwatcr

treatment devices or large points sources.

- Links simulate in-transit prwesses (deposition and resuspension ofsediments, biochemical transformation of pollutants, pollutant

adsorption to and release t&n sediments, etc.. .)

Figure 5.8 Link-node representation of pollutant generation and transport

85



5.5.3 Contaminants associated with different sediment sizes

In addition to estimating total pollutant loads, it is also important to know the relative loads associatedwith different sediment sizes. For example, if most of the pollutants are attached to the finersediments, it may be necessary to design stonnwater treatment methods to remove the finersediments (via vegetation in wetlands) rather than the total suspended solid loads (via settling inponds and sedimentation basins).

There is little data in the literature on contaminants associated with different sediment s izes.Table 5.3 shows metals and phosphorus loads from urban road data quoted by Dempsey et al. (1993)and Figure 5.9 shows the nutrient loads from a stormwater sample in Melbourne, Australia. The datacan vary across catchments (as indicated in the TP data in Table 5.3 and Figure 5.9) but theygenerally show that a large proportion of the pollutants are attached to the finer sediments. Forexample, the Melbourne data suggests that to effectively reduce nutrient loads from particulatepollutants in the catchment, the treatment facilities should target the removal of sediments down to atleast 50 j.un for TP and 10 pm for TN.

Event monitoring programs to estimate pollutant loads are now common in water qualitystudies, and some of these programs also determine the particle size distributions of the collectedsamples. However, depending on the objectives of the study, it may also be important to also assessthe contaminant load associated with different sediment sizes.

Table 5.3 Metals and phosphorus loads associated with different particlesizes (from urban road data analysed by Dempsey et al., 1993)

Contaminant

PbZncuTP

74 pm1424813

Percentage of pollutant in par-tide size finer than105pm 250 pm 840 pm

34 56 8860 87 9621 92 9835 58 69

2000 km95100loo83

loo

90

80

70

60

50

40

30

20

10

06a 0.1 1 IO loo loo0

Particle size (mm)

Figure 5.9 Nutrient load associated with different particle sizes from an urban stormwater sample inMelbourne, Australia (mean annual rainfall of 650 mm)

86



Conclusions

Important issues related to urban stormwater pollution are reviewed and discussed. In arid and semi-arid areas, the surface pollutant load is likely to be high. Due to the dry conditions and reducedvegetation in and areas, higher pollutant concentrations and loads during stormwater events in theseareas compared to the same elsewhere. Sediment loads transported during storm events are also

generally much higher in arid and semi-arid areas because sparse native vegetation in and areasoffers little protection against soil erosion. However, the total annual pollutant loads from urbancatchments in and and semi-and areas are generally less than those from temperate and humidareas, because annual runoff vdumes are lower and storm events are less frequent in and and semi-and areas.

The monitoring of stormwater quality is essential where accurate estimates of pollutant loadsare required. In addition to the pollutant loads, the relative loads associated with different sedimentsizes should be known. The GIS is a useful technology, especially in large and arid catchments,because it can facilitate the handling of large volumes of data and can carry out complex spatialoperations and link spatial and descriptive information.

Bibliography

ATHAYDE, D.N., SHELLEY, P.E., DRISCOLL, E.D., GABOURY, D. AND BOYD, G. (1983) Results ofthe Nationwide Urban Runoff Program. Environmental Protection Agency, Washington DC.,U.S.A., PB84-185537.

CHIEW, F.H.S. AND MCMAHON, T.A. (1999) Modelling Runoff And Diffuse Pollution Loads In UrbanAreas. Water Science & Technology, Vol. 39, pp. 241-248.

CHIEW, F.H.S., MUDGWAY, L.B., DUNCAN, H.P. AND MCMAHON, T.A. (1997a) Urban StormwaterPollution. Cooperative Research Centre for Catchment Hydrology, Melbourne, Australia,Industry Report 9715.

CHIEW, F.H.S., DUNCAN, H.P. AND SMITH, W. (1997b) Modelling Pollutant Buildup And Washoff:Keep It Simple. Proceedings of the 24th International Hydrology and Water ResourcesSymposium, Auckland, November 1997, New Zealand Hydrological Society, pp. 131-I 38.

DEMPSEY, B.A., TAI, Y.L. AND HARRISON, S.G. (1993) Mobilisation And Removal Of

Contaminants Associated With Urban Dust And Dirt. Water Science & Technology, Vd., 28,pp. 225230.DUNCAN, H.P. (1995) A Review Of Urban Storm Water Quality Precesses. Cooperative Research

Centre for Catchment Hydrdogy, Melbourne, Australia, Report 95/9.DUNCAN, H.P. (1999) Urban Stormwater Quality: A Statistical Overview. Cooperative Research

Centre for Catchment Hydrology, Melbourne, Australia, Report 99/3.HUBER, W.C. AND DICKINSON, R.E. (1988) Storm Water Management Model, Version 4: Users

Manual, University of Florida, Gainesville, U.S.A.JAMES, W. AND SHIVALINGAIAH, B. (1988) Continuous Mass Balance Of Pollutant Build-Up

Processes. In: Urban Runoff Pollution, Vdume 10 (Editors: H .C. Tomo, J. Marsalek & M.Desbordes), Springer-Verlag, Benin, pp. 243-271.

JOHANSEN, R.C., IMHOFF, J.C., KITTLE, J.L. AND DONIGAN, A.S. (1984) Hydrocomp SimulationProgram - Fortran (Hspf): User’s Manual Release 8.0. USEPA, Athens, Georgia, U.S.A.

MAKEPEACE, D.K., SMITH, D.W. AND STANLEY, S .J. (1995) Urban Stonnwater Quality: SummaryOf Contaminant Data. Critical Reviews in Environmental Science and Techndogy, Vol. 25,pp. 93-l 39.

NOVOTNY, V. SUNG, H.-M., BANNERMAN, R. AND BAUM, K. (1985) Estimating Nonpoint PollutionFrom Small Urban Watersheds. Journal of the Water Pollution Control Federation, Vol. 57, pp.339-348.

VAZE, J. AND CHIEW, F.H.S. (1999) Investigation Of The Relationship Between Event PollutantLoad And Rainfall And Runoff Characteristics. Proceedings of the Water 99 Joint Congress,Brisbane, July 1999, Institution of Engineers Australia, Vdume I, pp. 27-32.

WONG, K.M., STRECKER, E.W. AND STENSTROM, M.K. (1997) GIS To Estimate Storm-WaterPollutant Mass Loadings. Journal of Environmental Engineering, Vol. 123, pp. 737-745.

87



Chapter 6

Traditional methods of urban drainage

in arid and semi-arid regions

Urban drainage programs did not develop as quickly as drinking water or wastewater programssince they are capital intensive and expensive. The concept of “need” in urban drainage is hardto pin down, and is based on decisions aboutaffordable levels of service especially in arid andsemi-and dimates. Stormwater needs are based on flood protection, convenience, andprotection of water quality, dong with cordlary objectives such as urban beautification andimprovement of life. In water short areas, the capital needs have been for water harvesting andwater quality management.

Although urban drainage facilities are often neglected, the growing wncerns aboutenvironmental protection will increase the future needs for urban drainage, which include

combined and separate storm sewers. Storm runoff in drainage basins undergoing urbanizationis a major concern. Runoff vdumes from newly urbanized drainage basins are significantlyattered due to an increase in impervious surfaces (i.e. roadways). Studying available data andcharacterizing the temporal nature of and and semi-and land rainfall events are important s tepsin planning and designing urban drainage systems. Good planning generates good projectsand programs, and good projects and programs can wst4fectively stop and solve problems.Today’s use of computers allows for the manipulation of large amounts of data and for moreprecise solutions by the use of sophisticated algorithms.

The dimate of and and semi-and areas is characterized essentially by rainfall variabilityin both space and time coupled with high potential evaporation rates (chapter 2). Thephenomena need to be understood in order to explain the environmental constraints imposedupon human activities and in order to investigate common and and semi-and lands’ urbandrainage problems.

Effective disposition of stormwater is essential. Drainage systems have changed fromprimitive ditches to complex networks of curbs, gutters, and underground conduits. Modemurban drainage systems spotlight sewers: storm, combined and sanitary. Along with theincreasing complexity of these systems has wme the need for a more thorough understandingof basic hydrologic processes. Simple rules of thumb and crude empirical formulas aregenerally inadequate. The approximation of maximal rates of flow to be expected with somerelative frequency is not sufficient for many modem designs. It is, therefore, essential toaccount for all key hydrologic processes and combine them in composite models that yieldoutputs at points of interest in time and space. In, addition, demands by society for betterenvironmental wntrd require that water quality considerations be superimposed on estimatesof quantii for effective urban water management.

The fdlowing sections present the state-of-the-art of wmmon methods of urbandrainage in and and semiarid regions. In this regard, the methods used may differ according to

the size and shape of the catchment. Moreover, the hydrologic design diffe rs according to thetype of hydraulic structure used. Such structure range from small crossroad culverts, levees,and drainage ditches to urban storm drainage systems. Meanwhile, the traditional urbandrainage methods are evaluated and compared with modem methods in the following sections.

6.1 Small catchments

Runoff occurs when precipitation moves across the land surface, some of which eventuallyreaches natural or artificial streams and lakes. The land area over which rain falls is called thecatchment and the land area that contributes surface runoff to any point of interest is called awatershed. A large watershed can contain many smaller subwatersheds. In urban drainagesystems, serving areas up to few hundred acres (or hectares in size, the stormwater is usually

collected in the streets and conveyed through inlets to buried conduits that carry it to a pointwhere it can be safely discharged into a water body as shown in Fig 6.1. In some instances

88



stormwater is percolated into the ground using infiltration ponds but the underlying strata musthave a high permeability. Another method considers collection of stormwater by open ditchesbut such open channels are not always acceptable in developed areas. The banks of an openditch also may be eroded from a high groundwater table. Thus, an underground piping system(french drain) that removes the water without removing the soil partides may be necessary.Excessive soil loss reduces the carrying capacity of the soil and can cause subsidence. Sometypes of protective filter material and select aggregate placed adjacent to the erodable soils canprovide a solution to soil loss. A fabric filter must be specified to prevent piping, reduce fabricclogging, and pass the water. These criteria must all be present. A french drain is a perforatedor slotted p ipe placed in selected backfill aggregate material with a fabric filter surrounding theaggregate as shown in Fig. 6.2. Also, a french drain may consist solely of aggregate materials.The purpose of a french drain is to lower the water table or remave stormwater through thebank of a stormwater pond. Since surface runoff quantities are usually small in arid and semi-arid lands, the runoff could be collected along with sanitary wastewater in combined sewers.

Stormwater

fl #=unoff*...'.

#.“

to disposal

Fig. 6.1 Collection of runoff by storm sewers

Stormwa ter

Fig. 6.2 A French drain

6.1 .l Amount of storm runoff

The Rational Formula for estimating peak runoff rate was introduced in the United States in

1889. Since then it has become the most widely used method for designing drainage facilitiesfor small urban areas and highways. Description of this formula is given in chapter 4.Attempts to verify the Rational Method have not produced encouraging results.

89



Specially in arid and semiarid areas, where the spatial variation of infiltration is high, highuncertainty exists in the discharge coefficient. In addition, it is impossible to observe ormeasure a uniform or an average rainfall over catchment. Furthermore, times of concentrationdo not seem to accord with the time to peak. The reasons are inherent in adopting a simplemodel to express a complex hydrologic system. Yet the method continues to be used inpractice with results implying acceptance by designers, officials, and the public. The method iseasy to apply and gives consistent results. From the standpoint of planning, for example, themethod demonstrates in clear terms the effects of development: Runoff from developedsurfaces increases because times o f concentration decrease and runoff coefficients increase.

6.1.2 Design of storm sewers

Storm drainage can be divided into two aspects: runoff prediction and system design. Thedesign of storm drains conforms to the principles of flow in open channels, and Manning’s orsimilar type equation is used to calculate the required pipes sizes. The main rules governingselection of pipe size and slope are as follows:

1 Free-surface flow exists for the design discharges; that is, the sewer system is designedfor “gravity flow”

2.To avoid clogging, the minimum pipe diameter should preferably be 10 or 12 in. (25 to 30cm), although 8-k-r. 2O-cm) pipes are used in some cities.

3.The minimum velocity flowing full should be at least 2.5 Wsec (0.75 m/s). This is required toprevent or reduce excessive deposition of solid material in the sewers.

4.To prevent scour and other undesirable effects of high-velocity flow, a maximumpermissible flow velocity is also specified.

S.The design diameter is the smallest commercially available pipe having flow capacity equalto or greater than the design discharge and satisfying all the appropriate constraints.

6.Pipes sizes should not decrease is the downstream direction even though increased slopemay provide adequate capacity in the smaller pipe. Any debris that enters a drain must becarried through the system to the outlet, and the possibility of clogging a smaller pipe withdebris that may pass a larger pipe is too great.

7.Pipe slope should conform to the ground slope insofar as possible to use a smaller pipe by

exceeding the ground slope. If this makes the use of smaller pipe possible for somedistance downslope, it may be economic despite increased excavation.

8.Pipe grades are described in terms of the elevation of the invert, or inside, bottom of thepipe. Where pipes of different size join, the tops of the pipes are placed at the sameelevation, and the invert of the larger pipe is correspondingly lower than that of the smallerpipe. This does not apply to tributary drains, which may enter the main sewer through adrop manhole.

The design of storm sewers is a direct application of the principles from both hydraulicsand hydrology. Intensity-Duration-Frequency (IDF) curves are used to specify rainfallintensities. Watershed characteristics are used to calculate pipe or channel sizes necessary toconvey the calculated rates of flow. The fundamentals of the storm sewer design processusing the rational method are as follows:

6.1.2.1 Determination of storm sewer flow rates

A storm sewer is typically designed for a specific return period storm, usually 10 or 25 yr. Thisreturn period is used in conjunction with a local IDF curve to determine a rainfall. The durationused in the determination of the rainfall intensity is equal to the time of concentration of thecontributing watershed. The widely accepted storm sewer design regulations usually specify aminimum time of concentration, and if the watershed time of concentration is less than thespecified minimum, the specified minimum is used rather than the watershed time ofconcentration. In cases where a storm sewer inlet has upstream piping, the maximum of thewatershed time of concentration or the accumulated upstream travel time is used.

The intensity (I) of the rainfall is used in the rational equation (Q = CIA) to determinethe flow rate at the storm sewer inlet. The rational equation is thought traditionally to beapplicable because in most storm sewer design situations, the entire watershed is broken up

into a number of smaller subwatersheds each influent to a manhole or inlet. Flows into eachinlet are then summed in a downstream direction. These flow rates are used to determine the

90



size of the storm sewer piping. Pipe sizes should be selected to meet or exceed the requiredcapacity for the design storm. The use of the rational formula for each inlet and the sum ofthese flows are considered a conservative estimate for pipe sizing.

6.1.2.2 Hydraulic grade calculations

Once flow rates have been calculated throughout the system, the hydraulic grades in thesystems also should be calculated. Using the principles of hydraulics and a known tailwaterelevation at the outlet along with other physical parameters of the systems the hydraulic gradescan be calculated.

Within storm sewer pipe systems the slope of the hydraulic grade line (HGL) may becalculated using Manning’s formula:

(6.1)

where

v = velocity in meter per secondR = hydraulic radius in meters = slope of the hydraulic grade line as a fraction

In the case of open channel flow this slope assumes that flow is at the normal depth and theslope of the HGL is parallel to the slope of the pipe. With a known slope of the HGL, it ispossible to calculate upstream hydraulic grade with a known downstream hydraulic grade.

HG upstream =HG downstream + SL(6.2)

whereHG = hydraulic grade

s = slope of HGLL = length of piping

The term SL is also the head loss in the pipe section. Head loss at junctions, inlets, ormanholes can be calculated using the equation:

2

h,dL2g

(8.3)

wherehL = head lossK = head loss coefficient dependent upon geometryv = maximum velocity influent to junction

9 = gravitational constant

By summing head losses in an upstream direction the hydraulic grade at any point within thesystem can be calculated.

Storm sewer design follows a number of design constraints which must be met.Minimum velocities inside the storm sewer piping must usually be greater than 0.5 to 0.8 m/s.This ensures that materials do not deposit in the piping. The velocity must also usually be lessthan 3 to 5 m/s to prevent scouring. The pipe crown, or the top of the pipe, must be (typically)0.8 m below the level of the ground to prevent crushing or collapsing of the pipe under loads.The distance between the pipe crown and the ground elevation is referred to as pipe cover.Pipes must be sized large enough to prevent flooding. Surcharge conditions are oftendiscouraged in design.

91



6.1.3 French drains

The retention ability of a filter material to retain soil is a function of its openings and the size ofthe soil particles. Cedergren et al. (1972) reports on a criteria that states that the equivalentopening size of the fabric (standard test method for fabrics) divided by the nominal diameter of

soil particles for which 85% of the soil degradation is finer must be less than 2 to retain the soiland prevent excess piping. Particles larger than the fabric pore size will be retained, and someof the smaller ones will interact with the fabric filter and will be impacted. Other smaller oneswill be “bridged” between the fabric and the soil particles and further retention is possible.Clogging of the filter can be minimized by specifying the expected hydraulic gradient of thegroundwater during the selection process. The permeability of the fabric filter should be at least10 times greater than the protected soil, which should allow the soil water to drain. Properspecification of the size of pipe and trench depth below the expected water table is necessary ifthe water table is to be lowered. Darcy’s law can be used to approximate the seepage fromsaturated soil into the french drain

Since the water table will fluctuate, the area normal to flow will fluctuate. The maximumwater table elevation can be chosen, and the flow rate (size of trench) determined from thiscondition. Permeability of the soil can be estimated using laboratory or field permeability tests.Generally, lower values are chosen for determining drawdown time and higher values fordetermining pipe sizes.

