Distributed ecohydrological modelling to evaluate irrigation system performance in Sirsa district, India II: Impact of viable water management scenarios

Journal of Hydrology (2006) 329, 714–723

ava i lab le at www.sc iencedi rec t . com

journal homepage: www.elsevier .com/ locate / jhydro l

Distributed ecohydrological modelling toevaluate irrigation system performance inSirsa district, India II: Impact of viable watermanagement scenarios

R. Singh a, R.K. Jhorar b, J.C. van Dam a,*, R.A. Feddes a

a Department of Environmental Sciences, Wageningen University and Research Centre, P.O. Box 47, 6700 AA,Wageningen, The Netherlandsb Department of Soil and Water Engineering, CCS Haryana Agricultural University, Hisar, Haryana, India

Received 30 June 2005; received in revised form 16 March 2006; accepted 23 March 2006

Summary This study focuses on the identification of appropriate strategies to improve watermanagement and productivity in an irrigated area of 4270 km2 in India (Sirsa district). The fieldscale ecohydrological model SWAP in combination with field experiments, remote sensing andGIS has been applied in a distributed manner generating the required hydrological and biophys-ical variables to evaluate alternative water management scenarios at different spatial and tem-poral scales.

Simulation results for the period 1991–2001 show that the water and salt limited crop pro-duction is 1.2–2.0 times higher than the actual recorded crop production. Improved crop hus-bandry in terms of improved crop varieties, timely sowing, better nutrient supply and moreeffective weed, pest and disease control, will increase crop yields and water productivity inSirsa district. The scenario results further showed that reduction of seepage losses to 25–30% of the total canal inflow and reallocation of 15% canal water inflow from the northern tothe central canal commands will improve significantly the long term water productivity, haltthe rising and declining groundwater levels, and decrease the salinization in Sirsa district.

�c 2006 Elsevier B.V. All rights reserved.

KEYWORDSCanal water distribution;Water productivity;Net groundwater recharge;Salinization;Regional scale;SWAP;Bhakra irrigation system

0d

4

cf

022-1694/$ - see front matter �c 2006 Elsevier B.V. All rights reserveoi:10.1016/j.jhydrol.2006.03.016

* Corresponding author. Tel.: +31 317 484825; fax: +31 31784885.E-mail addresses: [email protected] (R. Singh), jhorar@yahoo.

om (R.K. Jhorar), [email protected] (J.C. van Dam), [email protected] (R.A. Feddes).

Introduction

Crop production in arid and semi-arid regions like HaryanaState (India) is very limited without supplemental irrigation.The surface water supply diverted from high precipitation

d.

Impact of viable water management scenarios 715

regions is also not sufficient to meet the crop water require-ments, and is often supplemented by groundwater. Duringpast decades, main factors that increased the crop yieldsin Haryana were improved irrigation, higher yielding varie-ties, and increased use of fertilizer and herbicides. For in-stance, the average wheat grain yield in Sirsa district(Haryana) has doubled in the period 1975–2000. Water pro-ductivity WPET (kg grain or seed per m3 of evapotranspira-tion) amounts 1.39 for wheat, 0.86 for rice and 0.22 forcotton. This corresponds to mean water productivity valuesfor Northwest India (Singh, 2005). The major constraint forfurther increase of food production in Haryana State is thelimited amount of surface water available for irrigation(Aggarwal et al., 2001). The scope of meeting the growingdemand of water for food production through more inten-sive use of water resources is diminishing, as its exploitationin most of the river basins is already on its maximum. Fur-ther, the urban, industrial and environmental users are call-ing for their legitimate share of water, which will certainlycut down the water availability for irrigated agriculture.Facing the less availability of water, the performance ofirrigated agriculture even deteriorates due to rising ground-water levels, causing waterlogging and salinization (Aggar-wal and Roest, 1996). In June 2000, about 5% of Sirsadistrict had a groundwater level within 3 m of the soil sur-face (Groundwater Cell, 2002). In contrast, the centralparts of Sirsa district groundwater levels showed a declineof 4–7 m in the period 1990–2000. Therefore ways mustbe found to develop a more productive and sustainablewater management in Sirsa district.

Problems such as waterlogging and soil salinization,though severe, are reversible (Jensen, 1996). As the prevail-ing ecohydrological problems are to a large extent causedby inefficient water management practices, the sustainabil-ity of irrigated agriculture might be improved when moreefficient water management strategies are recognized andimplemented. The impact of water management on regionalwater and salt balances and on water productivity largelydepends on the actual ecohydrological conditions of anarea. Also, problems due to certain water managementare often not recognized until they are well advanced.Therefore proper evaluation of alternative water manage-ment scenarios in advance is important.

