Book of abstracts - risorseidriche.dica.unict.it · Carlo De Michele Polytechnic of Milano ... Giuseppe Rossi University of Catania (Italy) ... Hotel Villa Diodoro, Via Bagnoli Croci,

International Workshop

U n i v e r s i t y o f

C a t a n i a D e p a r t m e n t o f C i v i l a n d

E n v i r o n m e n t a l E n g i n e e r i n g

Advances in statistical hydrology

Book of abstracts

TTaaoorrmmiinnaa,, IIttaallyy

MMaayy 2233--2255,, 22001100

http://www.stahy2010.it/

International Workshop, Advances in statistical hydrology, Taormina, May 23-25, 2010

i

Information

Statistical and probabilistic methods have been traditionally applied for addressing many

hydrological and geophysical problems, such as flood and precipitation frequency analysis,

drought analysis and prediction, spatial analysis of hydrological processes such as rainfall and

infiltration, stochastic simulation of hydrological processes such as streamflow and precipitation,

short and long range forecasting of hydrological processes, ocean waves prediction, etc. While

the field has been well established and much experience has been gained over the years, the

development of new statistical methods and tools and the advances in computing hardware and

software have further enlarged the scope of applications. In addition, the advances in other

neighboring fields such as atmospheric and climatic sciences have further expanded our

understanding and created new challenges in the field of statistical hydrology.

The workshop has the main objective to provide a focused forum for exploring the potential of

novel statistical methods and techniques to analyze hydrological and other geophysical

processes.

The workshop will address the following themes

Non stationarity and changes in Hydrology

Fitting and uncertainty assessment in statistical hydrological models

Time series analysis and multivariate methods for hydrological modelling and forecasting

Analysis and modelling of hydrological processes

Extreme events in geophysical processes

During the workshop, a round table devoted to highlight Professor V.M. Yevjvevich legacy to

modern hydrology and water resources management will take place. In particular, former

students and collaborators will revive both the scientist and the man through the review of his

main achievements and the recalling of personal memories.

Organizing committee

Antonino Cancelliere University of Catania, (Italy)

Giuseppe Rossi University of Catania, (Italy)

Salvatore Grimaldi University of Tuscia, (Italy)

Brunella Bonaccorso University of Catania, (Italy)

Vincenzo Nicolosi University of Catania, (Italy)

Leila Castiglione University of Catania, (Italy)


ii

Scientific committee

Hafzullah Aksoy Istanbul Technical University (Turkey)

András Bárdossy Universitätsbereich Stuttgart (Germany)

Paolo Burlando ETH Zürich (Switzerland)

Antonino Cancelliere University of Catania (Italy)

Pierluigi Claps Polytechnic of Torino (Italy)

Marcella Corduas University of Napoli "Federico II" (Italy)

Roberto Deidda University of Cagliari (Italy)

Carlo De Michele Polytechnic of Milano (Italy)

Anne-Catherine Favre University of Québec (Canada)

Mauro Fiorentino University of Basilicata (Italy)

Luis Garrote Technical University of Madrid (Spain)

Salvatore Grimaldi University of Tuscia (Italy)

Hoshin Gupta University of Arizona (USA)

Pierre Hubert IAHS Secretary General

Shih-Chieh Kao Purdue University (USA)

Demetris Koutsoyiannis National Technical University (Greece)

Witold Krajewski University of Iowa (USA)

George Kuczera University of South Australia (Australia)

Francesco Laio Polytechnic of Torino (Italy)

Alberto Montanari University of Bologna (Italy)

Taha Ouarda University of Québec (Canada)

Domenico Piccolo University of Napoli "Federico II" (Italy)

Giuseppe Rossi University of Catania (Italy)

Renzo Rosso Polytechnic of Milano (Italy)

José Salas Colorado State University (USA)

Daniel Schertzer National School of Bridges and Roads (France)

Uri Shamir Israel Istitute of Technology (Israel)

George Tsakiris National Technical University (Greece)

Lucio Ubertini University of Roma "La Sapienza" (Italy)

Workshop venue Hotel Villa Diodoro,

Via Bagnoli Croci, 75 98039

Taormina, Messina

Tel +39 0942 23312—Fax +39 0942 23391


iii

Workshop programme

Sunday May 23rd

17:00 Registration

19:15 Welcome cocktail

Monday May 24th

8:45 Introduction: welcome to the participants, presentation of the workshop

9:30 Invited lecture - Hydrologic synthesis using entropy theory (Vijay P. Singh, USA)

10:00 Session I: Non stationarity and changes in hydrology (Chairman: Demetris

Koutsoyiannis, Greece)

Pierre Hubert A study on the hydrological annual time series of the Niger

River at Koulikoro, Niamey and Lokoja stations

Marco Marani The detection of weekly preferential occurrences in rainfall

Ashish Sharma Simulating climate change impacts at the catchment scale:

The need for GCM nested bias correction and stochastic

downscaling

Maria Sunyer On the use of statistical downscaling for assessing climate

change impacts on hydrology

11:00 Coffee break

11:30 Session II: Fitting and uncertainty assessment in statistical hydrological models

(Chairman: Geoff Pegram, South Africa)

Samiran Das Examination of homogeneity of selected Irish pooling groups

Roberto Deidda A Multiple Threshold method for fitting the Generalized

Pareto distribution

Demetris Koutsoyiannis Mind the bias!

Mark Thyer Can hydrological model predictions be improved by

developing streamflow measurement error models using rating

curve data?

Steven Weijs Information theory, uncertainty and risk for evaluating

hydrologic predictions

Seth Westra An upper limit to seasonal precipitation predictability?

Iwona Markiewicz On accuracy of upper quantiles estimation

13:30 Lunch break

15:30 Session III: Time series analysis and multivariate methods for hydrological

modelling and forecasting (Chairman: Baldassare Bacchi, Italy)

Geoff Pegram Interpolation and simulation of daily rainfall observations

based on copulas and circulation patterns

Carlo De Michele Multivariate Extreme Value models in hydrology: a copula

approach


iv

Luis Mediero Design flood hydrographs from the relationship between flood

peak and volume

Sander Vandenberghe Properties and performance of a copula-based design storm

generator

Davide De Luca Stochastic models for rainfall nowcasting

Luciano Xavier PREVIVAZ – Improving weekly streamflow time series

forecasts with the current hydrologic state of the river basin

Baldassare Bacchi Rainfall event variables: dependence analysis by way of

copulas

17:15 POSTER SESSION

20:30 Social dinner

Tuesday May 25th

9:00 Round table: “The legacy of Prof. V. Yevjevich” (Coordinator: Giuseppe Rossi, Italy)

Tina Dasic, Jovan

Despotovic and Jasna

Plavsic

University of Belgrade

Jose D. Salas and Taha

Ouarda

Colorado State University, Université du Québec

Vijay P. Singh Texas A & M University

Francisco Gomide Universidade Federal do Paraná

Geoff Pegram University of KwaZulu-Natal

Giuseppe Rossi Università degli Studi di Catania

Nilgun Harmancioglu Dokuz Eylul University

11:15 Coffee break

11:45 Session IV: Analysis and modelling of hydrological processes (Chairman: Pierre

Hubert, France)

Tatiana Ghizzoni Multivariate skew-t description of flood events

Francesco Napolitano Study on the rainfall dependence structure using radar and rain

gauge data

Ons Oueslati Attempts to flow regime classification and characterisation in

Mediterranean streams using multivariate analysis

Ashish Sharma How does the El Niño Southern Oscillation (ENSO) and other

climate modes affect multi-scale temporal rainfall variability

in East Australia?

Li-Pen Wang Analysis of sub-daily rainfall sequences based upon a semi-

deterministic multiplicative cascade method

13:30 Lunch break

15:30 Invited lecture - Hydrologic Variability and Change: The role of Stochastic

Hydrology (José D. Salas, U.S.A.)


v

16:00 Session V: Extreme events in geophysical processes

(Chairman: Witold G. Strupczewski, Poland)

Taha Ouarda Use of Empirical Mode Decomposition for the prediction of

hydrometeorological extremes

Felix Francés Comparison of high return period flood-quantiles estimated

with upper bounded distribution functions, including non-

systematic information. Application to a Mediterranean river

Enrica Caporali Probabilistic definition of drought risk indicators for water

supply systems in Central Tuscany

17:00 Wrap up session and discussion

18:00 End of the workshop


vi

Table of Contents

INVITED LECTURES

Vijay P. Singh

Hydrologic synthesis using entropy theory 1

Jose D. Salas

Hydrologic variability and change: the role of stochastic hydrology 1

SESSION I: Non stationarity and changes in hydrology

Abrate T., Hubert P. & Sighomnou D.

A study on the hydrological annual time series of the Niger River at Koulikoro, Niamey

and Lokoja stations 2

Marani M.

The detection of weekly preferential occurrences in rainfall 3

Mehrotra R., Johnson F., Westra S. & Sharma A.

Simulating climate change impacts at the catchment scale: The need for GCM nested

bias correction and stochastic downscaling 3

Sunyer M. A., Madsen H. & Yamagata K.

On the use of statistical downscaling for assessing climate change impacts on

hydrology 4

SESSION II: Fitting and uncertainty assessment in statistical hydrological models

Das S. & Cunnane C.

Examination of homogeneity of selected Irish pooling groups 5

Deidda R.

A Multiple Threshold Method for fitting the Generalized Pareto Distribution 6

Papalexiou S. M., Koutsoyiannis D. &Montanari A.

Mind the bias! 6

Thyer M., Renard B., Kavetski D. & Kuczera G.

Can hydrological model predictions be improved by developing streamflow

measurement error models using rating curve data? 8

Weijs S. & van de Giesen N.

Information theory, uncertainty and risk for evaluating hydrologic predictions 8

Westra S. & Sharma A.

An upper limit to seasonal precipitation predictability? 8

Markiewicz I., Strupczewski W. G. & Kochanek K.

On accuracy of upper quantiles estimation 9


vii

SESSION III: Time series analysis and multivariate methods for hydrological

modelling and forecasting

Bárdossy A. & Pegram G.

Interpolation and simulation of Daily Rainfall observations based on Copulas and

Circulation Patterns 10

De Michele C. & Salvadori G.

Multivariate Extreme Value models in hydrology: a copula approach 10

Mediero L., Garrote L. & Jiménez A.

Design flood hydrographs from the relationship between flood peak and volume 11

Vandenberghe S., Verhoest N.E.C. & De Baets B.

Properties and performance of a copula-based design storm generator 12

De Luca D. L. & Versace P.

Stochastic models for rainfall nowcasting 12

De Souza S. A., da Serra Costa F., Nóbrega L., Xavier L., Piñeiro Maceira M. E.

& J. Machado Damázio

PREVIVAZ – Improving weekly streamflow time series forecasts with the current

hydrologic state of the river basin 13

Bacchi B. & Balistrocchi M.

Rainfall event variables: dependence analysis by way of copulas 14

SESSION IV: Analysis and modelling of hydrological processes

Roth G., Ghizzoni T. & Rudari R.

Multivariate skew-t description of flood events 15

Montesarchio V., Russo F., Napolitano F., Lombardo F. & Baldini L.

Study on the rainfall dependence structure using radar and rain gauge data 16

Oueslati O., De Girolamo, A.M, Abouabdillah A. & Lo Porto A.

Attempts to flow regime classification and characterisation in Mediterranean streams

using multivariate analysis 16

Pui A., Westra S. & Sharma A.

How does the El Niño Southern Oscillation (ENSO) and other climate modes affect

multi-scale temporal rainfall variability in East Australia? 17

Li-Pen Wang, C. Onof & Č. Maksimović

Analysis of sub-daily rainfall sequences based upon a deterministic multiplicative

cascade method 18

SESSION V: Extreme events in geophysical processes

Sam Lee T. & Ouarda T. B. M. J.

Use of Empirical Mode Decomposition for the prediction of hydrometeorological

extremes 19


viii

Botero B. & Francés F.

Comparison of high return period flood-quantiles estimated with upper bounded

distribution functions, including Non Systematic information. Application to a

Mediterranean river 19

Rossi G. ,Garrote L. & Caporali E.

Probabilistic definition of drought risk indicators for water supply systems in Central

Tuscany 20

ROUNDTABLE: THE LEGACY OF PROF. V. YEVJEVICH

Rossi G.

Biographical profile of Vujica Yevjevich 21

Dašić T., Plavšić J., Đorđević B. & Despotović J.

Teaching and professional activities in water resources engineering in Yugoslavia 25

Salas J. D.

In memory of professor Vujica M. Yevjevich (Contributions to Colorado State

University) 26

Ouarda T.

A comptemplation of the contributions of Professor Yevjevich 28

Singh V. P.

Bridging hydrologic theory and practice: a reflection on professor Yevjevich‟s

contributions 29

Gomide F. L. S.

Professor Yevjevich‟s contributions to the field of floods and droughts 30

Pegram G.

Contributions to the theory and practice in water storage 31

Rossi G.

International activities and transfer of knowledge in water resources 32

POSTER SESSION

Abouabdillah A., Oueslati O., De Girolamo A. M. & Lo Porto A.

Modeling the impact of climate change in a Mediterranean catchment 34

Arena C., Cannarozzo M. & Mazzola M. R.

Cross-correlation, homogeneity and uncertainty in regional models of yearly

streamflow 34

Aronica G. T. & Brigandì G.

A simple point rainfall stochastic model for single event generation using a

parsimonious approach 35

Bergaoui M.

Analysis of maximum rainfall daily. Station of Saida Manoubia, Tunis 36


ix

Boccafoschi A. & Rejtano B.

Flood predictions downstream of a river confluence by bivariate analysis of incoming

flows 36

Bocci C., Caporali E. & Petrucci A.

A geoadditive model of rainfall in Tuscany (Italy) 37

Bonaccorso B., Cancelliere A. & Rossi G.

Effect of trends on the estimation of hydrological quantiles 38

Buttafuoco G., Caloiero T. & Coscarelli R.

Assessing spatial uncertainty of reference evapotranspiration using stochastic

simulation in southern Italy (Calabria region) 39

De Michele C., Vezzoli R., Pecora S. & Zenoni E.

Multivariate frequency analysis of droughts via copulas 40

Di Domenico A., Laguardia G. & Margiotta M. R.

Investigating the propagation of droughts in the water cycle at the catchment scale 40

Di Piazza A., V. Noto L., Viola F. & La Loggia G.

Comparison of different methods for the extension of monthly runoff data 41

Dias Jardim Penna D. , Piñeiro M. E. & Machado Damázio J. M.

Definition of the streamflow scenario tree for long-term operation planning studies of

hydrothermal power generation system 42

Ellouze-Gargouri E. & Bargaoui Z.

Sensitivity analysis using multivariate approach for a rainfall–runoff model of

Geomorphological Unit Instantaneous Hydrograph type 43

Fatichi S., Ivanov V. Y. & Caporali E.

Simulating hydro-meteorological variables across a range of temporal scales with a

Weather Generator 44

Fernández Z. & Quiroga S.

Analysing hydrological risk and water policies implications on crop production in the

Ebro River basin in Spain 44

Ganora D., Laio F. & Claps P.

An approach to propagate streamflow statistics along the river network 45

Ghimire B. NS & Janga Reddyl M.

Development of stage-discharge rating curve in river using Genetic Algorithms and

Model tree 46

Gregersen I. B., Arnbjerg-Nielsen K. & Madsen H.

Parametric analysis of regional trends in observed extreme rainfall in Denmark 47

Hassanzadeh H., Eslamian S., Abdolhoseini M. & Grimaldi S.

Application of L-moment for Estimation of Quantile Mixtures 48

Herrera E., Nelsón D. & Neira O.

Statistical characterization of precipitation series in city Bogotá 48


x

Herrera E., Rico H. A., Ardila F. , Moreno J., Cárdenas E., Castañeda R. &

Dominguez E.

Analysis self-similarity with Hurst exponent and the autocorrelation function 48

Kim S. , Shin H., Kim T. & Heo J.-H.

Derivation of the Probability Plot Correlation Coefficient Test statistics for the

Generalized Logistic distribution 48

Kochanek K. , W.G. Strupczewski & Markiewicz I.

On feasibility of L-Moments Method for distributions with cumulative distribution

function and its inverse inexpressible in the explicit form 49

Kysely J., Gaal L., Beranova R. & Plavcova E.

Estimating precipitation extremes in climate change simulations by the region-of-

influence method 50

Laio F., Allamano P. & Claps P.

Revisiting the maximum-value test for the frequency analysis of extreme events 50

Loboda N.

Stochastic simulation of hydrological processes under water management

transformation and global climatic change 51

López de la Cruz J. & Francés F.

Influence of the North Atlantic Oscillation and the Western Mediterranean Oscillation

in the maximum flow events in Spain 52

Markus M.

Assessing changes in 100-year flood discharge estimates used in land use regulation in

the greater Chicago region, USA 52

Mehaiguene M., Meddi M., Toumi S. & Longobardi A.

Low flows quantification and regionalization in North West Algeria 53

Montesarchio V. & Napolitano F.

A single-site rainfall disaggregation model based on entropy 53

Nour El Houda B., Habib Snane M., Sfar Felfoul M. & Rached B. M.

The Impact of exceptional rain events on the development of gullies in geological

formation in Dhkekira watershed 54

Obregón N. & Ardila F.

Recovering dynamics of hydrological observed data via phase space reconstruction and

Genetic Programming 54

Peres D. J. , Nicolosi V. & Cancelliere A.

Long-term wave hindcast by neural networks using NCEP/NCAR Reanalysis wind fields 55

Picek J. & Kyselý J.

Homogeneity tests based on multivariate L-moments for extreme precipitation events 55

Plavcova E. & Kysely J.

Are biases in daily precipitation statistics in climate model simulations conditioned by 56


xi

large-scale atmospheric circulation?

Plavšić J. , Ferrari E. ,Despotović J. & Sirangelo B.

Peaks over threshold method in trend analysis of maximum daily precipitation in two

climatic regions 57

Ranzi R., Barontini S. , Ferri M. & Bacchi B.

A stochastic approach to assess hydro-geotechnical uncertainty in flood risk mapping 57

Rianna M. , Russo F. & Napolitano F.

Using the Stochastic Index method for calculation of flow duration curves on

ephemeral basins 58

Rico H. A., Moreno J., Cárdenas E. , Castañeda R., Herrera E. & Dominguez E.

Regionalization and geomorphological analysis of watersheds in Colombia 59

Ridolfi E., Montesarchio V., Russo F., Napolitano F. & Lombardo F.

Evaluation of rainfall network using entropy over the urban area of Rome 59

Rossi F., De Luca C., Furcolo P. , Villani P. & Vitolo C.

Extreme rainfall and floods in a Mediterranean area 60

Sarkar S. K. & Bhattacharya B. D.

Water quality analysis of the coastal regions of Sundarban mangrove wetland, India

using multivariate statistical techniques 61

Sebastianelli S., Russo F., Napolitano F. & Baldini L.

Comparison between radar and rain gauges data at different distances from radar and

correlation existing between the rainfall values in the adjacent pixels 62

Shabanlou S. & Rajabi A.

Evaluation of empirical relationships for estimation of instantaneous peak flow

Discharge in western catchments of IRAN 62

Tomei F., Antolini G., Tomozeiu R., Pavan V., Villani G. & Marletto V.

Analysis of precipitation in Emilia-Romagna (Italy) and impacts of climate change

scenarios 63

van den Berg M.J., Vandenberghe S., De Baets B. & Verhoest N.E.C.

Copula-based scaling of rainfall fields 64

van Nooijen R.R.P., Gubareva T. & Kolechkina A.G.

A comparison of fitting methods and tests for several distributions on hydrological

data 64

van Nooijen R.R.P. & Kolechkina A.G.

A comparison of fitting methods and tests for several distributions on computer

generated samples 64

Viola F., Noto L. V. & Cannarozzo M.

Regional flow duration curves for ungauged sites in Sicily 65

Vircavs V. & Veinbergs A.

Modelling of groundwater level fluctuations in agricultural monitoring stations with 66


xii

conceptual two dimensional mathematical model METUL

Volpi E., Lombardo F. & Napolitano F.

Study on scaling regimes of rainfall time-series from a dense raingage network 66

Westra S., Mehrotra R., Sharma A. & Srikanthan S.

Continuous rainfall simulation: estimation at ungauged locations 67

Westra S. & Sharma A.

Interpreting variability in global SST data using independent component analysis

and principal component analysis 68


1

Invited Lectures

Hydrologic synthesis using entropy theory

Vijay P. Singh

Department of Biological & Agricultural Engineering &

Department of Civil & Environmental Engineering

Texas A & M University

College Station, Texas 77843-2117, USA

e-mail: [email protected]

An entropy theory, comprising the Shannon entropy, the principle of maximum entropy, and the

concentration theorem, has been applied in recent years to a wide range of problems in hydrology ad water

resources engineering. From a physical point of view, applications in hydrology can be distinguished into

three classes: (1) physical, (2) statistical, and (3) mixed. This study focuses on the physical applications of

the entropy theory, wherein the theory is coupled with the laws of mathematical physics and solutions are

derived either in time or space domain rather than the frequency domain. A general framework for

aplication of the entropy theory is presented that can be employed to derive solutions to a wide range of

seemingly disparate problems. This framework is illustrated with a couple of example applications. The

entropy theory seems to have much potential in hydrology which remains yet to be fully exploited.

Hydrologic variability and change: the role of stochastic hydrology

Jose D. Salas

Department of Civil and Environmental Engineering

Colorado State University

This lecture will emphasize hydrologic variability and change as related to climate variability and change.

Examples will be provided where climatic processes represented by some indexes such as as the Pacific

Decadal Oscillation (PDO) and the Atlantic Multidecadal Oscillation (AMO) vary in a random fashion

with oscillations resembling cycles and trends.

Also some examples will be shown where variations of streamflows reflect those of large scale climatic

oscillations. We will also describe how stochastic methods can be useful for representing such variations.

Particularly we will explain the effect of persistence and alternative short and long memory models that

can be useful for describing and simulating the variability and change of hydroclimatic data. The lecture

will end with a number of examples.


2

Session I

Non stationarity and changes in hydrology

A study on the hydrological annual time series of the Niger River at

Koulikoro, Niamey and Lokoja stations

Tommaso Abrate

Organisation Météorologique Mondiale, Genève

E-mail : [email protected]

Pierre Hubert

UMR Sisyphe, Université P. & M. Curie, Paris VI

Daniel Sighomnou

Niger-HYCOS, B.P 729 Niamey (Niger)

Since the onset of the early ‟70 drought many researches have been conducted on the few long historical

series of hydroclimatological data available for Western Africa with the aim of understanding their

variability and identifying their discontinuities. The empirical observation that in the last decade of the

past century wetter conditions are again prevailing has reopened the debate about the non-stationarity of

climatic phenomena in this region.

There is a wide range of statistical tools aimed at identifying breaking points in time series. For this study

has been used the segmentation procedure designed by Hubert et al. (1987). This procedures applied to a

historical series of data yields an optimal partition of the original series into as many subseries as possible.

This procedure had already been previously applied to the discharge series of the rivers Senegal (at Bakel)

and Niger (at Koulikoro) between 1907 and 1984. It highlighted breaks around the same years: at Bakel in

1921-22, 1936-37, 1949-50 and 1967-68; at Koulikoro in 1923-24, 1932-33, 1950-51 and 1969-70. In

both cases the first and the third break correspond to increases in the flows, while the second and the

fourth with reductions. More recently the same procedure had been reapplied to the Bakel series, this time

ending in 2003. It showed in 1993-94 a new break, with an increase of the average in the period following

it.

For this study the historical series of the Niger River at Koulikoro (Mali), Niamey (Niger) and Lokoja

(Nigeria) have been used. The series, all ending in 2005/6 start respectively in 1907 (Koulikoro), 1915

(Lokoja) and 1929 (Niamey). Unfortunately the presence, except in Koulikoro, of wide gaps in data series

limits considerably the possibility of estimating reliable annual averages. It has been nevertheless possible

to get complete series of annual average discharges for Koulikoro from 1907 to 2005, for Niamey from

1944 to 2006 and for Lokoja from 1946 to 1992. The analysis of these historical series shows a series of

breaks as follows

• At Koulikoro : 1923-24 (positive), 1932-33 (negative), 1950-51 (positive), 1969-70 (negative),

1979-80 (negative) and 1993-94 (positive);

• At Niamey : 1950-51 (positive), 1970-71 (negative), 1982-83 (negative) and 1993-94 (positive).;

• At Lokoja : 1970-71 (negative), 1988-89 (positive, only in few series ending in 1992). A positive

break appears also around 1953-54 but only in one sub-series

These breaks, especially those identified at Koulikoro and Niamey, match well with those identified in the

Bakel series (Senegal) using the same procedure. The magnitude of the variation of the interannual means

between the alternating wet and dry periods are significant and similar. These results seem to hint that

phenomena causing non stationarity in hydrological series can have a subcontinental impact (e.g. the

negative discontinuity in 1970 visible in all the three station analyzed) or on the contrary being more


3

limited in their spatial coverage (e.g. the worsening of the drought in the early „80s well visible in

Koulikoro and Niamey but not in Lokoja). Further analyses using the same procedures on other long

historical series from Central and Western Africa could bring some more insight on the time and space

scales of the variability of hydroclimatic phenomena in this region.