For the aggregate that is used around the pipe, its permeability must be specified toensure that it is greater than the predicted soil. Furthermore, the aggregate must notdeteriorate over time. Some limestones do break down and form an impermeable material.Cross-sectional area and hydraulic gradients are usually determined by site conditions, thuspermeability of rock should be specified for desired flow conditions. During seepage, some finesoil particles do enter the aggregate, thus reducing the permeability.

6.1.4 Drainage system changes

Open channels (ditches) may be used to collect stormwater in small catchments. However, asareas urbanize, the drainage system will change due to increased runoff, increased velocities,and changing flow patterns caused by the use of stormwater management facilities.

In order to increase the capacity of the channel system to take more runoff and greatervelocities, the system must either be increased in size and/or stabilized. Stabilizationtechniques may be rigid designs using rigid materials such as concrete, or flexible designsusing vegetation or rock riprap. Flexible linings of erosion resistant vegetation and rock riprapare often preferred because of their aesthetic appearance. When vegetation is chosen as thepermanent channel lining, it may be established by seeding or sodding. Installation by seedingusually requires protection by using one of a variety o f temporary lining materials until thevegetation becomes established.

6.1.5 Infiltration systems

Infiltration is the soaking of runoff into the ground. It reduces flooding, as does detention, byreducing the volume of runoff that is discharged. Most infiltration systems are backfilled withcrushed stone. Runoff fills the void spaces of stone until it infiltrates the surrounding soil. Thusrunoff floi-ving into these systems is essentially remaved from the flood hydrograph and willhave no impact on downstream flooding. In addition, infiltration addresses water quality,groundwater levels, and surface water supplies. Unlike detention, infiltration puts the water in apart of the natural environment where it can be filtered, stored, and available for further use.Also, unlike detention basins, infiltration basins do not discharge runoff to the downstreamchannel system and thus, do not cause negative effects on the downstream channel system.

Although seeming to be the ideal solution to the problems associated with increasedurban runoff, infiltration systems are not without problems. Most of the problems are related tomaintenance. The infiltration capacity of the system must be maintained and as portions of thesystem lose this capacity they must be removed and new material added to the system.Because most of these systems are underground, it is often difficult to inspect the system todetermine if maintenance is required.

92



Another problem that is limiting the use of infiltration systems is their lack of acceptance withinmany municipal public works and engineering departments. Although design procedures areavailable, their use is still not extensive and these systems have “not proven themselves” to beeffective stormwater management facilities, in many municipalities. More details andconsiderations on applying o f these systems in arid and semiarid areas are given in chapter 7.

6.2 Large catchments

In large catchments s tormwater is usually wllected in the streets and conveyed through inletsto buried conduits that carry it to a point where it can be safely discharged into a stream, orocean. In some instances, stormwater is percolated into the ground using infiltration ponds.For such means of disposal to be practical, the underlying strata must have a high permeability.A single outfall may be used to convey the stormwater to the point of disposal or a number ofdisposal po ints may be selected on the basis of the topography of the area. The accumulatedwater should be discharged as close to its source as possible. Gravity discharge is preferablebut not always feasible, and pumping ,plants may be an important part of a city stormdrainagesystem.

Some communities require designers to provide detention basins or holding ponds withenough storage so that the outflow from the basins or ponds in a major storm is no greater thanthe peak outflow that would have occurred from the area prior to its development. This can beachieved in part by being careful not to overdesign the system. During intense stormssubstantial ponding in the streets will aid in reducing the peak outflows from the developedarea. Care must be taken, however, to guard against synchronization of the peak outflowsfrom the various subbasins.

Until recently, most aspects of urban drainage design have been accomplished byusing simplified formulas, nomographs, and “rules of thumb”. These techniques are often usedby engineers to design urban drainage systems in small catchments. In addition, mostdrainage design is limited to the site being developed without regard or analysis o f the effectsof urban runoff on downstream areas or on the overall drainage system. Recently, somemunicipalities have adopted ordinances which require detailed comprehensive drainagedesigns, including analysis o f downstream effects, and require the use of computer models toaccomplish the engineering calculations that are required to comply with the.ordinance.

6.2.1 Estimates of flow

The first step in the design of urban drainage works is the determination of the quantities ofwater that must be accommodated. In case of sewer sizing only an estimate of the peak flowis required, but where storage or pumping of water is proposed, the volume of flow must alsobe known. Drainage works are usually designed to dispose of the flow from a storm having aspecified return period. It is often difficult to evaluate the damage that results from urbanstormwater, especially when the “damage” is merely a nuisance. Hence, the selection o f thereturn period is often dependent on the designer’s judgment. In residential areas, there may belittle harm in filling gutters and flooding intersections several times each year if the floodinglasts only a short time. In a commercial district, such flooding may cause damage andinconvenience, and a greater degree of protection may be warranted. Areas of relatively goodnatural drainage need less protection than low areas, which serve as collecting basins for flowfrom a large tributary area. Return periods of 1 or 2 yr. in residential districts and 5 to 10 yr. incommercial districts are all that can be justified for the average city.

The most satisfactory method for estimating urban runoff is by simulation using acomputer. In this approach, flows are simulated throughout the system from available rainfalldata. For adequate definition of the lo-yr. event., at least 30 yr. of flow should be simulated.Output is the simulated flow at all key points in the system. From this output annual flow peakscan be selected and subjected to frequency analysis to define the design flow at each point.The effects of small storage reservoirs or pumping plants can be investigated in the simulation,and if more than one rainfall record is available the etfect of a real rainfall variation can beincluded. Calibration of the simulation model should be made against the nearest gaugedstream having soil character istics similar to those of the urban area under study. Parameter

adjustment accounts for the impervious area in the urban area. A preliminary layout of thedrainage system is required so that the effect of the improved drainage can be reflected in the

93



simulated flows. The simulation approach avoids the arbitrary assumptions of constant runoffcoefficient, uniform rainfall intensity, equal frequency of rainfall and runoff, etc. as used in theRational method.

6.2.2 Urban drainage models

The first computetized models of urban storm drainage were developed during the late 1960sand since that time a multitude of models have been discussed in the literature (Messman andLewis, 1996). The models applicable to design of storm sewer systems can be classified asdesign models, flow prediction models, and planning models.

Design models determine the sizes and other geometric dimensions of storm sewers(and of other facilities) for a new system or an extension or improvement to an existing system.The design computations are usually carried out for a specified design return period.

Flow prediction models simulate the flow of stormwater in existing systems of knowngeometric sizes or in proposed systems with predetermined geometric sizes. Most flowprediction models simulate the flow for a single rainfall event, but some can simulate theresponse to a sequence of events. The simulation might be for historical, real-time, or

synthetically-generated storm events. At least some simple hydraulics is considered in mostmodels. A model may or may not include water quality simulation. The purpose of a flowsimulation may be to check the adequacy and performance of an existing or proposed systemfor flood mitigation and water pollution control, to provide information for storm watermanagement, or to form part of a real-time operational control system.

Planning models are used for broader planning studies of urban stormwaterproblems, usually for a relatively large space frame and over a relatively long period of time.The quantity and quality of stormwater is treated in a gross manner, considering only the massconservation of water and pollutants without considering the dynamics of their motion throughthe system. Planning models are employed for such tasks as studies of receiving water qualityand treatment facilities. They do not require detailed geometric information on the drainagefacilities as do the first two groups of models.

6.2.3 Computer programs

Mathematical equations used in Hydrology and Hydraulics have been adapted to software usedon computer extremely well. As computers have become faster and more powerful, engineershave utilized the computer’s potential to solve more complex problems. Today, hundreds ofcomputer programs and computer tools exist which are useful for solving urban drainage andrelated problems.

Nine of the most frequently American used publicdomain urban stormwater packagesare presented in Table 6.1 (Viessman and Lewis, 1996). Models of similar popularity includeMOUSE, Hydroworks, and Hystem. Some of these models just simulate the urban rainfall-runoff process; others provide specifies on the type, size, and location of drainage andstormwater handling facilities. The model acronyms and dates of original release as softwareare shown in Table 6.1. These models are periodically updated, and the current version should

be requested when acquiring the code. Most are either single-event models or continuousmodels that are primarily used in a singleevent mode.A very widely accepted and applied storm runoff simulation model is the Storm Water

Management Model (SWMM) which was jointly developed by Metcalf and Eddy, Inc., theUniversity of Florida, and Water Resources Engineers (Metcalf & Eddy, 1971) for use by theU.S. Environmental Protection Agency (EPA). This model has been successfully used forurban drainage systems in and and semi-and regions (see chapter 9). Therefore, it will bediscussed in some detail.

The SWMM model is designed to simulate the runoff of a drainage basin for anypredescribed rainfall pattern. The total watershed is broken into a finite number of smaller unitsor sub-catchments that can readily be described by their hydraulic or geometric properties.This model has the capability o f determining, for short-duration storms of given intensity, thelocations and magnitudes of local floods as well as the quantity and quality of stormwater runoffat several locations both in the system and in the receiving waters. The original SWMM was anevent-simulation model, and later versions keep track of long-term water budgets.

94



Table 6.1 Frequently used Urban Stormwater Simulation Models in USA

Codename

Model name Agency originating Year

CHM

RRL

I LLUDAS

STORM

TR-55

DR3M

HYDRA

SWMM

UCURM

Chicago Hydrograph Method

Road Research Laboratory Method

Illinois Urban Drainage Area Simulator

Storage, Treatment, Overflow Runoff Model

SCS Technical Release 55

Distributed Routing Rainfall-Runoff Model

Hydrologic Component of HYDRAIN Package

Storm Water Management Model

U. of Cincinnati Urban Runoff Model

City of Chicago 1959

Road Research Lab 1962

Ill. Water Survey 1972

Corps. of Engineers 1974

scs 1992

USGS 1978

FHWA 1990

EPA 1971

U. of Cincinnati 1972

The fine detail in the design on the model allows the simulation of both water quantityand quality aspects associated with urban runoff and combined sewer systems. Only the waterquantity aspects are described here. Information obtained from SWMM would be used todesign storm sewer systems for stormwater runoff control. Use of the model is limited torelatively small urban watersheds in regions where seasonal differences in the quality aspectsof water are adequately documented.

The simulation is facilitated by five main subroutine blocks. Each block has a specificfunction, and the results of each block are entered on working storage devices to be used aspart of the input to other blocks.

The main calling program of the model is called the Executive Block. This block is thefirst and last to be used and performs all the necessary interfacing among the other blocks.The Runoff Block uses Manning’s equation to route the uniform rainfall intensity over theoverland flow surfaces, through the small gutters and pipes of the sewer system into the mainsewer pipes, and out of the sewer pipes into the receiving streams. This block also providestime-dependent pollutional graphs (pollutographs). A third package of subroutines, theTransport Block, determines the quality and quantity of dry weather flow, calculates the systeminfiltration, and calculates the water quality of the flows in the system.

A useful package of subroutines for water quality determination is contained in theStorage Block. The Storage Block allows the user to specify or have the model select sizes ofseveral treatment processes in an optional wastewater treatment facility that receives a user-selected percentage o f the peak flow. If used, this block simulates the changes in thehydrographs and pollutographs of the sewage as the sewage passes through the selectedsequence of unit processes.

The earlier version allowed simulation of any reservoir for which the outflow could be

approximated as either a weir or orifice, or if the water was pumped from the reservoir. Thenewer versions allow input of 11 points of any storage-outflow relation and routes hydrographsthrough natural or artificial reservoirs, including backwater areas behind culverts. Routing is bythe modified Puls method, which assumes that the reservoir is small enough that the watersurface is always level.

Evaporation from reservoirs is simulated by a monthly coefficient (supplied by the user)multiplied by the surface area. The Extran Block completes the hydraulic calculations foroverland flows, in channels, and in pipes and culverts. It solves the complete hydrodynamicequations, assesses surcharging, performs dynamic routing, and provides all the depth,velocity, and energy grade line information requested. Subcatchment areas, slopes, widths,and linkages must be specified by the user. Manning’s roughness coefficients can be suppliedfor pervious and impervious parts of each subcatchment.

Urban stem-r drainage components are modeled using Manning’s equation and the

continuity equation. The hydraulic radius of the trapezoidal gutters and circular pipes iscalculated from component dimensions and flow depths. A pipe surcharges if it is full, provided

95



that the inflow is greater than the oufflow capacity. In this case, the surcharged amount will becomputed and stored in the Runoff and Transport Blocks at the head end of the pipe. The pipewill remain full until the stored water is completely drained. Alternatively, the Extran Block canbe used to conduct a dynamic simulation of the system under pressure-flow conditions.

Necessary inputs in the model are the surface area, width of subcatchment, groundslope, Manning’s roughness coefficient, infiltration rate, and detention depth. Channeldescriptions are the length, Manning’s roughness coefficient, invert slope, diameter for pipes, orcross-sectional dimensions. General data requirements are summanzed in Table 6.2. A stepby-step process accounts for all inflow, infiltration losses, and flow from upstreamsubcatchment areas, providing a calculated discharge hydrograph at the drainage basin outlet.

Item 1.

Item 2.

Item 3.

Item 4.

Item 5.

Item 6.

Table 6.2 General Data Requirements for Storm Water Management Model (SWMM)(After Viessman and Lewis, 1996)

Define the Study Area. Land use, topography, population distribution, census tractdata, aerial photos, and area boundaries.Define the System. Plans of the collection system to define branching, sizes, andslopes; types and general locations of inlet structures.Define the System Specialties. Flow diversions, regulators, and storage basins.

Define the System Maintenance. Street sweeping (description and frequency),catch basin cleaning; trouble spots (flooding).Define the Base Flow (DWF). Measured directly or through sewerage facilityoperating data; hourly variation and weekday versus weekend; the DDWFcharacteristics (composited BOD and SS results); industrial flows (locations,average quantities, and quality).Define the Storm Flow. Daily rainfall totals over an extended period (6 months orlonger) encompassing the study events; continuous rainfall hyetographs,continuous runoff hydrographs, and combined flow quality measurements (BODand SS) for the study events; discrete or cornposited samples as available(describe fully when and how taken).

Three general types of output are provided by SWMM. If waste treatment processes

are simulated or proposed, the capital, land, and operation and maintenance costs are printed.Plots of water quality constituents versus time form the second type of output. Thesepollutographs are produced for several locations in the system and in the receiving waters.Quality constituents handled by SWMM include suspended solids (SS), settleable solids,biochemical oxygen demand (BOD), nitrogen, phosphorus, and grease. The third type ofoutput is hydrologic. Hydrographs at any point, for example, the end of a gutter or inlet, areprinted for designated time periods. The Statistics B lock will provide frequency analysis ofstorm events from a continuous simulation.

6.2.4 Urban storm drainage system

Surface waters enter a storm drainage system through inlets located in street gutters ordepressed areas that collect natural drainage. Catch basins under street inlets are connected

by short pipelines to the main storm sewer located in the street right-of-way, often along thecenter line. Manholes are placed at curb inlets, intersections of sewer lines, and regularintervals to facilitate inspection and cleaning. Pipeline gradients follow the general slope of theground surface such that water entering can flow downhill to a convenient point for discharge.Sewer pipes are set as shallow as possible to minimize excavation while providing 0.6 to 1.2 mof cover above the pipe to reduce the effect of wheel loadings. Common pipe materials usedare clay, tile or concrete. Sewer outlets that terminate in natural channels subject to tides orhigh water levels are equipped with flap gates to prevent backflooding into the sewer system.Backwater gates are also used on combined sewer outfalls and effluent lines from treatmentplants where needed.

6.2.4.1 Gutters

The discharge capacity o f gutters depends on their shape, slope, and roughness. Manning’sequation may be used for calculating the flow in gutter; however, the roughness coefficient n

96



must be modified to account for the effect of lateral inflow from the street. With the wide,shallow flow and the varying transverse depth common in gutters, the flow pattern is notsymmetric and the boundary shear stresses have an irregular distribution. For well-finishedgutters, n has a value of about 0.016 in the Manning equation. Unpaved gutters or gutters withbroken pavement will have much higher values o f n. Gutters are generally constructed with atransverse slope of 1 on 20. With such a transverse slope and a 15-cm curb height, the widthof flow in a gutter will be 3 m when there is no freeboard.

6.2.4.2 Inlets

Gutter flow is intercepted and directed to an underground storm-drain pipe system by dropinlets. There are two main types of inlets, with many commercial patterns available in eachtype. Grated inlets (Fig. 6.3) are openings in the gutter bottom protected by grates. A curbopening inlet (Fig. 6.3) is an opening in the face of the curb that operates much like a side-channel spillway. Curb opening inlets are feasible only where curbs have essentially verticalfaces.

The location of street inlets is determined largely by the judgment of the designer. Amaximum width of gutter flow of 1.8 m has been suggested as a suitable criterion for imporanthighways. Under this rule an inlet is necessary whenever the gutter flow exceeds the guttercapacity within the limiting 1.8 m width. In residential sections the ultimate in inlet spacingprovides four inlets at each intersection. With this arrangement flow travels in the gutter onlyone block before interception. A less expensive arrangement provides only two inlets at theuphill wmer of the intersection, and water is allowed to flow around two sides of the b lock.

6.2.4.3 Catch basins

At one time it was wmmon practice to provide catch basins (Fig. 6.4) at inlets to trap debrisand sediment and prevent their entry into the drain. The expense of cleaning catch basins canbe quite large, and it has become common practice to omit them from inlet design and dependon adequate velocities in the drain to prevent deposition of sediment.