In this study, we focus on the identification of appropri-ate water management options to improve water manage-ment and its productivity in Sirsa district, covering4270 km2 in the western corner of Haryana State (India).Similar to other arid and semi arid regions, the water man-agement is of very complex nature in the study area. Irri-gated agriculture is by far the largest consumer of water.The limited canal water supply is supplemented by ground-water pumping to irrigate two crops cultivated each year:the rabi (winter) crop (mainly Wheat) from November toApril, and the kharif (summer) crop (mainly Cotton or Rice)from May to October. Surface flooding is the most commonmethod of irrigation application. The key factors related towater management are scare and erratic rainfall, high evap-orative demands, marginal to poor groundwater quality, ris-ing and falling groundwater levels, sandy soils with lowwater holding capacity. Beside these factors, other factorslike delay and failure of monsoon, fluctuations in canalwater supply, conveyance losses from irrigation system

are further affecting the crop production and water produc-tivity in study area. First, feasible water management op-tions in Sirsa district are proposed and second, the impactof the proposed options on the regional water productivityand groundwater behaviour are quantified. The field scaleecohydrological model soil–water–atmosphere–plant(SWAP) (van Dam, 2000; Kroes and van Dam, 2003), coupledwith field experiments, remote sensing and GIS, is extendedin a distributed manner to produce the required hydrologi-cal and biophysical variables to evaluate alternative watermanagement scenarios at regional scale. Effects of spatialheterogeneity on the regional water and salt balances,and on the water productivity area analyzed running SWAPfor all combinations of weather–crop–soil–irrigation inthe study area (Singh et al., 2006). Although this distributedmodelling approach is based on certain assumptions andsimplification of real situations, alternative water manage-ment strategies can be evaluated by comparing and under-standing cause–effect relationships between predictedtheoretical water and salt balance and crop yields. To iden-tify appropriate recommendations for a more productiveand sustainable water management in Sirsa district, the sim-ulated results of the current and alternative water manage-ment scenarios are analyzed in terms of water productivity,net groundwater recharge and salt build-up at different spa-tial and temporal scales.

Current and alternative water management

The canal water distribution in Sirsa district (Fig. 1), knownas ‘warabandi’, follows the principle of equal water distri-bution in proportion to amount of land cultivated, withoutconsidering the actual crop, soil or drainage condition.The irrigation charges to the farmers are fixed and basedon crop type and actual area irrigated. This implies thatfarmers have to pay less if they irrigate less area with a cer-tain amount of canal water. Therefore the current systemmay lead to over-irrigation (Aggarwal and Roest, 1996). Thisover-irrigation might be contributing to the rising ground-water levels in saline groundwater areas, where reuse ofgroundwater is very limited. Based on a regional water man-agement analysis using simulation of water management inarid regions (Boels et al., 1996) and standard groundwatermodel package (Boonstra et al., 1996), Aggarwal and Roest(1996) recommended therefore a ‘water pricing’ and ‘de-mand driven’ system instead of the current warabandisystem.

However, water pricing and demand driven are probablynot a feasible solution. Implementation of the demand dri-ven system would require a complete change of the existinginfrastructure. The capacity to manage and maintain a de-mand driven system is not available in Sirsa district. Nava-lawala (1999) suggested that the prevailing canal watercharges are too low to have any influence on farmer’s deci-sions on canal water use. Hellegers and Perry (2003) alsoinvestigated as a feasible management tool to improvewater management in Sirsa district. Since the returns onthe water are on an average about 100 times the price ofwater delivery, they concluded that a social–political unac-ceptable increase in the water price is required to influencethe farmer’s decisions for irrigation water use. Berkoff

Figure 1 Location of Sirsa district showing main canal commands.

716 R. Singh et al.

(1990) advocated the rigid ‘warabandi’ system as a self-enforcing system. He concluded that a relative short, reli-able supply is very important for farmers to plan their croprotation, and to optimize crop return. Knowing that theycannot fully meet the crop water demands, farmers typi-cally plant such a mix of irrigated and non-irrigated cropsthat overall yield risks are minimal. Moreover, the farmersare able to check unauthorized use of water during theirturns.

In an earlier paper we analyzed water and salt balancesand water productivity in Sirsa district for the agriculturalyear 2001–2002 (Singh et al., 2006). In this study, we pro-pose feasible water management options for Sirsa districtwithin the existing irrigation infrastructure and maintainingthe warabandi system. The proposed water managementoptions are focused to improve the long term water produc-tivity, to halt the rising and declining groundwater levels,and to decrease salinization. First we will discuss the pro-posed water management options.

Improved crop management

Effects of improved crop management are indicated by theratio Yg/I, where Yg is the measured grain (or seed) yieldand I is the amount of irrigation applied. In Sirsa district,the ratio Yg/I was calculated using the measured values ofYg and I at 24 farmer fields and experimental fields duringthe agricultural year 2001–2002 (Malik et al., 2003). Atfarmer fields, the average value of Yg/I, expressed inkg m�3, amounted 0.94 for wheat and 0.35 for cotton. Atexperimental fields, the same ratio for the same cropswas much higher: 1.65 for wheat and 0.68 for cotton. Thehigh ratio Yg/I at experimental fields is attributed to im-proved crop varieties, efficient irrigation, optimal nutrient

supply, and better pest and disease control. If this improvedcrop management at experimental fields is adopted atfarmer fields, then in case of wheat 100 (1 � 0.94/1.65) =43%, and in case of cotton 100 (1 � 0.35/0.68) = 48% lesswater is required for the same crop yield. Also the positivecorrelation between simulated water and salt limited Ygand WPET (Singh et al., 2006) indicate that efforts to in-crease crop yields result in improved water productivity.Therefore improved crop management is one of theproposed options.