The detection of weekly preferential occurrences in rainfall

Marco Marani

Department of Hydraulic, Maritime, Environmental, and Geotechnical Engineering, University of

Padova, Padova, Italy. E-mail: [email protected], Tel: +39 049 8275449; fax +39 049 8275446

The detection of weekly preferential occurrences in atmospheric and hydrologic processes has recently

attracted much attention as a way to identify the signature of anthropogenic climatic changes. The

interpretation of previous analyses, however, is not unequivocal, in part as a result of a lack of widely

accepted statistical criteria.

Here, a general and exact method to detect the presence of weekly preferential occurrences is developed

and applied to long rainfall observations. The method makes use of the fact that, under the null hypothesis

of time homogeneity, the process of event occurrence in the different days of the week is equivalent to the

random distribution of a number balls (the wet days) in a set of boxes (the days of the week). The

departure from a homogeneous distribution is then characterized through the probability of the maximum

number of balls in a box, which can be computed exactly with no ad-hoc assumptions. The new method

shows that: i) preferential rainfall weekly occurrences emerge in all cases in the most recent period

analyzed (1990-2006), while they are absent, or are too weak to be detected, in previous years (before

1989); ii) the balls-in-boxes approach appears to be more sensitive than Pearson's test when deviations

from homogeneity are associated with just one day of the week, a common occurrence in connection with

day-of-the-week effects. The results presented help reconciling previous contrasting studies and contribute

compelling evidence that anthropogenic changes in local rainfall regimes have occurred over the past

century in urban and industrial areas.

Simulating climate change impacts at the catchment scale: the need

for GCM nested bias correction and stochastic downscaling

Rajeshwar Mehrotra, Fiona Johnson, Seth Westra and Ashish Sharma

Water Research Centre, School of Civil and Environmental Engineering

University of New South Wales, Sydney, Australia; Email: [email protected];

[email protected], [email protected]; [email protected], Tel: +61 2 9385 5140

A nonparametric stochastic downscaling framework for generation of daily rainfall at multiple point

locations for catchment scale climate change impact assessment under enhanced greenhouse conditions is

described. The framework is an extension of Modified Markov Model (Mehrotra and Sharma, 2007) that

generates daily rainfall at multiple point locations using exogenous inputs of relevant atmospheric

predictors as simulated by a General Circulation Model (GCM). An important aspect of the revision

presented here is the use of a Nested Bias Correction (NBC) rationale that transforms the atmospheric

predictors to impart an element of low frequency variability that helps simulate sequences having

appropriate representations of sustained extremes. The downscaling framework is formulated using 43

mailto:[email protected]






4

years of daily rainfall observations at 30 locations near Sydney, Australia, with reanalysis data being used

to represent observed atmospheric variables. Output of the CSIRO Mark 3.5 GCM (corresponding to the

SRES A2 emission scenario) is used to downscale the rainfall for year 2070 conditions. Significant

improvements are noted in the downscaled rainfall sequences compared to the case where simpler

downscaling formulations are used, especially with respect to the model‟s ability to simulated sustained

droughts and periods of high rain.

On the use of statistical downscaling for assessing climate change

impacts on hydrology Maria A. Sunyer, Henrik Madsen, Keiko Yamagata

DHI, Hørsholm, Denmark ([email protected])

It is well known that climate projections from climate simulation models are biased and cannot adequately

reproduce the variability in climate variables that are present at the local scale. Thus, there is a need for

statistical adjustments and downscaling of climate projections in order to make reliable impact

assessments at the local scale.

Statistical procedures (commonly referred to as statistical downscaling) have been introduced to relate the

climate projections at larger scale (from global circulation models or regional climate models) to climate

variables at the local scale. In this paper downscaling of regional climate model (RCM) projections is

considered for assessing climate change impacts on hydrological variables. Different downscaling

procedures based on a general change factor methodology are compared. The basic concept in change

factor methods is that climate model simulations are used to extract changes in different statistical

characteristics of climate variables from the present to the future climate (denoted change factors). The

change factors are then applied for adjusting local climate variables.

Three different procedures are considered: (i) mean correction (often referred to as the delta change

method), (ii) mean and variance correction, and (iii) stochastic weather generators (WG). Three different

WGs are compared: a Markov chain model, the LARS semi-empirical WG, and the Neyman-Scott

Rectangular Pulses (NSRP) model. The three WGs mainly differ in the way extreme events are generated.

Results from the comparison shows that the NSRP WG is better in representing changes in the distribution

of extreme precipitation events in accordance with the changes predicted by the RCM.

The paper considers climate change impacts on the hydrology in two case studies in Sealand, Denmark,

representing different spatial scales. Climate simulations from the HIRHAM RCM for the period 1950-

2100 based on the IPCC SRES scenario A1B are used in the analysis. Mean correction, mean and variance

correction, and the NSRP WG model are applied to downscale precipitation, temperature and potential

evapotranspiration. In the first study, the downscaled climate data are used as forcing to a MIKE SHE

integrated hydrological model of the North-Eastern part of Sealand for analysing the impacts of climate

change on the water resources in the region. In the second case study, the downscaled climate data is used

to analyze the risk of flooding due to climate change in an urban area in the Eastern part of Sealand.



5

Session II

Fitting and uncertainty assessment in statistical hydrological

models

Examination of homogeneity of selected Irish pooling groups Samiran Das and Conleth Cunnane

Department of Engineering Hydrology, NUI Galway, Galway, Ireland.

E-mail: [email protected], Tel: +353-91-492785; fax: + 353-91-495565

In regional flood frequency estimation a growth curve TXT relation is estimated from the data of a

homogeneous pooling group. One important aspect of the selection of a method of forming homogeneous

pooling groups is the investigation of how successfully the selection method can achieve homogeneity.

The homogeneity criterion implies that the TXT relation is the same at all sites in a pooling group. A

homogeneous pooling group of sites leads to a minimisation in the error of quantile estimators which is

the main aim of a regional flood frequency analysis. Tests for the homogeneity of regions/pooling groups

are usually based on a statistic that relates to the formulation of a frequency distribution model, e.g. the

coefficient of variation Cv (Wiltshire, 1986; Fill and Stedinger, 1995)) and/or skew coefficient Cs, or their

L-moment equivalents (Chowdhury et al., 1991; Hosking and Wallis, 1997) or of dimensionless quantiles

such as the 10-yr event, X10 (Dalrymple, 1960; Lu and Stedinger, 1992a). Hosking and Wallis (1993,

1997) proposed homogeneity tests based on L-moment ratios such as L-Cv alone (H1) and L-Cv & L-

skewness jointly (H2) which are widely used in flood frequency analysis although the former one is

recommended by the authors for having better power to discriminate between homogeneous and

heterogeneous regions. Recently a similar conclusion has been drawn by Viglione, Laio and Claps (2007).

In this paper a study, based on annual maximum series obtained from 85 Irish gauging stations, examines

how successful a common method of identifying pooling group membership is in selecting groups that

actually are homogeneous. Each station has its own unique pooling group selected by use of a distance

measure, 2

ln

2

ln

2

ln

lnlnlnlnlnln

BFI

ji

SAAR

ji

AREA

ji BFIBFISAARSAARAREAAREAdij

and with a minimum of 500 station years of data in the pooling group, which satisfies the 5T rule (FEH,

1999, 3, p-169) for the 100 year quantile. By simulation the sampling distribution of L-Cv (t2) in each

pooling group and the 95% confidence limits about the pooled estimation of t2 are obtained. The t2 values

of the selected group members are compared with these confidence limits both graphically and

numerically.

Of the 85 stations, only 1 station‟s pooling group members have all their t2 values within the confidence

limits, while 7, 33 and 44 of them have 1, 2 or 3, or more than 3 t2 values outside the confidence limits.

The outcomes are also compared with the heterogeneity measures H1 and H2. The H1 values show an

upward trend with the ranges of t2 values in the pooling group whereas the H2 values do not show any

such dependency. Overall the results show that even with a carefully considered selection procedure, as in

equation above, it is not certain that perfectly homogeneous pooling groups are identified. As a

compromise it is recommended that a group containing more than 2 values of t2 outside the confidence

limits should not be considered homogeneous.



6

A Multiple Threshold Method for fitting the Generalized Pareto

Distribution Roberto Deidda

Dipartimento di Ingegneria del Territorio, University of Cagliari, Cagliari, Italy.

E-mail: [email protected]; Tel. +39 070 6755324; fax +39 070 6755310,

Previous studies indicate the generalized Pareto distribution (GPD) as a suitable distribution function to

reliably describe the exceedances of daily rainfall records above a suitable optimum threshold, which

should be selected as small as possible to retain the largest sample while assuring an acceptable fitting.

Such an optimum threshold may differ from site to site, affecting consequently not only the GPD scale

parameter, but also the probability of threshold exceedance.

Thus a first objective of this paper is to derive some expressions to parameterize a simple threshold-

invariant three-parameter distribution function which is able to describe zero and non zero values of

rainfall time series by assuring a perfect overlapping with the GPD fitted on the exceedances of any

threshold larger than the optimum one. Since the proposed distribution does not depend on the local

thresholds adopted for fitting the GPD, it will only reflect the on-site climatic signature and thus appears

particularly suitable for hydrological applications and regional analyses.

A second objective is to develop and test the Multiple Threshold Method (MTM) to infer the parameters

of interest on the exceedances of a wide range of thresholds using again the concept of parameters

threshold-invariance. We show the ability of the MTM in fitting historical daily rainfall time series

recorded with different resolutions. Finally, we prove the supremacy of the MTM fit against the standard

single threshold fit, often adopted for partial duration series, by evaluating and comparing the

performances on Monte Carlo samples drawn by GPDs with different shape and scale parameters and

different discretizations.

Mind the bias! Simon M. Papalexiou, Demetris Koutsoyiannis

Department of Water Resources, National Technical Univerity of Athens, Athens, Greece.

E-mail: [email protected], Tel: +30-210772-2845; fax +30-210772-2832

A. Montanari

Faculty of Engineering, University of Bologna

Most statistical procedures, including parameter estimation and hypothesis testing, are based on the tacit

assumption that the observations are outcomes from independent random variables. This is generally not

true for geophysical processes, which usually exhibit a strong temporal dependence. Such dependence

implies substantial negative bias in the estimation of statistical parameters of dispersion, e.g., variance, as

well as parameters of dependence, e.g., autocorrelation. Failure to account for this bias leads to a distorted

picture of the underlying process and results in significant errors in modelling and prediction. Here we

demonstrate the impact of neglecting dependence in parameter estimators by using synthetic examples

from stochastic processes with short- and long-range dependence, as well as rainfall datasets that exhibit

high temporal dependence. We also propose a methodology to account for the bias therefore improving

model performances and design reliability.


7

Can hydrological model predictions be improved by developing

streamflow measurement error models using rating curve data? Mark Thyer

School of Engineering, University of Newcastle, Newcastle, Australia.

E-mail: [email protected] Tel: +61 2 4921 6057 fax +61 2 4921 6991

Ben Renard

Cemagref, Hydrology-Hydraulics, Lyon, France

Dmitri Kavetski and George Kuczera

School of Engineering, University of Newcastle, Newcastle, Australia.

Efforts to improve hydrological predictions are often thwarted by several sources of error, including (i)

data uncertainty, e.g, input (rainfall) errors and output (runoff) errors; and (ii) structural uncertainty in the

hydrological model. The Bayesian total error analysis methodology was introduced to overcome these

challenges and provides a framework which can explicitly account for each source of error. While several

studies have shown that accounting for rainfall error notably improves hydrological predictions, the runoff

error models used are based on unduly simplistic assumptions (e.g 10% random error). This study will use

streamflow rating curve data to develop probabilistic model of streamflow measurement error prior to the

calibration of the hydrological model. Several diagnostic tools aiding the runoff error model development

are presented. Different types of runoff error models (with both random and systematic errors) are

developed using the rating curve data. These runoff error models are incorporated into hydrological model

calibration using the BATEA methodology and a comparative analysis of the calibration results

(predictive time series, model parameters and state variables) is presented, thoroughly scrutinizing its

assumptions and comparing them to commonly error model assumptions. Several catchment case studies

are compared to evaluate the benefits under diverse hydrological regimes. The implications for improving

hydrological predictions via disentangling structural errors from data errors is discussed.

Information theory, uncertainty and risk for evaluating hydrologic

predictions Steven Weijs and Nick van de Giesen

Department of Water Resources Management, Delft University of Technology, Delft, The Netherlands.

E-mail: [email protected], Tel: +31 15 2782345; fax +31 15 2785559

Over the last decades, probabilistic forecasting has become increasingly popular in the field of hydrology.

Lacking enough information to completely eliminate uncertainty, probabilistic forecasts are intended to

reduce uncertainty of the user about the future events. A logical criterion for evaluating the quality of such

forecasts is therefore the remaining uncertainty about the truth, estimated from observations. Minimization

of this criterion can be used as an objective for calibration.

In this work, we make use of the information-theoretical definitions of uncertainty and information. We

demonstrate how Kullback-Leibler divergence can be used to measure remaining uncertainty in a model

outcome. Furthermore, we present a new three-way decomposition of this uncertainty, analogous to the

commonly used decomposition of the Brier Score and derived scores into uncertainty, resolution and

reliability. We show that these components are 2nd

order approximations of their information-theoretical




8

counterparts, which can be interpreted as inherent uncertainty, right information and wrong information.

Based on these insights, we argue that probabilistic model output is necessary to get the maximum utility

from forecasts for users (risk reduction), to correctly infer parameter distributions and to let a model learn

from data to improve predictions.

Utility, in contrast to information, is user dependent. The risk reduction that the forecast can provide to the

user depends on what is at stake for him. One might therefore argue that the user of a forecast should be

the one to evaluate it. Though this is true if evaluation is for his own purposes, we argue that calibrating or

improving models on the basis of a user-specified objective function is not always a good idea. The data

processing inequality tells us that if the information in observations is filtered through a user specific

mapping from information to utility, the model will have less data to learn from. This might require the

model structure to be reconsidered. User-oriented model calibration should therefore only be used in

combination with user oriented modeling.

An upper limit to seasonal precipitation predictability?

Seth Westra and Ashish Sharma


University of New South Wales, Sydney, Australia

Email: [email protected], [email protected],

Tel: +61 2 9385 5140, +61 2 9385 5679

The asymptotic predictability of global land-surface precipitation is estimated empirically at the seasonal

time scale with lead times from zero to 12 months. Predictability is defined as the unbiased estimator of

predictive skill using a given model structure assuming that all relevant predictors are included, thus

representing an upper bound to the predictive skill for seasonal forecasting applications. To estimate

predictability, a simple linear regression model is formulated based on the assumption that land-surface

precipitation variability can be divided into a component forced by low-frequency variability in the global

sea surface temperature anomaly (SSTA) field and which can theoretically be predicted one or more

seasons into the future, and a „weather noise‟ component which originates from nonlinear dynamical

instabilities in the atmosphere and which is not predictable beyond about 10 days. Asymptotic

predictability of global precipitation was found to be 14.7% of total precipitation variance using 1900-

2007 data, with only minor increases in predictability using shorter and presumably less error-prone

records. This estimate was derived based on concurrent SSTA-precipitation relationships, and therefore

constitutes the maximum skill achievable assuming perfect forecasts of the evolution of the SSTA field.

Imparting lags on the SSTA-precipitation relationship, the 3-, 6-, 9- and 12-month predictability of global

precipitation was estimated to be 7.3%, 5.4%, 4.2%, 3.7%, respectively, demonstrating the comparative

gains that can be achieved by developing improved SSTA forecasts compared to developing improved

SSTA-precipitation relationships. Finally the actual average cross-validated predictive skill was found to

be 2.1% of the total precipitation variance using the full 1900-2007 dataset, and was dominated by the El

Niño Southern Oscillation (ENSO) phenomenon. This indicates that there is still significant potential for

increases in predictive skill through improved parameter estimates, use of longer and/or more reliable

datasets and the use of larger spatial fields to substitute for limited temporal records.




9

On accuracy of upper quantiles estimation Iwona Markiewicz (Corresponding author)

Department of Hydrology and Hydrodynamics, Institute of Geophysics Polish Academy of Sciences,

Ksiecia Janusza 64, 01-452 Warsaw, Poland.

E-mail: [email protected],Tel: +48 22 6915804; fax: +48 22 6915915

Witold G. Strupczewski



E-mail: [email protected]

Krzysztof Kochanek




Flood frequency analysis (FFA) entails estimation of the upper tail of a probability density function of

annual peak flows obtained from either the annual maximum series or partial duration series. In practice

one deals with „false‟ model and is not able to assess the magnitude of model error of any estimated

statistics. To that end, the „true‟ PDF together with its parameter values should be known. The model bias

of large quantile estimates depends on both the estimation method and sample size. The interest is to find

a most robust estimation method for large quantiles in the case of both true and false distribution

assumption. In the paper the method of moments (MOM), the method of linear moments (LMM), the

maximum likelihood method (MLM) and the method based on mean deviation (MDM) are examined. The

last method is innovative, it has not been widely used so far. Basing on simulation experiments, the

accuracy of large quantile assessments obtained from the four estimation methods are compared for two-

parameter distributions log-normal, log-Gumbel and their three-parameter counterparts, i.e., three-

parameter log-normal and GEV distributions. The accuracy of large quantile estimate is expressed by the

relative root mean square error and the relative bias. Then the asymptotic and sampling errors of the upper

quantile estimates are assessed for the pair log-Gumbel as false and log-normal as true model. Comparing

the errors of the assessments of large quantile of various distributions with the same values of moments, it

is evident that their values depend on the sample size, on the distribution type, both „true‟ and „false‟, and

strongly depend on the estimation method. Hence, knowing both the parent distribution which is assumed

for describing the analyzed peak flow series and the properties of various estimation methods, one can

select an appropriate procedure of upper quantile estimation.





10

Session III

Time series analysis and multivariate methods for hydrological

modelling and forecasting

Interpolation and simulation of Daily Rainfall observations based on

Copulas and Circulation Patterns András Bárdossy

1 and Geoff Pegram

2

1Institute of Hydraulic Engineering, University of Stuttgart, Stuttgart, 70569, Germany

2Civil Engineering Program, University of KwaZulu-Natal, Durban, South Africa

Given a record of daily rainfall depths recorded for some years over a network of gauges, the purpose of

this work is to describe a method of spatial interpolation using two novel approaches. The first is to

describe the spatial dependence structure of the parameters of the rainfall data with copulas, whose

properties can be interpolated over the intervening space between the gauges with the help of Ordinary

and External Drift Kriging, both in the rain domain and also in the quantile domain. The second

innovation is to use daily Circulation Patterns (Sea Level Pressure anomalies) to refine the interpolation,

using the extra information they provided of each day‟s rainfall type (widespread or intermittent) and the

corresponding direction of the air flow. In addition to checking the performance of the procedure by

classical cross validation and split sampling, the estimation the uncertainty of the interpolated estimates is

also evaluated. The methodology is applied to a test area in Southern Germany on a network of

approximately 1000 good stations with 40 years of record.

Multivariate Extreme Value models in hydrology: a copula approach Carlo De Michele

Politecnico di Milano, Italy

[email protected]

Gianfausto Salvadori

Dipartimento di Matematica, Università del Salento, Lecce (Italy)

[email protected]

Multivariate Extreme Value models are a fundamental tool in hydrology in order to assess potentially

dangerous events. For instance, the analysis of the 2003 severe drought event, or of the 1994 catastrophic

food event, occurred over the Po river basin (Northern Italy) cannot be addressed completely by

considering only the data collected at a single river section, e.g. at Pontelagoscuro gauge station close to

the outlet.

As frequently stressed in hydrologic literature, the statistical analysis of multivariate extremes is difficult,

essentially due to (1) the complexity of the phenomena, (2) the reduced sample size of the actual

multivariate datasets, and (3) the availability of suitable multivariate probability distributions.

The target of this work is twofold. On the one hand we outline how, exploiting recent theoretical

developments in the theory of Copulas, new models can be easily constructed: in particular, we show how




11

a suitable number of parameters, having a physical meaning, can be introduced, a feature not shared by

traditional Extreme Value models. On the other hand, we suggest several strategies in order to estimate the

parameters of interest according to different criteria: these may use either a nearest neighbor or a nearest

cluster approach, or exploit the pair-wise relationships between the available gauge stations. An

application to food data is also illustrated and discussed.

Design flood hydrographs from the relationship between flood peak

and volume L. Mediero, L.Garrote

Department of Civil Engineering: Hydraulic and Energetic. Universidad Politécnica de Madrid, Madrid,

Spain.

E-mail: [email protected], Tel: +34 91 336 5261

A. Jiménez

Centre for Hydrographic Studies of CEDEX, Madrid, Spain.

Hydrologic frequency analysis is usually focused on flood peaks and there is a wide literature about this

topic, but flood volumes have not been so exhaustively studied. Flood volumes are needed to know flood

hydrographs, being these essential for designing some structures like dams. Joint distribution of flood

peaks and volumes in hydrographs, to obtain the conditional return periods, has been solved by means of

bivariate distributions with functions like Normal, Exponential and Gumbel. Typical flood series in arid

and semi-arid regions are very skewed, as most of Spanish basins. A Three-parameter function is required

to fit these peak and volume series. Bivariate distributions with three-parameter functions are not easy to

fit. Relationship between peak discharge and hydrograph volume has been analyzed to define the design

hydrograph in terms of dam safety.

An Annual Maximum Volumes (AMV) analysis has been carried out. Firstly, hydrographs have been

identified by a joint criterion based on independence between consecutive peaks and identification of the

start point of the rising and recession limbs. It has been seen that some of the hydrographs associated with

the maximum volume in a year have long durations and low peaks, but these hydrographs are not relevant

for the dam safety. The maximum hydrograph volume in a year should be that which leads to the

maximum reservoir water level after being routed. Homogeneity of AMV series has been tested by

homogeneity measures used in flood frequency analysis. Regionalization of L-moments has been achieved

in order to improve the fitting of the frequency curve from recorded data. Since the conditional return

period is represented by a bivariate distribution with two three-parameter distributions that can not be

fitted, a strong relationship between peak and volume has been found.

The analysis has been carried out over the mean daily discharge series recorded in all the unrouted gauge

stations in Spain. These data have been ranked in homogeneous regions. Regionalization of AMV in each

region has been achieved. Relationship between peak discharge and hydrograph volume has been

identified in the log-log space. For a T-peak discharge only a short range of volumes are feasible and a set

of hydrographs can be obtained by means of Monte Carlo simulation. These hydrographs are routed. A

maximum water level in the reservoir leads to a curve in the peak-volume space, which represent a set of

hydrographs that have the same risk for the dam. These hydrographs are the design hydrographs for the

dam.


12

Properties and performance of a copula-based design storm

generator S. Vandenberghe

1, N.E.C. Verhoest

1 and B. De Baets

2

1Laboratory of Hydrology and Water Management, Ghent University, Coupure links 653, B-9000 Ghent,

Belgium, Email: [email protected], Tel: +32 9 264 61 37; Fax +32 9 264 62 36 2Department of Applied Mathematics, Biometrics and Process Control, Ghent University, Coupure links

653, B-9000 Ghent, Belgium

In many hydrological and hydraulic applications, such as the construction of hydraulic structures or flood

risk assessment, the use of design storms can be very useful. In this study, we propose a methodology to

generate ensembles of design storms, defined by a storm depth and a storm duration for a given return

period and with a representative internal storm structure. Therefore, the concept of a copula-based

secondary return period is combined with the concept of a mass or Huff curve. The performed analyses

make use of storms, which are selected out of a 105-year 10-minute rainfall time series observed at Uccle,

Belgium.

Firstly, a copula is fitted to the dependence structure between the storm depth D and storm duration W. In

combination with some marginal distribution functions for D and W, the copula then allows for a random

simulation of a desired number of storm events having a specific secondary return period. Concerning this

return period, we provide a discussion on which random variables are best used to describe storms in the

context of hydrological design. Also attention is paid to the implications of the misspecification of the

copula on the calculation of the secondary return period.

Secondly, an algorithm is developed that allows for an easy generation of random internal storm structures

based on the analysis of Huff curves of the Uccle storms. These internal storm structures are then

superimposed to the randomly generated couples of (D,W) in order to obtain the actual design storms.

Special attention is paid to the independence between the internal storm structure and the secondary return

period.