6.2.4.4 Grated inletsFew inlets intercept all the flow that reaches them in the gutter unless they are at low pointsfrom which the water has no other route of escape. The most efficient grated inlets have barsparallel to the curb and a sufficient clear length so that water can fall through the openingwithout hitting a crossbar or the far side of the grate. Experiment has shown that this freelength ‘Y should be at least

X =.: y “‘(English units) or X = 0.94 VY”~ (SI units) (6.4)

where V is the mean approach velocity in the flow prism intercepted by the grate, and y is thedrop from the water surface to the underside o f the grate. A grate which satisfies these

requirements may be expected to intercept all the flow in the gutter prism crossing the grate.When water enters a curb-opening inlet, it must change direction. If the street has a low crown,the gutter flow will be spread out over a considerable width and a correspondingly greaterlength of inlet will be required for the change of direction to be accomplished. Curb-openinginlets, therefore, function best with relatively steep transverse slopes.

6.2.4.5 Manholes

Manholes are used in underground storm-drain systems to permit easy access to the pipes forcleanout and to serve as junction boxes for situations where there is a change in pipe size orslope or where several pipes join one another. Manholes are installed at intervals of not morethan 150 m along a line if the conduit is too small for a person to enter it. Manholes are usually

constructed of brick or concrete, and occasionally of concrete block or corrugated metal.

97



Preformed Fiberglas manholes are now available. The general design of brick ormanholes is shown in Fig. 6.5.

Sidewalk

Clay. tile. prc0ncrete vpe

Section A-ASection A-A

Maximum 1.25”4 MO.5 cm)r

ar length, x

Plan

(a) Grated inlet

i

E

\

r----

::_-__-

A A

Plan

(b) Curb-opening Inlet

Fig. 6.3 Some typical storm-drain inlets

RemovableCurb

acast-iron grate \

:=d . .y1

Concrete 0or brick J.

cStandard elbow ‘“,’

.Jo

a*.

Fig. 6.4 Inlet and catch basin

concrete

Fig. 6.5 Some typical manholes: (a) inside drop; (b) outside drop; (c) manhole for large pipe.

98



The bottom of the manhole is usually of concrete with a half-round or U-shaped troughfor the water. Tributary conduits intersecting above the grade of the main drain may be broughtdirectly into the manhole and flow allowed to drop inside (Fig. 6.5a), or a drop may beconstructed outside the manhole (Fig. 6.5b). The latter method is preferred where drops inexcess of 2 ft (0.6 m) occur in sanitary sewers to avoid splashing, which might interfere withwork in the manhole. Manholes are usually about 4 ft (1.2 m) in diameter at the conduit,

decreasing to 20 or 24 in. (0.5 to 0.6 m) at the top. If the conduit is less than 4 ft (1.2 m) indiameter, the manhole is usually centered over the pipe. For large conduits, .the manhole willspring from one wall of the pipe (Fig. 6.5~).

Manhole covers and cover frames are usually of cast iron, and the combin’ation willweigh 200 to 600 lb. (9Oto 270 kg). The light frames and covers are used only where subject tonegligible traffic loads, while the heavier combinations are employed on major highways.

6.2.5 System design

The urban drainage system consists of sewers, gutters, street inlets, and culverts. The designof each element is discussed below.

6.251 Storm sewer design

The design of storm sewers in large catchments is similar to that discussed earlier in section6.1.2 for small catchments, but estimation of flow in large catchments is often conducted usingcomputer simulation models such as the SWMM model (sec. 6.2.3).

6.2.5.2 Gutter design

Rainfall excess from a roadway watershed either is by overland flow or intercepted by someform of a gutter. Typical gutter shapes are of three types: Curb, V shaped, and curb withdepressed gutter. All three cross-sections are shown in Fig. 6. 6. Typical commercial areadesign is done using the curb type, whereas residential areas use both the curb and no-curb Vshapes. The no-curb type is primarily used where swales (ditches that both infiltrate andtransport) are used in a drainage plan, and usually average flow velocity is less than 2 fps. (0.6

mps). Details of design of the curb type gutter is given elsewhere (Wanielista et al., 1997)

1. Curb with stralghl cross-slope 2. NO curb. V-shaped gutter 3. Curb wth depressed gutter

Fig. 6.6 Typical gutter sections. (After FHWA, 1984).

In arid and semiarid areas, special considerations should be placed on the depositedsediments in these curbs. Such sediments should be controlled in order to have properdrainage.

From the top of a continuous slope to the end or bottom, the watershed area grows,and if the roadway width to the center line or cross-sectional slope peak remains constant, thearea increases linearly. As the area increases, the depth and width of flow increases. Velocityof flow in the gutter is not constant but will vary with the depth and width of flow. If thewatershed area vanes linearly with gutter distance, an average velocity may be at firstestimated as the velocity of flow at the midpoint in the watershed. This velocity is used toestimate time of concentration, and thus intensity of rainfall. Gutter design may be iterative.However, the gutter area must be capable of transporting the flow rate for any condition,otherwise overtopping or flooding will occur. Inlets to capture the water are designed toeliminate flooding.

99



6.2.5.3 Street inlet design

There are essentially two types of inlets for the collection of stormwater: a curb opening and agutter opening. These openings can also be used in combination. The typical curb and gutterinlets with grate are shown in Fig. 6. 7. The curb inlet is formed by an opening in the curb faceand generally has a depressed gutter section in front. The gutter inlet is formed by an openingin the gutter which is covered by a metal grate. These are usually square or rectangular inshape. However, in some applications, a slotted pipe that allows drainage to entercontinuously along its longitudinal axis is used (see Fig. 6. 8). The combination inlet is alsoshown in Fig. 6. 8.

Fig. 6.7 Perspective views of gutter and curb-opening inlets. (After FHWA, 1984)

Fig. 6. 8. Perspective views of combination and slotted drain inlets. (After FHWA, 1984)

Curb-opening inlets allow a capture of a specific maximum quantity and flow rate.They operate as weirs up to a depth equal to the opening height (h), then the inlet operates asan orifice at about 1.4 times the opening height. There is a transition section of flow from theweir to orifice flow rate

There are many different grate configurations that affect the “catch” efficiency.Empirical design procedures for interception capacity and efficiencies for seven grate types,slotted inlets, curb inlets, and combination ones are given in greater detail by the U.S. FederalHighway Administration, FHW A (1984). Also presented are additional details on roadwaygeometry, embankment, inlets, bridge deck inlets, unsteady-state flow, and IDF curvedevelopment.

6.2.5.4 Detention and retention storage fac ilities

Detention and retention storage facilities (ponds) have been extensively used in urban areas tocontrol runoff from developments. These facilities are normally designed to control the peakrate of runoff to the level which would have been expected from the site under predevelopedconditions. In most cases the design does not analyze the following:

The increased velocities at the outset of the facility.

100



The effects of the facility on the downstream channel system.The combined effects of several storage facilities within a drainage area.The design and use of detention and retention storage facilities concentrated on their

advantages including flood control, water quality improvement, and recreation and aestheticqualities. The negative aspects of these facilities include problems related to maintenance,water quality, aesthetics, and cost. In addition, the effec ts of these facilities on the downstream

channel system include increased velocity at the discharge, changes in the shape of thedischarge hydrographs resulting in higher downstream stages for longer time periods. Thiscreates streambank stabilization problems, and storing runoff for varying periods of time withinthe facility causing settling of suspected sediment which results in discharging runoff that willerode downstream channels in order to maintain its natural sediment carrying capacity.

Detention and retention storage facilities are important elements in any comprehensivestormwater management program but their design must take into account their effects on thedownstream channel system. Design must include measures to mitigate their negative aspectsand decrease downstream erosion and streambank degradation.

The effect of a detention facility on runoff flow rates can be shown most descriptively byobserving the inflow and outflow hydrographs for a detention pond, as shown in Fig. 6.9. Notethat although the total volume of runoff (area under the curves) is not reduced, the flow rateleaving the detention facility is significantly lower than the inflow rate. The objective, then, of

any on-site detention facility is simply to regulate the runoff from a given rainfall event and tocontrol discharge rates to reduce the impact on downstream drainage systems either natural orman made. Generally, detention facilities will not reduce the total volume of runoff but willsimply redistribute the rate of runoff over a certain period of time by providing temporary “live”storage of a certain amount o f the runoff. The volume of temporary live storage provided is thevolume indicated by the area between the inflow and outflow hydrographs ( Fig. 6.9).

The major benefti derived from properly designed and operated detention facilities isthe reduction in downstream flooding problems. Other benefits include reduced costs ofdownstream drainage facilities, reduction in pollution of receiving streams, and evenenhancement of aesthetics within a development area by providing the core of bluegreenareas for parks and recreation.

6.2.5.5 Stormwater culverts

Stormwater detention ponds are frequently connected together with culverts. A typical culvertis a hydraulically short (generally, less than a few hundred feet) conduit that conveysstormwater from one detention pond to another, or through the culvert, friction forces, inletlosses, and exit losses force water on the upstream side to pond at a deeper elevation. Thus,various inlet configurations and a variety o f materials are used in culvert design to decreasehead losses. In addition, there are a variety of culvert shapes as shown in Fig. 6.10. Theselected shape is based on construction cost, limitation on upstream water surface elevation,embankment heights, and flow-rate limitations.

Flow rates through a culvert may be improved by reducing friction losses at the inletside. Since upstream ponded areas and natural channel characteristics are usually wider thanthe culvert width (barrel width), there is a flow contraction at the culvert. A more gradual flowtransition will lessen the energy loss; leveled edges are more efficient than square ones, and

side or slope tapered inlets further reduce the flow contraction. Other factors affect flow ratesand headwater conditions, many of which must be examined when determining culvert sizes,inlet construction details, and headwater elevations.

&vales are vegetated open channels that infiltrate and transport runoff waters. Byincorporating the hydroiogic processes of runoff and infiltration. A swale design based onquantity is possible. Low velocities are important to prevent particle transport and loss of soil.The vegetation within the swales are very effective for the removal of solids and retention ofsoil. Erosion can be lessened by a vegetative area immediately after construction.

Swale volume must be available to contain the runoff waters. In highway designs forhigh speed situations, safety must be considered. Thus, a maximum depth of water equal toabout 1.5 ft. (0.5 m) and flow line slopes on the berms of 1 vertical/20 horizontal arerecommended. Along lower-speed highways or in some residential/commercial urban settings,steeper flow line berm slopes (1 on 6) are acceptable.

101



200

150

6ELs 100.z:!0

5C

c

Future condltlon runoff(Inflow to detentron has&n)

Existing cond4tion runoff

Outtlow fromdetcntlon basin

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21Time (hr)

Fig. 6.9. Typical detention system hydrographs

1 1

0 LJOircular Box (rectangular) Elliptical

Pips arch Metal box Arch

Fig. 6.10 Commonly used culvert shapes.

6.2.5.6 Infiltration ponds

Stormwater volume and pollutants in the stonnwater can be controlled by diverting thestormwater to infiltration ponds adjacent to the sewer line. The quantity diverted can beexpressed in terms of the depth of runoff over the entire watershed or the diversion of the runofffrom the first quantity (depth) o f precipitation. Examples of both criteria are to (1) divert the firsthalf-inch of runoff (over the total watershed) and (2) divert the runoff from the first inch ofrainfall. If the runoff coefficient for a watershed is 0.5, then the runoff from the first inch ofrainfall is the same as the first half-inch of runoff. Thus, the criteria may be consistent with oneanother.

The percolation pond can be built at less depth but occupy greater surface area for agiven volume of runoff. However, the less deep pond will infiltrate a depth (amount) ofstormwater in less time than a deeper pond because of the lesser depth. An infiltration rate of

1 in./hr can drain a 2Adeep pond in 24 hr, whereas a 4-ftdeep pond will take more than 24 hr.



Therefore, less deep ponds are more likely to recover storage capacity before the next runoffevent and the size of the pond to store the first specified amount of runoff from each and everystorm will be less.

6.3 Evaluation of urban drainage methods

There is a remarkable spatial variation in the frequency of high-intensity storms in arid andsemi-arid regions. Although the volume and intensity of individual (high-intensity) storms maynot be significantly different and the average discharge for similar sewer systems iscomparable, the number of sewer overflows may vary between 80 and 135% of the averagedepending on rainfall spatial variability. If combined sevvers are used, excessive hydraulic andpollutant loads reach the wastewater treatment plants causing several operational problems.The combination of collection, storage, and transport functions in these combined sewers issuboptimal from the water quality point of view. Some overflow wntrd structures such aswater sedimentation tanks may be used to handle combined sewer overflow. On the otherhand, use of separate stormwater sewers, although more costly, appear to be highly adequate.The stormwater could be discharged to detention pond systems and reused over the watershedfor irrigation. The reuse of stormwater reduces the vdume of discharged stormwater, thereby

decreasing the loss of a potential freshwater resource and decreasing the pollutant dischargefrom the system.

Since the majority of drainage facilities are for smaller drainage areas, there is theneed to adequately define the spatial and temporal distribution of local storms in arid and semi-arid regions. Rainfall induced floods may occur as a result of a severe storm over thecontributing watershed. These storms are classified as either local storms or general storms.Local storms are typically short duration, high intensity rainfalls of limited aerial distribution.General storms are large systems that are often associated with frontal activity. They are lowerintensity, longer duration storms that cover very large areas. Design of urban drainagesystems should, therefore, account for possible induced floods.

Stormwater pond systems are among the more adaptable, effective and widely appliedin urban drainage in developing areas. Their popularity can be attributed to their proven abilityto attenuate flows from design storms, economies of scale compared to other types of urban

drainage methods, high urban pollutant removal capability, community acceptance, and effecton adjacent land prices. In recent years, many communities have adopted regional stormwaterpond policies to achieve maximum stormwater benefits at the watershed scale at the least cost.A full description of the regional pond approach can be found in Hartigan (1986). However, theenvironmental impact of stormwater ponds should be evaluated in each case.

6.4 Traditional vs. modern methods

As indicated in chapter 2, the climate of arid and semi-arid areas is charactetized by rainfallvariability in both space and time coupled with high potential evaporation rates. This problem isaggravated by the highly localized nature of wnvectiie rainfall in and lands (UNESCO, 1977).Highly localized and intensive rainfalls produce surface runoff character&d by a high peak

discharge and high sediment loads. On the other hand, shortage of water resources is aserious environmental problem faced by all arid lands. It is appropriate then to adopt modem,rather than traditional, methods of urban drainage in order to achieve e ffective collection andutilization of stormwater runoff. On the other hand, modem rainwater catchment systemsinclude installation of roof systems which have distinct advantages of supplyinguncontaminated water directly to homes (i.e. water harvesting). More details are given inchapter 7.

6.4.1 Collection of stormwater

Traditionally, surface runoff is wllected along with domestic wastewater in a combined sewer.This was thought to be a less expensive way of stormwater collection but proved to be

troublesome and resulted in costly environmental problems. Use of a separate storm sewer

103



system to collect surface runoff is more cost-effective and allows for reuse of stormwater as avaluable water resource. The stormwater, as collected in a storm sewer, requires minimaltreatment and could be reused in irrigation. Modern materials and construction methods couldbe used to construct storm sewer lines knowing that stormwater is not corrosive and could beconveyed in sewers at minimal s lope.

In recent years there has been increasing concern for sewer maintenance and

rehabilitation as a long-term strategy. This concern will increase as stormwater systems takeon greater roles in water quality. Closed-circuit television (CTV) inspection is conducted todetermine the condition of the sewer interior and observe cracks and holes for quick repair.

6.4.2 Storage aid reuse of stormwater

Stormwater detention basins are storage facilities designed to control the peak rate of runoff inurban areas. Such storage facilities are important e lements in any comprehensive stormwater

management program. Their advantages as modern methods include flood control, waterquality improvement, and recreation and aesthetic qualities. Their importance as elements ofurban drainage in arid and semi-arid lands become apparent when increased urban runoffassociated with high-intensity storms occur in developed areas. In contrast, infiltration pondshave been traditionally used to reduce flooding by soaking of runoff into the ground. Thisresults in environmental pollution problems. Thus, infiltration ponds have not prove themselvesto be effective stormwater management facilities in many municipalities (Debo, 1995).A modern alternative to discharge of stormwater from wet detention pond systems is the reuseof the stormwater over the watershed for irrigation (Wanielista et al. 1997). The reuse ofstormwater reduces the volume of discharged stormwater, therefore decreasing the loss of apotential freshwater resource and decreasing the pollutant discharge from the system. Thestormwater can be used for a number of purposes including (1) irrigating open lands, (2)recharging groundwater, (3) supplementing water used for cooling purposes, (4) supplementingcar wash water, (5) enhancing and creating wetlands, and (6) supplying water for agriculturaluse.

The re-use pond differs from the typical detention pond in that instead of the temporarystorage being depleted using a discharge device, such as a weir or orifice, it is drawn downusing a re-use system. Even though the re-use system can be used to draw down the pond, a

discharge structure is still necessary for flood control.

Conclusions

The state-of-the-art of traditional methods of urban drainage in arid and semiarid climates isreviewed. In small size catchments, the rational method for estimating peak flowrates is

extensively used, but due to its idealistic assumptions (which can never be available in arid andsemiarid catchments) the accuracy of the produced results is unsatisfactory. Unless thecatchment size is small and the drainage design is preliminary, the rational method should notbe used. In large size catchments, existing computer models - especially the SWMM - areused and thought to be most applicable if the catchment is divided into small subcatchments,

and the model is applied to each individual subcatchment. However, due to problemsgenerated as a result of the high variability in infiltration and rainfall, and the long dry periodbetween two successive rainstorms, the accuracy of the produced drainage design is notsatisfactory. Further research is needed to realistically consider, fo r drainage design purposes,the processes of sheet erosion and excessive sediment transportation under the effect of flashfloods, and the high variability of infiltration and rainfall.