Reduction in seepage losses

Jacobs and de Jong (1997) made a detailed water balancestudy of two watercourses in Hisar district, next to Sirsadistrict. They concluded that seepage losses coupled withgroundwater flow and absence of natural drainage aremain factors for waterlogging and salinization in Hisar dis-trict. The seepage losses in Sirsa district are also high: 30–40% of the total canal inflow (Boels et al., 1996; Singhet al., 2006). Taking into account both the percola-tion Qbot from field irrigations and seepage losses QSL fromthe conveyance system, the water productivity reducedfrom WPET to WPETQ by 25–30% for the main crops (wheat,cotton, mustard and rice) in Sirsa district during the agri-cultural year 2001–2002 (Singh et al., 2006). Seepagelosses recharge to the groundwater, and might be reusedin good quality groundwater areas. However, reuse ofseepage water is limited in poor quality groundwaterareas, like the northern parts of Sirsa district, wheregroundwater levels are rising. The reduction of seepagelosses, especially in saline groundwater areas, thereforewill improve the water productivity and performance ofthe irrigation system.

Impact of viable water management scenarios 717

Canal water reallocation

In the period 1974–2000 the measured groundwater levelsin Sirsa district show a mean annual rise of 0.31 m y�1

(Groundwater Cell, 2002). Interesting is that the measuredgroundwater rise over the last 10 years (1990–2000) de-clined to 0.09 m y�1. The smaller rise of groundwater levelsis attributed mainly to increased groundwater pumping inorder to irrigate more water demanding crops such aswheat, rice and cotton, compared to the former crops gram(chickpea) and bajra (pearl millet). Since 1976 the totalnumber of tubewells in Sirsa district has increased by a fac-tor 4. However, groundwater pumping in the northern partsremains low due to the poor quality groundwater and theregroundwater levels keep rising. On the other side, over-exploitation of good quality groundwater to irrigate highwater consuming wheat and rice crops is resulting in thedeclining groundwater levels in central parts (Singh et al.,2006). Therefore, canal water reallocation from the risingto the declining groundwater level areas is explored to dis-tribute net groundwater recharge more evenly.

Implementation of alternative water managementscenarios

In this study, we followed a distributed modelling approachby running a field scale model for all combinations of weath-er–crop–soil–irrigation (simulation units). The ecohydro-logical SWAP model, including detailed crop growthprocesses, is extended in a distributed manner for simula-tions of hydrological and biophysical variables at all simula-tion units. Details of this approach developed for Sirsadistrict are described in Singh et al. (2006). Such a distrib-uted modelling may serve two purposes: first, to understandthe current situation and processes, and second, to evaluatealternative water management scenarios.

In this study, the five scenarios of Table 1 were imple-mented, and their consequences on the regional waterand salt balances, and on the water productivity in Sirsa dis-trict were evaluated over a period of 10 years (November 1,1991–October 31, 2001). The spatial model input data suchas weather, canal water supply, and number of tubewellsvaried with time were derived from the collected informa-tion (Malik et al., 2003). The variation in cropping pattern,cultivation area and groundwater quality over the 10 year

Table 1 Water management scenarios simulated for Sirsa distric

Scenario Description

1 Reference situation2 Increased (10–20%) crop yields

3 Reduced seepage losses at 25–30%4 Canal water reallocation

5 Combination of Scenarios 2, 3 and 4

BMB is the Bhakra Main Branch command in the northern part, SUK and Gof Sirsa district, respectively (Fig. 1).

period could not be included due to too limited data avail-able for the entire district. Therefore, the derived simula-tion units for the agricultural year 2001–2002 (Singhet al., 2006) were used for the analysis of the entire periodfrom 1991 to 2001.

Scenario 1, ‘business as usual’ or ‘reference situation’,shows the variability between meteorological years and re-lated canal water supply in Sirsa district. This scenarioquantifies the expected rise or decline in groundwater levelsand crop production with current water management, andserves as reference for alternative management.

Scenario 2, ‘increased (10–20%) crop yields’, was formu-lated to quantify the impact of improved crop husbandry.Improved crop varieties, optimal nutrient supply, and betterpest and/or disease controls are expected to improve po-tential crop yields in coming decades. Scenario 2, therefore,was implemented by increasing the crop input parametersmaximum CO2 assimilation rate and light use efficiency. Inorder to achieve 10–20% increased crop yields with thesame water supply, these crop input parameters (Singhet al., 2006) were increased by 5% for mustard, 10% forwheat and rice, and 15% for cotton (Fig. 2).

Scenario 3, ‘reduced seepage losses at 25–30%’, targetsthe salinization in the northern parts of Sirsa district. Theseepage losses from the conveyance system were estimatedat 34–43% of the net canal inflow in different canal com-mands during the agricultural year 2001–2002 (Singhet al., 2006). In scenario 3, the input parameters were ad-justed to reduce the seepage losses at 25–30%. Proper lin-ing and improved maintenance of the irrigation systemshould be able to achieve this reduction of seepage losses.