Finally, also an evaluation of the usefulness of the design storm generator is provided through the

evaluation of simulated discharges by means of a hydrological rainfall-runoff model.

Stochastic models for rainfall nowcasting

Davide Luciano De Luca, Pasquale Versace

Department of Soil Conservation, University of Calabria, Rende (CS), Italy.


Rainfall nowcasting constitutes an essential aspect in the realization of Early Warning Systems related to

mitigate the hydrogeological risk. In fact, many catastrophic hydrological events induced by rainfall, like

floods, flash floods and shallow landslides (debris and mud flows), are characterized by a very small

response time with respect to precipitation input.

Consequently, the implementation of suitable mathematical models for nowcasting of the precursor plays

an important role in the evaluation of occurrence probability referred to induced phenomena.

The paper describes the models developed at CAMILab laboratory of Department of Soil Conservation

(University of Calabria). In details, nowcasting can be carried out by using an at-site stochastic model

named PRAISE (Prediction of Rainfall Amount Inside Storm Events), a space-time stochastic model

named PRAISEST (Prediction of Rainfall Amount Inside Storm Events: Space and Time), and a coupled




13

meteorological and stochastic model named PRAISE-ME (Prediction of Rainfall Amount Inside Storm

Events with MEteo).

PRAISE is based on the assumption that the rainfall height H , to predict in the nowcasting temporal

interval, is correlated with a variable named Z , representing antecedent precipitation; a bivariate

probability distribution zhf ZH ,, is developed, which takes into account the mixed nature of random

variables, that is a finite probability in correspondence to the null value and infinitesimal probabilities in

correspondence to the positive values

PRAISEST constitutes the space-time generalization of PRAISE; the prediction of rainfall height H ,

referred to a spatial cell of size yx , depends on antecedent precipitation Z at the same site, and on

rainfalls of neighbour cells, represented by a variable named W . Similarly to PRAISE, a trivariate

probability distribution zwhf ZWH ,,,, is developed.

PRAISE-ME represents a Bayesian approach that allows the coupling between a stochastic model and

meteorological information. Output is constituted by a probability distribution of at site precipitation for a

given temporal interval of nowcasting, which is a revision of a prior density, deriving by the stochastic

model PRAISE, through the Bayesian integration with the prediction given by a meteorological model.

The improvement, with respect to PRAISE, is in the case of “outside storm event” before nowcasting.

Applications related to Calabria region are presented, to test performances of all the models.

PREVIVAZ – Improving weekly streamflow time series forecasts

with the current hydrologic state of the river basin Saulo Aires de Souza

1, Fernanda da Serra Costa

1,2, Luciano Nóbrega R. Xavier

1, Maria Elvira

Piñeiro Maceira1,2

, Jorge Machado Damázio1,2

1Electric Power Research Center – CEPEL, Rio de Janeiro, Brasil

2Rio de Janeiro State University - UERJ, Rio de Janeiro, Brasil

E-mail: [email protected], [email protected], [email protected], [email protected], [email protected], Tel:

+55 21 25986471; fax +55 21 25986471

Hydraulic generation provides more than 90% of Brazilian total power generation. In order to cope with

natural streamflow uncertainty and assure a more reliable energy supply, Brazilian Power System has a

great number of large dams with inter-annual regulation capacity. Streamflow forecasting is thus a

fundamental issue to enhance system‟s operation, increasing benefits and reducing overall costs.

The PREVIVAZ model is based upon univariate linear time series modeling, allowing the forecast of

weekly inflows to each hydroplant in a horizon up to 6 weeks. Taking into account reservoirs‟ inflow time

series, different univariate PARMA(p,q) models are analyzed, and the one that presents the minimum

RMSE are retained, being used for weekly forecasting. This model has been used in the short term

hydrothermal planning of the Brazilian Power System since 1998. During this period, it was observed that

for some hydroplants, the forecasts provided by the model indicate hydrological unlike values when faced

to the recent observed inflows.

PREVIVAZ model was recently improved by taking into account the inflows time series‟ rise and

recession limbs. By the analysis of inflow time series, the empirical probability density function of inflow

variations between two consecutive weeks is built. Probabilities of minimum and maximum probable

inflow variations are to be defined from the watershed hydrologic behavior. Minima and maxima inflow

variation limits are then calculated taking into account these threshold values, the inflow variations

probability density function, and the last observed inflows. These two limits are then considered on the

choice of the best PARMA(p,q) model to be used thereafter.






14

This new procedure was evaluated in 4 Brazilian watersheds, one located on North region (Tocantins

river) and the other ones on South region (Paraná, Uruguai and Jacuí rivers). Streamflow time series in

Brazilian South region are known to present weak temporal correlation structure thus a weak previsibility,

and they were therefore object of a closer attention herein. Model results were evaluated based on RMSE

and Nash-Sutcliffe metrics, and different inflow variation‟s probabilities of non-exceedance were taken

into account. It has been verified that the new procedure yields to better results on the two considered

metrics, and that the error reductions for the 3 watersheds on Brazilian South region were remarkable.

Moreover, more restrictive variation limits yields to a better performance, which could be particularly

seen in recession periods.

Rainfall event variables: dependence analysis by way of copulas Baldassare Bacchi

Department of Civil Engineering Architecture Land and Environment, University of Brescia, Brescia,

Italy.


Matteo Balistrocchi

Department of Civil Engineering Architecture Land and Environment, University of Brescia, Brescia,

Italy.


Three precipitation time series recorded in different Italian climates were analyzed for extracting samples

of individual rainfall events. So, the total rainfall depth, the wet weather duration and the interevent dry

time were computed for every event and assumed to be realization of as many random variables. The

dependence structures of the three pairs were then analyzed within the framework of the copula approach.

A not negligible positive concordance was evidenced by the volume-duration pair, for which the best

sample frequency fit was obtained through the Gumbel-Hougaard copula function. The interevent time

though showed to be reasonably independent of both the volume and the duration, suggesting the use of

independence copula functions. Test statistics were also performed for appraising the goodness-of-fit. The

trivariate joint probability function was finally derived, by assuming Weibull probabilistic models for the

three marginal distributions.




15

Session IV

Analysis and modelling of hydrological processes

Multivariate skew-t description of flood events Giorgio Roth

University of Genoa, Genoa, Italy

E-mail: [email protected]; Tel: +39 010 3532983; Fax: +39 010 3532948

Tatiana Ghizzoni

MunichRe, Munich, Germany

Roberto Rudari

Cima Foundation, Savona, Italy

One of the main consequences of the demographic and economic development and of markets and trades

globalization is represented by risks cumulus. In most cases, the cumulus of risks intuitively arises from

the geographic concentration of a number of vulnerable elements in a single place. For natural events,

risks cumulus can be associated, in addition to intensity, also to event‟s extension. In this case, the

magnitude can be such that large areas, that may include many regions or even large portions of different

countries, are stroked by single, catastrophic, events.

Among natural risks, the impact of the flooding hazard cannot be understated. To cope with, a variety of

mitigation actions can be put in place: from the improvement of monitoring and alert systems to the

development of hydraulic structures, throughout land use restrictions, civil protection, financial and

insurance plans. All of those viable options present social and economic impacts, either positive or

negative, whose proper estimate should rely on the assumption of appropriate – present and future – flood

risk scenarios. It is therefore necessary to identify proper statistical methodologies, able to describe the

multivariate aspects of the involved physical processes and their spatial dependence.

In hydrology and meteorology, but also in finance and insurance practice, it has early been recognized that

classical statistical theory distributions (e.g., the normal and gamma families) are of restricted use for

modeling multivariate spatial data. Recent research efforts have been therefore directed towards

developing statistical models capable of describing the forms of asymmetry manifest in data sets. This, in

particular, for the quite frequent case of phenomena whose empirical outcome behaves in a non‐normal

fashion, but still maintains some broad similarity with the multivariate normal distribution. Fruitful

approaches were recognized in the use of flexible models, which include the normal distribution as a

special or limiting case (e.g., the skew‐normal or skew‐t distributions).

The present contribution constitutes an attempt to provide a better estimation of the joint probability

distribution able to describe flood events in a multi‐site multi‐basin fashion. This goal will be pursued

through the multivariate skew‐t distribution, which allows to analytically define the joint probability

distribution. Performances of the skew‐t distribution will be discussed with reference to the Tanaro River

in Northwestern Italy. To enhance the characteristics of the correlation structure, both nested and

non‐nested gauging stations will be selected, with significantly different contributing areas.


16

Study on the rainfall dependence structure using radar and rain

gauge data Valeria Montesarchio, Fabio Russo, Francesco Napolitano

Dipartimento di Idraulica, Trasporti e Strade, Sapienza Università di Roma, Italy.


Federico Lombardo

Dipartimento di Scienze dell’Ingegneria Civile, Università degli Studi – Roma Tre, Italy.

Luca Baldini

Istituto di Scienze dell'Atmosfera e del Clima, Consiglio Nazionale delle Ricerche, Italy.

Rainfall processes are characterized by high variability in space and time. Integrating information coming

from different sources, such as rain gauges and weather radars, can be useful to better understand the

rainfall behaviour in the space-time domain. In fact, rain gauges can directly provide quantitative rainfall

measurements for fine time scales, but the information is only punctual. On the other hand, weather radars

provide high space-time resolution data, but the rainfall amounts are indirectly estimated.

The paper investigates on properties of rainfall over the North Lazio Region, in Central Italy. The

variability of rainfall is examined at time scales from 30 min to 12 hours using a dense rain gauge network

and weather radar data.

The dependence structure of rainfall (time and space correlation between time series for different time

scales) is examined. The analysis are performed first separately for the rain gauge data set (considering the

correlation between pairs of rain gauges) and the weather radar data set (considering the correlation

between pixels). Then, the correlation between radar and gauge data is examined. The study ends with a

discussion of results in order to highlight correlation patterns and decorrelation distances useful in a

framework of stochastic model for multisite rainfall simulation.

Attempts to flow regime classification and characterisation in

Mediterranean streams using multivariate analysis Oueslati O.

Water Research Institute, National Research Council, Bari, Italy

Department of Engineering and Physics of the Environment, University of Basilacata, Potenz, Italy

E-mail: [email protected] / [email protected], Tel: +39 0805820511; fax +39 0805313365

De Girolamo A.M


E-mail: [email protected], Tel: +39 0805820511; fax +39 0805313365

Abouabdillah A.


Department of Science, Engineering and Technologies for the Environmental Management University of

Tuscia, Viterbo, Italy


Lo Porto A.



Mediterranean rivers are characterized by the irregularity of flow and harsh hydrological fluctuations.

Extreme seasonal variation of the flow regimes often causes a marked pattern of zero or low flow.








17

However, until now, no river classification was available for the whole Mediterranean water courses

whereas many classifications was developed at country or regional scale. In this study, a long-term flow

records from 60 sites across the Mediterranean basin were analyzed to classify streams into flow regime

groupings. It is considered that these catchments provide a good representation of various flow regimes

prevailing across the Mediterranean basin. Six hydrological variables based on flow intermittency,

variability and predictability were used for this classification. Cluster analysis indicated that streams could

be classified broadly into perennial and intermittent. To get more verified information about the

hydrological regimes of Mediterranean rivers, 38 hydrological indices (HIs) were derived for each stream.

The principal component analysis (PCA) was then applied for the two river types and for all streams

combined to identify subsets of HIs that best describe each stream type while minimizing redundancy. The

results from the PCA applied on the three sets of streams allow us to extract three principal components,

each associated with a fundamental characteristic. First and second principal components for the

intermittent cluster describe respectively the intermittence and variability facets of the flow regime while

for perennial cluster they describe high flow condition and rate of change (rise rate and fall rate). Finally

linear and canonical correlation coefficients were calculated between the significant components extracted

for each stream type and catchment descriptors (area, latitude and longitude). This allows us to reveal the

descriptors that influence the spatial variability of the flow regime in the Mediterranean area.

How does the El Niño Southern Oscillation (ENSO) and other

climate modes affect multi-scale temporal rainfall variability in East

Australia?

Alexander Pui, Seth Westra, Ashish Sharma



Email: [email protected], [email protected], [email protected],

Tel: +61 9385 5040, +61 2 9385 5140, +61 2 9385 5679

While there have been several well documented studies showing shifts in seasonal mean rainfall

conditional on to ENSO phase, the nature of rainfall variability at finer temporal scales remains poorly

understood. Fine scale temporal rainfall variability is important for a range of hydrological applications

such as design flood estimation (via estimation of both short-duration storm bursts and associated

antecedent moisture conditions), agricultural purposes (number of wet days per growing season) among

others. As such, pertinent questions regarding the behaviour of daily to sub-daily rainfall characteristics

conditional to the ENSO phenomenon as well as other low frequency climate modes namely the Indian

Ocean Dipole and Southern Annular Mode are addressed in this study. A simple linear regression model is

formulated to estimate the „elasticity‟ of daily and sub-daily rainfall as a function of each of these climate

modes. For example, given a one standard deviation departure from the mean of a common index of

ENSO (the Southern Oscillation Index (SOI)), is the change in total seasonal rainfall amount mostly

attributable to changes in wet day amounts or rainfall per wet day? Similarly, are changes in daily rainfall

caused by greater number of individual rain events, or simply the same number of rain events that persist

for longer period (longer wet spells)?

Utilizing data from 166 high quality daily rainfall stations (1920 – 2001) across the continent, the effect of

ENSO on East Australian rainfall was found to be most pronounced during the winter and spring seasons,

with a 17.8% and 17.0% increase respectively per +1 standard deviation from the mean SOI value.

Interestingly, changes in rainfall amount within each season were found to be driven more predominantly

by changes in number of wet days rather than rainfall per wet day, with the notable exception of the




18

summer season. Finally, at a sub daily scale (using data from 47 sub daily stations), we found that ENSO

appeared to influence only changes to number of wet spells and length of wet spells, with negligible

changes to rainfall intensity per wet spell. This result helps to explain the observed changes at lower time

resolutions, provides some preliminary insight into the physical mechanisms associated with ENSO, and

has important ramifications for flood design via the effect on antecedent catchment conditions by the

reduction of time between consecutive wet spells during La Niña events. Results from the effect of other

climate modes, both in isolation and when combined, will also be discussed.

Analysis of sub-daily rainfall sequences based upon a deterministic

multiplicative cascade method Li-Pen Wang

*, C. Onof and Č. Maksimović

Department of Civil and Environmental Engineering, Imperial College London, UK. *E-mail: [email protected], Tel: +44 778 0962088

Multiplicative cascade methods have been widely developed to model spatial and temporal distributions of

rainfall in literature by investigating their scale-invariant features (Deidda et al., 1999; Tessier et al.,

1993). Random multiplicative cascade methods are the mainstream in this research area to simulate

cascading by constructing rainfall generators to distribute rainfall volumes on successive regular

subdivisions of an interval in a randomly multiplicative manner (Molnar and Burlando, 2005). However,

the type and parameters of generators strongly depend upon the probability distribution of specific rainfall

features. This indicates that a generator may radically alter when a different rainfall field/sequence starts.

Thus, it is difficult to find a promising and general generator.

A possible approach to deriving a general generator could be carried out by re-investigating the forming of

multiplicative cascades. It is a process to generate randomness by subdividing a unit set into smaller and

smaller subsets according to a fixed set of contracting ratios and at the same time subdividing the

associated unit measure by another set of contracting ratios. In other words, a random process could be

implemented based upon a self-similar process with deterministic fragmentation ratios. This can be simply

termed by the generating equation:

11

)(

b

i

qi

qi ps

,

where si and pi denote the ratios that are used to respectively fragment the unit set and measure into b

components, and τ(q) is the scaling feature curve. A deterministic cascade method has been implemented

(Wang et al., 2009) by solving this equation. However, the associated results show that it is unable to

consistently reproduce statistics of the selected rainfall events due to insufficient understanding of the

structure of the deterministic cascade. This research thus will focus on analysing the behavioral altering of

rainfall sequences over the scales of interest (e.g. 5 minutes to 1 day) by investigating the probability

distribution of rainfall volumes between successive cascades/timescales. Rainfall data in 5-min resolution

from two rain gauges in London (Greenwich and Stansted) will be used to implement the proposed study.

This work is expected to improve the applicability of a deterministic cascade method in sub-daily rainfall

modelling.



19

Session V

Extreme events in geophysical processes

Use of Empirical Mode Decomposition for the prediction of

hydrometeorological extremes Tae Sam Lee, Taha B.M.J. Ouarda

Canada Research Chair on the Estimation of Hydrometeorological Variables, INRS-ETE, 490 De La

Couronne, Québec, QC, Canada. email: [email protected], Tel : +(418) 654-3842; fax : +(418)

654-2600

Nonstationary oscillations in climatic variables and indices have been the focus of many studies on the

possible causes of the internal ocean-atmosphere coupled system and the external solar radiation heating.

Since climate indices and their associated hydro-meteorological variables might contain nonstationary

oscillation processes, it would be useful to be able to divide the intrinsic nonstationary oscillation into a

finite number of components since those components are able to predict the future evolution.

We propose in the present work to extract nonstationary oscillations of hydrometeoroloigical variables by

using a decomposition analysis called the empirical mode decomposition (EMD). The important

components are then modeled with a nonstationary oscillation resampling (NSOR) technique. For this

objective, time series with long records are required. The theoretical background for the proposed

approach is first presented. A first application is presented for the regional precipitation of Eastern

Canada for which the averaged point measurements are scaled by subtracting the mean and dividing the

standard deviation. In a second example, the future evolution of extreme streamflows at two stations in

the Province of Quebec, Canada, is studied by using the long-term patterns of climatic indices. Results

indicate that NSOR and EMD lead to well modeled future long-term patterns. However, the indirect

approach to finding the interconnection sometimes gives rise to a high prediction uncertainty.

Comparison of high return period flood-quantiles estimated with

upper bounded distribution functions, including Non Systematic

information. Application to a Mediterranean river Blanca Botero

Facultad de Ingeniería y Arquitectura, Universidad Nacional de Colombia sede Manizales. Manizales,

Colombia.

E-mail: [email protected], Tel: +57 6 8879300 ext 50317; fax +57 6 8879388

Félix Francés

Instituto de Ingeniería del Agua y Medio Ambiente, Universidad Politécnica de Valencia, Valencia, Spain.

Three distribution functions with upper bound were used to estimate high return period flood-quantiles in

three typical Mediterranean rivers. The parameters of the EV4 (Extreme Value with four parameters),

LN4 (Lognormal with four parameters) and TDF (Transformed distribution function) distribution

functions were estimated by Maximum likelihood method, using Systematic and additional Non

systematic information available.




20

The present paper compares the different flood-quantile estimated with each one of the upper bounded

distribution functions, and in addition with other two common distributions, the TCEV and GEV (Two

Component extreme value and General extreme value).

Results of the frequency analysis with Non systematic information, using distribution functions with upper

bound were successful. Comparison among high return period flood-quantile estimates shows a different

performance between quantiles estimated with the different distribution functions.

Probabilistic definition of drought risk indicators for water supply

systems in Central Tuscany

Giuseppe Rossi

Department of Civil and Environmental Engineering, University of Firenze, Firenze, Italy.

E-mail: [email protected], Tel: +39 0554796306; fax +39 0554796306.

Luis Garrote

Department of Hydraulic and Energy Engineering, Technical University of Madrid, Madrid, Spain.

Enrica Caporali


Drought indices, particularly the meteorological ones, can describe the incoming and the persistency of

droughts, especially in natural systems. However they have to be used cautiously when applied to water

supply systems. They show little correlation with water shortage situations, since water storage play an

important role in water resources management. For that reason a more complex system of indicators is

required in order to identify situations when there is risk of water shortages. In water supply systems there

is great variability on the natural water resources that flow in rivers, aquifers or into the reservoirs and also

on the demands. These quantities can only be defined probabilistically. This great variability is faced

defining some threshold values, expressed in probabilistic terms, that measure the hydrologic state of the

system. They can identify specific actions in an operational context in different levels of severity, like the

normal, pre-alert, alert and emergency scenarios. The threshold values are defined considering the

probability to satisfy a given fraction of the demand in a certain time horizon, and are calibrated through

discussion with water managers. A simplified model of every water resources system is built to evaluate

the threshold values and the management rules. The probability distribution of demand shortages is

estimated through Monte Carlo simulation, considering an ensemble of hydrological inflows for each

storage level in the reservoirs. The threshold values are validated with a long term simulation that takes

into account the characteristics of the evaluated system. The levels and volumes in the different reservoirs

are simulated using 20-30 years time series. The critical situations are assessed month by month in order

to evaluate optimal management rules during the year and avoid conditions of total water shortage. The

methodology is applied to the urban area Firenze-Prato-Pistoia in central Tuscany, in central Italy. The

catchment of the investigated area has a surface of 1231 km2 and, accordingly to the census ISTAT 2001,

945˙972 inhabitants.



21

Roundtable: The legacy of Prof. V. Yevjevich

BIOGRAPHICAL PROFILE OF VUJICA YEVJEVICH*

by Giuseppe Rossi, University of Catania

Born on October 12 , 1913 in Kasidoli, a small Serbian village, he was the son of

a minister of the Serbian Eastern Orthodox Church. After attending the primary

school in Priboj and the high school in Sarajevo, he graduated in Civil

Engineering at the University of Belgrade in 1936. After the military service and

a professional experience of hydraulic engineering works in Skopje, Macedonia,

he attended the School for Hydraulic Engineers of University of Grenoble, with

a fellowship by the French Government.

When Great Britain declared war to Germany and Italy in September 1939, after

Germany invaded Poland, he left London, where he was visiting some hydraulic

laboratories, and suddenly returned to Grenoble and then, in Yugoslavia, just

before the France-Italy frontier was closed due to the war declaration of Italy to

France.

Until 1941 he was involved, as hydraulic engineer, in several hydraulic works in the Vardar River basin.

After the invasion of Yugoslavia by German forces, he was a prisoner of war and spent the period from

April 1941 to September 1943 in German and Italian prisoner‟s camps, where he had the opportunity to

read a relevant number of books on technical issues in water resources area thus also improving

knowledge of languages. “The hard war times, - Yevjevich remarks in his autobiography‟s manuscript -

became a forced educational bonanza to me”.

When the armistice was signed between the Italian government and the Allied Forces, he left the

Cortemaggiore camp, before German troops re-occupied the camp, and went to Rome, where he spent

nine months, under the false name of Giovanni Bossera. During his stay in Rome he improved his

knowledge of Italian language and had contacts with the “Resistence”, contributing to help persecuted

political men through the preparation of false identity cards, as he liked to remember. After the liberation

of Rome, in June 1944, he was transferred to the island of Vis in Dalmatia, where the Provisional

Yugoslav Government had its liberated territory, and, later, to Belgrade.

From 1944 to 1958, he carried out an intense professional activity in Yugoslavia, in the fields of design

of hydropower plants, planning of water resources systems and hydraulic research. He was director of the

Water Resources Research Institute of Belgrade from 1949 to 1956 and advisor for water resources and

hydropower problems in the Union of Yugoslav Electric Power Authorities. Since 1946 he taught

hydrology and hydroelectric power engineering at the University of Belgrade and in 1955 he obtained the

Doctor degree by the Serbian Academy of Sciences and Arts. In 1956 he published a book on hydrology

(in Serbo-Croatian), which can be considered one of pioneering attempts to shift hydrology from a

descriptive discipline to a scientific one.

After the invitation of U.S. National Bureau of Standard, he moved in 1958 to Washington, D.C., - with

his wife Mirijana and the three daughters-, for a research period, which was planned to be 18 months,

while it lasted 48 years! His first researches dealt with water outlet waves of dam breach and analysis of

hydrological time series, which in the following period at U.S. Geological Survey evolved into different

subjects of stochastic hydrology and an extensive annotated bibliography on “Flood routing methods”.

* translated from “L‟Acqua”( Journal of the Associazione Idrotecnica Italiana) , no. 3, 2006


22

Afterwards, he moved to Colorado State University (CSU), Department of Civil Engineering, where he

developed a graduate and research program in hydrology and water resources. According to his

autobiography‟s manuscript, the decision to stay and work in U.S.A, instead of returning to Yugoslavia,

was not easy and derived mainly from the difficulties he encountered as a full professor at the University

of Belgrade, likely due to his contrasts with the communist party of his country.

During the period spent at CSU (almost twenty years), he contributed to increase the university prestige

both from the academic and research point of view, in particular in the fields of hydraulics, hydrology and

water resources, attracting a very large number of Master and Ph.D. students from all over the world. It is

worth to recall that, in the period from 1962 to 1980, 100 Hydrology Papers were published presenting the

results of researches and dissertations carried out at CSU. The Hydrology Papers represented also the

initial kernel of the book‟s production of Water Resources Publication, a publishing company founded by

Prof. Yevjevich and his wife in 1971, which in the following years published many highly appreciated

books on water and related environment topics and also released some specialized computer software.