Bibliography

AGNEW, C. AND ANDERSON, E. (1992). Water Resources in the Arid Realm. RoutledgePublishers, London, England.

ASCE (1970). Design and Construction of Sanitary Storm Sewers. American Society of Civil

Engineers, Manuals and Reports on Engineering Practice No. 37, New York, NY.

104



ASCE (1982). Gravity Sanitary Sewer Design and Construction. American Society of CivilEngineers, Manuals and Reports on Engineering Practice No. 60, New York, NY.

CEDERGEN, H.R., O’BRIEN, K.H. AND ARMAN, J.A. (1972). Guidelines for the Design ofSubsurface Drainage Systems for Highway Structural Sections. Report No. FHWA-RD-72-30, Federal Highway Administration, Washington, D.C.

CHOW, V.T., MAIDMENT, D.R. AND MAYS, L.W. (1988). Applied Hydrology. McGrawHill,

Inc., New York, NY.DEBO, T.N. (1995). Urban Channel Systems - The Engineering Issues of Impact Mitigation.

In: Stormwater Runoff and Receiving Systems, Henicks, E.E. ed., CRC LewisPublishers, New York, NY.

FHWA (1984). HEC No.12, Federal Highway Administration, U.S. Departement ofTransportation.

FRENCH, R.H. (1990). Hydraulics/Hydrology of Arid Lands. American Society of CivilEngineers, New York, NY.

HARTIGAN (1986). Regional BMP Master Plans. In: Urban Runoff Quality - Impact andQuality Enhancement Technology, B. Urbonas and L.Roesner, ed., American Societyof Civil Engineers, New York, NY, 351-365.

LINSLEY, R.K., FRANZINI, J.B., FREYBERG, D .L. AND TCHOBANOGLOUS, G. (1992).Water-Resources Engineering. 4th ed., McGraw-Hill, Inc., New York, NY.

MACAITIS, W.A. (1994). Urban Drainage Rehabilitation Programs and Techniques. AmericanSociety of Civil Engineers, New York, NY.

METCALF AND EDDY, INC. (1971). Storm Water Management Model. Metcalf and Eddy,Inc., University of Florida, Gainesville, and Water Resources Engineers, Inc. Vol. 1,Environmental Protection Agency, Washington, D.C.

VIESSMAN, W., JR. AND LEWIS, G.L. (1996). Introduction to Hydrology. 4th ed., HarperCollins College Publishers, New York, NY.

UNESCO (1977). Development Of Arid And Semi-Arid Lands: Obstacles And Prospects. MABTechnical Note 6, UNESCO, Pans.

WANIELISTA, M., KERSTEN, R. AND EAGLIN, R. (1997). Hydrology: Water Quantity andQuality Control. 2”d ed., John Wiley 8 Sons, Inc., New York, NY.

105



Chapter 7

Sustainable solutions for urban drainage problems

in arid and semiarid regions

The previous chapters discussed variety of problems of urban drainage in arid climates.Problems related to data collection, to stormwater prediction, and to maintenance of drainagefacilities have been highlighted. It has been clarified that although stormwater runoff isnormally of small quantities in arid areas, its flashy nature can cause serious environmental

hazards. Proper management of such water can be an effective measure of environmentprotection and, in the mean time, can provide significant amount of water to be added tosustainable water sources. In this chapter, some sustainable solutions for urban drainageproblems would be discussed. These solutions are water harvesting, infiltration potential andnatural drainage.

7.1 Water Harvesting

Water harvesting is an old method used to capture rain water, store it, then make it availablefor different uses and particularly for drinking and cooking (Appan, 1997). It helps to savemoney on water bills and reduce the dependence on municipally supplied water. In arid andsemi-add areas, where shortage of water supply exists, the harvested water can be asignificant addition to available water sources. A rain water harvesting system consistsmainly of a collection area, a conveyance system, and a storage area. Figure 7.1 showsschematic presentation for a typical rain water harvesting system.

ee

bb b

b b b bb

bd

b b

Figure 7.1. Schematic presentation for a typical rain water harvesting system

196



The collection area is a surface, preferabiy impervious, that can capture and/or carry water towhere it can be used immediately or stored for later use. It can be either of the followings:. Roofs. Driveways. Parks. Contoured surface

. Furrows, Channel, Lakes, Pools . .etcThe size of the collection surface and its physical characteristics (such as absorption,

cover, slope, roughness . . etc.) are important factors affecting its efficiency.The function of the conveyance system is to carry water to the storage site. This can be

the drain pipes from the roof or channels on the ground, which may be lined with hardimpermeable surface or ordinary local soil material. The conveyance system can be fewmeters long to hundreds of meters depending on the distance between the collection areaand the usage area but the closer both sides the better from the point of view of water lossesand cost of construction.

The storage area can be as simple as a container such as a drum or a barrel placedunder a rain gutter downspot or it can more sophisticated such as a tank (steel, aluminum,concrete or fiber glass . . etc) on the ground with a shut off system and pumping facilities, alsofiltration may be added if the water to be used for human usage. The size of the storage

system depends on the required amount of water and the expected rainfall and the size of thecollection area. The longer the water is kept in storage the greater are the generatedproblems of stagnate, odorous, breeding of flies and insects, in addition to problems of healthhazards.

The storage system should be free from in/out leakage to avoid water infection and waterlosses, respectively. For individual houses, small capacity storage system ( lo-20 m3 ) maybe used. Due to the high cost involved in the construction and maintenance of storagesystems of harvested water for human usage, two different ways of storage may be adopted;one for human consumption which should be clean, hygienic, and another for gardening, floorwashing, toilets . etc. Obviously, the more sophisticated the system the higher the cost.

7. 1.1 Design of rainwater harvesting system

The quantity of harvested rainwater is mainly affected by the following factors:. Intensity and duration of rainfall. Size of collection area. The physical characteristics of the collection area. The rate of water losses by different ways (evaporation, leakage, infiltration, interception

. .etc). The storage capacity. Water demand

Considering zero losses, the amount of harvested rainwater can be identified by simplecalculations or from Table 7.1. The losses (such as the absorption of the collection surface,evaporation rate, and losses from the conveyance and storage systems) should be minimizedin order to maximize the efficiency of the rainwater harvesting system.

The quality of harvested water depends, among others, on the location of the storagesystem. Collected rainwater from roofs is generally higher in quality than the one receivedthrough municipal pipe, especially where the ground and the water are affected by chemicalsarid pesticides and many other pollutants. The salt content of the rainwater is in the range of30 ppm, compared to 360 - 500 ppm for city water and 2400+ppm for some well water. Thecomplexity of the harvested rainwater depends on the purpose of use. In places of high airpollution, the rainwater is not safe for drinking without treatment but it may be used for toiletflushing, agriculture, driveway washing . .etc.

7. 1. 2 Maintenance of rainwater harvesting systems

Maintenance of the rainwater harvesting system should be made frequently every 3-4 months

and consists of:

107



. Keeping gutters, pipes, channels and the storage screen clean from dirt, leaves, entry ofmosquitoes.. .etc.

. Checking the performance of the filters (in some cases the Ultra-violet lamp for watersterilization) and replacing when needed.

. The storage (tank) should be deaned frequently, especially when it empties in the non-rainy period.

. Periodic testing of water for the contamination by bacteria

. Leakage from the conveyance and storage systems should be examined frequently andproperly treated.

Table 7.1. Harvested rainwater volume “m3 y for different rainfall depths and differentsizes of collection area

RoioWI io mm

Sizeof cokctiosurface in m*

100

150

200

250

300

350

350

400

450

500

600

700

800

!wo

1ooo

-ii-2.53.75

5.0

6.25

7.50

8.75

8.75

10.0

ll.2!

.2.5C

15.0

17.5

--ii-7.5011.25

15.0

18.75

22.50

26.25

26.25

30.0

33.75

3730

45.0

52.50

loo

10.0

15.0

20.0

25.0

30.0

35.0

35.0

40.0

45.0

50.0

60.0

70.0

-is-

12.5

18.75

25.0

31.25

37.50

43.75

43.75

50.0

56.25

62.50

75.0

B7.50

00.0

12.50

.25.0

-iii

15.0

22.9

30.0

37.H

45.0

52s

52-H

60.0

67.54

75.0

90.0

m5.a

20.0

35.0

50.0

-iii-

17.50

26.25

35.0

43.75

52.50

61.25

61.25

70.0

78.75

87.50

105.0

22.50

40.0

57.50

75.0

200

20.0

30.0

40.0

50.0

60.0

70.0

70.0

80.0

90.0

100.0

120.0

140.0

60.0

80.0

,OO.O

7.1.3 Costs of rainwater harvesting systems

The installation cost of the rainwater harvesting system is low and varies from country tocountry depending on the materials used and the cost of labor. It is estimated3 (Fok, 1998)that if one constructs his own system the cost will be from US$ 10 to US$250/m , dependingon the materials used. In developing countries it may even cost very much less. The onlymajor cost will be the storage.

The cost has been found (WHO, 1981) depends on the user per capita annualincome, and is also strongly depends on the scarcity of water resources and the cost of otheralternatives (desalination, wastewater recycle or import). Table 7.2 shows the average costaccording to the affordable income.

The following Table 7.2 indicates that per-capita annual income less than US$50, theuser has no choice but to use ponds as the water storage tank or fetch water from othersources.

Table 7.3 shows a comparison of rainwater harvesting system to other methods. The

lower range of income is for self constructed or by teamwork. As in most developing

108



countries as well as in the and countries of the Middle East is more than US$ 150, thus mostof the people in these countries can afford the cost of rainwater harvesting system.

Table 7.2. Attributes of rainwater catchment system (WHO, 1981)

Table 7.3. Cost comparison of rainwater harvesting systems against other watersupply techniques (Lee et al., 1991)

Cooperative Effort Source of Water

7.1.4 Rainwater harvesting systems in arid and semi-arid areas

Due to the shortages of water supply, rainwater harvesting have been exercised in manycountries of arid climates. It has been reported (US National Academy of Sciences, 1974)that rainwater harvested in Israel with as little as 24 mm rainfall at Negev Desert and this lowrainfall yielded significant quantity of water. In Iraq, although there are a number of rivers butthe country suffers from fresh water in some regions, so people used to collect rainwater fromthe roof through gutters and pipes and then to the storage, which is any thing, that can beoffered. In Libya, in places of severe salt water intrusion exists and affect the quality o f water

supply, rainwater harvested from roofs in most cities. In Iran, rainwater harvesting was usedin many places to supply water for agriculture and human usage. The water storage, called“Abanbar” is shown in Figure 7.2. “Abanbar” is actually excavated cistern which is also called“Haze”. Stairs in the collection canal can have different types as shown in Figure 7.3. The“abanbar has openings on top to allow inside air circulation (Figure 7 .4). Figure 7.5 showsviews for the abanbar systems, which received their design long time ago but still in use insome parts of Iran to date. The most common shape of the “abanbar” is cylindrical, coveredby stones or bricks in the form of dome. In cases that amount of harvested rainwater is notsufficient to fill the cistern, water from river flow, spring water or “qanat” (long infiltrationgalleries) is directed into the “abanbar.” Similar systems are employed in many places in theworld. Figure 7.6 shows a typical rainwater harvesting system in Srilanka. It can be seen thatthe rainwater is diverted from the roof to two tanks. Water can be raised manually or by usinga pump for utilization. Feasibility study (Zuhair et al., 1999) have shown that rainwater

harvesting in the Arabian Gulf States can provide significant amount o f fresh water fordifferent usage. For example; In Kuwait: 12 % of water demand for landscape agriculture;

109



Oman (Muscat): 27% of water demand for industry; In Saudi Arabia (Abha):ll% of waterdemand for industry and landscape irrigation; In UAE (Al-Ain):l6% of water demand foragriculture.

Ventilation Cap

Yntilation Shaft

Doorway

-- . r

Stairs

Figure 7.2. Main components of the Abanbar

Figure 7.3. Some types of staircases in the Abanbar

110



-. .- . . . *. s . * .

Figure 7.4. Air circulation system in the Abanbar

Figure 7.5. Entrance to the Abanbar system

111



Figure 7.6. A Typical rainwater harvesting system in Srilanks

7. 1.5 Types of rainwater harvesting collecting systems

There are a variety of rainwater harvesting systems which can be systematically distinguishedaccording to their hydraulic properties as:

l The “total flow type” (Figure 7.7). The total runoff flow is confined to the storage tank,passing a filter or screen before the tank. Overflow to the drainage system only occurswhen the storage tank is full. It is important, that in the case of a clogged screen or filter,that there is no overflow before the tank.

Q Utilization

Qb b b

Figure 7.7. Total flow type rainwater usage system

+ The “diverter type” (Figure 7.8) which contains a branch installed in the vertical rainwaterpipe after the gutter or in the underground drainage pipe. The collected fraction isseparated from the total flow at this branch and a surplus is diverted to the sewerage

system; most of these branches contain a fine-meshed sieve diverting most of theparticles to the sewer. These devices are a typical invention of the period, when

112



rainwater usage was only looked on to save drinking water and the diversion ofstormwater to a sewer was the usual and accepted habit. The ratio of efficiency of thediverting devices decreases with increasing flow. So, during heavy rain, most of therunoff is diverted to the sewerage system. At low precipitation rates, a minimum flow isdiverted to the sewer and the efficiency decreases to zero (Graf, 1995).

Utilization

Qb b b

bbb b

:::::::::.___:*. ._VJmk :.

Figure 7.8. Diverter type rainwater usage system

l The “retention and throttle type” (Figure 7.9); The storage tank here provides anadditional retention volume, which is emptied via a throttle to the sewer (Mall-Beton,1999).

Q Qb b b

bbAb b

Utiliizition C-----r,

Retention volume

Consumption volume

( Throttled overflowSewer -

Figure 7.9. Retention and throttle type rainwater usage system

l The “infiltration type” (Figure 7.10). Local infiltration from the surplus tank overflow is apossible alternative to the diversion to the sewer. It has been shown (Herrmann et al.,1999a) that by the combination of rainwater usage and local infiltration, the natural localwater balance can be restored and maintained independent of the infiltration capacity ofthe soil, and independent of available surface for infiltration facilities.It has been found that the most effective cleansing process for roof runoff is natural

sedimentation in the storage tank. Therefore, the simplest method of treatment is to avoidturbulent mixing within the tank to prevent the sediment from mixing within the water column.A fine-meshed filter in the pressure pipe after the pump is not necessary and notrecommended. Chemical disinfection is absolutely not necessary and would only result in the

113



formation of carcinogenic chlorinated hydrocarbons if done by chlorine. The only furtherdevice to recommend is a sieve of 0.5 - 1.0 mm in the pipe before the pump, to preventresidues of the plumbing from entering the pump and the installation. After first operation, thesieve may be removed to reduce pressure losses in the suction pipe.

Utilization 4

n

. . Infiltration

Figure 7.10. Usage and infiltration type rainwater usage system

7. 1.6 Efficiency of rainwater harvesting systems

A long-term simulation (10 years of precipitation data) has been made (Herrmann et al.,1999b) to identify various hydraulic factors in connection with the rainwater harvestingsystems. In the simulation, the following expressions are called:Storaae tank (if situated underwound called cistern): Watertight tank for the collection andstorage of roof runoff.Storaae volume or tank volume: Volume of a tank which is emptied only by consumption of

rainwater.Retention volume: Volume in a tank which is emptied continuously via a throttle outletindependent of consumption. The retention volume does not serve for consumption purposesbut is for buffering and retention of peak flows of the roof runoff.Service water or cistern water: The water taken out from a rainwater storage tank, pumpedand distributed by pipes in a building for consumption. When there is a lack of rainwater, thestorage tank or a special refill tank before the pump is filled up by drinking water refill from thepublic network.Rainwater: This is the portion of service water derived from roof runoff, the rest is calleddrinking water refill (refill).Roof runoff (runoff): Volume of the total runoff from the connected roofs.Overflow: When the storage tank is full, the surplus of roof runoff is diverted as overflow.Drinkina water: The water consumption of a system connected household or unit taken from

the public network, including the refill.Svstem efficiencv: The percentage of service water provided by collected rainwater.Collection efficiency: The percentage of the roof runoff which is consumed.Effective area “4:: The horizontal projection of the roof.

Based on the assumption that the average water consumption in household isdependent on the age of installation, household appliances and varies between 100 and 145liter per capita per day, the simulation provided results, which are summarized in Figs 11 to13. Figure 7.11 shows the relationship between system efficiency for a range of servicewater consumption rates, storage volumes and roof areas. The results were used to calculatethe specific daily consumption rate, in mm/d, defined as the daily service water consumptionto the effective roof area.

Further, the storage volume is related to the roof area , to give the specific storagevolume, which is used for the sizing of centralised stormwater retention tanks. The derived

data set is represented in Figure 7.12.

114



50

012345678 012345678

J! i I0 2 4 6 8 10 12

1

0 2 4 6 8 10 12 14

Figure 7.11. Relationship between system efficiency and storage volume

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4

Specific rainwater consumption [mm/d]Figure 7.12. System efficiency and specific consumption for various specific storage

volumes (Herrmann et al., 199913)

lspection of Figure 7.12 reveals that the system efficiency depends on the specificconsumption rate, but does not depend on the roof extent. For a private household, theaverage drinking water saving will be between 30% and 60% using a 4 - 6 m3 tank,depending on the consumption habits and the available roof area. When the maximumsystem efficiency is desired, the necessary tank volumes can be taken from Table 7.4. Figure7.13 gives the measured consumption data of a house, where two adults and two children livein Germany.