Scenario 4, ‘canal water reallocation (15%) from the BMBto the GHG and SUK’, was formulated to target both the ris-ing groundwater levels in the northern parts, and the declin-ing groundwater levels in the central parts. In this scenario,15% of the measured canal water inflow in the northern Bha-kra Main Branch (BMB) command was reallocated to the cen-tral Ghagger (GHG) and Sukchain (SUK) commands. Thisextra canal water inflow received by GHG and SUK was di-vided in proportion to their Cultivable Command Area(CCA), the agricultural area attached with canal waterrights.

Scenario 5, ‘combination of Scenarios 2, 3 and 4’, evalu-ates the impact of integrated efforts on the regional waterproductivity, the rising and declining groundwater levelsand the salinization.

t

Required practical action

Business as usualImproved crop varieties, optimal nutrient supply, effectivepest and disease controlLining and maintenance of water distribution systemDiversion of a part (15%) of canal water supply in northernparts facing rising groundwater levels to the central partsfacing declining groundwater levelsDifferent actions as described for scenarios 2, 3 and 4

HG are the Sukchain and the Ghagger command in the central part

0

10

20

30

40

50

5 10 15

% increase in A (kg ha-1 hr -1) and

ε (kg ha-1 hr -1/ J m 2 s -1)

% ni esaercni Y

gah not(

1-)

WheatMustardCottonRice

Figure 2 Sensitivity of the simulated grain (or seed) yields Yg(ton ha�1) to the crop parameters maximum CO2 assimilationrate A and light use efficiency e. Remaining model inputparameters, including data as weather, irrigation, and initialand boundary conditions, were the same as for the ‘referencesituation’.

718 R. Singh et al.

Results and discussion

Business as usual

The average grain (or seed) yields Yg (ton ha�1) in Sirsa dis-trict over the years 1991–2001 amounted 4.1 for wheat, 1.2for cotton (seed), 1.2 for mustard (oilseed) and 3.2 for rice(Department of Agriculture, Sirsa). In this study, SWAP sim-ulated the water and salt limited Yg taking into account thewater and salt stress, but not the nutritional, pest and dis-ease stress. In case of ‘business as usual’ (Scenario 1), thewater and salt limited Yg during the same period (1991–2001) were simulated 1.2–2.0 times higher than therecorded Yg, especially for rice and cotton (Fig. 3). This

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

Wheat Rice

ah not( dleiy )dees ro( niarg porC

8.00

9.00

1-)

Figure 3 Average (±standard deviation) grain (or seed) yields Yg (tthe ‘business as usual’ (Scenario 1) in Sirsa district over the periodthe distributed SWAP simulations, while actual Yg are based on the

indicates substantial nutritional, pest or disease stresseson crop production in Sirsa district.

The mean annual water and salt budget for the entireSirsa district and its four main commands (BMB, SUK, GHGand FB) over the period of 10 years (1991–2001) is pre-sented in Table 2. The rainfall P varied from 149 to388 mm y�1 with a mean value of 256 mm y�1. The canalwater inflow Qcw amounted 1889 million m3 y�1 with a stan-dard deviation of 213 million m3 y�1. About 41% of this an-nual Qcw was lost as seepage QSL from the conveyancesystem, which contributes to the groundwater recharge.The canal water irrigation Icw provided 45% of the total irri-gation I for crop production, and the rest came from ground-water pumping Igw. The total water inflow of 839 mm y�1

(P + I) exceeds the evapotranspiration ET of 722 mm y�1.The percolation Qbot was calculated at 109 mm y�1, whichcorresponds to 13% of the total water inflow.

The net groundwater recharge varies over time andspace, and depends on rainfall, canal water inflow, ground-water pumping, crop water use or ET, percolation from thefield irrigations, seepage losses from the conveyance systemand rivers, and regional lateral groundwater inflow.Neglecting regional lateral groundwater flow, we estimatedR as:

R ¼ P þ Q cw � ET� DW ð1Þ

where R is the net groundwater recharge (LT�1), ET is theactual evapotranspiration (LT�1), P is the rainfall (LT�1),Qcw is the total canal water inflow, and DW is the changein the soil water storage (LT�1). For entire Sirsa districtthe analysis resulted in average R = �27 mm y�1. The mea-sured mean groundwater level in Sirsa district did rise from10.7 m below surface in 1990 to 9.8 m in 2000. Using aspecific yield of 12% for the aquifer in Sirsa district (Boon-stra et al., 1996), this measured groundwater level rise of90 mm y�1 corresponds to average R = +11 mm y�1. Thedifference between the simulated (�27 mm y�1) and mea-sured (+11 mm y�1) average net groundwater recharge isattributed to a number of factors which were not includedin the simulation: lateral groundwater flow, seepage fromthe Ghagger river, and change in cropping pattern and

Mustard Cotton

Actual (recorded)

Water and salt limited

on ha�1) of the main crops (wheat, rice, mustard and cotton) forof 10 years (1991–2001). Water and salt limited Yg are based onrecords of the Department of Agriculture in Sirsa.