In 1979, after retiring from CSU, he was director of the Water Resources Institute of the private university

George Washington (Washington, D.C.), until 1987, continuing his cooperation with CSU. After the end

of his academic career, he intensified his international activities until 1998, giving several lectures and

short courses in different countries, coordinating and organizing a series of Advanced Study Institutes

(ASI) and Advanced Research Workshops (ARW) within NATO programs on water subjects, in particular

in countries of Eastern Europe.

The very long period of teaching in graduate and post-graduate programs has been always oriented to

transfer not merely the scientific and technical contents of disciplines, but also an example of professional

rigour and integrity, of hard work and responsible attitude.

The scientific production of Prof. Yevjevich includes 23 books and more of 200 papers published in

refereed journals and proceedings of scientific meetings. The researches carried out during his early

staying in Yugoslavia are mainly related to hydraulics and economic aspects of hydropower plants, and to

statistical hydrology and earth and rock dams topics. A significant contribution in the field of hydraulics

was given on the topic of unsteady flow in channels, both through the cited bibliography on “Flood

routing methods” and several papers on resolution of the de Saint Venant equations (4 numbers of

Hydrology Papers on “Flood routing through storm drain” and three chapters in the volumes on “Unstead

flow in open channels”). Another subject, particularly important for Balcanic region, has been the analysis

of water resources in karst environments: indeed, he published several papers from 1953 to 1985 and two

Symposia proceedings books “Karst hydrology and water resources”(WRP,1977) and “Karst water

resources” (1985).

The most innovative scientific results however were attained in hydrology, which he considered not as a

stand-alone discipline, but as a study field oriented to solve the problems of water resources and hydraulic

defence and, in the last years, also of water quality within the environmental protection framework. The

hydrologic research was initiated in Yugoslavia with the study of annual high and low flows and the

preparation of a textbook in hydrology. In USA he started extensive researches on the application of

stochastic process in hydrology and contributed to the “Handbook of Applied Hydrology”, edited by Ven

Te Chow (1964) as author of chapter 8.II on “Regression and correlation analysis”. The continuous

efforts and outstanding results of teaching and research at CSU are documented by many pionering papers

in important journals, as well as the already mentioned Hydrology Papers and by the books “Probability

and Statistics in Hydrology” and “Stochastic Processes in Hydrology”, published by Water Resources

Publication in 1972.

The most remarkable contributions to the progress of hydrology include at least the following three

aspects: i) the development of advanced methods for time series analysis (particularly to deal with trend,

and periodic, intermittent and stochastic components), ii) the proposal to apply the “run” concept for an

objective definition of drought (this is one of the most cited papers in the hydrological literature), iii) a

systematic framework of basic principles and methods to study floods and to select adequate measures for

flood defence (discussed in several papers presented at conferences and symposia and in the proceeding


23

books of NATO ASI “Coping with Floods”, Erice (TP), 1994 and “Defence from Floods and Floodplain

Management”, Budapest, 1994). Other significant contributions were given also in the field of planning

and operation of reservoirs and of application of system analysis tools in water resources.

During his long scientific activity that lasted more than 60 years, Vujica Yevjevich has contributed to

demonstrate that some alternative approaches, apparently not compatible with one another, can be

harmonized in a higher synthesis. First, the choice between theoretical and applied research: in many

cases Yevjevich‟s researches, even if motivated by real life, practical engineering problems, have

requested the use of advanced theoretical tools, for example for dealing with the unsteady flow problems

in open channels through approximation procedures to solve the de Saint Venant equations, or for

studying the time-space variability of hydrological processes through advanced stochastic models.

Second, the choice between local and global perspective: the initial input for several investigations carried

out by the students of Master or Ph.D. programmes at CSU, derived by the problems of a specific basin or

region but the adopted methodologies, and the achieved final results, presented a more general validity,

applicable also to other contexts; this occurred for example for the studies on karst water resources or the

analyses of floods and droughts.

Third, the controversy between the preeminent role of the technical disciplines dealing with water

problems, such as hydraulic and hydrology, and the necessity of an advanced multi-disciplinary and

interdisciplinary approach. The opinion of Prof. Yevjevich was well-balanced: though he was educated as

a civil engineering with specialization in hydraulics and hydraulic plants and his scientific production was

mainly devoted to develop hydrology as a basic support to the planning of hydropower plants first and to

the management of multi-purpose water supply systems later, he believed that other disciplines, such as

law, economics etc, were essential to tackle water resources issues. In particular these disciplines should

contribute to the analysis of water resources problems since the beginning, helping to understand the

nature of the problem and to find robust solutions instead of being applied only after a solution has been

found on a technical basis. This deep belief is shown by several general lectures at international

conferences on the future of the research in hydrology and water resources, as well as by the design of the

post-graduate courses (e.g NATO meetings) and books, where he carefully selected, within the large set of

experts of many countries with whom he had good relationships, lecturers and/or authors to deal with

legislative, economic, environmental and social aspects.

The international contacts and the permanent links with many students of different countries, enabled the

development of important research projects in cooperation with several universities and organizations in

water field all over the world. This led mature experts to strongly cooperate and young researchers to

build new ideas and to test their capability.

In particular, his friendship links with some Italian scientists, like Arnaldo Maria Angelini, (president of

ENEL -Electric Power Company) or Emanuele Guggino Picone, (director of the Hydraulics, Hydrology

and Water Resources Institute in Catania) led to a long scientific cooperation with several Italian

organizations. Besides he participated in several Italian meetings as an invited lecturer: e.g. the

Symposium on Climatic Change, Rome, 1961; the Conference on “Floods: prediction and soil defence”,

Roma, Accademia dei Lincei, 1969, and within the cooperation activities with University of Catania, the

Seminar “Floods and droughts”,1980; the NATO ASI “Operation of complex water systems”, Erice,1981;

the Accademia dei Lincei Symposium “System science and technology in water resources management”,

held in Roma in1984; the NATO ASI “Coping with floods, Erice 1992, and the LXI Meeting of the

Società Italiana per il progresso della scienza “Water: current status and perspectives”, held in Catania

in1991). The long standing cooperation between Colorado State University and the University of Catania

started with him, and included the development of an innovative tool for continuing education (.“A

videotape Course on “Management of Water Resources: A System approach”), and some research

programs (e.g. on Conjunctive use of water and Regional drought analysis) as well as the exchange of

several students and professors.


24

Prof Yevjevich was awarded the Honoris Causa Graduate at Lund University (Sweden), the “Ven Te

Chow Hydraulic Engineering Award” of ASCE, the “International Hydrology Award of IAHS” and the

“Chow Memorial lecture” at the V World water Congress of IWRA, an Association which he contributed

to found.

He passed away on 26 March 2006 at Highlands Ranch near Denver (Colorado). In spite of his age (92

years), and of his fierce struggle against Parkinson disease, he was still strong as an oak, enlarging its

branches to more wide intellectual interests, as proved by the publication of “Quo Vadis America”(1999),

devoted to an analysis of socio-political processes in United States observed in the last half century, and in

the last period by the drafting of his autobiography, carried out in the last years of his life-time.

According to the statement of the Denver Post published on the funeral‟s day, “ work was life for V.Y.

tireless scientist full of ideas”! This assertion reminded me a comment recently read on the meaning of the

work in the life of Van Gogh: “Van Gogh, son of a Calvinist pastor and a preacher himself in young age,

through his hard work was a witness of faith, although not explicitly religious, able to speak deeply to

person‟s heart “ (P. Stefani) . In my opinion, the tireless work in scientific research, teaching and

professional commitments, carried out by prof. Yevjevich during his long life, represents an unusual

expression of faith –may be not religious- on a subject, the Water, which is one of the most precious goods

for Mankind and for the Earth.


25

Teaching and professional activities in water resources engineering

in Yugoslavia

Tina Dašić, Jasna Plavšić, Branislav Đorđević and Jovan Despotović

Faculty of Civil Engineering

University of Belgrade, Serbia

Professor Vujica Yevjevich was, without any doubt, one of the greatest scientists in the fields of

hydrology and water management in Yugoslavia. His professional career began in 1937, when he

graduated at the Faculty of Civil Engineering in Belgrade. Although he worked in Yugoslavia only till

1958, professor Yevjevich gave outstanding contributions to the development of water power and

hydrology of the country.

After the Second World War professor Yevjevich took a series of responsible professional duties in

Yugoslavia, as a designer, researcher and consultant for water resources and hydropower issues. He was

founder of Hydropower Institute, as a part of the Federal Ministry of Electric power, and in the period

from 1949 till 1956 he was appointed at the position of the Director. The Institute, with its hydraulic

laboratory, was one of the most prominent laboratories of that kind in the world. Professor Yevjevich

played a decisive role in development of Yugoslav hydropower after the war. In the same time (1946 -

1957) he was teaching courses Hydrology and Waterpower Engineering, at the Faculty of Civil

Engineering in Belgrade. In both cases Professor Yevjevich performed pioneering role of forming those

scientific disciplines. At the same time, he worked as a researcher and organizer of all hydropower

projects in Yugoslavia (Vlasina HPPs, HPP Zvornik, HPP Perucica, HPP Mavrovo, HPP Jablanica, etc.).

He provides methodological, the world's recognized contributions to the science of waterpower utilization.

A large number of professional and scientific papers (in Serbo-Croatian) resulted from professor

Yevjevich activities in the period 1948-1956. Beside over 30 scientific papers he wrote and published six

books, among which: "Water Management" (1946), "Water Resources Schemes", "Hydrology", "Water

Power Resources of Yugoslavia" (in two languages, English and Serbo-Croatian), etc.

His book "Hydrology" (1956) was the first scientific monograph published in the field of hydrology in the

country (and the only one for the next 20 years). In that book he sets hydrology as the exact - engineering

scientific discipline, instead mostly descriptive science, as it was before.

In 1955, he submitted a doctoral thesis on the methodology of investigation of hydropower potential to the

Serbian Academy of Sciences and Arts, which awarded him the degree of Doctor of Engineering. At that

time awards of doctor degree by academies of sciences were considered to be a higher standing than

similar degrees awarded by universities. Yevjevich was the first and the last hydraulic engineer to be

awarded that degree of Doctor of Engineering by the Serbian Academy of Sciences and Arts. The

chairman of his doctoral committee was Dr. Milutin Milanković, academician, best known for his theory

of ice ages and long term climate changes (Milankovitch cycles).

In the field of water resource analysis, with the previously mentioned contributions, it is important to note

that professor Yevjevich was among the first to have introduced and discussed issues of risk and

uncertainty in water management planning, Hurst's phenomenon, a systematic approach to modeling and

so on.

In 1958 professor Yevjevich with his family moved to U.S.A., where he continued his work and become

word known hydrologist.


26

In memory of professor Vujica M. Yevjevich (Contributions to

Colorado State University)

Jose D. Salas, Professor

Department of Civil and Environmental Engineering

Colorado State University

Dr. Vujica Yevjevich, Emeritus Professor of Civil and Environmental Engineering at Colorado State

University (CSU) passed away on March 26, 2006 after a long battle with Parkinson‟s disease. He was 92

years old.

Dr. Yevjevich, originally from Yugoslavia was invited in 1958 by the United States Government

to work with the U.S. Bureau of Standards as a foreign scientist and later with the U.S. Geological Survey

in Washington D.C. In 1960, Dr. Yevjevich was offered a faculty position in Civil Engineering at

Colorado State University as Professor-In-Charge of the Hydrology and Water Resources Program. After

reaching the federally mandated (for professors and CEO‟s of corporations) retirement age of 65, he

retired from CSU in 1979.

Professor V. Yevjevich has been one of the most influential men in the field of Hydrology and

Water Resources and was considered the father of Stochastic Hydrology, a field that has become a major

area of scientific and engineering activity in water resources for the past several decades. His

contributions encompass most of the elements of the hydrological cycle. A notable achievement has been

his work on characterizing and organizing the underlying attributes of hydrological processes, such as

annual and monthly precipitation and streamflows, into elementary components including trend,

intermittency, periodicity, and stochasticity. His seminal approach paved the way for developing many

mathematical models and modeling strategies for stochastic simulation in water resources, which have

become standard techniques for planning and management of water resources systems. Also, major

contributions of his research endeavors have been in the area of drought and storage analyses. For

example, his 1967 paper on modeling continental droughts by using the theory of runs opened a new

direction in the field for defining, characterizing, and analyzing multiyear droughts. His approach made

the analysis simpler, more logical, and more practical, and opened alternative paths for drought analysis

using empirical and simulation techniques, and analytical closed-form solutions. In the 1960‟s, 1970‟s,

and beyond, Dr. Yevjevich and his students expanded such work to include the modeling of critical

droughts and developed the mathematical framework for characterizing regional droughts. This work

have had worldwide impact in that it has been applied for characterizing droughts in many countries such

as Portugal, Italy, Chile, Japan, etc. Equally important has been his work on water storage by using the

theory of range and deficit analysis and related issues such as the ability of stochastic models to preserve

short-range and long-range statistics. His work in the 1960‟s and 1970‟s led the direction on the

mathematical description of storage properties and several formulas are now available for estimating the

storage capacity of reservoirs, which are useful for sizing reservoirs at the preliminary and pre-feasibility

project stages.

Dr. Yevjevich dedicated a significant part of his life to educating undergraduate and graduate

students at Colorado State University (CSU) as Professor of Civil Engineering and Professor-in-Charge of

the Hydrology and Water Resources Graduate Program and at George Washington University as Professor

and Director of the International Institute of Water Resources. At CSU he developed an internationally

renowned program in Hydrology and Water Resources where he advised over 130 graduate students. In

2005, Dr. Yevjevich donated funds to Colorado State University establishing an annual award to an

outstanding graduate student in Civil Engineering emphasizing hydrology and water resources. A prolific

author, he published 23 books, wrote over 200 professional articles, and was the founder of Water

Resources Publications, a company that publishes and distributes texts on water resources. The research

from the Hydrology and Water Resources Program was published not only in peer reviewed journals and

presented in scientific and technical conferences, but also in the well-known Colorado State University

Hydrology Papers. His books on “Probability and Statistics in Hydrology” and “Stochastic Processes in


27

Hydrology” published in 1972, articulated water resources processes in simple probabilistic terms and

brought a tremendous impetus to the field. At the time of his death, he was finishing his autobiography

that includes many ideas for research in hydrology, as well brief descriptions of the several stages of his

life history, all of which are human stories worthy of archiving. As a persona, every discussion with a

colleague, every faculty meeting, and every professional conference that he was a part of was enlivened

by his presence. He spoke his mind freely and forcefully, not with malice, but with the goal of engaging

in issues. Many graduate students may recall entering his office with trepidation, soon replaced by respect

and friendship as the moment passed, being later aware and proud having been one of his students. He

was looked to for advice and counsel by almost every person or group or organization that he touched,

whether in America or in one the multitude of his international contacts.

Throughout the years he has been awarded many national and international prizes and honors such

as, the American Society of Civil Engineers (ASCE) Ven Te Chow Hydrologic Engineering Award, the

International Hydrology Award of the International Association of Hydrologic Sciences, Doctor Honoris

Causa, University of Lund, Sweden, Member of the Yugoslav Academy of Engineering, and Honorary

Membership of the International Association of Hydraulic Research. He is recognized as one of the most

influential engineers from Serbia and was also elected to the Serbian Academy of Engineering. A

museum bearing his name at his place of birth in Priboj will store most of his professional works that are

not at the University of Belgrade or Colorado State University.

In recognition to the invaluable contributions made by the late Dr. Yevjevich to Colorado State University

(CSU) graduate education in Hydrology and Water Resources, CSU‟s hydrologic community created a

website (http://www.engr.colostate.edu/ce/facultystaff/yevjevich/Yevjevich_index.htm) to remember his

life and his achievements. His commitment to graduate education at CSU will be always remembered and

appreciated. Some of his former students and colleagues have written memories of his life in various

languages. Part of Dr. Yevjevich‟s legacy has been the well known Hydrology Papers series. The entire

collection of 100 papers in pdf format can be found in the referred website.

http://www.engr.colostate.edu/ce/facultystaff/yevjevich/Yevjevich_index.htm


28

A comptemplation of the contributions of Professor Yevjevich Taha B.M.J. Ouarda, Chairman

Canada Research Chair on the Estimation of Hydrometeorological Variables

INRS-ETE, 490, de la Couronne

Quebec (Quebec), G1K 9A9, CANADA

[email: [email protected]]

“I would say that professor Yevjevich is like a diamond: He is sharp, hard, brilliant, and he

will capture your attention and your admiration. He is made out of carbon just like the rest of

us, but he is rare and so well structured”.

I met professor Yevjevich at CSU as a Ph.D. student. My choice for CSU to pursue my graduate studies

was solely motivated by the prestige that the university had acquired over the years in the field of water

resources; Prestige for which Professor Yevjevich deserves a large part of the credit. When I arrived at

CSU, Professor Yevjevich had already retired but continued to be present at CSU from time to time. The

little interaction I had with him was influential for my carrier. I still remember some of the discussions I

had with him, and how I left the discussion feeling as if I had just discovered the meaning of life. I truly

believe that Professor Yevjevich helped me learn HOW to think (although at times he may have sounded

as if he were trying to teach me WHAT to think).

It is funny how I always called my professors by their first names, so I was working with John, Pepe,

Darrel and Neil. However, I never dared to call professor Yevjevich by his first name, even after I

finished my Ph.D. I don‟t know how he would have reacted to that. Today, what I recall the most are

those moments where we met and talked “another language than English”. I felt appreciated, respected

and treated like a friend. I still cherish those moments.

When I finished my Ph.D. I decided to pursue my carrier and my research activities in the field of

statistical hydrology. While my decision was partially motivated by the desire to explore new directions

and to avoid recycling what I did in my Ph.D. in the field of stochastic optimization, I have to admit that

the influence of Professors Yevjevich and Salas (Pepe) on my decision was, without any doubt, very

significant. They implemented in me the love of this branch of hydrology, the awareness of its impacts

and contributions, the belief that there is still much to be done in this direction, and the desire to dedicate

part of my life to it. In a certain sense, Professor Yevjevich oriented my professional life.

Today, I still use some of his papers and books as standard teaching material in my classes. For instance,

the books “probability and statistics in hydrology, 1972” and “stochastic processes in hydrology, 1972”,

published by Water Resources Publications are compulsory reading material for my graduate students. I

always felt that they laid an excellent background on earlier developments and prepared the students for

the more recent material in the field. Some of Professor Yevjevich‟s publications represent unavoidable

benchmarks in the field. For instance I can mention the following titles “Fluctuations of wet and dry years,

Analysis by serial correlation, 1964”, “The application of surplus, deficit and range in hydrology, 1965”,

“The statistical measure of hydrologic time series, 1967”, “Mean range of linearly dependent normal

variables with application to storage problems, 1967”, “Stochasticity in geophysical and hydrological time

series, 1971”, “Generation of hydrologic samples, case study of the Great Lakes, 1975”, and

“Relationships among water storage variables, 1987”. I always find it amazing how all these publications,

some of which date before my birth, are still top-of-the-list today. The book that my students know as the

red bible of time series modeling in hydrology (because of the terms I use, following Pepe) is also co-

authored by Professor Yevjevich (Applied modeling of hydrologic time series, by J.D. Salas, J.W. Delleur,

V. Yevjevich and W.L. Lane, 1980).



29

When I was student at CSU, the older students used to say that Professor Yevjevich has always been very

difficult to work with. I would respond that he seemed fair, straightforward and rigorous. One day a

student asked me how I would describe professor Yevjevich. After a short reflection, I answered that “I

would say that professor Yevjevich is like a diamond: He is sharp, hard, brilliant, and he will capture

your attention and your admiration. He is made out of carbon just like the rest of us, but he is rare and so

well structured”. That generated a few smiles. Today, I still wouldn‟t change a word in that statement.

Bridging hydrologic theory and practice: a reflection on professor

Yevjevich’s contributions

Vijay P. Singh

Department of Biological & Agricultural Engineering &

Department of Civil & Environmental Engineering

Texas A & M University

College Station, Texas 77843-2117, USA

Professor V. Yevjevich‟s contributions to the theory and practice of hydrology, hydraulics and water

resources engineering are monumental. It is therefore difficult to summarize them in a short note.

Hydrologic theory and hydrologic practice are complimentary and they feed each other. Using hydrology

in a broad sense encompassing hydraulics and water resources, Dr. Yevjevich‟s contributions can be

summarized, for purposes of discussion, into three categories: (1) research, (2) education, and (3)

technology transfer and practice. His theoretical or research contributions dealt with all major elements of

the hydrologic cycle. More specifically, his contributions to modeling of droughts (regional and

continental), modeling of monthly and annual precipitation, stochastic analyses of streamflows, storage

theory, and stochastic simulation in water resources are widely recognized. He is also noted for his work

on analyses of trends, intermittency, periodicity, and stochasticity in precipitation and streamflow time

series. His leadership in the 1960s and 1970s greatly contributed to the development of the field of

stochastic hydrology as we know today. On the other hand, he also contributed to deterministic side of

hydrology: open channel flow, flood routing, and karst aquifers.

Many of Dr. Yevjevich‟s theoretical contributions have found their way in hydrologic practice. For

example, his work on droughts is now a standard practice in defining and characterizing droughts. His

work on storage theory contributed to the current practice of sizing reservoirs. He contributed to the

practice of water resources engineering in a multitude of projects undertaken by the U.S. Army Corps of

Engineers, the U.S. Bureau of Reclamation (USBR), the U.S. Geological Survey, the U.S. Department of

Agriculture, the Federal Highway Administration, and a number of consulting companies worldwide. He

worked on flood analysis for the Tulsa Corps of Engineers Office; he introduced statistical and stochastic

methods in planning and management activities of the USBR, Denver; he evaluated annual flows of the

Colorado River; he developed dual cavity for mining of Potash in the Sasketchewan region of Canada for

Kalium Chemicals Consortium; redesigned limestone kilns in Corpus Christy, Texas, to extend the life of

brick walls; analyzed extreme floods for Santa Clara County, California; and estimated flooding in urban

watersheds of the Washington D.C. area. Dr. Yevjevich‟s professional work took him to many countries,

including Brazil Canada, Italy, Peru, Portugal, Spain, Turkey, USA, and Yugoslavia.

In the education area, his contributions span the development of graduate program in hydrology and water

resources at Colorado State University, development of graduate level courses in statistical and stochastic

hydrology, teaching, advising a large number of graduate students (more than 130), and mentoring junior

faculty members. Many of students have become leaders in their own right He shared his knowledge


30

through publication of books (23), and professional articles (>200). He also contributed to the

furtherance of hydrology through the founding of Water Resources Publications, a company that publishes

and distributes texts in the water resources area. He organized international conferences, chaired sessions,

gave keynote addresses, taught at short courses and institute and gave invited lectures all over the word.

He was a man of unique attributes-he was highly disciplined, knew a lot of languages-French, German,

Italian, Russian, Spanish, besides, of course, English and his mother tongue, Serbian and Yugoslavian. He

was sophisticated and well mannered. He spoke his mind fearlessly and forcefully, and made his presence

known. With his passing an era has gone, but place in hydrology is preserved for a long time to come.

Professor Yevjevich’s contributions to the field of floods and

droughts

Francisco Luiz Sibut Gomide

Professor Titular

Universidade Federal do Paraná, Brasil

Professor Vujica Yevjevich‟s contributions to the practice and theory of Water Sciences are outstanding.

Equally outstanding is his influence over an extraordinary number of water professionals from several

countries all over the world. Thanks to “Dr. Y”, as he was called in Fort Collins, Colorado State

University has quickly become one of the most active and famous universities in the field of Hydrology

and Water Resources. Starting in the sixties, and for many decades, CSU not only attracted students from

other American states, but also from foreign countries: from 1962 to 1980, under Dr. Y‟s leadership,

several authors published one hundred “Hydrology Papers”.

Professor Yevjevich is considered the father of Stochastic Hydrology, due to his pioneering method of

analysis of hydrological time series and to the innovative application of probability and statistics to the

study of the two extreme situations of the hydrologic cycle: droughts and floods. Droughts, for centuries,

forced people to migrate in search of food and water to survive. Floods, still today, cause damages and

losses of human lives. Both are important departures from normal hydrologic situations. One needs special

attention and rigorous scientific analysis to find out means of living harmoniously with these recurring

hazards and to protect humanity from their impacts. Professor Yevjevich devoted more than 60 years of

professional life to droughts and floods as research topics. It all started with the study of annual high and

low flows in Yugoslavia.

In 1967, Professor Yevjevich applied the statistical theory of runs to the analysis of drought events,

leading the field on drought definition and framework analysis. This (Hydrology Paper 23) is one of the

most cited papers in the hydrological literature. In the opposite extreme of the hydrologic cycle, several

contributions on methods to study floods and measures for flood defense derived from European-

American cooperation. “Coping with Floods”, co-authored with Giuseppe Rossi and Nilgun Harmancioglu

is a representative example.