115



110

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 22 2.4

Spccil3crainwatcrconsumption [n&d]Figure 7.13. Ovefflow and specific rainwater consumption for various specific

storage volumes (Herrmann et al., 1999b)

Table 7.4. Tank volumes for maximum system efficiency

7.1.7 Modelling of rainwater collection systems

There are three types of rainwater reservoir sizing models. These are the critical period,Moran and behavioral models. Critical period methods identify the use sequences of flowswhere demand exceeds supply to determine the storage capacity. The sequences of flows ortime series used in this method are usually derived from historical data. In Moran method(Moran, 1959) a system of simultaneous equations are used to relate reservoir capacity,demand and supply. The analysis is based upon queuing theory. Behavioral modelssimulate the operation of the reservoir with respect to time by routing simulated mass flowsthrough an algorithm which describes the operation of the reservoir. The operation of therainfall collector will usually be simulated over a period of years. The input data which is intime series form are used to simulate the mass flow through the model and will be basedupon a time interval of either a minute, hour, day or month. More details of these models canbe found elsewhere (McMahon and Mein, 1978).

The general wnfigration of a rainwater collection system is illustrated in Figure 7.14.

As described by Jenkins et al. (1978) two operation algorithms can be identified; namely, theyield after spilling “YAS” and the yield before spilling “YBS”.

116



b

AI

yt s v

‘1 A\ Y

c

Figure 7.14. Illustration of rainwater collection operating systems

The YAS operating rule is:

-c-

DtYt = min

N-1

(7.1)

Vt-, +Qt-YtVt = min (7.2)

s - Yt

where R1 s the rainfall (m) during time interval “t”, Qt , Vt , Y r and D t are the rainwater runoff(m3 ),volume in store (m3 ), yield from store (m 3 ) and demand (m 3), respectively, during thetime internal. S, and A are store capacity (m 3, and roof area (m2), respectively.

The YAS operating rule assigns the yield as either the volume of rainwater in storagefrom the preceding time interval or the demand in the current time interval whichever is thesmaller. The rainwater runoff in the current time interval is then added to the volume ofrainwater in storage from the preceding time interval with any excess spilling via the overflow

and then subtracts the yield.

The YBS operating rule is:

-=E

QYt = min

M-7+ Qt(7.3)

-c

Vt-, + Qt -YtVt = min (7.4)

S

The YBS operating rule assigns the yield as either the volume o f rainwater in storage from thepreceding time interval plus the runoff in the current interval or the present demand whicheveris the smaller. The rainwater runoff in the current time interval is then added to the volume of

the rainwater in storage from the preceding time interval before subtracting the yield and

117



allowing any excess to spill via the overflow. More details about these operating rules andtheir applications can be found elsewhere (Jenkins et al., 1978; among others).

7.2 Infiltration potential and natural drainage

As described in the previous chapters, soils of arid catchments are subject to fluctuations indaily temperature and this leads to its disintegration. This increase the infiltration potential ofthese catchments to the degree that they can be used as sustainable solutions for urbandrainage problems. To avoid the hazards of flash floods, water can be directed to basins,ditches, bonds . . etc in these catchments where it will infiltrate and recharge groundwaterreservoirs. This process can be one or more of the following benefiis:

. Protect urban facilities from the hazards of flash floods

. Conserve and dispose of runoff and flood water

. Eliminate the decline in groundwater levels

. Reduce (or prevent) salt water intrusion

. Allow heat exchange by diffusion through the groundThe methods used for the above process are very much the same used for direct

groundwater artificial recharge. In these methods, flood is directed to the permeable groundsurface where it infiltrates through the unsaturated zone to slowly reach the undergroundwater table. Popular techniques include spreading and ponds. Spreading techniques are byfar the most widely used methods in the and areas of the Middle East. They are classicallydefined as including the following: 1) basin; 2) furrow or ditch; and regulated wadi (orflooding).

Basin is the most popular of all spreading methods. It is used in Oman and SaudiArabia. It requires the building of dikes or small dams (rockfill or earth dams 0.25 to 0.5 mhigh, depending on the expected vdume of flood) spaced at regular intervals (10 to 15 m).Stormwater is directed to the first basin and then it spills from one basin to the next one.

Furrow (or ditch) spreading is suitable in rough or sloping terrain where they can belaid with non-silting slopes. It has been used in south of Oman to divert rural floods. Themaintenance cost of this technique is less than that of the basin spreading.

Wadi spreading is accomplished by widening a drainage channel to accommodatethe flash floods of short duration. It has been reported successfully in many arid places in theworld.

In case of low vertical permeability of soils, flash stormwater runoff can be directed towhere it can be taken by a recharge well to groundwater reservoirs. The recharge well ishaving similar characteristics to a pumping well. It can be of gravel-pack or non-gravel-packtype depending on the geological formation o f the well location. The major problem o f suchtype of well is the clogging caused by air entrainment , presence of suspended material,growth of micro-organisms and chemical reaction between the recharge stormwater and thegroundwater.

Pits and shafts are other alternatives to recharge wells. They are shallower andrelatively larger than wells. They are opening, usually excavated by a dry excavation andterminated some distance above the water table. They may be cased, uncased, or back-filledwith gravel, and may be round or rectangular in cross-section. The purpose of the shaft is to

transmit the directed stormwater from the surface to groundwater reservoirs by passing slowlypermeable layers in the soil or upper geologic strata.

The popularity of the pits is mainly due to its low wst. The basic factors for asuccessful pit are correct shape and high permeability formation. Successful pits are where athick layer of slowly permeable material exists near the surface. On the other hand, pits andshafts suffer from the same problems restricting well recharge, namely clogging.

7.2.1. Comparison between different direct techniques

The best method is the one that can maintain a high infiltration rate at an economical leveland with sustained desirable quality of water. With that definition in mind, it would be difficultto identify a particular method as being the best for all locations and at all times. Though,

spreading techniques may be described as the most widely used ones because of the

118



following advantages: 1) It requires smaller area; 2) Discharges water is of low turbidity andsilt content; 3) Ability to recharge confined aquifers; and 4) Undesirable mixing of recharged

and pumped waters can be prevented by locating wells at different depths.Spreading through basin is considered by many sources as the most commonly

practiced technique. However, such a practice may not be feasible in areas of compactedlayer which forms a barrier preventing direct downward percolation. For such a case, well or

trench recharge would form two of two possible alternative methods. Trench recharge isfavored in the Middle East, being more efficient hydraulically and economically in case of veryshallow subsurface barriers. However, wells are more competitive as the thickness of the

barrier increases.

7.2.2 Managerial aspects

To achieve optimum benefits from any of the above techniques, other than engineeringfactors, economical, social, health, legal, and political factors must be considered. Luckily the

literature contains considerable amounts of publications on some of these aspects especiallythe engineering, economical, and legal factors. Unfortunately, most of the developments in

these fields overlooked the interaction between these factors and their influences on the

performance of recharge systems. For example, proper location of the water disposal orrecharge sites would require the inspection of the following factors: the characteristics ofdiverted flash stormwater, factors affecting infiltration rates and movement of water within the

water table (topographical, hydrogeological, hydraulics . . .etc), operation and managementproblems, water quality consideration, economic consideration, social constraints, legal

aspects, and health hazards.Engineering aspects are meant to inspect the influence of various engineering

factors on the selection, design, operation and maintenance of stormwater disposal and waterrecharge systems. Inspection of engineering aspects by undertaking a pilot project beforeindulging in an expensive large scale installation could be recommended as a good practice.Pre-construction investigation could also be attained through numerical methods. However,the factors that would mean most, for avoiding the hazards of flash stormwater runoff, can bedescribed as being the physical and chemical characteristics of floods as well as the

geological condition of water disposal area.Hydrogeolog ical and groundwater considerations have considerable influence on the

selection of a stormwater disposal scheme. The extent of these influences depend on factorssuch as the relationship between the physical and mechanical properties of the pervious

media and desirable infiltration rates, the efficiency of the constituent material to effect or toassist chemical and biological improvement, the ability of the water bearing deposits to storerecharge water temporar ily and subsequently to permit groundwater movement at acceptablerates and over adequate retention times and the efficiency of abstraction works to recover the

recharge water (Ineson, 1970).Economic aspects re fer to the economic evaluation of the stormwater disposal

system. It is considered as vary important for rational resources management. However,numerous difficulties can be met towards real applications of that aspect. These difficulties

are mainly related to the wide variations in estimating the values of damages due to flash

stormwater floods, and the water values with respect to its various utilizations. Nevertheless,items contributing towards economic evaluation include: 1) Tangible benefits to considercosts of damages due to flash floods; 2) Intangible benefits, mainly social benefits; 3) Value of

water; 4) Direct benefits of recharged water; 5) Institutional and financial arrangements; and6) Cost components of the water disposal and recharge systems.

Conclusions

To overcome the problem of water shortage and, in the mean time, to avoid the damages duej to flash floods in arid and semiarid areas, some sustainable solutions are offered. The

solutions consists of using the techniques of rainwater harvesting in the urbanized areas andadopting the methods of water spreading over the infiltrating surface of catchments.

119



Aspects of design and maintenance of rainwater harvesting systems are reviewed withrecommendations to ensure proper operations. Due to the distinguished climatecharacteristics (high concentrations of atmospheric dusts together with large variations intemperature, humidity and rainfall amounts) in add climates, periodic testing of harvestedwater together with periodic cleaning and testing of various parts of the rainwater harvestingsystem are highly advisable. In addition, a guide to the estimated costs is given. The

rainwater harvesting systems, being implemented in many parts of the world, have beenfound to be a substitute to sign&ant amounts of water in the Middle Eastern and countries.Unpaved surfaces of catchments are of high infiltration rates. This is, as described in

chapter 3, due to the destruction of soils and soil cracks generated as a result of seasonaland daily fluctuations in temperature in addition to the long dry period between twosuccessive rainstorms. Such characteristics can be utilized in a useful way. Flash floods canbe directed to these infiltrating surfaces, spread over them, and allow floodwater to infiltratethrough the soils to the groundwater reservoir. This way, floodwater can be savedunderground and damages due to the flood can be prevented to a great extent.

Both rainwater harvesting and flood spreading over the infiltrating catchment surfacescan contribute significantly with considerable amounts of water to available water resources inarid climates.

Bibliography

APPAN, A. (1997) “Rainwater-Catchment System Technology: Concept, Classification,Methodologies And Application,” first Brazilian Conference on Rainwater Catchment,November.

FOK, Y. S. (1998) “Rainwater Catchment Systems Development Guidelines,” Proc. The 25rhAnnual Conference on Water Resources Planning and Management, June, Chicago,Illinois.

GRAF, 0.(1995), “Regenwasserfiltereinrichtung, Gebraauchsmuster, Rollennr. 295 02 895.5Vol. 24, Deutsches Patentamt.

HERRMANN, T., KAUP, J. AND HESSE, TH.(1999a), “Innovative Water Concept Applied AtAn Urban Multi-Story Building,” Proceedings, 8’” International Conference on Urban

Storm Drainage, Vol. 3, Sydney, 30 August - 3 September, pp. 1296-1303.HERRMANN, T. AND SCHMIDA, U.(1999b), “Rainwater Utilisation In Germany: Efficiency,Dimensioning, Hydraulic And Environmental Aspects,” Urban Water, Vol. 1, No. 4, pp.307-316.

INESON, J. (1970) “Hydrogeological And Groundwater Aspects Of Artificial Recharge,”Conference on Groundwater, University of Reading, September.

JENKINS, D., PEARSON, F., MOORE, E ., SUN, J. AND VALENTINE, R. (1978) “FeasibilityOf Rainwater Collection Systems In California,” Contribution no. 173, California WaterResources Center, University of California.

LEE, D. J., LEUNG, P. P., FOK, Y. S., AND CHU, S. C (199l),“Opportunity For RainwaterCistern System In Rural Economic Development,” Proc. 5’h International Conferenceon Rainwater Cistern, Keeling, Taiwan, China, August.

MALL-BETON (1999) “Personal Message On November Fifth,” Mall-Beton GmbH, HufingerStr. 3945, D-78166 Donaueschingen.

MCMAHON, T. A. AND MEIN, R. G. (1978) “Reservoir Capacity And Yield, “ Development inWater Science, Amsterdam: Elsevier.

MORAN, P . A.(1959), “The Theory Of Storage,” London: Methuen.UNITED STATES NATIONAL ACADEMY OF SCIENCES (1974) “More Water For And

Lands,” National Academy of Sciences, Washington D. C., Library of CongressCatalog No. 10058.

WORLD HEALTH ORGANIZATION (1981) “The International Drinking Water Supply AndSanitation Decade Directory,” Thomas Telford Ltd., London.

ZUHAIR, A., NOUH, M., EL-SAYED, M. (1999) “Flood Harvesting In Selected Arab States, “Final Report No. M31/99, Institute of Water Resources, 217~.

120



Chapter 8

Urban drainage maintenance and management issues

in arid and semiarid regions

8.1 General

The efficacy of urban stormwater management systems is fundamental to sustainable operation bothas a stormwater drainage system as well as an ecologically-based stormwater system. Themaintenance and management practices thus need to be carefully considered when formulating astormwater management strategy for a given urban catchment. The key to sustainable operation of astormwater management system lies in the proper selection and design of the various components ofthe system. Discussion on design considerations are thus included in this chapter on urban drainagemaintenance and management issues.

Traditional approaches to stormwater management have been based on a singlemanagement objective that considers stormwater as a source of potential hazard to public safety.Stormwater management was essentially that of stormwater drainage using two general methods, ie.

(i) conveyance of stormwater to receiving waters in an hydraulically efficient manner; and (ii) detentionand retardation of stormwater. Maintenance and management practices are thus traditionally directedat ensuring that the various components of a drainage system retain their respective design dischargecapacities. Typical maintenance activities include channel bank protection, channel and basindesilting and removal of debris following flood events. In arid and semi-arid areas, the mostprominent issue in the maintenance of stormwater drainage systems is that of sediment and grosspollutant control, often involving a higher frequency of stormwater drain inlet clean-outs and scouringof stower pipes.

A stronger emphasis on the managing urban stormwater for multiple objectives, includingstormwater quality improvement and ecosystem protection, has led to the development of ewlogical-based stormwater systems. These systems are significantly different from conventional systems andrequire different maintenance and management practices. For example, constructed wetlands andinfiltration systems have been shown to be effective in removing stormwater pollutants. The efficacy

of constructed wetlands is reliant on the sustainable role of aquatic macrophytes for stormwatercleansing. The efficacy of stonnwater infiltration systems is directly related to the maintenance of thehydraulic conductivity of the infiltration media. There are also many maintenance and managementissues that are not related to the operation of these systems as stormwater management measuresbut are necessary to ensure that they remain an integral part of the urban landscape and continue toenhance the amenity o f the surrounding urban environment.

Fundamental to the sustainable operation of these system is,their appropriate selection. Inarid and semi-arid areas, the provision of adequate catchment runoff to sustain vegetated systemssuch as wetlands and grass swales or wet detention systems requires careful consideration. And andsemi-and regions are characterised by low rainfall and high evaporation, with many regions having adistinct dry and wet season with the majority of the mean annual rainfall occurring in a small numberof high rainfall events. These conditions are thus less conducive to stormwater systems with largepermanent pools and dense vegetation. In selecting the most appropriate stormwater management

practices fo r individual catchments, it is not enough to just consider the mean annual rainfall of theregion but also the seasonal variation of rainfall.

121



Vegetated systems can often be sustained in semi-and areas where the seasonal variation ofrainfall is not excessive. In most cases, the catchment area to individual stormwater treatmentelements would tend to be larger than in tropical or temperate regions. This may result in lessemphasis on near source control systems and stormwater management systems may involve acombination of convention stormwater drainage elements (eg. pipes and pits) reticulating intostomnnrater quality improvement elements in the vicinity of the receiving waters or aquifer recharge

zone.In arid areas, there would be a tendency to adopt infiltration systems with an emphasis on water

conservation through aquifer storage and recovery schemes (see chapter 7).

8.2Stormwater detention and retention systems

The provision o f detention and retention basins is one technique available to manage stormwaterrunoff. A distinction is made here regarding stormwater detention and stormwater retention from asurface stormwater drainage context. Stormwater detention is defined as the temporary storage ofstormwater (ie. detention) for subsequent discharge, at a lower rate, to the receiving waters.Stormwater retention is the removal, by infiltration to groundwater or by evapotranspiration, of

stormwater and thereby preventing their discharge to the receiving waters.The key operating mechanism of these systems is the temporary storage of stormwater.

Their efficacy is affected by the storage capacity and discharge capacity of the outlets. Solids,ranging from gross pollutants to fine particulates, are transported in stormwater and a prominentmaintenance issue in detention and retention systems is the management of the long-term effects ofaccumulated sediment and gross pollutant loads on the stormwater detention or retention system.

Stormwater retention systems can indude Aquifer Storage and Recovery (ASR) schemespractised in arid and semiarid areas where stormwater is either injected, or allowed to percolate, intogroundwater aquifers for subsequent recovery by pumping during the drier periods. These systemsare designed to operate for frequent storm events and have little effects on peak discharges of largeevents unless they form part of a flood retarding basin. Stormwater retention systems are perhapsbest suited for such source areas as roofs and car parks with pre-treatment of runoff from thesesurfaces for removal of gross pollutants and sediment providing a means of reducing the maintenance

requirements of the retention system.