Table 2 Annual water and salt budget (‘Business as usual’ or Scenario 1) for the entire Sirsa district and its four maincommands: BMB, SUK, GHG and FB (see Fig. 1 for different canal commands)

Spatial scale/componenta

Sirsa district Sirsa district BMB SUK GHG FB

Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD

Water budget(106 m3 y�1)

Water balance (mm y�1)

P 1083 369 256 87 248 111 292 98 113 33 324 104Qcw 1889 213 446 50 549 84 190 95 382 111 328 43I 2471 123 583 29 542 39 744 53 910 24 486 23Icw 1123 130 265 31 318 51 115 58 244 72 204 51Igw 1348 117 318 28 224 27 629 50 666 75 283 35T 2205 162 520 38 505 40 668 54 635 44 468 41ET 3061 152 722 36 694 40 909 49 879 29 669 46Qbot 463 153 109 36 89 41 123 33 143 20 132 57DW 27 151 6 36 6 43 4 35 2 19 9 41QSL (% of Qcw) 41 41 42 39 36 38R �117 292 �27 69 97 82 �432 85 �385 108 �26 82

Salt budget(106 ton y�1)

Salt balance (mg cm�2 y�1)

ICi 3.1 0.2 73 6 64 6 113 8 120 12 59 7QbotCbot 2.3 0.9 53 22 39 20 67 30 104 29 56 29DC 0.8 1.1 20 25 25 23 46 32 16 29 4 32

Mean values (±standard deviation) are based on the distributed SWAP simulations for the unsaturated zone (0–300 cm) over the period of10 years (1991–2001), and apply to the entire area (cropped as well as bare soil). The total area of Sirsa district is about 4270 km2, out ofwhich 69% is under cultivation.a P is the rainfall, Qcw is the canal water inflow, I is the irrigation, Icw is the canal irrigation, Igw is the groundwater irrigation, T is the

actual transpiration, ET is the actual evapotranspiration, Qbot is the percolation (negative upward), DW is the change in soil water storage,R is the net groundwater recharge, QSL is the seepage loss (% of Qcw) from the conveyance system, C is the salt concentration, and DC isthe change in salt storage.

Impact of viable water management scenarios 719

cultivation area between 1991 and 2000. Moreover, duringthis study the SWAP simulated crop growth was affectedonly by water and salt stress, assuming absence of any nutri-tional or pest/disease stresses. This means that the simu-lated ET is higher, thereby resulting in a lower estimatedR. For entire Sirsa district the small measured and simulatedvalue of R indicate that the water balance for entire Sirsadistrict is nearly closed. However, between the main com-mands large differences of R and salt build-up DC are calcu-lated (Table 2), which suggests that interventions arerequired for a more sustainable water management in Sirsadistrict.

Regional water productivity

The impact of the five considered scenarios on the waterproductivity of the main crops is depicted in Fig. 4. Waterproductivity WP was computed as kg grain (or seed) yieldYg/m

3 water used for transpiration T (=WPT), evapotranspi-ration ET (=WPET), or ET plus percolation Qbot from field irri-gations and seepage losses QSL from the conveyance system(=WPETQ) (Molden, 1997; Molden et al., 2003).

The average simulated water and salt limited WPT, WPETand WPETQ values during the period from 1991 to 2001 underScenario 1 ‘business as usual’ were higher than those ob-tained during the agricultural year 2001–2002 (Singhet al., 2006). This is caused by the low rainfall amounts in

2002. In addition, high air temperatures and evaporative de-mands in the year 2002 resulted into low WPT, WPET andWPETQ values, especially for summer crops (rice/cotton).

In Scenario 2, ‘increased (10–20%) crop yields’, the in-creased crop input parameters maximum CO2 assimilationrate and light use efficiency resulted in an increased cropyield of 12% for cotton, 15% for wheat, 16% for mustardand 17% for rice. The enhanced crop growth implies moreleaf area development and more transpiration. At the simu-lated 12–17% increased Yg, the transpiration T increased by2–8% for different crops. There was hardly any increase inevapotranspiration ET, especially for summer (cotton/rice)crops. Even for winter (wheat/mustard) crops, the ET in-creased by 2–3% only. Soil evaporation E decreased whileT increased at enhanced leaf area development. The calcu-lated percolation Qbot from field irrigations decreased:about 6% for rice and wheat, and about 15% for cottonand mustard. Ultimately, the simulated water and salt lim-ited WPT, WPET and WPETQ values for different crops ob-tained in Scenario 2, were 8–19% higher than thoseobtained in Scenario 1 (Fig. 4).

The reduction in seepage losses QSL (Scenario 3) has asimilar impact on the T and ET, while it acts opposite onQbot as compared to the increased crop yields Yg (Scenario2). Reduction in seepage losses from the conveyance systemimplies more canal water irrigation. In Scenario 3, thereduction in QSL at 25–30% of the total canal inflow (Table

Figure 4 Simulated water and salt limited water productivity of the main crops (wheat, rice, mustard and cotton) under the fivescenarios (Table 1) in Sirsa district. Values are based on the distributed SWAP simulations for the unsaturated zone (0–300 cm) overthe period of 10 years (1991–2001), and apply to the entire area under a particular crop.

720 R. Singh et al.

2) provided 23% higher canal irrigation Icw over entire Sirsadistrict as compared to Scenario 1. The calculated ground-water irrigation Igw decreased by 8% only, and total irriga-tion I increased by 6%. This increased I resulted in 6–52%higher Qbot for different crops in Scenario 3 as comparedto Scenario 1. The increased use of good quality canal waterin Scenario 3, however, also resulted into 4–9% higher sim-ulated Yg for different crops as compared to Scenario 1. De-spite increased Qbot, the higher simulated Yg in combinationwith reduced seepage losses from 41% to 27% of total canalinflow in Sirsa district increased the WPETQ by 6–12% for dif-ferent crops (Fig. 4). The WPT and WPET values obtained inScenario 3 were also slightly higher than those obtained inScenario 1.