In 1976, Professor Yevjevich visited Brazil and lectured a course on Stochastic Hydrology - the first ever

presented in South America - and changed the history of the Brazilian electricity sector. The concept of

“guaranteed energy, for a given level of risk” replaced the traditional concept of “firm energy” and since

then, reservoirs are designed and operated according to sophisticated probabilistic methods (the national

integrated system operates more than 120 hydropower plants and more than 60 reservoirs with total

storage of 200 million MWh). Nowadays, Stochastic Hydrology is not only a vitally important discipline

for the Brazilian electricity sector, but an essential part of modern Water Resources Engineering.


31

The scientist and his achievements cannot be separated from the man, his strong personality and his acid

sense of humor. Fearless critical of governmental bureaucracy, Professor Yevjevich often mentioned the

“hydro-illogical cycle” to explain why providences, projects and hydraulic works come after the

occurrence of natural disasters such as floods and droughts, and not before, as it should, to try to avoid

them. In 1999, the 86-year-old Professor Yevjevich published “Quo Vadis, America?” an analysis of the

American socio-political process and an eloquent demonstration of the range of his intellectual interest

and curiosity. In this book, Professor Yevjevich, outspoken as always, speaks about everything: global

warming, environmental protection, floods, droughts, specialists, generalists and politicians. Pure

Yevjevichian vintage.

Contributions to the theory and practice in water storage

Geoffrey GS Pegram PhD, PrEng, FSAICE, MAGU, AMASCE

Director: Pegram & Associates (Pty) Ltd

Professor Emeritus and Senior Research Associate

Civil Engineering, University of KwaZulu-Natal, DURBAN, 4041, South Africa

[email protected]

In 1976, I was introduced to the famous (controversial) Dr Vujica Yevjevich when he visited South Africa

to give a series of lectures. I never looked back after my first sabbatical which I spent at CSU in 1977 with

him; he was very influential in my attitude to stochastic hydrology - I recall his iconoclastic statements:

“Smash the Holy Cows” and “during time of drought, work on floods and in times of plenty, work on

storage”.

What prompted this interest in storage reliability and multivariate time series analysis? Some background;

Emlyn Lloyd (my PhD supervisor) was a highly respected author on Storage reliability using Markov

chains. I was drawn by Lloyd‟s writing style, which was to make mathematics entertaining and

understandable and we had 2 fruitful years together - 1970 to 72. I was fascinated by the way Markov

chains were so neatly “nested” even if their application to storage reliability is slightly esoteric and

simulation is much more flexible in real-world situations. Nevertheless the concepts of reliability and the

precision one needs to express this concept are well aided by the MC approach.

When Dr Yevjevich visited Pretoria in 1976, I asked him if I could visit CSU for a sabbatical - that

dumped my family in the exciting world of the Foothills campus of Fort Collins for six months - I

collaborated with Dr Y, Duane Boes and Pepe Salas. Pepe & I helped put together the Third International

Hydrology Symposium and we co-authored a couple of papers with Dr Y and Duane Boes. I also got to

work with Dave Woolhiser with whom I published on rainfall modelling - using Markov chains, indeed!

Foothills campus was the place in 1977.

On my return to South Africa I finished the last of the CSU Hydrology papers, No. 100 co-authored with

Dr Y, Duane and Pepe. It referred to many illuminati from among Dr Y‟s students: Francisco Gomide &

Brent Troutman in particular stand out. Then in 1984 South Africa experienced the start of a very serious

national drought and I was drawn in to the practical implementation of the reliability of systems of

interconnected reservoirs in a practical way. I drew heavily on the work that Dr Y, Pepe, Duane & I had

done and worked on multisite streamflow modelling and storage reliability for our Department of Water

Affairs through consultants BKS; we recorded our efforts in two papers and a Book. The programs are

still being used as the standard after 25 years.

Then during the 1990s there was a lull in storage research; I moved into weather radar. However, I worked

with Tom McMahon in Melbourne since 2000 and we published a string of papers on reservoir storage–

yield relationships using a global streamflow database.

The presentation will highlight some of the advances made since the heady days of 1977 and connect the

dots in my career as a stochastic hydrologist‟s contribution to reservoir reliability, acknowledging my debt

to Dr Y and his intellectual heritage.


32

International activities and transfer of knowledge in water resources

Giuseppe Rossi

University of Catania

Prof. Yevjevich‟s deep interest for the international dimension in his professional life, goes back

to the period of his education. After his graduation in Civil Engineering at University of Belgrade (1936),

the military service and a professional experience in Yugoslavia, he studied in France at the University of

Grenoble ( September 1938-June 1939) to improve his background in hydraulic engineering and he visited

some British hydraulic laboratories to learn from these advanced research experiences abroad. Also the

period spent as a prisoner of war in various camps in Germany and Italy (December 1941 to September

1943) was an opportunity for him to improve foreign languages and develop the knowledge of technical

literature in water engineering. He wrote in his autobiography: “The hard war time became a forced

educational bonanza to me”

His aptitude for developing personal links with people of every country, a good knowledge of

many languages and his regular participation to international scientific meetings led him to several

international contacts. The long teaching period at Colorado State University Master and Ph.D. programs

fostered the development of relationships with several young researchers all over the world, which were

eventually strengthened by frequent trips to give lectures at universities and other institutions.

The main outcomes of the international activities of Prof. Yevjevich can be grouped into :

i) international research programmes and scientific meetings,

ii) international post-graduate education activities (particularly NATO meetings)

iii) publication of books by authors from several countries.

All these activities, in Yevjevich‟s vision, had a common motivation: the importance of knowledge-

transfer in professional training and continuing education, where the international exchange of

competences and experiences is imperative.

Among the main research international programs it may be worthwhile to mention: i) “Conjunctive

use of surface and groundwater resources, in cooperation between CSU, Servicio Geologico des Opras

Publica of Spain and University Politécnica de Valencia (group of prof. Andrés Sahuquillo); ii)

“Conjunctive use of multiple sources of water and its role in regional development”, developed in

cooperation between CSU and University of Catania (group leaded by prof. Emanuele Guggino); iii)“

Karst hydrology and water resources, in cooperation between CSU and Yugoslavian research centers; iv)

“The system approach to analysis of hydrologic processes and environment”, developed with the financial

support of U.S. National Science Foundation and Japan Society for the Promotion of Science at the

University of Hawaii, Honolulu.

Several scientific meetings were also organised by prof Yevjevich e.g.: IAHR Congress, Fort Collins,

1967; 1st International Hydrology Symposium, Fort Collins, 1967; US-Japan seminar in Hydrology,

Honolulu, 1971; 2nd

International Hydrology Symposium, Fort Collins, 1972; U.S. Yugoslavia

Symposium on Karst Hydrology and Water resources, 1976; 3rd

International Hydrology Symposium, Fort

Collins, 1977; US-Uruguayan Symposium on Experimental Watersheds, Montevideo,1983; US-Italy

Seminar on Urban Storm Drainage, Cagliari, 1991

Among the international post-graduate Courses, a pioneering activity was the Videotape Course on

“ Water Resources Management: A System Approach”, carried out by CSU in cooperation with the

University of Catania and funded by FORMEZ (the Agency for the professional education in South Italy).

It was a continuing education activity aimed to improve skills of the technical personnel responsible of

water governance and management. The Course, consisting of 30 lectures, was prepared in English and


33

in Italian in 1975-77 and was employed in post-graduate courses in 1978-1980. My personal involvement

in such Course, gave me the opportunity to appreciate Prof. Yevjevich‟s outstanding scientific skills and

his excellent organizational capacities.

The organization of many educational events for international experts represented a remarkable

activity to improve the knowledge on key issues of hydrology and water resources and to favour the

exchange of the best practices. The main NATO meetings, organized by prof. Yevjevich with many

colleagues of several countries, include:

- Lisbon, Portugal 1980, NATO ASI “Coping with Droughts”

- Erice, Italy, 1981, NATO ASI “ Operation of Complex Water Systems”

- Erice , Italy, 1992, NATO ASI “ Coping with Floods”

- Budapest, Hungary 1994, NATO ASI “ Defence from Floods and Floodplain Management

- Varna, Bulgaria,1994, NATO ASI “ New Technologies for Water Supply Systems”

The proceedings of the Nato meetings were published by Kluwer, Springer, WRP.

Foreign authors invited to publish in Water Resources Publication include: B. Djordjevich from

Yugoslavia, R.B.Allen, M.S.Basson, G.G.S.Pegram and J.A. van Rooyen from South Africa, D.M.Hicks

and P.D. Mason from New Zeland, T.A.McMahon from Australia, I.Becchi, M.Benedini, C.Cao, and

R.Rosso from Italy, F. Askar, B. Bobèe and J.Marsalek from Canada

The international activities of Vujica Yevjevich demonstrated also that some apparently alternative

approaches can be harmonized in a higher synthesis:

First, the choice between theoretical and applied research: in many cases Yevjevich‟s researches,

though motivated by practical engineering problems, have requested the use of advanced theoretical tools,

for example the unsteady flow problems in open channels required advanced solutions to the de Saint

Venant equations, or the studyi of the time-space variability of hydrological processes oriented to the use

of innovative stochastic models.

Second, the choice between local and global perspective: the initial input for several investigations

carried out by CSU students of Master or Ph.D. programmes focussed on the problems of a specific river

basin or region but the adopted methodologies presented a more general validity, applicable also to other

contexts; e.g. the studies on karst water resources or on floods and droughts.

Third, the controversy between the leading role of the technical disciplines to deal with water

problems and the necessity of an advanced multi-disciplinary and interdisciplinary approach.

The opinion of Prof. Yevjevich was well-balanced: though he was educated as a civil engineering and

published significant scientific contributions in hydrology, hydropower plant design, hydrology and water

resources planning, he believed that other disciplines, such as law, economics etc, were essential to tackle

water resources issues. In particular these disciplines should help to understand the nature of the problem

and to find robust solutions instead of supporting a narrow solution, founded only on a technical basis.

This view is shown in the opinions expressed in several general lectures at international conferences on

the future of water resources as well as by the design of the contents of several NATO meetings and

books, where he carefully selected lecturers and/or authors for dealing with legislative, economic,

environmental and social aspects from a large set of experts of many countries with whom he had good

relationships.

A conclusive remark refers to the role and meaning of the work in the experience and life‟s

vision of Prof. Yevjevich. As already observed with reference to Van Gogh‟s painting activity, it is

possible to consider that the tireless work carried out by Prof. Yevjevich during his long life, represents an

unusual expression of faith, - may be not religious - on a subject, the Water , which is one of the most

precious goods for Mankind and for the Earth.


34

Poster session

Modeling the impact of climate change in a Mediterranean

catchment

Aziz Abouabdillah, Ons Oueslati, AnnaMaria De Girolamo, Antonio Lo Porto

Water Research Institute, National Research Council, Bari, Italy;


During the last decades, the Mediterranean region is suffering more and more from droughts. It has been

recognized as one of the most vulnerable regions in the world to climate change. Understanding the

impact of climate change on various components of water cycle is an important challenge for long-term

sustainable management of water resources. In this paper, the integrated hydrological model “Soil and

Water assessment tool” (SWAT 2005) was used to study the impact of future climate on water resources

of a Mediterranean watershed. Future climate scenarios for periods of 2010-2039 and 2070-2099 were

generated from the Canadian Global Coupled model (CGCM 3.1) for scenarios A1B, B1, and A2. These

CGCMs data were then statistically downscaled to generate future possible local meteorological data of

precipitation and temperature in the study area. SWAT model was run first under current climate (1986-

2005) and then for the future climate period to analyze the potential impact of climate change on flow,

evapotranspiration, soil moisture, and crop yield of durum wheat and olive across this catchment. Finally,

Richter et al.‟s Indicators of Hydrologic Alteration (IHA) were used to analyze the flow regime alterations

under changing climate. The main results indicate that this catchment would suffer a combination of

increased temperature and reduced rainfall that will reduce water resources in this area. Consequently,

summer droughts would be intensified. Different spatial responses to climate change were observed in the

catchment for near future simulations. Higher altitude regions would experience an increase of the total

water yield, while a reduction is foreseen for lower parts. For far future, a noticeable decrease would

affect water resources in all part of the catchment.

Cross-correlation, homogeneity and uncertainty in regional models

of yearly streamflow Claudio Arena, Marcella Cannarozzo, Mario Rosario Mazzola

Department of Hydraulics and Environmental Applications, University of Palermo, Palermo, Italy.


The focus of the paper is on regional methods for yearly streamflow analysis. The index flow procedure,

an extension to annual flows of the well know index flood method, has been introduced some year ago

(Viglione et al., 2006) and has been applied to the frequency analysis of annual streamflow in Sicily using

different nuances of the method (Cannarozzo et al., 2009, Arena and Mazzola, 2008).

The procedure assumes that a single probability distribution function (a growth curve) may adequately

describe the pattern of yearly streamflow at different sites, provided that they are scaled by some measure

of central tendency (e.g. the mean) and that the sites belong to a homogeneous region. Application of the




35

procedure hence requires 1) the identification of homogeneous regions (frequently based on whether L-

moments‟ observed variability may be ascribed to sample variability only), and 2) the estimation of mean

annual streamflow by multiple regression analysis with some significant climatic and morphologic

covariates. Once a model has been selected, parameter uncertainty concerns both the parameters of the

growth curve and the parameters of the regression model of mean annual flow.

The paper investigates the role of cross-correlation in the homogeneity analysis of regions that may

feature the same growth curve of yearly streamflow and in the uncertainty analysis of the estimated

quantiles. Investigation is performed on the data set of yearly streamflow at 59 gauge stations (and

reservoirs) in Sicily.

In general terms, the effect of cross correlation is to reduce the amount of available effective information,

thereby increasing uncertainty of estimated parameters and quantiles. The role of cross-correlation in the

analysis of homogeneous regions in flood-related issues has been recently acknowledged by Castellarin et

al. (2008).

Incorporating cross-correlation in the homogeneity and uncertainty analysis first requires the availability

of a multivariate model of yearly streamflow also allowing for the (relatively high, especially in the

instance of semi-arid basins) asymmetry of the growth curves of yearly streamflow. The paper uses a

commonly employed procedure (e.g. Hosking and Wallis, 1988, Bayazit and Önöz, 2004) to build up such

a multivariate model for three areas in Sicily that have proved homogeneous to the classical Hosking and

Wallis (H-W) (1993) homogeneity assessment method. Such multivariate models are then used to both

incorporate cross-correlation in the H-W homogeneity assessment procedure and to estimate parameters of

the growth curve using Markov Chain Monte Carlo methods providing measures of parameter variability

and uncertainty. Comparisons with results obtained using the traditional assumption of independence

among time series show the impact of cross-correlation on the various assessments related to yearly

streamflow.

A simple point rainfall stochastic model for single event generation

using a parsimonious approach

Giuseppe T. Aronica, Giuseppina Brigandì

Department of Civil Engineering, University of Messina, Messina, Italy.


The paper describes a very simple stochastic model to derive single synthetic events at a point. Rainfall

events are generated using a Monte Carlo technique with characteristics in terms of shape, duration and

rainfall amount derived by statistical analyses of the available historic records. Because of its simplicity

and of the reduced number of input necessary to calibrate the model, it could be very useful if it is used as

a rainfall generator for a rainfall-runoff model at ungauged sites, where precipitation time series with long

duration and high temporal resolution are required to know the temporal evolution of the rain storm.

Model structure is characterized by two separate modules: (a) Intensity-Duration module (statistical

description and generation of storm characteristics using a multivariate model); (b) Storm Shape module

(generation of within-storm temporal characteristics as time step intensity variations using simple

statistical descriptors).

Storm duration and average intensity of the observed events were fitted on the basis of the historical data

by using Weibull and Lognormal distributions respectively. Instead, for the characterization of storm

shapes, dimensionless events were considered and fitted on the basis of the historical data using Beta

distribution. They have been than correlated to each other using the Frank Copula. The model has been

calibrated using sub-hourly rainfall data collected at Palazzolo Acreide raingauge station, in the Sicily,

Italy, where 10 minutes recorded rainfall data for a period of 5 years were available. The simulated events

were checked by comparison with measured data showing the good performance of the model.



36

Analysis of maximum rainfall daily. Station of Saida Manoubia,

Tunis

Bergaoui Mohamed Address: IRESA 30 Rue Alain Savary 1002 Tunis

[email protected]

The purpose of this study is to determine "rain project" used for the sizing networks of storm drainage.

Thus, our approach is based on the analysis and study of rainfall data in terms of normality and

stationarity. We first checked distribution normality of precipitation series using Kolmogorov-Smirnov

test and a graphical way. Then the frequency analysis of maximum daily rainfall is made while the

quantiles of scarce spectrum are determined. Finally, the study of stationary series and the development of

rain project are developed. Based on a series of long observation period, we have highlighted the non

stationarity of the series. Several breaks in studied series are detected. According to Hubert and al.(1989),

this segmentation procedure can be considered as a test on the stationarity of the series. Such a method is

appropriate for changes research of the mean. We found out that general trend of series is downward.

However, the daily rainfall maxima picks are observed in recent decades, reflecting a change in climate.

Maxima daily rainfalls are due to intense storms and short durations. In Tunis, we recorded 193 mm of

rain in 90 minutes in 2003. This rain has a return period of 500 years. These exceptional rains are then

processed in order to define the rain project. A "rain project" is a rain fictitious defined by a synthetic

hyetograph and statistically representative of real rain. Results of this study are interesting: they help to

diagnostic the behavior of the network of storm drainage (size and emptying time) and reduce the impact

of these events during flooding periods.

Flood predictions downstream of a river confluence by bivariate

analysis of incoming flows

Antonio Boccafoschi

Department of Civil and Environmental Engineering, University of Catania, Catania, Italy.


Bartolomeo Rejtano



The paper deals with the statistical prediction of flood flows downstream of the confluence of two sub-

basins. The research refers in particular to the circumstance which occurs when significant series of

annual maxima of flows are available for both the converging branches, whereas data are inadequate to

cover the reach downstream of the confluence.

Provided that the contributions of the sub-basins are someway correlated, a common approach to predict

the peak flow downstream of the confluence for the desired return period is to sum the predicted peaks of

the converging branches. This implies the conservative hypotheses that each year the return periods of

annual maxima of the two sub-basins are the same, i.e. the two series are correlated perfectly, and also that

the annual maxima of the converging branches occur at the same time. These hypotheses don't hold

necessarily, and might result to be too conservative, thus leading to overestimate the predicted





37

downstream flow. The research aims to remove the first one of the two hypotheses, thus reducing the

overestimation effect.

The proposed approach assumes that the annual maxima in the converging branches are mutually

dependent, with a specific imperfect degree of correlation. The approach aims to obtain the frequency

distribution of the peak flow downstream of the confluence from the frequency distributions of the

converging streams and from the correlation coefficient of annual maxima.

The hydrological problem implies the statistical problem of obtaining the frequency distribution of the

sum of two correlated variables X e Y when their marginal distributions and the correlation coefficient are

known. The distribution of the sum is obtained by numerical integration from the joint distribution of X

and Y, assumed that it has been obtained previously from the marginal distributions of X and Y and from

the correlation coefficient. The numerical algorithm which was developed is presented as it has been

specified for the case of a joint bivariate distribution expressed by the logistic Gumbel model, with

standard Gumbel marginal distributions.

The sense of the proposed approach is in the possibility of evaluating the correlation on the basis of short

sub-series of contemporary data for both branches, while the marginal distributions may be obtained from

the complete series. An other case for appropriate use of the methodology is when the number of common

years with flow data is too small, while the correlation degree can be estimated indirectly, for example on

the basis of rainfall data.

The validation of the procedure is obtained from an example where data are available in contemporary

years for the converging branches and for the downstream reach. The results show that the proposed

procedure leads to lower values than the gross sum of the branch predictions. The resulting values are still

overestimated, but they are much closer to the values obtained from the plain analysis of downstream data.

A geoadditive model of rainfall in Tuscany (Italy)

Chiara Bocci

Dipartimento di Statistica "Giuseppe Parenti", University of Firenze, Firenze, Italy

E-mail: [email protected], Tel: +39.055 4237263; fax +39 055 4223560

Enrica Caporali


E-mail: [email protected], Tel: +39 055 4796321; fax +39 055 495333.

Alessandra Petrucci

Dipartimento di Statistica "Giuseppe Parenti", University of Firenze, Firenze, Italy


Rainfall is considered a highly valuable climatologic resources and is a fundamental component of any

water resource assessment strategy. Rainfall characterization has important repercussions, among others,

on climate change scenarios, flood risk mitigation, water resources availability and drought assessment. In

this work, the application of a geoadditive model to analyze the spatial distribution of rainfall is explored.

Geoadditive models, introduced by Kammann and Wand (2003, Applied Statistics, 52, 1-18), analyze the

spatial distribution of the studied variable while accounting for the explicit consideration of linear and

nonlinear relations with relevant explanatory variables, as well as the spatial correlation described by a

standard spatial autocorrelation function. Under the additivity assumption they can handle the covariate

effects by combining the ideas of additive models and kriging, both represented as linear mixed model.

This approach, based on the generalized mixed model/splines paradigm, has achieved a valuable success

during the last decade as useful tool with which to study the spatial distribution of climate variables as

well as in other contexts. We investigate here the possibility to apply a geoadditive model to rainfall data.

The study area is the Tuscany Region, in Central Italy. The rainfall dataset is composed by the time series





38

of about 700 recording rain gauges, spatially distributed over an area of about 23.000 km2. The record

period covers mainly the second half of 20th century.

Effect of trends on the estimation of hydrological quantiles Brunella Bonaccorso, Antonino Cancelliere and Giuseppe Rossi



Estimation of quantiles of hydrological variables, i.e. values corresponding to fixed non-exceedence

probabilities or return periods, is traditionally carried out by assuming that the variables are identically

distributed and by fitting a probability distribution function to an observed sample. Generally, consistent

estimators are applied, leading to the paradigm that the longer the available sample the better the

estimation of quantiles.

Recent concerns about potential changes in present and future climate, however have led to challenge the

hypothesis of stationary series. In particular, more and more evidence is produced in literature about the

presence of non stationarities in many meteorological and hydrological records around the world in the

form of trends and/or jumps in the statistics of the series. Regardless of the causes, the presence of non

stationarities in the available sample requires to dramatically modify the procedures for estimating

probabilistic properties of hydrological time series. Despite several methods have been developed and

applied to model non stationary series, very few studies have addressed the problem of how non

stationarity affects the error of estimation of quantiles.

Regardless of the way non stationarities are modelled, dropping the stationarity hypothesis appears to have

direct impacts on at least two aspects: first of all, in a non stationary setting, the paradigm “the longer the

sample, the better the estimation” does not hold anymore. One may expect on the other hand that too short

a sample also should lead to larger errors of estimation. Therefore, the existence of an optimal sample

size, where optimal refers to the sampling properties of the estimators, can be postulated. The second

aspect is related to the same concept of return period, which loses in this case its traditional meaning of

expected interarrival time and therefore would require a new formulation.

Furthermore, when one suspects the presence of a trend in a series, the question arises as to whether

detrending the series, i.e. assuming a parametric form for the trend and removing it prior to the estimation

of the distribution parameters, leads to an improved estimation of quantiles. Indeed, due to the

uncertainties related to the choice of the correct parametric form of the unknown trend, as well as to the

sampling variability related to the estimation of its parameters, the estimated quantiles may not necessarily

be affected by a smaller error with respect to the case when the trend is neglected. Thus the analyst is left

with the dilemma whether to detrend a potential trend, or to neglect it.

In the paper, preliminary analyses oriented to assess how the presence of trend in precipitation series

affects the sampling properties of the estimated quantiles are illustrated. In particular, sampling properties

of precipitation quantiles, namely bias and Mean Square Error (MSE) are investigated with respect to the

size of the estimation sample, assuming a trend in the parameters of the underlying distribution. Also the

effect of preliminary trend removal is investigated and compared to the case when trend is neglected.

Analytical results are derived for the cases of simple distributions, while more complex cases are

investigated numerically by simulation.



39

Assessing spatial uncertainty of reference evapotranspiration using

stochastic simulation in southern Italy (Calabria region)

Gabriele Buttafuoco

Institute for Agricultural and Forest Systems in the Mediterranean, National Research Council, Rende CS

(Italy)


Tommaso Caloiero

Research Institute for Geo-Hydrological Protection, National Research Council, Rende CS (Italy)

Roberto Coscarelli

Research Institute for Geo-Hydrological Protection, National Research Council, Rende CS (Italy)

Environmental management decisions require an accurate computation of water balance.