8.2.1 Flood retarding basins

Stormwater detention system for flood management commonly adopted in stormwater managementpractices range from small on-site detention systems (discussed in Section 8.2.3) to large regionalstormwater retarding basins. The use of regional stormwater retarding basins is wmmon in urbancatchment management and these basins can either have a permanent water storage component (ie.a wet detention basin) or are complete ly dry during non-flood periods (ie. a dry detention basin), withdry retarding basins being more appropriate in arid areas. Both types of basins have the potential toserve multiple objectives in addition to their primary flood mitigation function.

Flood retarding basins are often designed to operate optimally under large storm events witha wrnmon design criterion being the attenuation of the 100 year Average Recurrence Interval (ARI)post-urbanisation peak discharge to a level comparable to the 100 year ARI pm-urbanisation peakdischarge. In catchments with sandy soils, infiltration basins have been utilised as flood retardingbasins owing to the high infiltration rates of the in-situ soils.

From a hydrologic and hydraulic petfomance perspective, the maintenance andmanagement issue for these systems is principally one of maintaining the hydraulic characteristics ofthe basin outlet by preventing blockage of the outlet structure by flood debris or the clogging of theinfiltration basin. The use of trash racks for prevention o f blockage of outlet structures by flood debrisis standard practice in retarding basin design and regular inspection of the outlet structure shouldform part of the management plan for the basin. For infiltration systems, the planting of vegetation onthe floor of the basin can often reduce the risk of clogging of the infiltration media with the rootsystems of the vegetation continually maintaining the porosity of the underlying soil. Maintenance

considerations under such circumstances would include the need to sustain the vegetation over

122



extended dry periods. Other maintenance issues include regular inspection of the embankment andspillway for indication of piping failure.

Dry detention basins often serve as playing fields and recreation parks in addition to its floodmitigation function as illustrated in Figure 8.1. A traditional design practice in dry detention systems isthe provision o f a low flow channel/pipe to by-pass dry weather flows and runoff from frequent events

thus avoiding the frequent inundation of the storage area. Management of sediment deposition in theretarding basin in the vicinity of the inlet to the retarding basin is also a wmmon maintenancerequirement. Associated with the conjunctive use of the basin is the need to maintain suitable grasscover of the basin.

Wet detention basins are wmmonly~utilised for water pollution control, to satisfy ewlogicaland conservation objectives, and provide public passive recreational amenities. Common wetdetention basins in urban catchments are ponds and wetlands build within retarding basins asillustrated in Figure 8.2. In the design of these systems, it is imperative that consideration be given toensuring a right balance between the volume of catchment runoff and that of the permanent pool toensure that the pool would not dry up frequently. Maintenance and management issues of thesesystems are more complex than conventional retarding basins owing to the need to ensure aminimum water quality standard in the permanent waterbody. In addition to the requirements toensure the hydrologic and hydraulic functions of the system, common water quality managementobjectives for wet detention basins include maintaining a high dissolved oxygen level in the water and

reducing gross pollutant (particularly anthropogenic litter), organic loads (eg. leaf litter, sewage etc)and hydrocarbon inflow to the permanent water-body.Maintaining a regular inflow to the permanent water-body of a wet detention pond is an

important management technique to ensure regular flushed and well mixed water-body. In somecases, it may be necessary to artificially mix and aerate the water-body. A wmmon design over-sightis the incompatibility of the catchment area to pond area, with the pond area being too large for asustained permanent waterbody both in terms of water quality and hydrology. The consequence ofthis over-sight is the tendency for stagnant water-body, low dissolved oxygen and algal blooms.

Managing wildlife in the waterbody is sometimes necessary to avoid excessive organicloading of the system which may lead to depleted dissolved oxygen and anaerobic conditions at thewater/bed sediment interface leading to contamination of the waterbody by metals and nutrientreleased from the sediment.

123



#2.2

Figure 8.2igure 8.2 An urban pond within a retarding basin (Wet Detention Basin) cann urban pond within a retarding basin (Wet Detention Basin) cansatisfy a number of other stormwater management and landscapeatisfy a number of other stormwater management and landscapeobjectives.bjectives.

Stormwater quality improvement facilities

The need to address the quality of urban stormwater runoff in recent times have led to the adoption ofstormwater quality best management practices involving the use of constructed wetlands andinfiltration systems as integral part of the stormwater management system. Constructed wetlands are

less suited to arid regions but are widely used in semi-arid regions for stormwater quality control.

Constructed wetlands are systems designed to detain stormwater for an extended period to enableremoval of pollutants, particularly those associated with fine suspended particulates. These systemsare typically most appropriate for higher rainfall regions but have been extensively used in semi-andregions for stormwater quality improvement in Australia. In semi-arid regions, these systems aredesigned as predominantly ephemeral wetlands which fills during storm events and drains completely(except for a small permanent pool area) a few days after the storm event.

Wetland systems are often designed to operate effectively for frequent storm events and largeevents are often allowed to by-pass the vegetated zone of the wetland to prevent resuspension ofdeposited material and scouring of the system (Wang et al., 1998). Figure 8.3 shows the impact ofsediment load on a constructed wetland without a high flow by-pass flowpath.

Wetland systems are low maintenance systems and not no maintenance systems. Careful designof these system can significantly reduce maintenance requirements and cost. Partitioning oftreatment components in a wetland system allows for maintenance of individual components to betargeted, eg. gross pollutant trap followed by the inlet zone/sedimentation basin, and then themacrophyte zone.

gross pollutant trap - designed for the removal of natural and anthropogenic gross litter. Grosspollutant loading in urban catchments can be high and maintenance frequency of gross pollutanttraps is often in terms of months. Figure 8.4 shows a typical gross pollutant load generated froman urban catchment following a storm event, with the expected gross pollutant load generatedfrom a typical urban area is of the order of 0.4 m3/ha/yr (Allison et al., 1998). The maintenanceoperation is dependent on the type of trap and the pollutants removed can normally be safelydisposed of in municipal landfills. Gross pollutant traps with a permanent pool can cause odourproblems and maintenance frequency may need to be increased. Easy access for frequent andefficient maintenance operation is an important design consideration in siting gross pollutant

traps.

124



Figure 8.3 txcesswe slltatton ana scounng of a constructeo wetlana owrng IO lnaoequare prowsronfor high flow management.

IFigure 8.4 Gross pollutants generated from a typical urban catchment during a stc m event.

in/et zone/sedimentation basin - designed for sedimentation of coarse to medium size particles.Typical sediment load from an urban catchment is of the order of 0.5 m3/haQr, but can be as high as1.5 m3/ha/yr, with the majority of the sediment being in the coarse to medium size fraction.

125



Catchments in and and semi-and regions have been known to produce higher sediment loads owingto inadequate vegetative cover of open spaces and bank erosion of ephemeral waterways. Theremoval of the coarse to medium-sized fraction of sediment exported from the catchment is animportant pm-treatment of stormwater prior to its discharge into the vegetated zone of the wetlandsystem. This prevents excessive smothering o f the wetland and allows the finer particulates to be

removed by the vegetation in the wetland through the mechanisms of enhanced sedimentation,filtration and surface adhesion promoted by the wetland macrophytes.It is well established that the majority of urban stormwater contaminants are associated with

the finer fraction of suspended solids. The expected maintenance frequency of the inletzone/sedimentation basin of a wetland system is between 5 to 10 years, depending on the geologyand level and maturity of development in the catchment. The maintenance operation involves the useof mechanical excavation of deposited sediment and vehicle access is an important designconsideration of the inlet zone. Deposited coarse to medium-sized sediment can be disposed of inmunicipal landfill.

Care needs to be taken to ensure that the sedimentation basin is not overdesigned, resultingin longer than required detention period. Longer than desired detention period promotes settling offiner material and associated contaminants (eg. metals) as well as the deposition of excessive organicmaterial leading to possible reduced redox potential in the sediment and subsequent release of

sediment bound contaminants. Excessive deposition o f fine particulates may result in higher thanacceptable metal concentrations in the deposited material and consequently required special disposalconditions.

2 macrophyte zone - designed for trapping and settling of fine particulates using wetlandvegetation to promote enhanced sedimentation processe s (Wang et al., 1999) and typicalmaintenance operation o f this zone includes weed control and removal of dominant macrophytespecies which may alter the hydrodynamic flow character istics of the wetland. The removal ofdeposited material and vegetation biomass is expected to be at a frequency of between 15 to 25years and deposited sediment can be expected to have elevated metal and nutrientconcentrations and may need to be disposed of as prescribed waste. Water level manipulationsmay be necessa ry as a means of controlling excessive dominance of macrophyte species as wellas promoting the rapid degradation of organic matter.

In constructed wetlands, the major long term management strategy for vegetation is to ensurethat the different vegetation zones receive a hydrologic regime that will allow the target specie(s) for aparticular zone to survive naturally and have a competitive advantage over potentially invasivespecies. A major element of vegetation management is to ensure as natural a hydrologic regime aspossible. Plants selected for the wetland should be predominantly native species. Most naturalhydrologic regimes are variable and result in water level fluctuations and wetting and drying cycles.The normal water level of a system has to be able to vary up, but particularly down, in a relativelyseasonal way to ensure good vegetation wver and stability. In constructed wetland system thedesign of the outlet structure is critical to achieving variable hydrologic regimes that are well matchedto the requirements of vegetation. Weir outlets tend to miminise water level variation, whereasperfora ted riser outlets and siphons maximise variation and allow some wntroi of the hydrologicregime and the wetland vegetation.

Harvesting of emergent aquatic macrophytes in stormwater treatment systems is not required as

a pollutant removal mechanism. The major role of vegetation in pollutant removal during event flowsis its role in enhancing sedimentation processes and providing surface area for the trapping andfiltering of fine particles. During low flows vegetation provides a surface for the growth of biofrlms.Pollutant uptake and transformation by bioflms is an important stormwater treatment process duringlow flows. Consequently the harvesting of vegetation could potentially decrease the treatmentperformance of systems.

However some large, deepwater species, may over time develop a canopy that can contain asubstantial amount of senescent, standing biomass that may limit new growth and result in a patchyvegetation distribution. Consequently harvesting may occasionally be required to ensure vigorousand even growth across the flow path. Elecause of the vital role of vegetation in stormwater treatmentsystems it is not advisable to harvest the whole system at one time, should harvesting be required.The system should be progressively harvested over time to ensure uniform hydraulic resistanceacross the flow path and to encourage vegetation diversity.

126



Infiltration systems have been used as detention systems where stormwater is infiltrated through afilter media (eg. gravel trench) but subsequently collected by perforated pipes beneath the filter mediafor discharge to the receiving waters. Like constructed wetlands, such systems are designed tooperate effectively for frequent storm events with design standards often being in the 0.5 year to 2year ARI range, treating up to 90% of the mean annual stormwater runoff volume. Best practices instormwater management could include the linking of these detention systems in series to promoteboth water quality improvement and flood mitigation outcomes. Infiltration systems are the mostcommonly adopted stormwater best management practice in arid regions and are often incorporatedinto groundwater aquifer storage and recover schemes to optimise water conservation andstormwater recycling.

Maintenance requirements of infiltration systems can be significantly reduced with proper pre-treatment of stormwater runoff prior to their discharge into these systems. The key managementissue is that of sediment loads and the consequential clogging o f the infiltration medium. Sedimentload generated during land development is significantly higher than from a matured urban catchmentsand infiltration system should ideally no be exposed to sediment load generated during thewnstruction phase of land development. This requires implementation of environmentalmanagement plans for construction sites which aim to retain sediment on-site using sedimentationbasins, geotextile filters and grassed buffer strips. The maintenance of infiltration basin operating in amatured urban catchment would involve periodic (eg. annual) tilling of the top 5 cm of the infiltrationbed to break up the organic crust formed on the surface (see Figure 8.5). Often, the introduction ofsurface vegetation (either grass or native plants) can help maintain the permeability of the infiltrationbed.

Infiltration systems for road and car-park runoff management have often been designed withlandscape functions as shown in Figure 8.6. The watering of grass wver to maintain aesthetic valueswill also yield beneficial stormwater quality outcomes.

Figure 8.5 Organic crust formed on the bed surface of an on-site infiltration system bed can leadto clogging of the system.

127



8.2.2.1. On-site detention basins

On-site detention basins are localised systems built within individual properties and are oftenassociated with urban consolidation of an already urbanised catchment. Often the existingstormwater drainage infrastructure cannot be upgraded without significant expense and urbancatchment managers have to resort to source control of increased peak discharge at the individualbuilding lot level. These systems often deteriorate in their effectiveness over time owing to poormaintenance with the most common contributing factor being the siltation o f the system and cloggingof the outlet. Typical sediment load from roof areas is of the order of 0.1 to 0.2 m3/ha/yr.

One of the key management practice associated with maintaining the efficacy of on-sitedetention system is regular inspection by local government officers. As these systems are installedwithin individual lots of private ownership, the efficacy of these system are affected by change inproperty ownership. The function (and even the presence) of these facilities are not necessarily madeaware to new owners and thus their maintenance requirements are overlooked. One mechanism by

which this can be managed is the keeping of a register of on-site detention basins by the localmunicipalities and the mandatory clean-out of these units as property ownership changes.Storm runoff from roofs should be cleared of leaves and any other roof-litter prior to discharge

into on-site detention tanks. Storm runoff from paved areas including courtyards, walkways,driveways, carriageways, car parks, etc., should under no circumstances be passed directly to on-site detention tanks but instead should be pm-treated by passing it across a grassed buffer strip (asillustrated in Figure 8.6) or through a sand/loam filter at least 200 mm thick and covered with grass.

8.2.2.2. On-site retention systems

The use of porous pavements in hard standing surfaces such as car parks can significantly reducethe runoff rate and sediment load generated from an urban catchment. A typical section of a porous

pavement and illustration of a car park build with porous pavement is shown in Figure 8.7.

128



r

GRANITECURBING GRANITECURBING

SUBGRADE SUBGRADE

Figure 8.7a Schematic illustration of a stormwater infiltration system beneath a car park

Figure 8.7b Porous concrete used to manage the quantii and quality of car park stormwaterrunoff.

Maintenance practices to ensure the efficacy of this system is the annual dislodgement ofdeposited particles by high pressured hosing of the porous pavement

It should. be noted that not all catchments are suited to on-site retention systems and the keyto sustainable operation of these systems lie in the appropriate utilisation of this technology. Thusdiscussion herein on appropriate maintenance and management practices is more to do with designconsiderations.

Soils with low hydraulic wnductivities do not necessarily preclude them from being suitablefor on-site retention system even though the required infiltration area may bewme uneconomical.

129



However, these soils are likely to render them more susceptible to clogging if the stonnwater inflowhas not undergone some degree of pm-treatment to remove litter and sediment.

On-site retention systems should not be placed near building footings to avoid any likelyinfluence o f continually wet subsurface on the structural integrity of these structures. Identification ofsuitable sites for on-site retention systems should also include avoidance of steep terrain and area ofshallow soil cover over rock. An understanding of the seasonal variation of the groundwater table isalso an essential element in the design of these systems.

8.3. Channel management

Stormwater conveyance systems in urban areas include the conventional underground pipe networkand constructed open channels (often either grass-lined, rock-lined or concrete line or a combinationof these features) to the use of grass swales and hybrid channels involving a natural and vegetatedlow flow channel with a grassed bench to increase the discharge capacity of the channel as illustratedin Figures 8.8 and 8.9.

p-- ‘T, ’ .‘,?$ -.&j+ p3,k

Figure 8.8 Concrete-lined open channels have been aconventional approach to urban storrnwater drainage.

Figure 8.9 Grass swales are alternative drainage systems which provides bothigure 8.9 Grass swales are alternative drainage systems which provides bothstormwater quantity and quality management.tormwater quantity and quality management.

130



Maintenance and manageme nt issues related to the effii of stormwater conveyancesystems are associated with desiting and removal of gross pollutants deposited in the undergroundstormwater pipes and pits, and in the channel and adjoining benches. There is generally a need forfrequent cleanout stormwater pits of deposited sediment and storm debris to maintain the dischargecapacity of the stmer drainage network.

In grass swales, maintenance will involve regular mowing of the grass and w?ed control.There are water quality benefits in maintaining a lush grass wver over the swale but this will need tobe balanced against cost implications in the provision o f water during extended dry periods to sustainthe vegetation.

Conclusions

The practice of maintenance and management of urban stormwater drainage is reviewed withemphasis on applications in arid and semiarid climates. Because of the low rainfall and vegetation,and high evaporation in arid and semiarid areas, the most prominent issue in the maintenance is thatof sediment and gross pollutant control. Consideration should be given to flash floods. In selectingthe most appropriate stonnwater managemen t practices for individual catchments, it is not enough to

just consider the mean annual rainfall of the region but also the seasonal variation of rainfall. Insandy soils, infiltration systems are suitable as stormwater best management practice in and regionsand can be incorporated into groundwater aquifer storage and recover schemes to optimise waterconservation and stormwater recyding. Constructed wetlands systems, designed to enable removalof pollutants particularly those associated with fine suspended particulates, are most .appropriate forhigher rainfall regions but may be used in semiarid areas. In these areas, the systems could bedesigned as predorninantiy ephemeral v&lands which fills during storm events and drains completely(except for a small permanent pool area) a few days after the storm event.