In Scenario 4, 15% of the measured canal water inflow inthe northern BMB command was reallocated to the centralGHG and SUK commands. This reallocation resulted into16% less canal irrigation Icw in BMB in Scenario 4 as com-pared to Scenario 1 (Table 2). The groundwater quality inthe northern parts of Sirsa district is relatively poor, whichrestricts groundwater pumping. The groundwater pumpingIgw in BMB under Scenario 4, therefore, was increased by11% only, and hence the total irrigation I was decreasedby 5%. This lower amount of irrigation and increased useof poor quality groundwater increased the water and saltstress on crops in BMB, which accounts for 54% of the totalcrop area in Sirsa district. As a result, Scenario 4 has aslightly adverse impact on the simulated Yg, and subse-quently on the WPT, WPET and WPETQ values for all crops(Fig. 4). This reduction of 15% canal water inflow in BMBmight be compensated by the reduction in seepage lossesin Scenario 3. The impact of these integrated actions on

the regional water productivity was quantified in Scenario5, which combined Scenarios 2, 3 and 4. The simulatedwater and salt limited WPT, WPET and WPETQ values for allcrops in Scenario 5 were the highest among all scenarios(Fig. 4). For instance, WPET in Scenario 5 was 12% higherfor wheat and cotton, 16% for mustard and 21% for rice ascompared to the reference situation (Scenario 1). This sig-nificant gain in WPT, WPET and WPETQ values was mainlydue to the simulated 12–17% increased Yg (Scenario 2). Im-proved crop management in terms of timely sowing, optimalnutrient supply and effective pest and disease control is ex-pected to achieve these higher simulated water and saltlimited Yg, and WPT, WPET and WPETQ values in Sirsa district.Also the reduction in seepage losses QSL at 25–30% (Scenario3) contributed to these improved WP values.

Net groundwater recharge

For the five considered scenarios of Table 1, the net ground-water recharge R was calculated using the simulated waterbalances in Sirsa district and its four main commands overthe years from 1991 to 2001. For ‘business as usual’ (Sce-nario 1), the calculated R of 97 mm y�1 in BMB correspondsto a rise in groundwater levels in the northern parts of Sirsadistrict. This is due to the low pumping of poor qualitygroundwater Igw in this region, which is compensatedthrough a relatively high canal water inflow Qcw (Table 2).On the other hand, along the Ghagger river belt existsover-exploitation of good quality groundwater. The esti-mated mean annual Igw in SUK and GHG was almost twotimes Igw for entire Sirsa district (Table 2). Hence, the R

Impact of viable water management scenarios 721

in SUK and GHG was estimated at �432 and �385 mm y�1,respectively, corresponding with a strong decline of ground-water levels in the central parts. Therefore, the BMB, SUKand GHG commands need more attention to halt the riseand decline in groundwater levels.

In Scenario 2, the increased crop yields Yg imply a highercrop growth and leaf area development, which increases thecrop transpiration T and decreases the percolation Qbot. Asa result, Scenario 2 has a slightly favourable impact in BMB,and decreased the R from 97 to 90 mm y�1. The calculated Rin the central commands also slightly decreased from �432to �442 mm y�1 in SUK, and from �385 to �396 mm y�1 inGHG (Fig. 5). This means that increased crop yields will re-duce the rise in groundwater levels in the northern parts,while it will accelerate the decline in groundwater levelsin the central parts. The reduction in seepage losses (Sce-nario 3) has a similar impact as Scenario 2 on the groundwa-ter behaviour in Sirsa district. In Scenario 3, the reduction inseepage losses decreased the R from 97 to 71 mm y�1 inBMB, and presents a significant contribution to halt the riseof groundwater levels in the northern parts.

Scenario 4, ‘canal water reallocation (15%) from the BMBto the GHG and SUK’, was mainly formulated to counteractthe rise in groundwater levels in BMB, and the decline ingroundwater levels in GHG and SUK. In Scenario 4, the real-location of 15% of the measured canal water inflow Qcw fromBMB to GHG and SUK provided about 16% less canal irrigationIcw over the entire BMB as compared to Scenario 1 (Table 2).On the other hand, the Icw was increased by 130% over SUKand 57% over GHG. As a result, the high groundwater pump-ing Igw in SUK and GHG in Scenario 1 (Table 2) decreased by18% in Scenario 4. Further, the evapotranspiration ET in SUKand GHG increased by 2–3% only. As a result, the estimatedR in BMB decreased from 97 to 33 mm y�1, while R in thecentral commands increased from �385 to �182 mm y�1

in SUK, and from �432 to �213 mm y�1 in GHG (Fig. 5).The integrated impact of the proposed measures (Sce-

nario 5) further decreased the calculated R in BMB from97 to 1 mm y�1. Scenario 5 has a slightly unfavourable im-pact on SUK and GHG as compared to Scenario 4, but a sig-

Figure 5 Net groundwater recharge R (mm y�1) over the entire SirFig. 1) under different scenarios (see Table 1) in Sirsa district. Meaunsaturated zone (0–300 cm) over the period of 10 years (1991–20

nificant favourable impact as compared to Scenario 1.Referring to Fig. 5, we conclude that canal water realloca-tion (15%) from BMB to SUK and GHG (Scenario 4) in combi-nation with reduction in seepage losses (Scenario 3),especially in BMB, is an attractive option to halt the risinggroundwater levels in the northern and the declininggroundwater levels in the central parts of Sirsa district.The recently constructed ‘Ottu weir’ dam on Ghagger rivernear Sirsa town will further increase the amount and reli-ability of the canal irrigation, and contribute to halt thedeclining groundwater levels in the central commands.