Evapotranspiration is one of the major components of the water balance and has been identified as a key

factor in hydrological modelling. For this reason, several methods have been developed to calculate the

reference evapotranspiration (ET0). Whatever model is used, the errors in the input will propagate to the

output of the calculated ET0. Neglecting information about estimation uncertainty, however, may lead to

improper decision-making. One geostatistical approach to spatial analysis is stochastic simulation which

draws alternative, equally probable, realizations of a regionalized variable. Differences between the

realizations provide a measure of spatial uncertainty and allow to carry out an error propagation analysis.

Among the evapotranspiration models, the Droogers and Allen‟s modification of Hargreaves-Samani

model was used, which includes besides temperature and solar energy, a precipitation term to improve ET0

estimates in arid regions.

The objective of this paper was to assess spatial uncertainty of a monthly reference evapotranspiration

model resulting from the uncertainties in the input parameters (mainly temperature and precipitation data)

in southern Italy (Calabria region).

Temperature data were simulated by using external drift simulation with elevation as external drift, while

precipitation data were simulated by using sequential Gaussian simulation.

The ET0 was then estimated for each set of the 500 realizations of the input variables, and the ensemble of

the model outputs was used to infer the reference evapotranspiration probability distribution function. This

approach permitted to delineate the areas characterized by greater uncertainty, to improve supplementary

sampling strategies and ET0 value predictions.



40

Multivariate frequency analysis of droughts via copulas Carlo De Michele


[email protected]

Renata Vezzoli


[email protected]

Silvano Pecora, Enrica Zenoni

HydroMeteoClimate Service, Hydrology Area, Parma, Italy

[email protected], [email protected]

In Literature the assessment of droughts is conventionally based on the development of indices: SPI, SRI,

PDSI, RDI. However the statistical description of these indices is not always straightforward, and the

same is for the determination of their return periods.

Droughts, like floods, are extreme expressions of the river discharge phenomenon. Droughts, like floods,

can be identified using a threshold on the flow discharge: a flood is a period when discharges are equal or

larger than the flood threshold, while a drought is defined as a period when discharges are equal or less

than the drought threshold.

Here a drought is intended as a multivariate event described not only by its duration but considering also

the severity, and intensity. A multivariate statistical characterization of drought events is proposed using

the theory of copulas (Salvadori et al. 2007), and occurrences of drought quantified via the concept of

secondary return period.

The case study of Po river at the river sections of Piacenza, Cremona, Boretto, Borgoforte and

Pontelagoscuro, is illustrated and commented.

Investigating the propagation of droughts in the water cycle at the

catchment scale

Antonella Di Domenico

DIFA, Università degli Studi della Basilicata, Potenza, Italy.

E-mail: [email protected], Tel: +39 0971205370

Giovanni Laguardia

DITIC, Politecnico di Torino, Torino, Italy.

Maria Rosaria Margiotta

DIFA, Università degli Studi della Basilicata, Potenza, Italy.

The time series of the water fluxes and of the storages within a catchment produced by means of

simulations with a weather generator and a hydrological model are evaluated in order to assess the effect

of droughts‟ propagation on water resources.

The study area is the Agri basin, Southern Italy, closed at the Tarangelo gauging station (507 km2).

Once calibrated the weather generator on observed data, a 100 years time series of precipitation has been

produced. The drought statistics obtained from the synthetic data have been compared to the ones obtained

from the limited observations available.

The hydrological model has been calibrated based on observed precipitation and discharge. In particular,

we have assessed its capability in reproducing the flow duration curve. From the model run on the







41

synthetic precipitation we have obtained the time series of variables relevant for assessing the status of the

catchment, namely total runoff and its components, actual evapotranspiration, and soil moisture.

The obtained data have been processed for the calculation of drought indices. In particular, we have

chosen to use their transformation into standardized variables. The Standardized Precipitation Index

(McKee et al., 1993) has been calculated for different averaging periods.

The characteristics of the series have been assessed by means of autocorrelation analysis and correlation

among each other. Their drought statistics, namely number of events, duration, and deficit volumes, have

been assessed.

The characteristic time scale, as assessed by means of autocorrelation analysis, and the SPI averaging

period for which the maximum correlation is reached for the different time series tend to increase as a

result of the filtering effect exerted by the different catchment storages.

As a consequence, the number of drought events tends to decrease and their duration to increase under

increasing storage.

Comparison of different methods for the extension of monthly runoff

data

Annalisa Di Piazza, Leonardo Valerio Noto, Francesco Viola, Goffredo La Loggia

Department of Hydraulic Engineering and Environmental Applications, University of Palermo, Palermo,

Italy.


The availability of reliable and long time series of runoff data is fundamental for most of the hydrological

analyses and for the assessment and the management of water resources. However, hydrologic data set are

often characterized by a short duration and also suffer from missing data values, called gaps, mainly due

to malfunctioning of gauging stations for a specific period. In order to overcome this problem and obtain

long and continuous runoff time series, different models and methods have been previously developed and

proposed. The most of these models, used to extent the streamflow record, are conceptual, empirical,

regressive models based on the rainfall input. The main aim of the present work is to provide a

comparative analysis of different procedures proposed to estimate to fill the data gap in the monthly runoff

series.

In particular, different algorithms have been used: deterministic methods as simple linear regression,

multiple regression and artificial neural network. In particular the neural network method, considered as a

robust tools for modeling many of complex non-linear hydrologic processes with no prior assumptions on

the variables, i.e. linear or non-linear as in regression methods, is a flexible mathematical structure capable

of modeling the rainfall-runoff relationship. In this paper the application of radial basis function and

multilayer perceptron is proposed. In particular different configurations of this model, in terms of input

variables and supervised training methods, have been here developed. The above mentioned methods have

been applied to monthly runoff time series of some sub-basins of the Imera basin (Sicily, Italy) provided

by Osservatorio delle Acque. The assessment of the parameters concerning the different interpolation

methods has been carried out using a subset of the available runoff dataset while the remaining subset has

been used to compare the results provided by the different algorithms. The performances of the used

interpolation methods have been finally assessed by comparing them using different indexes starting from

the validation subset.



42

Definition of the streamflow scenario tree for long-term operation

planning studies of hydrothermal power generation system

Débora Dias Jardim Penna

Electric Energy Research Center, Rio de Janeiro, Brazil.

E-mail: [email protected], Tel: +55 21 2598 6450; fax +55 2598 6482

Maria Elvira Piñeiro Maceira

Electric Energy Research Center and State Univesity of Rio de Janeiro, Rio de Janeiro, Brazil


Jorge Machado Damázio

Electric Energy Research Center and State Univesity of Rio de Janeiro, Rio de Janeiro, Brazil.


Synthetically generated multivariate hydrological sequences are currently use to explicitly represent the

uncertainty concerning streamflow in the optimization model runned for the long-term operation planning

studies of the large Brazilian interconnected hydrothermal power generation system. The Brazilian power

system is predominantly hydro dominated. Its reservoirs have large storage volumes with multiyear

regulation capability, and many of them are disposed along the same river. In this way, the inflows to the

reservoirs possess spatial and temporal dependence, and as a consequence, there is a link between an

operation decision in a given stage and the future consequences of this decision. The set of all possible

realizations of the streamflow stochastic process throughout the planning horizon forms a scenario tree,

usually using horizon of 10 years. This tree represents the entire probabilistic universe on which are

calculated the optimal operation strategies. As the scenario tree of the long-term operation planning

problem has a high cardinality, in practice it is impossible to visit the complete tree due to computational

effort. Therefore, only a portion of the tree (sub-tree) is covered. Currently the sub-tree is selected using

the Monte-Carlo method with classical simple random sampling. This paper describes a method for

defining the sub-tree to be visited during the calculation of the optimal operating strategy for the Brazilian

hydro-thermal power system by stochastic dual dynamic programming in order to obtain more robust

results from this operation policy with regard to variations in the number of scenarios of forward and

backward simulations, and variations in the sample hydrological scenario. Two proposals for definition of

the sub-tree are analyzed: (i) substitution of the simple random sampling by latin hypercube sampling or

descriptive sampling in the multivariate streamflow scenario generation model, and (ii) application of

multivariate statistical techniques to develop criteria that allow grouping similar objects (clustering

techniques). The proposals can be applied together or separately. The results obtained by applying the

proposed techniques in the representation of the streamflow stochastic process in the long-term operation

planning model will be presented considering real cases of the Brazilian interconnect system monthly

energy operation program (PMO).





43

Sensitivity analysis using multivariate approach for a rainfall–runoff

model of Geomorphological Unit Instantaneous Hydrograph type

Emna Ellouze-Gargouri 1,2

1Laboratoire de Modélisation en Hydraulique et Environnement, Ecole Nationale d’Ingénieurs de Tunis

BP 37 Ecole Nationale d’Ingénieurs de Tunis, Le Belvédère 1002 Tunis.

2Institut Supérieur des Etudes Technologiques de Radès, Département de Génie Civil, Tunisie.

E-mail: [email protected]; Tel: +216 23 356 050

Zoubeida Bargaoui1

1Laboratoire de Modélisation en Hydraulique et Environnement, Ecole Nationale d’Ingénieurs de Tunis

BP 37 Ecole Nationale d’Ingénieurs de Tunis, Le Belvédère 1002 Tunis.

A new approach is proposed to carry out the sensitivity analysis (SA) of a model‟s inputs and parameters.

It is proposed to study the Nash unit hydrograph combined with Geomorphological Instantaneous Unit

Hydrograph (GIUH) model sensitivity to a given parameter or input (effective rainfall characteristics and

geomorphological indices), using a bivariate analysis of inputs and outputs and parameter and outputs.

The SA is carried out with tools based on rank statistics: Kendall- and Q-Q plots. The

interest of this approach is, firstly the pre-selection of the most discriminate model‟s inputs and

parameters with Kendall-plots, and then their classification in hierarchical order impact with the measure

of dependence by Kendall‟s To achieve SA, Monte Carlo Simulations (MCS) are used and different

sampling methods are adopted. Thus, the relative error distributions of model outputs (peak flow, peak

time and base time) are constructed for each parameter set. The results show that duration has the most

significant impact on peak flow; highest-order stream length and Manning roughness coefficient have the

same weight, maximum and average intensities play an identical role, and finally the variability of event

and Horton‟s ratios have the least significant role. Moreover peak time is mainly sensitive to Manning

roughness coefficient and the variability of the event, fairly sensitive to duration and Horton‟s ratios and

not very sensitive to average and maximum intensities. It is worth noting that base time has an equivalent

behavior except toward Manning roughness coefficient which has not an important influence. In addition,

peak flow is the most sensitive output compared with peak time and base time. Moreover, this analysis

shows that the choice of a distribution a priori of parameters may not has an influence on output

sensitivity.



44

Simulating hydro-meteorological variables across a range of

temporal scales with a Weather Generator

Simone Fatichi


E-mail: [email protected], Tel: +39 055 4796306

Valeriy Y. Ivanov

Department of Civil and Environmental Engineering, University of Michigan, Ann Arbor, USA.


Enrica Caporali



Long continuous series of meteorological variables are required by many ecological, hydrological,

agricultural and geomorphological applications. Long series of hydro-climatic variables are needed for

planning water resource management or to study crop and ecosystems responses to climate forcing.

Observational time series are often too short or discontinuous to achieve these purposes or allow modeling

approaches. This study presents an application of a new weather generator Advanced WEather Generator,

AWE-GEN, capable of simulating a wide range of hydro-climatic variables at the hourly time scale. The

generator operates at the point scale and reproduces precipitation, cloudiness, air temperature, vapor

pressure, wind speed, atmospheric pressure, and solar radiation partitioned into the diffuse and direct

beam components as well as into visible and near-infrared spectral bands. In the presented analysis, time

series of meteorological variables observed at the station of Firenze University and the long-term

observations of annual precipitation at a neighboring location are used. Forty years of hourly

meteorological variables are reproduced starting from 8 years of observations. The developed tool

satisfactorily reproduces climate characteristics at a given location. The reproduction of higher order

statistics, including the extremes, is verified for the simulated variables. A method to account for inter-

annual variability of precipitation, allows one to preserve this important characteristic of climate, reducing

the over-dispersion problem present in weather generator. A comparison of the observed and simulated

climatic variables shows the excellent performance of AWE-GEN to match statistics across a wide range

of temporal scales and to preserve the principal causal relationships among the variables.

Analysing hydrological risk and water policies implications on crop

production in the Ebro River basin in Spain

Zaira Fernández and Sonia Quiroga

Department of Statistics, Economic Structure and International Economic Organization, Universidad de

Alcala, Spain; E-mail: z.fernandez@ uah.es and [email protected], Tel: +34 918855198; fax: +34

918854201

The increasing pressure over water systems on the Mediterranean enlarges the existing water conflicts and

difficult the water supply for agriculture. In this context, one of the main priorities of agricultural research

and public policy is the adaptation of crop yields to contingencies such as droughts and floods, climate

change, decreased resource availability and more severe environmental standards.

This paper is focused on the Ebro River basin in Spain, which suffers recurrent drought events during

summer resulting in conflicts over water irrigation. So it is crucial for this Mediterranean region, were

irrigation reaches up to 90% of total water consumption, to measure the risk of climate variability to




mailto:z.fernandez@%20uah.es



45

agriculture and to implement water demand policies that promote an efficient allocation and use of the

resource in the farms of the region.

Our methodological approach includes four steps. First, we use panel data econometrics to estimate the

impacts of rainfall and irrigation water on crop yields. This study is not limited to general crop production

functions and considers the linkages between economic and biophysical aspects, which potentially may

have an important effect on crop productivity. We use traditional statistical models of yield response to

address how environmental, hydrological, technological, geographical and economic variables affect the

yield of the main Mediterranean crops in the Ebro river basin. Second, this study takes into consideration

the effects of those interactions and analyzes the gross value added of the region and its interaction with

the aggregate crop production and price. In few words, we try to understand the interactions between

agricultural production and profit functions focusing on water demand. Third, we assess the structural

adjustments that could allow coping with increased water restrictions for the agricultural sector. This can

be useful in a context of adaptation to the risk of the climate variability and increased environmental

pressure. And finally, we simulate projections for different policy scenarios. We use Monte Carlo method

to characterize statistical properties of crop yield in response to precipitation patterns or policy

adjustments. These simulations are useful to evaluate probabilistic risk.

An approach to propagate streamflow statistics along the river

network Daniele Ganora

Dipartimento di Idraulica, Trasporti e Infrastrutture Civili, Politecnico di Torino, Torino, Italy.

E-mail: [email protected],Tel: +39 011 0905684; fax +39 011 0905698

F. Laio

Politecnico di Torino, Torino, Italy.

P. Claps

Politecnico di Torino, Torino, Italy.

The problem of the lack of data for predicting streamflow statistics in catchments is often handled by

means of regional statistical procedures. However, this approach does not preserve the information related

to the natural hierarchy between gauged stations deriving from their location along the river network. This

information is particularly important when one wants to estimate runoff at a site located immediately

upstream or downstream a gauged station. In this case, a possible alternative is to estimate the variable

directly, on the basis of the corresponding statistics calculated at the gauged station. The closest the

estimation point is to the gauged station, the greater is the expected quality of this estimation procedure.

This approach, referred to as stream estimation method to underline that it is applied to points along a

stream network, requires to identify a suitable formula to compute the variable at the ungauged site. This

formula can be based on a set of basin characteristics, or, in alternative, on a regional estimate (local

estimation coupled with a regional model). Then, a criterion to assess the reliability of the stream model

and its domain of application is defined and, finally, the accuracy of the approach is evaluated through the

assessment of the standard deviation of its estimates. In this way it is possible to compare the variance of

the stream estimates against the variance of other models, if any, and thus to choose the most accurate

method (or to combine different estimates).

The stream estimation method has been applied in the Northwestern part of Italy to improve the results of

an existing regional model for flood frequency estimation. In particular, the variables involved have been

the L-moments, that can be used to summarize the behavior of the frequency curve. Firstly, the gauging

stations present in that area have been organized according to their location along the river network, in

order to identify all the couples of connected stations. Then, each station has been used to predict the L-

moments of its linked counterpart, using a simple formula for the information transfer. The comparison of



46

the results with the sample data shows that the stream estimation approach can be effectively used to

improve the regional estimates for the L-moment of order one (the mean), in particular when the area-ratio

of a pair of gauged-ungauged basins is less or equal to ten. The method, in contrast, does not provide

significant improvements to the estimation of higher-order L-moments.

Development of stage-discharge rating curve in river using Genetic

Algorithms and Model Tree Bhola NS Ghimire

Department of Civil Engineering, IITB, India.


Dr. M. Janga Reddyl

Department of Civil Engineering, IITB, India

Discharge measurement in rivers is a challenging job for hydraulic engineers. A graph of stage versus

discharge or the line through the data points represents the stage-discharge relationship, also known as

rating curve. The stage-discharge relationship is an approximate method employed for estimating

discharge in rivers, streams etc. For various hydrological applications such as water and sediment budget

analysis, operation and control of water resources projects, the accurate information about flow value in

rivers is very important. Stages are easy to measure as compared to the measurement of discharge in

rivers. The stage-discharge relationship at a particular river cross-section, even under conditions of

meticulous observation, it is not necessary unique as rivers are often influenced by several other factors

which are neither always understood, nor easy to quantify. This is due to the fact that in reality, discharge

is not a function of stage alone. Discharge also depends upon longitudinal slope of river, geometry of

channel, bed roughness etc. However, the measurement of these parameters at even and every time step

and section is not possible. Hence there is a need to establish the accurate relationship between stage and

discharge. The conventional parametric regression methods usually fail to model these relationships.

This paper presents the use of genetic algorithms (GA), a search procedure based on the mechanics of

natural selection and natural genetics, and Model Tree (M5), a data driven technique for dealing with

continuous class problems, that provides structural representation of the data and piecewise linear fit of

the classes, for river hydrology to establish the stage-discharge relationship. The results obtained are

compared with the other methods such as gene-expression programming (GEP), multiple linear

regressions (MLR) and classical stage-discharge rating curve (RC). To measure the performance of

models, statistical measures such as coefficient of determination and root mean square error are used. The

results obtained from the GA based model as well as MT based model are found to be much better than

the other methods.


47

Parametric analysis of regional trends in observed extreme rainfall

in Denmark

Ida Bülow Gregersen, Karsten Arnbjerg-Nielsen

DTU Environment, Technical University of Denmark, DK-2800 Kgs. Lyngby, E-mail:

[email protected], [email protected]

Henrik Madsen

DHI Water • Environment • Health, DK-2970 Hørsholm, Denmark, E-mail: [email protected]

It has become commonly accepted that the annual precipitation over Northern Europe will increase in the

future and that the precipitation patterns are changing so extreme rainfall will become more frequent and

more intense. Several projects have applied the outcome of climate models to investigate how the extreme

precipitation will change in the future, with the general conclusion that the frequency of the extreme

events will increase and that the return period of the very extreme events will decrease. The spatial

resolution of the recent climate models is, however, still too coarse for accurate simulation of extremes. It

is therefore of high interest to evaluate if changes in measured extreme precipitation can be observed and

how this information can be used to downscale extreme precipitation characteristics from climate model

simulations.

This paper presents a new statistical approach to evaluate changes in the extreme rainfall over time,

looking both at the annual number of extreme events and the mean intensity of the extreme events for

durations ranging from 1 min to 24 hours. The analysis is based on measurements from 70 rain gauge

stations in Denmark that have been active in the period between 1979 and 2009. When periods of rain

gauge malfunction has been taken into account, the present data material represents 1428 station-years.

Because any development over time can be blurred by both the periods of breakdown and for some

stations a short period of measurements, the statistical models are based on all 70 stations. The annual

number of extreme events is modeled by a Poisson rate model using the annual number of measurement

days for each station as a parameter in the model. The mean intensity of the extreme events is modeled by

ordinary linear regression. To reduce the very high variability in the data the mean annual precipitation at

each station and an index value for the North Atlantic Oscillation are included in the models, both

showing significance. The correlation with the North Atlantic Oscillation is, however, shown to be

dependent on the modeled intensity. The study shows that the annual number of extreme events increases

significantly over time for all durations, for the mean intensity this is only the case for duration between 5

and 180 minutes. The models are, however, only capable of describing a small amount of the total

variability in the data. By extrapolation of the estimated rates of increase the characteristics of future

extreme rainfall is found and a „climate factor‟ is calculated as the ratio between the future and present

characteristics. The climate factor for the number of extreme events is between 2.8 and 3.7 and between

1.5 and 1.8 for the mean intensity of the extremes. These results are of course affected by the uncertainty

of extrapolation, but the estimated climate factors are considerably larger than what have been predicted

by downscaling results from regional climate models.




48

Application of L-moment for estimation of quantile mixtures

Hadi Hassanzadeh

Department of Water Engineering, University of Isfahan Technology, Isfahan 84156-83111 IRAN.


Saeid Eslamian


E-mail:[email protected], Tel: +98-311391-3432; fax +98-311391-2254

Mohammad Abdolhoseini


Visiting Scientist, Faculty of Agriculture - Tuscia University, Italy


Salvatore Grimaldi

Water engineering section, GEMINI Department, Tuscia University, Italy


L-moments are linear combinations of order statistics which are robust to outliers and virtually unbiased

for small samples, making them suitable for frequency analysis, including identification of distribution

and parameter estimation. L-moments are appropriated alternatives to the conventional moments. This

paper develops the use of L-moments proposing new parametric families of distributions that can be

estimated by the method of L-moments. Because of the analogy to mixtures of densities, this class of

parametric families is called quantile mixtures. Two parametric families, the logistic-polynomial quantile

mixture and the generalized logistic-polynomial quantile mixture are developed. The proposed quantile

mixtures are applied to frequency analysis of annual maximum precipitation and annual maximum

discharge, and compared to the logistic and generalized logistic distributions.

Statistical characterization of precipitation series in city Bogotá Eddy Herrera Daza

Mathematics and Statistics, MSc, PhD student in Engineering, Pontificia Universidad Javeriana, Bogotá

D.C. Colombia,


Nelsón Obregòn Neira

PhD, MSc, IC, Director Engineering, Director Geophysical Institute, Professor Faculty of Engineering,

Pontificia Universidad Javeriana, Professor Faculty of Engineering Pontificia Universidad Nacional de

Colombia, Bogotá DC Colombia. [email protected]

This paper proposes an analysis of rainfall time series, to be applied to stations in Bogotá with the purpose

of optimizing the work of estimating missing data. The methodology is based on extensive statistical

characterization of observables in order to facilitate the process of identifying and selecting best models to

approximate the missing data problem. Among the statistical attributes include those intentional linear and

nonlinear characterizations and those who study the dynamic properties of the generators of the

observable systems.






49

Analysis self- similarity with Hurst exponent and the autocorrelation

function Eddy Herrera


D.C. Colombia,


Hugo Armando Rico T.*


Felipe Ardila*


Jaime Moreno*


Eder Cárdenas*


Raúl Castañeda*


* Master's student in Hydrosystem’s. Civil Engineering Department. Pontificia Universidad Javeriana,

Bogotá, Colombia.

Efraín Dominguez

PhD. Technical Sciences (Hydrology, Water Resources and Hydrochemistry). Pontificia Universidad

Javeriana. Bogotá D.C., Colombia. [email protected]

The degree of self-similarity of a series can be expressed using only one parameter which expresses the

rate of decrease of the autocorrelation function (FAC), this parameter is called Hurst parameter. In this

paper we present an analysis of the Hurst exponent, different methods of estimation and establish the

fundamental relationship with the autocorrelation function.

Derivation of the Probability Plot Correlation Coefficient Test

Statistics for the Generalized Logistic Distribution

Sooyoung Kim

Department of Civil and Environmental Engineering, Yonsei University, Korea


Hongjoon Shin



Taesoon Kim



Jun-Haeng Heo



The selection of an appropriate probability distribution is very important in hydrology to estimate the

accurate design rainfall. An appropriate probability distribution is chosen by the goodness of fit tests such

as Kolmogorov-Smirnov test, Cramer von Mises test, 2 -test, and probability plot correlation coefficient

(PPCC) test. The PPCC test has been known as a powerful and easy test among the goodness of fit tests.












50

Generally, the PPCC test statistics are calculated by considering significance levels, sample sizes, plotting

position formulas, and shape parameters of a given probability distribution included a shape parameter. It

is important to select an exact plotting position formula for a given probability distribution because the

PPCC test statistics are defined from correlation coefficient values based on the selected plotting position

formula. After Cunnane (1978) defined the plotting position that related with the mean of data and

proposed the general formula that can be applied to various probability distributions, various plotting

position formulas have been developed for considering the effect of coefficients of skewness related with

the shape parameter for a given distribution. In this study, the PPCC test statistics are derived by using a

plotting position formula contained a term of a coefficient of skewness that can be considered the effect of

shape parameters for the generalized logistic distribution. In addition, the derivation of the PPCC test

statistics is performed by considering various sample sizes, significance levels, and shape parameters of

the generalized logistic distribution. And then, the power test to estimate the rejection ratios of derived

PPCC test statistics is performed and the comparison between derived plotting position formula and other

plotting position formulas such as Gringortern, Cunnane, and Blom is accomplished.