Bibliography

ALLISON, R .A, W ALKER, T.A, CHIEW , F.H.S., O’NEILL, I.C. AND MCMAHON, T.A. (1998) “From

Roads to Rivers: Gross Pdlutant Removal from Urban Watervvays”, Report 986 CooperatiieResearch Centre for Catchment Hydrology, May 1998,97p.CAMP DRESSER AND MCKEE (1993). “California Ston-nwater Best Practice handbooks: Municipal”,

Prepared for California Stonnwater Quality Task Force.STORMWATER COMMITTEE OF VICTORIA ww “Urban Stormwater - Best Practice

Environmental Management Guidelines”, CSIRO Publishing, 288~.WONG, T.H.F., BREEN, P.F., SOMES, N.L.G. AND LLOYD, SD. (1998) Managing Urban

Stonnwater using Constructed Wetlands, Industry Report 98/7, Cooperative Research Centrefor Catchment Hydrology, November 1998,4Op.

131



Chapter 9

Case studies

Many models, varying in complexity, are used by professional engineers fo r design of urbanfacilities. Not only are there many models, there are many versions of the same model.Moreover, the several comparative tests that have been conducted to determine the relativeaccuracy of the better known models, while informative, were inconclusive. The fact is that allof the major models are useful. Each is based on well-known and acceptable flowrelationships. Their water quality sections are much less well defined than their hydrologicand hydraulic sections; but even so, they are valuable tools. It should be obvious that no

single model is uniquely suited for all types of problems, or even for a specific problem, nor isthere a universal model. Table 9.1 shows the relative suitability of five different models to avariety of typical problems. A review of these models is given elsewhere (Yen et al., 1976).

Table 9.1. Model Application to Urban Drainage Problems

Application

Selection ofcritical rainfall

Rational Unit STORM SWMMMethod Hydrograph

Unsuitable Unsuitable Very good Good

events

Preliminaryanalysis of

Fair

urban areasDetailedanalysis ofurban areas

Analysis ofdetention

Design of

Poor Poor Fair Very good

Unsuitable Unsuitable Very good Good

Unsuitable Unsuitable Good Verv moddetention

Analysis of

- -

1 Unsuitable j Unsuitable 1 Unsuitable 1 Goodsurchargedsewer systems

Prediction of Good Fair Poorpeak flows in

small systemsDesign of sewer Fair (for Fair Poor Goodsystems (open - smallpipe flow) areas) !

Good

GOOd

Unsuitable

GoodI I

ILLUDAS

Poor

Very good

GOOd

It has to be mentioned here that the model selected in Table 9.1 were for examplepurposes, and other models could have been equally well included in the table. The selectiondoes not imply that these models are better but rather that they are among the relative fewwhich receive wide use. Tomo (1979) has identified some of these models and haspostulated reasons for their selection. The applications can be one or more of the followinggeneral categories:. Planning, the evaluation o f future alternatives. Such applications usually start with the

use of a simple model and get progressively more complex, as necessary.

. Analysis, the evaluation of existing systems or the detailed study of a plannedalternative.

132



. Design, the final sizing or configuring of a system for which planning or analysis iscomplete.The dividing lines between these categories are by no means clear, and many

applications could fall into more than one category, depending on the point of view of themodel user. However, from Table 9.1, the EPA SWMM model suits a wide range ofapplications. This conclusion is reached by other investigators (Marsalek et al., 1975; Sueichi

et al., 1984; among others). In and climates, this particular model is also proved to be as oneof the best applicable models (Nouh, 1991, 1997). So, the SWMM model is used in thischapter to present the results of case studies in arid climates. The main emphasis o f thechapter will be in providing examples of the range of problems, which have been addressed inchapter 3. The intention is neither to describe the models nor to provide comparativeevaluations.

The main categories of problems considered in this chapter are problems due to:n Duststorm effects on runoff water quality3 Catchment sized High infiltration ratesJ High temporal and spatial variations of rainfallLli High concentrations of transported suspended sedimentsReal data from three residential catchments in Saudi Arabia were used. The data

included rainwater and runoff samples collected during duststorms events. Briefcharacteristics of the data are shown in Table 9.2.

Table 9.2. Characteristics of Catchments and Storms

Data Name of catchment

Rabia Siteen Homadi

Catchment: Size (IO’ ml) 110.45 117.50 917.00Perviousness ratio (%) 39.18 40.40 43.28Land use ResidentialInitial infiltration (mmlhr) 48 86 83Constant infiltration (mmlhr) 5.3 17.6 18.5Infiltration decay rate (min-‘) 0.05 0.17 0.15

Depression storage (mm) on:pervious area 3.80 10.80 10.56impervious area 2.70 2.90 3.05

Manning’ n Varied from 0.017 to 0.37

Rainwater: Number of rainstorms 64 73 54Average depth (mm) 54.95 51.82 49.46Average duration (min) 30.10 21.60 31.50Average dry period between twosuccessive rainstoms (days) 49.70 48.16 51.45Average spatial coefficient of variation 1.06 1.11 1.23Average temporal coefficient of kurtosis -1.14 -1.03 -1.17

Number of collected samples in:Rainwater 317 306 326Stormwater 428 432 408

Duststorms: Number of investigated duststorms 96 96 96Average concentration of TSP (&m3) 2780 2310 2460

Rainfall was measured with automatic gages to the nearest 0.25 mm. Dischargeswere determined by means of current meters (Curley, type 622) and water level recorders(SIAP, type ID 5755). It has been reported (McLaren Ltd., 1979) that the minimum speed ofresponse o f the current meters is generally of the order of about 0.03 mls. At this speed theuncertainty of the reading is about + or - 15%. At higher speeds the uncertainty improves

and reaches a value of + or - 2% at speeds of 0.25 m/s or higher. Wflhin the range of flow

133



velocity in the investigated catchments, the average uncertainty in current metermeasurements is or the order of + or - 4%, and that in the evaluated discharges is in theorder of + or - 7%. More information regarding the errors of measurements in theinvestigated catchments are reported elsewhere (Nouh and El-Laithy, 1988a). Size andslope of the catchments were determined by means of 1:25,000 topographic survey maps.The infiltration capacity rates were determined using the available rainfall-runoff records and a

procedure proposed by Whelan et al. (1952).The total suspended particulate “TSP” were evaluated during each of the investigated98 duststorms a t the selected sites in the considered catchments. This was made by usingcalibrated high volume air samplers located outdoors at the third floor level in the sites. Thesamplers were operated simultaneously during each duststorm for a period of about 38 hrs(depending on the duration of the measured duststorm). The air samples were wllected onpm-weighted glass fiber filters. The glass fiber filters were preconditioned prior to weighing,both before and after sampling in a desiccator. The high volume filters were weighted on astandard analytical balance and the particle sizing filters were weighted on an ultra micro-balance with a readability of 0.1 pg. More information about the sampling techniques arereported elsewhere (Nouh et al., 1988b; Rowe et al., 1985).

The average spatial coefficient of variation and the average temporal coefficient ofkurtosis were used to measure the spatial and temporal variations of rainstorms over

catchment, respectively. These coefficients were computed as follows: (1) For eachrainstorm over a catchment, the recorded rainfall depths at each rain gage were used todetermine the mean and the coefficient of kurtosis “K” for rainfall intensities at the gage site;(2) The means computed from all gage sites were used to determine the coefficient ofvariation for the means “SCV” (e.g. standard deviation of the means/ mean of the means).This coefficient, which describes the dispersion of the storm, is taken as a measure of thespatial distribution of the rainstorm over the catchment; (3). The coefficients o f variation fromall rainstorms were averaged and named “average spatial coefficient of variation - ASCV.”Similarly, coefficient of kurtosis computed in (I), which describes the peakedness of thestorm, is taken as a measure of the temporal variation of a rainstorm at the gage site, and theaverage of the computed coefficients of kurtosis from all sites, called the storm temporalcoefficient of kurtosis “SK, is a measure of the temporal variation of the rainstorm over thecatchment. The coefficients of kurtosis from all rainstorms were averaged and called the

average temporal coefficient of kurtosis “ASK’. This coefficient (ASK) is used as a measureof the temporal variation of the rainstorms over the catchment.The stormwater runoff samples from every rainstorm were collected every 15 minutes

at the basin outlet. The collected samples were then analyzed for selected nutrients, ions,and trace metals following the standard procedures that are documented in many of referencetexts (Quinby-Hunt et al., 1988; Straub, 1989). The average of the concentrations of eachconstituent is taken as a representative of the concentration of the constituent in the collectedrunoff samples. The selection o f the type constituents was based on the results of previousstudies [Black, 1980; Ellis et al., 1987; Marsalek and Ng, 1989; Nouh, 1991; Straub, 1989)that such constituents normally exist in urban stormwater runoff.

The stormwater samples, produced from rainstorms of almost the same depth, andspatial and temporal distribution over catchment, were divided into groups according to theirSK values (measure of temporal variability) of duststorms. The samples within each group

were further subdivided according to their SCV values (measure of spatial variability) of theduststorms. The temporal and spatial distributions of duststorms over catchment weredetermined as follow:

1. During each duststorm and at each site of every catchment, the TSP was evaluatedevery hour (about 31-42 evaluations were made, depending on the duration of theduststonn), from which the mean value of TSP “pg/m3” and the coefficient of kurtosis “K”were computed. The coefficient of kurtosis was computed by the method of moments,and is taken as a measure of temporal variation of the duststorm at the site.

2. The means o f TSP from all sites were used to compute the coefficient of variation “SCVof the duststorm over the catchment (standard deviation of the means/ mean of themeans). This coefficient, which describes the dispersion of the duststorm, is taken as ameasure for the spatial distribution of the duststorm over the catchment.

3. The coefficients of kurtosis, determined from the individual sites were averaged to give

SK, and is taken as a measure of temporal distribution of the duststorm over thecatchment.

134



4. The SCV and SK values computed from all duststorms were averaged as ASCV andASK, and are used to measure the average spatial and temporal variability, respectively,of all duststorms over the catchment.

9.1 Effect of duststorms on stormwater quality

9.1.1. Average TSP level

To show the general influence o f duststorm, represented by the concentration of TSP, on thequality of stormwater runoff, the mean concentration of TSP in duststorms having almost thesame spatial and temporal variability over the Siteen catchment (i.e. similar SCV and SKvalues) was determined. This was plotted vs the corresponding average concenbation ofpollutants; namely Nitrate Nitrite N02IN03, Dissolved Carbon DC, Calcium Ca, TotalPhosphorus TP, Sodium Na, Chloride Cl, Cadmium Cd, Lead Pb, Zinc Ze, and Iron Fe, and isshown as a typical example in Figure 9.1 using the Steen catchment data. Similar resultswere obtained using the data from the other two catchments. Inspection of Figure 9.1indicates that the increase in the concentration of dusts leads to an increase in the

concentration of the investigated pollutants. The concentration of some pollutants is almostdoubled with the increase in the wncentration of TSP. It may be also observed that the rateof the increase in the concentration of ions (“Ca, Na, and Cl”), followed by that of metals (“Cd,Pb, Ze, and Fe”), is larger than that of nutrients (“N02IN03, DC, and TP”). Because thedynamic etfects of rainstorms on the concentration of pollutants in runoff is negligible (therunoff were produced from rainstorms of almost the same depth, ASCV, and ASK values) itmay be concluded that dusts play an important rule in transporting pollutants from one placeto another.

+Pw100 81eflO

0 !....i....i....i....i*...i*...i 0

500 1000 1500 2000 2500 3000 3500

lSP CONCENlRAllON (ug/cu.m)

Figure 9.1. Influence of mean concentration of TSP (pg/m3) on mean concentration ofpollutants (mg/l) in stormwater runoff (duststorms

(ASCV=0.9to1.0,ASK=0.8to0.9;andrainstorms:ASCV=1.2to1.40,ASK=1.0to1.20).

9.1,2 Spatial variation of duststorm

As mentioned earlier, the spatial variation of a dust storm over a catchment is measured bythe duststorm coefficient of variation SCV. A trial was made to plot SCV for duststonns of

135



similar temporal variability “SK” and similar mean TSP wncent-ration vs the concentrations ofpollutants in the stormwater produced from rainstorms of similar characteristics. Figure 9.2shows one of these plots for the Siteen catchment. Data from the other catchments producedvery similar results. Inspection of Figure 9.2 reveals that the concentration of pollutantsincreases, with a decreasing rate, as the SCV value increases. The large spread of TSPconcentrations about its mean value (reflected by the large SCV value) allows in general for

larger amounts of dusts over the catchment, resulting in larger amounts of pollutants to becarried with the dusts, and then transported with the stormwater runoff to the drainagefacility. The decreasing rate of the increase in the concentration of pollutants is due to thedecreasing rate of dust concentration increase relative to the mean concentration of TSP overthe catchment. It also can be generally noticed that the increase in the concentration of ions,followed by that of trace metals, is larger than that of nutrients.

-=-N02/N03 +DC *TP s-Ca/lO

*Na/lOO +Cl/lOO +Cd*lOO ++Pb*lO

+Ze*lO +Fe

10

7

4

I0.5 1.5 2.5 3.5 4.5 5.5 6.5

COEFFICIEN T OF VARIATION OF DUSTSTORM

Figure 9.2. Influence of duststorm coefficient of variation on concentration of pollutants (mg/l)in stormwater runoff

(dust&on-w Mean TSP = 1700 p9/m3, SK = 0.8 to 0.9; and rainstorms: ASCV = 1.2 to 1.40, ASK = 1 O to 1.20 ).

9.1.3 Temporal variation of duststorm

The concentration of the pollutants in stormwater runoff samples, produced by rainstorms ofsimi!ar spatial and temporal variability over catchment and under the effect of duststorms ofsimilar spatial variability over the catchment, was plotted against the duststorm coefficien t ofkurtosis. Typical p lot is presented in Figure 9.3 for the Siteen catchment. The data from theother catchments produced similar results. The figures indicates that there is a considerableinfluence of the duststorm coefficient- of kurtosis, which describes the peakedness of theduststorm, on the concentration of the pollutants. The concentration of the pollutantsincreases, with an increasing rate, as the coefficient of kurtosis “SK increases. The increasein the wncentration of pollutants can be related to the increase in the concentration of TSP asthe peakedness of the dust storm increases. Comparing the results of Figure 9.2 with those

of Figure 9.3, it can be seen that the influence on the concentration of pollutants of thetemporal variation of duststorms is more significant than that of the spatial variation of the

136



duststorms. This may be explained by the fact that the dynamic effect, on carryingatmospheric pollutants at a site, of a duststorm of rapidly increasing concentration of TSP islarger than that of a duststorm of slowly increasing concentration of TSP. As Figure 9.3indicates the wncent-ration of a pollutant increased about two times (on average) as thecoefficient of kurtosis increased from 0.80 to 1.43. The concentration of the ions, followed by

that of the metals, was noticeably increased more than that of the nutrients.

3-N02fN03 +DC

*TP -+Ca/lO

*Na/lOO -.-cl/100

*TP

*Na/lOO

+Cd*lOO

-+Ca/lO

-.-cl/100

++Pb*lO

10

87

t4

8 10.8 0.9 1 1.1 1.2 1.3 1.4 1.5

COEFFICIE NT OF KURTXXIS OF DUSTSTORM

Figure 9.3. Influence of duststorm coefficient of Kurtosis on concentration of pollutants (mg/l)in stormwater runoff

(duststonns: Mean TSP = 1700 p4/m3, SCV = 0.8 to 0.9; and rainstorms: ASCV = I .2 to 1.40, ASK = I .O to 1.20 ).

9.1.4 Catchment size

Stormwater runoff samples, produced from rainstorms of similar spatial and temporaldistributions and under the effect of duststorms of similar spatial and temporal distributions,were analyzed for the selected pollutants. The general pattern of the changes in theconcentration of all pollutants with the catchment size was almost the same. To show this atypical example of such general pattern, the mean concentrations of one nutrient “N02/N03”,one metal “Ze”, and one ion “Na” in each of the three considered catchments were evaluated,and are plotted in Figure 9.4. It can be seen that the concentration of all pollutants increasesas catchment size increases. In addition, the increase, and also the rate of such increase, inthe concentration of Sodium “Na”, followed by that of Zinc “Zen, is larger than that o fNitrate/Nitrite “NOZ/N03”. This can be explained by the fact that the amount of solid particles(which acts as a carrying agent for the pollutants) generated by a duststorm over a large sizecatchment is greater than that generated by the duststorm over a small size catchment. So,the rule of catchment size in the changes of pollutants in stom-rwatermff is its relation with

the amount of solid particles that are generated&y~&s&erred by rainfall and

then stormwater to drainage facilities.

137



I(N02M03)* 10

m&*10

=Na/lO I-

Rabia Siteen

NAME OF CAXXMENT

Figure 9.4. Variation of concentration of pollutants with catchment size(duststomw mean TSP = 2100 pg/m3, ASCV = l.lOto 1.20, ASK = 1.10 to 1.20; and rainstorms: ASCV= 1.2 to

1.40,ASK=l.Oto1.20).

9.2 Application of SWMM model

For each catchment (Table 9.2) an equal number of rainstorm events was split into twohalves. The first half was used to calibrate the model, whereas the second half was used toexamine the model performance. Several hydraulic variables; namely, peak discharge Q,time to peak discharge T, and total runoff volume VM, were considered. The followingmeasures were used to asses the performance:

0 The ratio “RAT” between the calculated and observed values of the variable. This is

expressed for single storm events as the mean of all the storms on a catchment, toindicate the average performance of the model in the catchment.0 The standard deviation DEV of the individual values about the overall mean, to

describe the scatter in the RAT ratio.0 The absolute error ABS between the calculated and observed values, expressed as a

percentage of the observed value. As in RAT, the ABS is expressed for single stormevents as the mean of all the storms on a catchment.

Accuracy of the model-i _

The accuracy of the model in predicting the hydraulic variables was evaluated using 59 stormevents, and the result obtained are shown in Table 9.3. The variation of the accuracy with

various catchment and climate characteristics is described below.