For the entire Sirsa district, interesting is the decrease ofestimated R from �27 mm y�1 in Scenario 1 to �50 mm y�1

in Scenario 5 (Fig. 5). The decrease of estimated R by�23 mm y�1 in Scenario 5 is higher than the measured an-nual R of 11 mm y�1 (see ‘‘Business as usual’’). This wasmainly due to an increase in simulated ET by 20 mm y�1 inScenario 5 as compared to Scenario 1 (Table 2). This in-crease in ET is mainly caused by a reduction in seepagelosses (Scenario 3) and increased crop yields (Scenario 2).

Salt build-up

In Sirsa district, irrigation with poor quality groundwater isthe major source of salinization. In shallow groundwater le-vel areas, also capillary rise of poor quality groundwatercontributes to the salt build-up. Rainfall and irrigation withgood quality canal water cause salt leaching. In this study,the mixed use of canal water and groundwater for irrigationwas simulated over the period of 10 years (1991–2001), andsalt build-up DC in the upper (0–300) cm soil profile was cal-culated according to:

DC ¼ PCp þ ICi � Q botCbot ð2Þ

where DC is the change in salt storage (ML�2), C is the soluteconcentration (ML�3), and subscript ‘p’ refers to rainfall, ‘i’to irrigation, and ‘bot’ to bottom flux.

For ‘business as usual’ (Scenario 1), there was a DC of2.0 ton ha�1 y�1 over entire Sirsa district (Table 3). Thehigh DC of 4.6 ton ha�1 y�1 in SUK is a result of low canal

sa district and its four main commands (BMB, SUK, GHG and FB;n values are based on the distributed SWAP simulations for the01), and apply to the entire area (cropped as well as bare soil).

Table 3 Salt build-up DC (ton ha�1 y�1) over the entire Sirsa district and its four main commands (BMB, SUK, GHG and FB; Fig. 1)under the five scenarios (Table 1) in Sirsa district

Scenarios/spatial scale Scenario 1 Scenario 2 Scenario 3 Scenario 4 Scenario 5

Sirsa district 2.0 2.2 1.3 2.0 1.6BMB 2.5 2.8 1.7 3.0 2.6SUK 4.6 5.2 4.4 3.8 3.5GHG 1.6 2.0 0.6 0.5 �1.0FB 0.4 0.6 0.1 0.4 0.2

Mean values are based on the distributed SWAP simulations for unsaturated zone (0–300 cm) over the period of 10 years (1991–2001), andapply to the entire area (cropped as well as bare soil).

722 R. Singh et al.

water inflow Qcw and high groundwater pumping Igw (Table2). Despite the high Qcw and the low Igw, BMB also showeda significant DC of 2.5 ton ha�1 y�1. This is caused by rela-tively saline groundwater in the northern parts of Sirsadistrict.

In Scenario 2, the simulated increased crop yields Yg de-creased the percolation Qbot by 6–15% for different crops ascompared to Scenario 1. The decreased Qbot implies lessleaching of salts, and hence Scenario 2 resulted in a slightlyincreased DC in Sirsa district and its four main commands(Table 3). In Scenario 3, the reduction in seepage lossesQSL implies a higher on-farm canal irrigation, which will re-sult into low groundwater pumping. As a result, the totalsalt inflow of 7.3 ton ha�1 y�1 over entire Sirsa district inScenario 1 (Table 2) was decreased to 6.8 ton ha�1 y�1 inScenario 3. In addition, the Qbot for different crops in-creased in Scenario 3, which increased the leaching of salts.Therefore Scenario 3 has a significant favourable impact onthe DC of entire Sirsa district and its four main commands(Table 3). In Scenario 4, canal water reallocation (15%) fromBMB to GHG and SUK increased the use of poor qualitygroundwater, and decrease the leaching of salts in BMB,and vice versa in GHG and SUK. Therefore, Scenario 4 re-sulted into a slightly increased DC in BMB, while it de-creased in GHG and SUK. For BMB, the increased DC inScenario 4 might be neutralized by the decreased DC in Sce-nario 3. Therefore the DC in BMB in Scenario 5 was2.6 ton ha�1 y�1, which was almost equal to that in Scenario1 (Table 3). In addition, Scenario 5 has a significant favour-able impact on the DC in GHG, SUK, FB and over entire Sirsadistrict.