On feasibility of L-Moments method for distributions with

cumulative distribution function and its inverse inexpressible in the

explicit form

K. Kochanek (corresponding author), W.G. Strupczewski and I. Markiewicz.

Department of Hydrology and Hydrodynamics, Institute of Geophysics,

Polish Academy of Sciences, Ksiecia Janusza 64, 01-452 Warsaw, Poland.

Fax: 48 22 6915915; e-mails:[email protected], [email protected], [email protected]

Because of its convenient properties, the L-moments method, LMM, (or the probability weighted

moments method) has become more popular than maximum likelihood in applications to hydrologic

extremes, both thanks to its computational simplicity and good performance for small samples. However,

to apply the LMM as a method of parameters estimation, the relationship between the L-moments and

distribution parameters must be known. It is relatively easy to get it when the cumulative distribution

function or its inverse (i.e. quantile function) are expressed in an analytically closed form. There are also

distributions for which, using tiresome and time consuming complex mathematical derivations, one can

eventually find formulae for the second, third or sometimes even forth order linear moments. But

problems arise when one need linear moments of higher orders whose analytical derivation is even more

difficult (or impossible) but which are indispensable to calculate linear moments ratios or LH-moments

and their ratios (especially useful for distributions‟ upper tail estimation). In this paper we introduce a

useful solution to circumvent these obstacles which allows to calculate linear moments of any level with

accuracy sufficient for the practical hydrology. The idea is based on the finding that expressions for linear

moments and their ratios for both distributions expressible and inexpressible in inverse form are of the

similar structure as the conventional moments and their ratios (which are known). We propose two

calculation procedures, namely Monte Carlo simulations and the numerical integration. As an example,

we will show how to use the LMM for distributions that have no explicit inverse form and are

characterised by one shape parameter, including the two- and three-parameter Inverse Gaussian

distribution which is more and more popular in flood frequency analysis (FFA) and, according to the latest

research, is among the best models for flood frequency analysis in Poland.


51

Estimating precipitation extremes in climate change simulations by

the region-of-influence method

Jan Kysely 1, Ladislav Gaal

1,2, Romana Beranova

1, Eva Plavcova

1,3

1 Institute of Atmospheric Physics AS CR, Prague, Czech Republic

2 Slovak University of Technology, Bratislava, Slovakia

3 Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic


Future scenarios of high quantiles of precipitation amounts are examined in an ensemble of regional

climate model (RCM) simulations with high resolution (25 km). The region-of-influence method is

adopted as a pooling scheme in the frequency analysis of one-day and multi-day precipitation extremes.

The method takes into account data from a pooling group (set of gridboxes) when fitting the extreme value

distribution in any single gridbox. „Homogeneity‟ of the pooling groups, i.e. the condition that the

distributions of extremes are identical apart from a gridbox-specific scaling factor, is tested by a built-in

regional homogeneity test. The performance of the region-of-influence method is compared with a single-

gridbox analysis.

Differences between scenarios of changes for winter (DJF) and summer (JJA) seasons, and the

dependence of results on the driving GCM are examined. We show that the application of the pooling

scheme in the frequency analysis efficiently reduces (random) variations in the estimates of parameters of

the extreme value distributions in individual gridboxes that result from large spatial variability of heavy

precipitation. The method represents a useful and straightforward tool for „weighting‟ data from nearby

gridboxes within the estimation procedure in high resolution climate change simulations.

Revisiting the maximum-value test for the frequency analysis of

extreme events

Francesco Laio, P. Allamano and P. Claps

Department of Hydraulics, Transports and Civil Infrastructures, Politecnico di Torino, Torino, Italy.


Validation of probabilistic models based on goodness-of-fit tests is an essential step for the frequency

analysis of extreme values. Standard goodness-of-fit tests compare a suitable representation of the

empirical frequency curve to the hypothetical probability distribution H. A common drawback of these

tests is that their outcome is mainly determined by the behaviour of the hypothetical model in the central

part of the distribution, while the behaviour in the tails of the distribution, which is of course very relevant

in hydrological applications, is relatively unimportant for the results of the test. The maximum-value test,

originally proposed as a technique for outliers detection, is a suitable, but seldom applied, technique to

overcome this problem. The test is specifically targeted to verify if the maximum (or minimum) values in

the sample are consistent with the hypothesis that the distribution H is the real parent distribution.

However, usual applications of this test fail to consider the effects of parameter-estimation: when

parameters are estimated on the same sample used for verification the power of the test is relevantly

decreased, in particular when small samples are considered. We propose here a simple, analytically

explicit, technique to suitably account for this effect, based on the application of L-moments estimators of

the parameters. We demonstrate with an application to artificially generated samples the superiority of this




52

modified maximum-value test with respect to the standard version of the test. We also show that the test

has comparable or larger power with respect to other goodness-of-fit tests (e.g, chi-squared test,

Kolmogorov-Smirnov test, Anderson-Darling test), in particular when dealing with small (sample size

lower than 30), highly skewed, samples.

Stochastic simulation of hydrological processes under water

management transformation and global climatic change

Loboda Nataliia

Department of Hydroecology and Water research, Odessa State Environmental University, Odessa,

Ukraine

E-mail:[email protected], Tel: +38 048 78527181; fax +38 0482 326764

Main problems of calculation, prediction and forecasting of hydrological processes in Ukraine are lack of

observed runoff data and significant influence of water management. The main approach applied in this

study is the stochastic modelling. Various stochastic models are widely used in hydrology (Kottegoda

1980; Yevjevich 1987; Ratkovich 1993; Bras and Rodriguez-Iturbe 1993; Lawal et al. 1997; Srinivas and

Srinivasan 2001; Salas et al. 2003; Bolgov 2005; Prairie et al. 2007). The stochastic simulation model of

water resources has been developed for evaluation and prediction of probabilistic characteristics of annual

runoff.

The first block of the model includes the evaluation of runoff on meteorological data. The input of the

model is meteorological data, and the output is the calculated runoff. Runoff, calculated by

meteorological data, is termed „climatic runoff‟. The theoretical base for the calculations of the climatic

runoff is water-heat balance equation. Climatic runoff is identified with the natural zonal runoff of the

rivers. For small and medium-size rivers, where formation of runoff depends not only on climatic

conditions but also on the underlying surface factors, methodology for determination of regional corrector

coefficients for transition from climatic runoff to the natural one is developed. Comparison of annual

runoff characteristics determined with the help of the model with the actual set of observational data on

the rivers showed their satisfactory agreement. Spatial generalization (mapping) of the annual normal

climatic runoff is of importance for the Ukraine because predominant part of the runoff has been changed

by water management activities.

The second block of the model is generation of natural annual runoff series. The numerical simulation of

natural runoff series is based on stochastic models of river annual runoff fluctuations (Ratkovich 1993;

Bolgov 2005). The problem of insufficiency of runoff data under natural conditions were solved by the

water-heat balance method using meteorological data (Loboda 1998; Loboda and Chinh 2004).

The third block of the model includes stochastic simulation of life conditioned annual runoff series. The

water-management system, i.e. watershed has been considered under external (climatic) and internal

(human-induced) impact. The stochastic model was performed to model water management effect. The

natural runoff was considered as a stochastic process. The external factors (precipitation, evaporation) are

also stochastic. The runoff transformations, that have place within the watershed under water

management, include stochastic components. To retain the stochastic stationarity for the generated series

of life conditioned runoff, all non-random are defined as constant in every case of numerical simulations.

The output of the model is operator that describes the transformation of system's initial state, i.e. a natural

flow, under water management activity. The mathematical representation for this operator is the response

functions, represented as relationship between runoff statistical parameters and non-random characteristics

of size of water management activity under certain climatic conditions. The model was applied in

predictions of Ukrainian water resources under global climatic change scenarios (Loboda 1998), in

restoration of natural runoff series with using first component of EOF method (Loboda, Glushkov and


53

Khohlov 2006); for evaluation and prediction state of water resources under water management

transformation and global warming.

Influence of the North Atlantic Oscillation and the Western

Mediterranean Oscillation in the maximum flow events in Spain

Jesús López de la Cruz

Department of Hydraulic Engineering and Environmet, Polytechnic University of Valencia, Valencia,

España. E-mail: [email protected]

Félix Francés García

Department of Hydraulic Engineering and Enviroment, Polytechnic University of Valencia, Valencia,

España. E-mail: [email protected]

The flow rates in the Spanish peninsula are characterized by large inter-annual variability, especially in

the south of Spain. One way to identify this variation with climatic indices, that can have an impact on the

study area. This article discusses the possible influence of the levels of the North Atlantic Oscillation

(NAO) which has been shown in various studies presented a domain in climate regimes in the region and

Western Mediterranean Oscillation (WeMO) in the caudal maximum instantaneous and maximum daily

average in several basins of Spain. To study the information was taken from 67 monitoring stations for

flow in natural regime. The results show that there is a strong correlation between peak events that occur

in winter (December to March) with the negative phase of NAO index which is significant to the basins of

southern and western Spain, while there are correlations significant for the eastern basin (Mediterranean)

and the northern basins. In the case of WEMO index showed significant correlations with the

Mediterranean basin and north, which were obtained as for the NAO index for winter. There is a strong

impact of the NAO index and WEMO in winter peak flows, and it has an impact almost zero for the

periods presented in events outside of winter.

Assessing changes in 100-year flood discharge estimates used in land

use regulation in the greater Chicago region, USA

Momcilo Markus

Illinois State Water Survey, Institute of Natural Resource Sustainability, University of Illinois at Urbana-

Champaign, USA.

E-mail: [email protected], Tel: +(217) 333-0237; fax +(217) 333-2304

The national regulations in the United States, administered by the Federal Emergency Management

Agency and implemented by local jurisdictions, rely on 1%-annual-chance exceedance flood event, i.e.,

the base flood. The floodplain maps need to be updated as conditions changed, whether due to land use

development, hydraulic alterations, long-term hydrologic variability, reassessment, or climate change.

Developing approaches to accurately account for changing conditions becomes a critical task for

hydrologists. In this presentation, apparent increases -- and decreases -- in 100-year flood flow estimates

in two different watersheds in the greater Chicago region are illustrated and discussed in the context of

regulatory implementation. The first example presents the combined effects of urbanization and increases

in heavy precipitation on the increasing floods on several watersheds in the suburban Chicago area. The

results indicated that for precipitation analysis, the uncertainty based on sample selection is critical,





54

followed by the assumed distribution and selection of statistical region. The other example shows

decreasing floods based on the climate change in the mostly rural Pecatonica River watershed. It was

found that the long-term trends in annual peak discharges are related to significant seasonal changes, not

detectable on an annual scale. The seasonal shifts also were discussed. In conclusion, due to the increasing

complexity of flood frequency analysis in a changing environment, floodplain managers need to apply

more complex non-standard statistical approaches, and potentially data mining. Finally, the results

obtained in this study can teach us valuable lessons about the ways to study the hydrologic effects under

different future scenarios.

Low flows quantification and regionalization in North West Algeria

M. Mehaiguene, M. Meddi, S. Toumi

LERP - University Center Khemis Miliana, Khemis Miliana, Algeria


A. Longobardi

Department of Civil Engineering, University of Salerno, Fisciano, Italy.

Low flow hydrological regimes and their characterization are crucial for an efficient development of water

resources management tools, especially in those areas stressed by the combination of a dry climate and an

excessive water demand, such as the Mediterranean basins. Consequently, a lot of effort has been made by

the worldwide scientific community toward the prediction of low flows features in ungauged catchments.

The aims of the present study are: (1) the quantification of low flows characteristics, both at the annual

and seasonal scale, in particular with regard to the BFI, one of the most important low flow features, and

(2) the application of a simple regional approach to predict the BFI at ungauged sites. The investigated

area is located in North West Algeria and has an extension of about 130000 Km2. The relevant extension

and a poorly dense monitoring network, which consist in 24 hydrometric stations, require the

identification of homogeneous regions for further application of simple linear regional regression models.

Regions delineation is accomplished both through the Principal Component Analysis method (PCA) and

the cluster analysis, based on geographic, physiographic and climatic data. Five homogeneous and

geographically contiguous areas have been identified, and comparisons between global and local regional

linear regression models are made. Comparisons indicate a large percentage of BFI explained variance in

each area.

A single-site rainfall disaggregation model based on entropy

Valeria Montesarchio, Francesco Napolitano

D.I.T.S., Sapienza University of Rome, Rome, Italy.


Since the first development of Shannon entropy theory (1948) many applications interested hydrology and

water resources. Recently, Koutsoyiannis showed how entropy can be used to explain and model

hydrological phenomena, such as the observed dependence properties of the rainfall occurrence process,

including the clustering behavior and persistence. In this paper an entropy approach is presented, in order

to disaggregate daily rainfall amounts to hourly level. First of all an investigation of basic statistical

properties (maximum and mean values, variances, skewness, probability and length of dry intervals, and

dependence structure of rainfall) is performed for different time scales, ranging from 1 to 24 hours. Then,

using the entropy concept, properties like proportion of dry intervals and rain amount are evaluated. The



55

analysis are performed using a data set of 20 rain gauges from North Lazio Region, in Central Italy, for

the years 1993-2008. Finally, the disaggregated data set are evaluated against the available hourly data and

the results are discussed.

The Impact of exceptional rain events on the development of gullies

in geological formation in Dhkekira watershed

Boughattas Nour El Houda

LTSIRS, ENIT, Tunis, Tunisie.

E-mail: [email protected], Tel: +216 71 874 700; fax +216 71 872 729

Mohamed Habib Snane

LTSIRS, ENIT, Tunis, Tunisie

Mennoubi Sfar Felfoul

INAT, Tunis,Tunisie.

Boussema Mohamed Rached

LTSIRS, ENIT, Tunis, Tunisie.

High return period monthly rainfalls of September-October 1969 have initiated extensive gully

development in many lithologic formations of the semi-arid region of central Tunisia. Monthly rainfall of

that magnitude have not happened since then although in the 1969-2000 period yearly rainfall has

exceeded that of 1969. The impact of these September-October 1969 rainfall events on gully density

increase in the watershed of Oued Dhkekira (327ha) was quantified using the variation in gullies length as

measured with Arc View GIS on the rectified air photographs taken in 1962 and 1974. Photos rectification

was done with ERDAS IMAGINE using the digital elevation model (DEM) and the coordinates of photos

fiducials and minimum 4 tics marks on each photograph. Most of these gullies are located on the Souar

lithologic formation dated from middle to late Eocene. This formation is characterized by large strata of

gypsum clay, marl and sandy clay intercalated with closed banks of limestone and friable sandstone.

Gullies length measurement were also made on the same watershed using the air photographs taken in

1989 and 2000.The number of gully that has been monitored in this watershed increased from one period

to another due to the newly created gullies. Comparison of gully density variation in 1974-1989 and 1989-

2000 periods has shown that gully incision in these two periods was not as important as during the 1962-

1974 period. Statistical analysis of monthly rainfall events was performed on the monthly rainfall series of

the years 1962 to 2000 and a relationship was established between the rate of increase in gully density and

the mean annual rainfall provided by monthly rainfalls exceeding a threshold value.

Recovering dynamics of hydrological observed data via phase space

reconstruction and Genetic Programming

Nelson Obregón

Institute of Geophysics, Javeriana University, Bogotá, Colombia.

E-mail: [email protected], Tel-Fax: +57 1 3208320

F. Ardila

MSc Program on Hydrosystems, Javeriana University, Bogota, Colombia.

This work presents advances of a new approach aimed at recovering dynamics of hydrologic observed

data. Specifically the model is based on phase space reconstruction via Takens‟ Theorem and statistical

attributes of nonlinear time series analysis. Under this framework a new genetic approach is conceived in




56

order to explore dynamical orbits such that a set of coupled nonlinear differential equations describing

dynamics of observables can be sought. The model is tested on well-known chaotic dynamical system and

hydrological registered data. The results suggest that this novel approach is suitable not only to conceive

and explore dynamical properties on observed times series, but also to advance toward to proper

hydrological forecast in the phase space.

Long-term wave hindcast by neural networks using NCEP/NCAR

Reanalysis wind fields David J. Peres, Vincenzo Nicolosi and Antonino Cancelliere



Wave information is of fundamental importance for almost all coastal engineering purposes. Proper

information is rarely available from buoy measurements: for instance, in Italy, the set up of an efficient

national buoy network has been possible only in recent times. In general only 10-20 years of observations

are available, often inadequate for a good estimation of design waves characterized by high return periods.

To overpass this lack of information, both simple empirical or complex physically-based models can be

used to derive wave data from surface-wind data. The strongest drawback of these traditionally employed

models is the impossibility to take explicitly into account buoy measurements information when available.

Furthermore, difficulties in gathering wind-fields data discourage indirect derivation of sea-state

parameters series to increase the length of observed time series, so applications are generally developed

for cases of total absence of wave measurements.

In this paper, surface-wind data of the NCEP/NCAR Reanalysis I project, available on the world wide

web, are used. Supervised Artificial Neural Networks are employed to reconstruct 6-hourly significant

wave-height time series in the Mediterranean for a point in deep water near Catania from 1948 to 2008,

where buoy measurements are today available from 1989 to 2005.

The obtained results have shown good suitability of the reanalysis wind-data for indirect wind-wave sea

state parameters derivation and good performance of the Neural Network modeling technique. This was

also confirmed by the comparison with other long-term wave reconstructions, such as the ECMWF‟s ERA

40 one.

Homogeneity tests based on multivariate L-moments for extreme

precipitation events

Jan Picek

Department of Applied Mathematics, Technical University of Liberec, Liberec, Czech Republic.


Jan Kyselý

Institute of Atmospheric Physics, Prague, Czech Republic

In regional frequency analysis, the examination of the regional homogeneity, i.e. the condition that

distributions of extremes are identical except for a site-specific factor, represents an important step of the

procedure. If more than one variable is of interest, e.g. one-day and multi-day precipitation, homogeneity

tests are usually applied and evaluated independently. The aim of this study is to identify homogeneous

regions according to multivariate statistical characteristics of precipitation extremes in the Czech



57

Republic. Precipitation totals at more than 200 stations over 1961-2007 form the input dataset. Statistical

multivariate tests for regional homogeneity of Chebana and Ouarda are utilized. In the present paper,

practical aspects are investigated jointly for maximum annual k-day precipitation amounts, with k = 1, 3, 5

and 7 days. Simulation experiments are proposed to evaluate stability of the test results. A comparison is

carried out with the classical homogeneity test of Hosking and Wallis based on several types of

regions.The study is supported by the Czech Science Foundation under project P209/10/2045.

Are biases in daily precipitation statistics in climate model

simulations conditioned by large-scale atmospheric circulation?

Eva Plavcova and Jan Kysely

Institute of Atmospheric Physics AS CR, Prague, Czech Republic


Regional climate models (RCMs), which are the most widely used tools for simulating regional scenarios

of climate change, suffer from many deficiencies in the reproduction of daily precipitation statistics in

control (recent) climate. Little attention has been paid to the question whether the biases are conditioned

by characteristics of large-scale atmospheric flow, and/or whether they are related to biases in the

reproduction of circulation patterns. The present study examines links between atmospheric circulation,

characterized by indices that measure flow direction, strength and vorticity, and statistics of daily

precipitation over central Europe. The relationships are compared between observed data (re-analysis and

gridded E-Obs dataset) and an ensemble of control simulations of high-resolution RCMs driven by several

global models (GCMs). We focus on low precipitation amounts in spring-summer season (early vegetation

period, AMJ), and heavy precipitation in winter (DJF) and summer (JJA), i.e. events with direct impacts

associated with droughts and floods. In addition to extremes, full empirical pdfs of precipitation,

conditioned by circulation types, are compared between observations and RCM simulations using

goodness-of-fit tests. This allows identifying whether or not systematic precipitation errors in model

outputs are related to specific circulation conditions.



58

Peaks over threshold method in trend analysis of maximum daily

precipitation in two climatic regions

Jasna Plavšić

Faculty of Civil Engineering, University of Belgrade, Belgrade, Serbia.

E-mail: [email protected], Tel/fax: +381 11 3370206

Ennio Ferrari

Dipartimento di Difesa del Suolo, Università della Calabria, Rende, Italy.


Jovan Despotović

Faculty of Civil Engineering, University of Belgrade, Belgrade, Serbia.


Beniamino Sirangelo

Dipartimento di Difesa del Suolo, Università della Calabria, Rende, Italy.


The objective of the present study is to investigate possible nonstationarity in daily rainfall series in two

different climatic regions, southern Italy and Serbia, using the peaks over threshold (POT) method. The

nonstationarities are looked for in both components of the POT method, peak counts and exceedances.

The number of peaks is modeled using an appropriate discrete process (Poisson or binomial/negative

binomial) by taking into account both seasonal (inter-annual) variability and long-term trend. Series of

peak exceedances are tested for the presence of linear trend and homogeneity. The exceedances are

modeled with an appropriate probability distribution (exponential or Weibull), with time-varying

parameters if proven necessary to take long-term trend into account. The parameters are then assumed to

have linear trend with time and a method for their estimation is devised. The series are also modeled using

the standard POT approach without the nonstationarity assumptions and the results are compared to the

nonstationary approach in terms of quantile estimation. Moreover the effect of the choice of the threshold

level on the identification of nonstationarity is discussed.

A stochastic approach to assess hydro-geotechnical uncertainty in

flood risk mapping

Roberto Ranzi

DICATA, University of Brescia, Brescia, Italy.

E-mail [email protected], Tel: +39 030 371 1291; fax: +39 030 371 1312

Stefano Barontini

DICATA, University of Brescia, Brescia, Italy.

Michele Ferri

Autorità di Bacino dei fiumi dell'Alto Adriatico, Venezia, Italy.

Baldassare Bacchi

DICATA, University of Brescia, Brescia, Italy

Several major rivers in the world have a long and complex history of training in order to protect the plain

and the cities from inundations and it is difficult to predict the position of levees‟ failures can occur and

the width and depth of breaches. As a consequence, flood risk mapping is affected by major uncertainties

as the volume and duration of inundations strongly depend on the mechanism, position and the geometry

of the collapse of rivers‟ embankments.






59

In view of an investigation on the uncertainties in flood risk mapping, a stochastic approach based on an

inhomogeneous PRP-Poisson Rectangular Pulses model of levee‟s breaches occurrence and geometry

coupled with a simplified method to determine the flooded area is developed. To calibrate some of its

parameters we collected information on major floods and statistics of levees breaches occurred in the Po,

Adige, Brenta, Piave and Tagliamento rivers over the last two centuries. We provide the statistics of 225

historical breaches occurred in the 1801-2000 time period in the floodplain course of the Po river, which is

324 km long (from upstream to downstream, 80 km are meandering, 100 km are braided and the others are

sinuous--straight). The first two stretches are characterised also by lateral inflow from tributaries, while

the last stretch is mainly characterised by routing and floodplain inundation processes.

Three dominant levee collapse mechanisms (overtopping, erosion, piping) were considered. The highest

number of breaches was registered in the meandering course, with overtopping strongly (78%) as the most

frequent dominant mechanism. Three different samples of historical floods were considered, separated by

two important floods (1857 and 1879). A significant decrease in the total number of breaches (per year

and per kilometer) was observed since the second half of the 19th century, as a consequence of flood

directives issued in that period and to the levees restoration after the floods. A decrease in the percentage

of the overtopping was observed in the second period (77% in the first half of the 19th century, 56% in the

1857-1879 window and again 74% up to the 1951 flood) jointly to the increasing of piping (respectively

9%, 32% and 18% in the three time windows). Statistics on the breach occurrences and geometry and

flood inundations in the Adige, Brenta, Piave and Tagliamento rivers are also presented and discussed in

view of the calibration of the inhomogeneous PRP model and the resulting flood risk mapping.

Using the Stochastic Index method for calculation of flow duration

curves on ephemeral basins

Maura Rianna, Fabio Russo and Francesco Napolitano



The flow river regime can be described using flow duration curves (FDC) and annual flow duration

curves. FDCs represents frequency distribution of discharges, and can be derived from gauged data,

AFDCs allow to introduce confidence intervals for the median AFDC, and to assign return periods to

individual AFDCs. But in semiarid environments with ephemeral flows or little basins, where is really

important have information about the flow duration curves because is possible to have environmental

problems caused by pollutant discharges, the presence of zeros in a registration creates a discontinuity that

makes difficult to model this curves using common distributions.