Table 9.3. Summary of the Accuracy of Performance of the Model

Name of Hydraulic Number of RAT DW ABSCatchment Variable storms

n nr

n IQ 1 59 1 0.82

nomaal I

IT

138



9.2.1 Effect of catchment size

A typical example on the effect of catchment size on the accuracy is shown in Figure 9.5. Itcan be seen from the figure as well as from Table 9.3 that the accuracy decreases as size ofcatchment increases. This is reflected not only by the values of BAT but also by the DEV andABS values. With the increase of catchment size the spatial variability of rainstorms and

infiltration increases sharply to the degree that the model components can not explain (seechapter 3). Table 9.3 also shows that the accuracy of predicting time to peak flowrates andtotal runoff volumes is better and less sensitive to variation in catchment size than that ofpredicting peak flowrates.

1 1 I I I I 1 II

Siteen

Totai rainfall depth = 86 mm

Dly period= IO days

Homadi

Total rainfall depth = 86 amDry periob 1Q4 days

0 40 a0 120 160 200 240 2ao

~ min

Figure 9.5. Calibrated and Observed Hydrographs in Small (up) and Large (down)

Size Catchments

139



9. 2. 2 Effect of infiltration rates

The accuracy of the model was evaluated in catchments of different rates of infiltration, and atypical result is shown in Figure 9.6. It can be seen that there is a considerable effect ofinfiltration on the accuracy o f the model. As the infiltration rate increases the accuracy o f themodel decreases. This fact is also apparent in Table 9.3. The level of accuracy for peak

flowrates and time to peaks is higher than that for total runoff volumes. The low accuracy fortotal runoff volumes may be due to the large amount of infiltrated water, which isunderestimated by the model. Obviously, the accuracy o f the model decreases as theperviousness ratio of the catchment increases.

10

a

6

o 6z

4

2

0

/ c

Rabia

Total rainfall depth = 84 mm

Siteen:

Total rainfall depth = 82 mmDry period = 12.8 days

\

h

\

o-- -i) CALIBRATED

1 ~OBsERvEn\

\

4

0 40 a0 I20 160 200 240 280TIMF, min

Figure 9.6. Calibrated and Observed Hydrographs in Catchments of Low (up) andHigh (down) Rates of Infiltration

140



9. 2. 3 Effect of rainfall characteristics

In this chapter, fwe characteristics of rainfall over catchment were considered and their effectson the model accuracy were evaluated. These characteristics are the spatial variability ofrainfall, the temporal variability of the rainfall, the total rainfall depth, the duration of rainfall,

and the dry period between two successive rainstorms over catchment. The results of theevaluated accuracy are discussed in the following.

9. 2.3.1 Spatial variation of rainfall

As it is mentioned earlier that the coefficient of variation of a rainstonn is taken as a measureof the spatial variability of the rainstorm over the catchment. Figure 9.7 shows the results ofprediction with two rainstorms of different spatial variation. It is apparent that the accuracydecreases as the spatial variability of rainstorm increases. In large size catchments, wherethe spatial variation is expected to be large, the accuracy of the model may reachunacceptable level.

1 I C-J I

0 30 60 90 120 150 180 210

TIME. min

CALIBRATEDWSERVFII

I I I I I I 1

1Coefficient of Rainstorm Variation = -0.82

hCoefficient of RainstormVariation = - 1 58

Figure 9.7. Calibrated and Observed Hydrographs from Rainstorms of Low (up) andHigh (down) Spatial Variations on Rabia Catchment

(Coefficient of Kurtosis = - 0.86; Total rainfall depth = 62 mm; Dry period = 18 days)

141



9.2. 3. 2. Temporal variation of rainfall

The effect of the temporal variation of rainstorm on the accuracy of the model was evaluatedusing rainstroms of almost the same depth, duration, and spatial variation, and a typicalresults are shown in Figure 9.8. As it has been mentioned earlier, the coefficient of kurtosis of

rainstorm over catchment is taken as a measure of the spatial variation. It can be seenclearly that the model accuracy decreases as the temporal variation of rainstorm increases.The high temporal variation means rapid rise of hyetograph, and in such a case the modelsbecome inaccurate in estimating hydrograph components.

i-

I -

I -

Coefficient of Kurtosis = - 1.42

Coeffiient of Kurtosis = - 0.53

90 120

w min

180

-

7

Figure 9.8. Calibrated and Observed Hydrographs from Rainstorms of High (up) andLow (down) temporal variations on Rabia Catchment

(Coefficient of rainsto rm variation = 1.12; Total rainfall depth = 55 mm; Dry period = 15 days)

142



9.2.3.3 Dry period between two successive rainstorms

Chapter 3 describes the effect of dry periods between two successive rainstorms on soilmoisture and infiltration. To demonstrate the effect of the dry period on the accuracy of themodel, rainstorms of almost the same, depth, duration, coefficients of variation and kurtosiswere used in the SWMM model to predict hydrographs, which were compared with theobserved ones on the Homadi catchment. A typical result is shown in Figure 9.9. Inspectionof the figure indicates that the accuracy decreases as the dry period between two successiverainstorms increases. As explained earlier in chapter 3, as the dry period increases the soilmoisture decreases rapidly with much of soils disintegration. This results in sharp increase ininfiltration. Such increase with its high spatial variability (characteristics in arid areas) is nottaken realistically by the SWMM model.

I I 1 I I . I 1 I

Dry period = 2.70 days

50 150 200

TIME, min

250 300 350

Figure 9.9. Predicted and Observed Hydrographs from Rainstorms of Short (up) and

Long (down) Dry Periods on Homadi Catchment(Coefficient of variation = 0.92; Coefficient of kurtosis = -0.85; Total rainfall depth = 65 mm)

143



the depth of rainfall increases.

In case of small depth of rainfall, especially received after a long dry period, most of therainwater infiltrate to the soils and produce insignificant storm runoff. To investigate the effec tof the rainfall depth on the accuracy of the model, rainstorms of similar duration, dry period,and spatial and temporal variations over the Homadi catchments were used. The generatedhydrographs from the calibrated model were compared with the observed ones, and a typicalexample is shown in Figure 9.10. It can be seen that the accuracy of the model increases as

0.5 -l

Total rainfall * 12 mm

0.4PREDICTED

omEFtvm

0.3

)I

c\

E 0.2

0

i

IL

< a1

e! I.# \0

I .’ \4

\

0p! 0.0 I

c

b

x20 A Total rainfall depth = 68 mm

/ \ Ao--*

t, -0EsERvED

Ii

0z 1.5 -

3

oi

1.0 Is \

O-0 1 d0 40 80 120 160 200 240 280

TIME, min

Figure 9.10. Predicted and Observed Hydrographs from Rainstorms of Small (up)

and Large (down) Total Rainfall Depths on Homadi Catchment(Coefficient of variation = 0.92; Ccefticient of kurtceis = - 0.6; Dry period = 16 days)

9. 2. 3.4 Total depth of rainfall



9. 2. 3. 5 Duration of rainfall

It has been found (Hino et al., 1988) that rainstorm duration has significant effects onhydrographs predicted by models. These effects can be magnified in and climates due totheir particular soils and rainfall characteristics (see chapter 3). In the following, the effect ofrainstrom duration on the accuracy of the SWMM model in and climates was evaluated.

Similar rainstons on the Homadi catchments were used to calibrate the model and toexamine its accuracy. A typical results of predicted and observed hydrographs on thecatchment is shown in Figure 9.11. It is apparent that the rainstorm duration has significanteffects on the hydrograph components predicted by the model. Wiih the decrease inrainstorm duration, the accuracy o f the model decreases. This may be explained by theinfiltration process of the small amount of rainwater through dry soils, which has notconsidered realistically by the model.

5

4_ Rainstorm Duration = 55 min

3

Raingtorm Duration = 20 minaingtorm Duration = 20 min

o--* -a-ED- OBSERVED

3ry

2

i

‘\

x,

0I I

0 20 40 60 80 100 120 140

llME$ min

Figure 9.11. Predicted and Observed Hydrographs from Rainstorms of Long (up)

and Short (down) Duration on Homadi Catchment(Coefficient of variation = 0.95 ; Coefficient of kurtceis = -0.9; Dry period = 12 days)

145



9.2.4 Effect of duststorms

It is mentioned earlier that heavy duststorms is an important climate characteristics in themajority of arid catchments (see chapter 2) and that these storms have significant effects onthe runoff and sewer flow properties (see chapter 3). Earlier evaluations in this chapter (seesection 9. 1) confirm the significant impacts of duststorm properties on stormwater quality. It

is anticipated, therefore, that duststorms might have significant effects on the accuracy of thehydrograph components predicted by the SWMM. This anticipation was confirmed byobserving hydrographs during duststorms of different characteristics, and these hydrographswere compared with the corresponding ones predicted by the model. A typical comparison isshown in Figure 9.12.

5

4

3

L

.c

.2E

u

2

= I

u

a!

0

0 0

c

cl

r

= 4

IL

Lr.

0 3

z

3

c

2

I

0

ATSP = 2360 fig/m3

w-4 PREDICTED

H OBSECRVED

h -TSP = 830 ug/m3

I \ d”\ PREDICTED

I

II

\

\

- OBSERVED

P 4.

v./I

/p

-

0 40 a0 120 160 200 240 280

Tlh4E, min

Figure 9.12. Predicted and Observed Hydrographs from Rainstorms of High (up) and

Low (down) Duststorm Concentrations in the Homadi Catchment(Total rainfall depth = 72 mm ; Dry period = 63 days)

146



9. 3 Effect of urbanization

As chapters 4 and 5 indicate, urbanization has significant effects on stormwater. Theincrease in urbanization increases the runoff volume and peak discharge, but decreases the

time of concentration of the produced hydrograph. Such effec ts are mainly due to the

decrease in the infiltration capacity, surface detention, and roughness of catchments. Sincestormwater quality is affected by the produced hydrographs, and urbanization is playing asignificant role in shaping these hydrographs, it is anticipated that stormwater .quality isaffected to great extent by urbanization.

Nouh (1995) was able to evaluate the stormwater quality, before and after

urbanization, of catchments in Saudi Arabia. The changes in the concentrations of selectedchemical constituents (mg/l) were expressed in dimensionless format relative to the measuredconcentrations of the same in the unurbanized catchments, and the following regression

equation was developed:

(9.1)

where h is the percentage change in the concentration of a constituent, cp s the percentage

change in urbanization, IX and 3 are regression parameters.

Mean and maximum concentrations of a constituent were considered. Accordingly, Eq.9.1

becomes for the mean and maximum concentrations, respectively, as:

Amean = amean $0 pmean

and

Amax = a,, ppmax

The mean and maximum concentrations of a constituent in the stormwater runoff is denoted

by him as E,,,==” and E,,,~~ , respectively. The results are given in Table 9.4. Inspection of

Table 9.3 reveals that the increase in urbanization generally increases the concentrations ofnutrients, ions, and trace metals in the stormwater runoff of the arid catchments. The

coefficient of multiple determination for the mean, R2 mean and that for the maximum, R2 maX,

indicate that there is strong association between the changes in urbanization and those in theconcentrations of the constituents in the stormwater runoff. However, such association with

the mean concentrations; with the exception of Barium “Ba”, is stronger than that with the

maximum concentrations of the constituents. The increase in the concentrations of theconstituents is due to the human activities which increase with the increase of urbanization.

Another set of data was used (Nouh, 1995) to verify the developed regression model(Eq.9.1). The absolute maximum difference between the observed concentration of aconstituent and the corresponding one estimated by Eq.9.1 was evaluated at differenturbanization levels, and the results are given in Table 9.5. Inspection of the table indicates

that the regression model may be used with reasonable accuracy in arid catchments.

Conclusions

The accuracy of stormwater prediction in arid catchments by using the SWMM was evaluated.Several catchment and rainstorm characteristics were considered. It has been found that theaccuracy of the model decreases as catchment size increases. However, the accuracy ofpredicting time to peak flowrates and total runoff volumes is better and much less sensitive to

variation in catchment size than that of predicting peak flowrates. The model providesunsatisfactory results in large-size arid catchments.

147



Table 9.4. Variation of regression parameters with urbanization

Table 9.5. Absolute maximum error (%)

Rate of infiltration is found to be another factor affecting the hydrograph prediction bythe model. As the infiltration rate increases the accuracy of the model decreases. All rainfallproperties have significant effects on the accuracy of prediction. The increase in the spatialand/or the temporal variation of rainstorm over catchment decreases the accuracy of the

/ 148



model. The larger the depth of rainfall and/or the duration o f rainfall the higher the accuracyof the model is produced. The dry period between two successive rainstorms also hassignificant effect on the model accuracy. The longer the dry period the lower the accuracy o fthe model is produced. The existence of duststorm affects the accuracy; the higher theconcentrations of dust in atmosphere the lower is the accuracy of the model. The increase inthe spatial and/or temporal variability of the duststorm also decreases the accuracy of themodel.

On the water quality side, duststorms affect the stormwater quality to a great extent,Not only the mean concentration of total suspended particulate “TSP” but also the spatial aswell as the temporal variations of the duststorm over catchment have significant effects on thestorrnwater quality. The high the concentrations and/or the spatial or the temporal variationsof the dust the higher are the concentrations of chemical constituents in the stormvvater.Such effects increase with the increase in catchment size.

Urbanization has significant effects not only on the produced hydrograph componentsbut also on the runoff water quality. With the increase in urbanization, the concentrations ofthe chemical constituents in the stormwater increase. A regression type model has beenproposed to estimate the e ffects of urbanization on the stormwater quality in arid catchments.

Bibliography

BLACK, P. E. (1980) ‘Water Quality Patterns During A Storm On A Small Parking Lot,” WaterResources Bulletin, Vol. 16, No. 4, pp. 615-620.

ELLIS, J. B, REVITT, D. M, HARROP, D. 0, AND BECKWITH, P . R. (1987) “TheContribution Of Highway Surfaces To Urban Stormwater Sediments And MetalLoadings,” The Science of the Total Environment, Vol. 59, pp. 339349.

HINO, M., ODAKA, Y., NADAOKA, K., AND SATO, A. (1988) “Effect Of Initial Soil MoistureContent On The Vertical Infiltration Process - A Guide To The Problem Of Runoff-Ratio And Loss,” Journal o f Hydrdogy, Vol. 102, pp. 267-284.

MARSALEK, J., DICK, T., WISNER, P., AND CLARKE, W. (1975) “Comparative EvaluationOf Three Urban Runoff Models,” Water Resources Bulletin, Vd. 11, No. 2, pp. 366-328.

MCLAREN INTERNATIONAL LIMITED (1979) Water and Agricultural DevelopmentStudies,” Technical Report Prepared for the Ministry of Agriculture and Water,Riyadh, Saudi Arabia.

MARSALEK, J, AND NG, H. F. (1989) “Evaluation Of Pollution Loadings From UrbanNonpoint Sources: Methodology And Application,” Journal of Great Lakes Research,Vol. 15, No. 3, pp. 444-451.

NOUH, M., AND EL-LAITHY, A . (1986a), “Construction Damages Due To Floods In WadiAddilah,” Technical Report Prepared on the Saudi Arabian National Project NumberAR-562 for the King AbdulAziz City for Science and Technology, Riyadh, SaudiArabia.

NOUH, M, ROW E, D, AL-DOWALIA, K, AND MANSOUR, K. (1986b), “Indoor-OutdoorSuspended Particulate Matter Concentrations At Four Sites In Riyadh, Saudi Arabia,”International Journal of Environmental Studies, Vol. 28, pp. 163-168.

NOUH, M. (1991) “Urban Stormwater Quality In Arid Catchment,” Invited Lecture,Proceedings of the International Conference on Urban Drainage and NewTechnologies, UNESCO/IAHR, Dubbrovnik, Former Yugoslavia, June, pp. 355364.

NOUH, M. (1995), “Effect Of Urbanization On Runoff Quality In Arid Catchments,” 2ndInternational Conference on Innovative Technologies in Urban Storm Drainage -NOVATECH 95, Lyon, France, June, pp. 623626.

NOUH, M. (1997) “Urban Stormwater Prediction In And Climates,” Water ResourcesEngineering, Vol. 18, No. 2, pp. 2642.

SUEICHI, T., MORIOKA, T., AND KAWAMOTO, S. (1984) Comparison Among Models ForStormwater And Pdlutants Runoff In View Of Their Reliability, Sensitivity, AndApplicability,” Proceedings, 3ti International Conference on Urban Drainage,Goteborg, Sweden.

TORNO, H. C. (1979) “Model Application In EPA Planning Programs,” ASCE National

Convention, Boston, Mass., April.

149



QUINBY-HUNT, M. S, MCLAUGHLIN, R. D, AND QUINTANILHA, A. (1986), Instrumentationfor Environmental Monitoring, Vol. 2: Wate r, 2nd Edition, John Wiley & Sons, NewYork.

ROWE, D., NOUH, M., AL-DOWALIA, K., AND MANSOUR, M. (1985),“lndoor-OutdoorRelationship Of Suspended Particulate Matter In Riyadh, Saudi Arabia,” Journal ofthe Air Pollution Control Association - JAPCA, Vol. 35, 24-26.

STRAUB, C. (1989) Practical Handbook of Environmental Control, CRC Press Inc., Florida.WHELAN, D., MILLER, L., AND CAVALLERO, J. (1952) “A Method For Determining Surface

Runoff By Rooting Infiltrated Wate r Through The Soil Profiles,” US Forest ServiceNortheast, Paper 54, USA.

YEN, B. C., WENZEL, H. G., MAYS, L., AND TANG, W. (1976) “Advanced Methodologies for

Design of Storm Sewer Systems,” W ater Resources Center Research ReportNumber 112, NTIS.