Therefore this work combines the stochastic index model, that relates FDC and AFDC of daily flow

series, and that enables to reproduce the mean, the median and variance of AFDC without regard

persistency and seasonality of the series for recreate the curve of the non-zero period, with the theory of

total probability, that allows to evaluate the percentage of time the river is dry. Next is necessary also

apply this model to ungauged stations, regionalizing the information and using a cross-validation

procedure for evaluate robustness of results.

This procedure is applied to a nested catchment in the Lazio Region.



60

Regionalization and geomorphological analysis of watersheds in

Colombia

Hugo Armando Rico T.*


Jaime Moreno*


Eder Cárdenas*


Raúl Castañeda*


Eddy Herrera


D.C. Colombia, E-mail: [email protected]

* Master's student in Hydrosystem’s. Civil Engineering Department. Pontificia Universidad Javeriana,

Bogotá, Colombia.

Efraín Dominguez

PhD. Technical Sciences (Hydrology, Water Resources and Hydrochemistry). Pontificia Universidad

Javeriana. Bogotá D.C., Colombia. [email protected]

Water as a strategic element in the development of a region, is a resource that basically depends on proper

management of watersheds. This is why knowledge and geomorphological analysis of these represents an

important input for the plans of integrated river basin management in a given territory. This paper first

presents the results of the delimitation process, geomorphological characterization of all bounded basins

and a second part of this study the regionalization, through data analysis.

Evaluation of rainfall network using entropy over the urban area of

Rome

Elena Ridolfi, Valeria Montesarchio, Fabio Russo, Francesco Napolitano



Federico Lombardo


In many hydrological applications, rainfall data acquisition takes a great importance. Hydrological

modeling accuracy is strongly dependent on the quality and quantity of the input information. This input is

usually provided by raingauge networks whose density is a key parameter for proper observations of

rainfall fields. The finer the space-time rainfall resolution, the higher the accuracy of the hazard

nowcasting, as flash floods as well as monitoring of sewer systems.

This work focuses on testing the adequacy of a high density raingauge network to cover the urban area of

Rome. First, the number of the raingauges in the network is examined by evaluating the coefficient of non

transferred information and the corresponding isocorrelation contours. Using the tool of informative

entropy it is possible to evaluate if a certain raingauge of the network adds new information in terms of

rainfall record or it has redundant information, so that it is possible to evaluate the optimum number of

raingauges of the network. Then, the adequacy of the length of rainfall record measurements of every

single raingauge is tested by calculating the space-time variance.









61

This method is useful not only in terms of gained accuracy of rainfall information and flood prediction,

but also in terms of money savings, because redundant raingauges could be removed.

Finally, a comparative analysis of the rainfall fields at different time scales is performed on the study area

by grouping rainfall data according to the seasonality.

Extreme rainfall and floods in a Mediterranean area

Fabio Rossi, Carmine De Luca, Pierluigi Furcolo, Paolo Villani and Claudia Vitolo

Department of Civil Engineering, University of Salerno, Fisciano, Italy.

E-mail: [email protected], Tel: +39 089 964075; fax +39 089 964099 1 now at River Team, Halcrow Group Limited, Crawley office,United Kingdom.; Email: [email protected], Tel:

+44 (0)1293 434532

This paper examines the recent advances in the analysis of the extreme rainfall and floods in the

Mediterranean area. In the „90s the National Group for Defence from Hydrogeological Disasters for the

Italian rivers developed a regional methodology that later became the VAPI procedure (Rossi, F. and P.

Villani, 1994). The procedure is based on the TCEV probabilistic model (Rossi et al., 1984). The TCEV

distribution assumes individual rainfall or flood peaks can be expressed as a mixture of two exponential

components. A regionalized TCEV distribution was shown to accurately reproduce the observed

distribution of skewness, a statistic test particularly useful for flood frequency analysis.

Recently, advances in meteorology showed that the variability of extreme rainfall can depend on the

presence of several meteorological processes with different characteristics. The baroclinic cyclogenesis is

the dominant meteorological phenomenon in the Mediterranean. When a baroclinic wave impacts on an

orographic obstacle, like the Apennines, it causes a baroclinic lee cyclone generated at the lee of the

obstacle. The baroclinic structure generates ordinary extremes (more frequent and less severe on average).

Other two mechanisms exhibit a greater variability and tend to generate more severe and less frequent

extremes: the first mechanism consists of isolated convective cells of small spatial scale (tens of

kilometers) mostly associated with warm season thunderstorms; the second mechanism consists of tropical

cyclones, defined as hurricane-like cyclones in the Mediterranean environment, which have a spatial scale

greater (hundreds of kilometers) than the isolated convective cells. This phenomenon strongly depends on

the interaction between the atmosphere and the sea.

In this study a new methodological approach is proposed based on:

a) a procedure for “a priori” identification of meteorological structure of the events. In this

procedure, three homogeneous time series can be identified in each site. Moreover, each of the

series can be considered independent of the others and then studied separately;

b) the application of Power Extreme Value (PEV) distribution, a parametric family of power law

transformations. The PEV distribution is applied to a single series of homogeneous data by

considering a hierarchical approach to estimate the regional parameters.

Finally, an application of the approach on the maximum annual daily rainfall time series in Campania is

provided.




62

Water quality analysis of the coastal regions of Sundarban mangrove

wetland, India using multivariate statistical techniques

Santosh Kumar Sarkar and Bhaskar Deb Bhattacharya

Department of Marine Science, University of Calcutta, 35 Ballygunge Circular Road, Kolkata – 700019,

India; E-mail : [email protected]; [email protected]

Multivariate statistical techniques, including cluster analysis (CA), canonical correspondence analysis

(CCA) and factor analysis (FA) were worked out by applying statistical softwares (MINITAB 16,

STATISTICA 6, XLSTAT) to evaluate spatiotemporal variations and to interpret a large and complex

water quality data set collected form the coastal regions of Sundarban mangrove wetland, India. The

wetland is formed at the estuarine phase of Ganges (Hugli) river which have immense impact on its

hydrological setup due to occurrence of the southwest monsoon (September - October). The data was

transformed to log(n+1) function to allow the high variation of the parameters to exert same influence on

the calculation of similarities. The multivariate statistical analysis has been successfully applied to

interpret the complex data set and to assess the water quality data to get a better understanding of the

ecological status of the estuarine complex. Clustering dendrogram indicated that the seven sampling

stations can be classified into three individual groups based on the similarity of the salinity, dissolved

micronutrients (nitrate, phosphate and silicate) and chlorophyll pigments (chlorophyll a, b and c) at about

88.8% level of similarity. Canonical Correspondence Analysis (CCA) revealed high phosphate load

originating from the upstream discharges and high organic load from the downstream mangrove swamps.

High dissolved oxygen (DO) found to be more associated with high transparency and low turbidity.

Subject to the variation of salinity gradients, the seven stations were again grouped revealing the riverine,

estuarine, marine-estuarine and marine environment located along the stretch of Hugli estuary. Analysis of

variance (ANOVA) gave significant result (0.78; p<0.005) for spatial variations. The sorted rotated factor-

loading pattern of all the seven stations extracts four factors, which account for a total cumulative variance

of only 76.9% of which factor 1 is responsible for 38.9%, which implies positive loadings of salinity,

transparency, pH, chlorophyll a and chemical oxygen demand (COD). Thus factor 1 represents the

spatiotemporal effects of salinity and organic pollution. The second factor which represents 16.40% of the

total variance is due to turbidity, pH, nitrate and chlorophyll c. Nitrate is an important component of

agricultural pesticides applied to the crop fields adjacent to the estuary and normally comes form runoff,

and thus this factor represents nitrogenous nutrient pollution. The third factor represents 12.1% of the total

variance related to biochemical oxygen demand (BOD) and can be termed as indicator of high

microorganism growth, as mangrove waters containing organic matter regulates microorganisms. The

fourth factor represents seasonal effects of water temperature at 9.5% level of variance. The study

revealed the major causes of water quality deterioration which were related to untreated or semitreated

wastes from domestic, agricultural and industrial sources along with discharge of dredged materials, storm

water runoff, aerial fall out, oil spills, boating and other nonpoint sources.




63

Comparison between radar and rain gauges data at different

distances from radar and correlation existing between the rainfall

values in the adjacent pixels Stefano Sebastianelli, Fabio Russo, Francesco Napolitano



Luca Baldini

Istituto di Scienze dell'Atmosfera e del Clima, Consiglio Nazionale delle Ricerche, Italy.

Rainfall intensity data in pixels very far from radar are less correlated than values in pixels near the radar,

because at those distances the width of a range-bin is comparable or bigger than the pixel width, so in a

pixel there are one or just a few rainfall intensity values. Vice versa, near the radar, there are many range-

bins which belong to a single pixel, so there is great correlation between rainfall intensity values for

contiguous range-bins of two adjacent pixels.

Moreover, the signal returned from precipitation at far distance from antenna radar can be partially filled

or be filled with mixed phase or ice particle.

All these phenomena can influence the goodness of rainfall estimates, introducing errors which increase

with distance from radar. The objective of this work is to quantify these errors.

For this aim is possible to compare the rainfall data obtained by rain gauges at different distances from

radar with rainfall radar data at the same distances, verifying the correlation existing between the rainfall

values in the adjacent pixels and how the difference between radar and rain gauges data changes.

The radar data utilized in this work have been obtained from Polar 55C radar located in Rome Tor Vergata

during 2008 and 2009.

Evaluation of empirical relationships for estimation of instantaneous

peak flow Discharge in western catchments of Iran

Saeid Shabanlou

Department of Water Engineering, Kermanshah Branch Islamic Azad University , Iran..


Ahmad Rajabi

Department of Water Engineering, Kermanshah Branch Islamic Azad University , Iran..


For defining the hydrologic criteria in the regions which data and information is not available or

insufficient, the important problem is to develop those relationships and create mathematical models so

that the existing data can be transferred to other sites. One of the most important parameters in design of

hydraulic structures is the instantaneous peak discharge. To record this value, a limnograph instrument is

required and due to its expensive cost, it is not suitable for small catchments.

In this research, 20 empirical formulas that have been developed all over the world to estimate the peak

discharge, are use in 11 hydrologic stations of western watersheds of iran including Do-ab and Pole-chehr

in Gamasiab catchment; Do-abe mereg, Pole-kohne, and Ghorbaghestan in Qarah-sou catchment; Pole-




64

kashkan, Pol-dokhtar, and Afarine in Kashkan catchment; Abshine in Qare-chai catchment; and Holilan in

Saimareh catchment. Concurrent peak discharge data during a 31 years period at about 11 stations were

selected. flood frequency analysis was performed on these peak discharge s by 8 probability distribution

functions and after the best fit test , the peak discharges with return periods of 2, 5, 10, 25, 50, 100, 500

and 1000 year were estimated by Log Pierson type 3 distribution. Homogeneity test was done by

Dalrymple method. As a result, it was found that all the stations are hydrologically homogeneous.

The physiographic characteristics and climatologically data required for testing and developing the

empirical formula s such as drainage area ,the main channel length , slope of the main channel between

10% and 85% of its length, catchment mean elevation, stream frequency, drainage density, time of

concentration, mean annual precipitation, and maximum 24-hour and 30 minute rainfall rates with

associated return periods were determined from 1:250000 and 1:50000 topographic maps with ArcGis or

by using the SAS statistical software, and the empirical formulas were calibrated for the study region.

Then, by comparing the coefficient of determination, standard error, mean error and residuals graph for

each return period, the best formula was recommended. For 2, 5, 10, 25 and 50– year return periods,

equation 15(Illinois State) and for 100–year return period, equation 13(Missouri State) was recommended.

For 500-year return period, formula no. 5(developed in Canada) and for 1000-year return period, equation

18(developed in USA) were advised.

The comparison between estimated and measured peak discharges showed that model no. 2 has the

maximum error (114.75%) and model no. 12 has the minimum error (22.58%). For the recommended

models, the mean error for 2 to 1000-year return periods was between 21.67% to 34.67%.

Analysis of precipitation in Emilia-Romagna (Italy) and impacts of

climate change scenarios

Fausto Tomei(1)

, Gabriele Antolini(1)

, Rodica Tomozeiu(1)

, Valentina Pavan(1)

, Giulia Villani(2)

and

Vittorio Marletto(1)

(1) Arpa Servizio Idro-Meteo-Clima, Emilia-Romagna, Bologna, Italy ([email protected])

(2) DEIAGRA, Department of Agricultural Economics and Agricultural Engineering University of

Bologna, Bologna, Italy

A hydroclimatic regional atlas for the period 1961-2008 was produced in the framework of the Eraclito

project, funded by the Italian region Emilia-Romagna (22,000 km2). The data used for the atlas come from

a unified regional database, completed in 2009 after collecting data from different sources. Daily data

were spatially interpolated on a triangular irregular grid (TIN) with more than 4000 cells whose dimension

is inversely proportional to orographic variability. Climatological computations refer to two distinct

periods: 1961-1990, as climatic reference period according to WMO conventions and the following

available period, 1991-2008. Variation maps of annual and seasonal amounts between the two periods

were produced. A generalized decrease of precipitations is observed in winter, spring and summer,

whereas a significant increase is observed in autumn. These trends are evident also for the number of rainy

days. A general temperature increase is observed all over the region.

Within the national Agroscenari project, we generated climate change scenarios of seasonal temperature

and precipitation over the period 2021-2050 against 1961-1990 for several Italian areas. A statistical

downscaling technique, applied to the ENSEMBLES experiments (A1B scenario), is used to reach this

objective. The projections under scenario conditions for the areas of Emilia-Romagna region show that

significant increases in temperature could be expected to occur in all seasons, whereas the pattern of

precipitation is more complex with a slightly decrease especially in winter and summer and an increase in

autumn. The projections seem to be consistent with the signal identified in the 1991-2008 period. The


65

projections for the above areas were then used as input in a weather generator in order to produce

synthetic series of daily data. These series were used to feed a water balance and crop growth model

(Criteria) to evaluate the impact of climate change in water table depth and in irrigation crop water needs.

We chose kiwifruit as reference crop, which is characterised by high water need and is widespread in the

hills of the Romagna subregion.

Copula-based scaling of rainfall fields

M.J. van den Berg1, S. Vandenberghe

1, B. De Baets

2, N.E.C. Verhoest

1

1Laboratory of Hydrology and Water Management, Ghent University,Coupure links 653, B-9000 Ghent,

Belgium, Email: [email protected]

2Department of Applied Mathematics, Biometrics and Process Control, Ghent University,Coupure links

653, B-9000 Ghent, Belgium

The estimation of the distribution of precipitation fields, based on a coarse scale field, has progressed

much with the advent of fractal methods. In a downscaling framework these methods assume that the

distribution of the field as a whole remains valid at a subpixel level, regardless of the value at coarse scale.

However, preliminary results readily show that this is not the case. Instead, a dependency on the rainfall

intensity at the coarse scale is observed, giving rise to a systematic deviation from the field wide

distribution at fine scale. Through the use of a copula the bivariate joint distribution can be found between

the coarse scale intensity and the subpixel distribution at fine scale. Using this, the distribution of

subsamples contained within a coarse scale pixel can be estimated. Moreover, it is shown that this

approach outperforms the classical fractal scaling methods.

A comparison of fitting methods and tests for several distributions

on hydrological data

R.R.P. van Nooijen

Faculty of Civil Engineering and Geo Sciences, Delft University of Technology, Stevinweg 1, 2628 CN,

Delft, The Netherlands.


T. Gubareva

Laboratory of Hydrology and Climatology, Pacific Institute of Geography FEB RAS, Radio, 7, 690041

Vladivostok, Russia

A.G. Kolechkina

Aronwis, Leeuwenberg 16, 2635 GD Den Hoorn (ZH), The Netherlands.

In hydrology extrapolation of times series data on extremes is often accomplished by fitting a probability

distributions to the series. There are many different probability distributions in use for this purpose,

several different fitting methods and a number of different ways to evaluate the usefulness of the resulting

fit. In this paper we study the interaction between fitting method, the evaluation method and the purpose

of extrapolating to as yet unobserved extremes for a number of real data sets. While we tried to use off the

shelf routines and software wherever possible, but we still needed to write some custom routines to create

a consistent whole. The software is based on an open source package. The custom routines will be made

available upon request.




66

A comparison of fitting methods and tests for several distributions

on computer generated samples

R.R.P. van Nooijen

Faculty of Civil Engineering and Geo Sciences, Delft University of Technology, Stevinweg 1, 2628 CN,

Delft, The Netherlands.


A.G. Kolechkina

Aronwis, Leeuwenberg 16, 2635 GD Den Hoorn (ZH), The Netherlands.

In hydrology probability distributions are fitted to data in different contexts and with different techniques.

The context may be the study of annual maxima or rainfall intensity-duration-frequency curves or even

synthetic unit hydrographs. There are many different ways to evaluate the usefulness of the resulting fit. In

this paper we conduct a series of Monte Carlo experiments to study the interaction between fitting method,

evaluation method and purpose. We use samples both with and without noise and/or outliers. While we

tried to use off the shelf routines and software wherever possible, but we still needed to write some custom

routines to create a consistent whole. The software is based on an open source package. The custom

routines will be made available upon request.

Regional flow duration curves for ungauged sites in Sicily

Francesco Viola

Dipartimento d’Ingegneria Idraulica ed Applicazioni Ambientali, Università degli Studi di Palermo,

Italia.


Leonardo Valerio Noto


Italia.


Marcella Cannarozzo


Italia.


Flow duration curves are simple and powerful tools to deal with many hydrological and environmental

problems related to water quality assessment, water-use assessment and water allocation. Unfortunately

the scarcity of streamflow data enables the use of these instruments only on few gauged basins.

A regional model is developed here for estimating flow duration curves at ungauged and unregulated

basins in Sicily. Due to the complex ephemeral behavior of the examined basins, in this study the lower

and the upper half of daily flow duration curves are analyzed separately using two probability density

function, whose parameters were fitted on 24 basins.

Regional regression equations are developed to describe the model parameters in terms of morphological

basin characteristics. The study adopts a jack-knife cross-validation procedure to evaluate the uncertainty

of regional flow duration curves.






67

Modelling of groundwater level fluctuations in agricultural

monitoring stations with conceptual two dimensional mathematical

model METUL

Valdis Vircavs and Arturs Veinbergs

Department of Environmental Engineering and Water Management, Faculty of Rural Engineering, Latvia

University of Agriculture, Jelgava, Latvia ([email protected]),([email protected])

Abstract

Shallow groundwater is a major natural resource and an integral part of the hydrological cycle. Surface

water and shallow groundwater interactions could be studied with regard of modelling. The objective of

the presented study is to investigate the overall status of the fluctuations of groundwater level and to

evaluate the possibility to use modelling for simulation of groundwater level. Study is to establish required

data sets for groundwater level modelling using METUL (meteorological data and water level).

Water level of groundwater in unconfined aquifer is determined by a variety of factors, where the

petrographic properties of the vadose and groundwater saturated zone, and the regional hydrological,

hydrodynamic conditions and soil water balance are the major natural factors. In agricultural fields

additional to these natural factors, groundwater fluctuations are also highly influenced by anthropogenic

impacts influences, in particular land use, subsurface drainage.

Monitoring of groundwater level in Latvia at three agricultural run-off monitoring stations in 10

observation wells started in year 2006. Conceptual groundwater level model METUL by Krams and

Ziverts (1993) was calibrated using measurements of the daily groundwater level fluctuations for the

period 2006–2009. This model is site oriented two-dimensional mathematical model based on daily

weather data (temperature, precipitation, and relative humidity).

Study on scaling regimes of rainfall time-series from a dense

raingage network

Elena Volpi, Federico Lombardo



Francesco Napolitano


The need of understanding and modeling the high space-time variability of rainfall fields produced a large

amount of literature in the last thirty years. A parameter parsimonious approach to this problem is based

on the empirical detection of some regularities in hydrological observations, such as the scale-invariance

properties of rainfall (e.g. Lovejoy and Schertzer, 1985). Models following this approach are based upon

the assumption of a power law dependence of all statistical moments on the scale of aggregation. That

means scaling properties can provide simple relationships to link the statistical distribution of the rainfall

process at different spatial and temporal scales, in the ranges of which the power-law assumption can be

verified (Marani, 2005).

This work focuses on the analysis of the scaling properties of rainfall time series from a high density rain

gauge network covering the urban area of Rome. The network consists of 24 sites, and the gauge record at

each site has 10-minute time resolution and about 16-year length (1992-2007). In the hypotheses of

stationary monthly rainfall series and spatial homogeneity of the rainfall fields over the study area, the



68

scale-invariance properties in the time domain of the studied rainfall are investigated within some specific

scale intervals (temporal scaling regimes) by using different methods: q-moments, PDMS, autocovariance

structure. Furthermore, a multiplicative random cascade model (Rupp et al., 2009) is calibrated on each

scaling regime and then the statistical properties of the simulated time series are validated with the

observations.

Continuous rainfall simulation: estimation at ungauged locations

Seth Westra , Rajeshwar Mehrotra, Ashish Sharma



Email: [email protected]; [email protected]; [email protected], Tel: +61 2 9385 5140

Sri Srikanthan

Water Division, Australian Bureau of Meteorology

GPO Box 1289, Melbourne, Victoria, Australia

Email: [email protected] Tel: +61 3 9669 4513

Continuous simulation of extended rainfall sequences is becoming an increasingly important tool in

rainfall-runoff modelling and design flood estimation. Recently, a nonparametric approach has been

developed in which continuous (sub-daily) rainfall fragments are conditioned to daily rainfall on the

current day and the rainfall state (wet or dry) on adjacent days using a k-nearest neighbour conditional

resampling algorithm, with this method performing well in simulating both Intensity-Frequency-Duration

characteristics and the joint distribution between extreme storm events and the antecedent moisture

conditions leading up to the storm. A limitation of this approach, however, is that extended pluviograph

records at the location of interest are required, limiting applicability to only a small number of locations

across Australia where such records are available.

In this paper we propose an extension in which sub-daily fragments from nearby pluviograph records can

be substituted for at-site data, widening the applicability of the method to all locations for which extended

daily data is available or for which such data can be stochastically generated. The method was developed

by using a two sample, two dimensional Kolmogorov-Smirnov (2S2D-KS) test to determine whether the

daily/sub-daily rainfall scaling at any two stations were statistically similar. This was repeated by

considering the similarity between all pairs of 213 pluviograph stations distributed throughout Australia,

totalling more than 22000 station pairs. A logistic regression model was then formulated to identify

physiographic factors which would maximise the probability that the daily/sub-daily scaling between two

stations will be statistically similar. The outcome of this analysis was that latitude, longitude, elevation

and distance to coast were all important predictors, with a multivariate statistical model showing the

relative contribution of each. This forms the basis for the selection of „nearby‟ stations to substitute for at-

site stations when applying the continuous simulation approach described above to ungauged locations.

Testing suggests that the non-parametric continuous simulation approach performs well in representing

the historical rainfall pattern at ungauged locations, and as such comprises a potentially viable approach in

the generation of extended synthetic sequences for use in rainfall-runoff modelling and design flood

estimation.






69

Interpreting variability in global SST data using independent

component analysis and principal component analysis

Seth Westra and Ashish Sharma



Email: [email protected], [email protected],

Tel: +61 2 9385 5140, +61 2 9385 5679

Component extraction techniques are used widely in the analysis and interpretation of high-dimensional

climate datasets such as global sea surface temperatures (SSTs). Principal component analysis (PCA), a

frequently used component extraction technique, provides an orthogonal representation of the multivariate

dataset and maximizes the variance explained by successive components. A disadvantage of PCA,

however, is that the interpretability of the second and higher components may be limited. For this reason,

a Varimax rotation is often applied to the PCA solution to enhance the interpretability of the components

by maximizing a simple structure. An alternative rotational approach is known as independent component

analysis (ICA), which finds a set of underlying „source signals‟ which drive the multivariate „mixed‟

dataset.

Here we compare the capacity of PCA, the Varimax rotation and ICA in explaining climate variability

present in globally distributed SST anomaly (SSTA) data. We find that phenomena which are global in

extent, such as the global warming trend and the El Ni˜no-Southern Oscillation (ENSO), are well

represented using PCA. In contrast, the Varimax rotation provides distinct advantages in interpreting more

localized phenomena such as variability in the tropical Atlantic. Finally, our analysis suggests that the

interpretability of independent components (ICs) appears to be low. This does not diminish the statistical

advantages of deriving components that are mutually independent, with potential applications ranging

from synthetically generating multivariate datasets, developing statistical forecasts, and reconstructing

spatial datasets from patchy observations at multiple point locations.




70

NOTES


71

NOTES


72

NOTES


73

NOTES


74

NOTES


75

NOTES


76

NOTES


77

NOTES

Book of abstracts - risorseidriche.dica.unict.it · Carlo De Michele Polytechnic of Milano ... Giuseppe Rossi University of Catania (Italy) ... Hotel Villa Diodoro, Via Bagnoli Croci,

Documents