Concluding remarks

Sirsa district has an almost closed water balance, as the totalannual water inflow, accounting for both rainfall and canalwater inflow, is almost equal to the simulated annual evapo-transpiration. This is also confirmed by themeasured ground-water levels, which show a marginal annual net groundwaterrecharge of 11 mm y�1 during the period from 1990 to2000. At the same time, large differences occur of the calcu-lated net groundwater recharge and salt build-up in differentcanal commands. Therefore interventions are required for amore sustainable water management in Sirsa district.

The average simulated water and salt limited grain (orseed) yields of the main crops (wheat, mustard, cottonand rice) in Sirsa district were 1.2–2.0 times higher com-

pared to the average recorded yields over the period1991–2001. This indicates that there exists substantialnutritional, pest and/or disease stress on crops in Sirsa dis-trict. Efforts to increase crop yields will certainly improvethe water productivity. In this study, the simulated 12–17% increased grain (or seed) yields increased the waterproductivity WPET by 12% for wheat and cotton, 14% formustard and 18% for rice as compared to the reference sit-uation ‘business as usual’.

The proposed strategy to reduce seepage losses from thecurrent 36–42% to 25–30%, improved the simulated waterand salt limited WPT, WPET and especially WPETQ by 6–12%for different crops. In addition, this strategy contributedsignificantly to the decrease of net groundwater rechargein the northern part of Sirsa district. The reduction in seep-age losses from the conveyance system, therefore, isstrongly recommended, especially in the northern partswith relatively poor groundwater quality.

Canal water reallocation (15%) from northern to centralparts will have a slightly adverse impact on the simulatedwater and salt limited WPT, WPET and WPETQ values, andon the salt build-up in the northern parts. The decreased ca-nal water inflow in the northern parts may partly be com-pensated by the reduction in seepage losses. Further, thereallocation of water from the northern parts to the centralparts will have a tremendous favourable impact to decreasethe net groundwater recharge in the north, and to increasethe groundwater recharge in the central parts. Canal waterreallocation (15%) from northern to central parts in combi-nation with the reduction in seepage losses at 25–30%,therefore, is suggested to halt the rising groundwater levelsin the northern, and the declining groundwater levels in thecentral parts of Sirsa district.

Simulation of the combined effect of activities i.e. im-proved crop husbandry, reduction in seepage losses and ca-nal water reallocation significantly improved the waterproductivity values, the large spatial differences of netgroundwater recharge and salt build-up as compared tothe current situation. The simulated evapotranspirationover entire Sirsa district increased by 20 mm y�1 for thecombined scenario, which exceeds the measured annualgroundwater recharge of 11 mm y�1. This means that theintegrated impact of the proposed alternative water man-agement scenarios will significantly contribute to a moreproductive and sustainable water management in Sirsadistrict.

The physical based ecohydrological SWAP model, whencombined with field experiments, remote sensing and GIS,

Impact of viable water management scenarios 723

demonstrated its capabilities to analysis the impact of alter-native water management scenarios on a regional irrigationsystem. The developed distributed modelling framework isable to quantify the net groundwater recharge, which isthe most important component of proper regional ground-water management. Spatio-temporal quantification of netgroundwater recharge from the unsaturated zone may serveas input to regional groundwater flow models. A combina-tion of both a distributed ecohydrological model and a re-gional groundwater flow model should be investigated forimproved regional irrigation and groundwater management.

Acknowledgements

This research in Sirsa district (Haryana), India was financedby the Dutch Ministry of Agriculture through the WATer PRO-ductivity (WATPRO) project from January 2001 to November2003. The authors thank the anonymous reviewers for theirconstructive comments.

References

Aggarwal, M.C., Roest, C.W.J., 1996. Towards improved watermanagement in Haryana state. Final Report of the Indo-Dutchoperational Research Project on Hydrological Studies. Chaudh-ary Charan Singh Haryana Agricultural University, Hisar. Inter-national Institute for Land Reclamation and Improvement,Wageningen, Alterra Green World Research, Wageningen, TheNetherlands, 80pp.

Aggarwal, P.K., Roeter, R.P., Kalra, N., van Keulen, H., Hoanh, C.T.,Van Laar, H.H. (Eds.), 2001. Land Use Analysis and Planning forSustainable Food Security: with an Illustration for the State ofHaryana, India. Indian Research Institute – International RiceResearch Institute–Wageningen University and Research Centre,New Delhi – Los Banos – Wageningen, The Netherlands, 167pp.

Berkoff, J., 1990. Irrigation management on the Indo-GangeticPlain. Technical Paper 129. Washington, DC, World bank.

Boels, D., Smit, A.A.M.F.R., Jhorar, R.K., Kumar, R., Singh, J.,1996. Analysis of water management in Sirsa District in Haryana:model testing and application. Report 115. Alterra Green WorldResearch, Wageningen, The Netherlands, 50pp.

Boonstra, J., Singh, J., Kumar, R., 1996. Groundwater model study,Sirsa District, Haryana. Indo-Dutch operational research projecton hydrological studies. Technical Report. International Insti-tute for Land Reclamation and Improvement. Wageningen, TheNetherlands, 87pp.

van Dam, J.C., 2000. Field-scale water flow and solute transport.SWAP model concepts, parameter estimation and case studies.PhD thesis, ISBN 90-5808-256-3, Wageningen University, Wagen-ingen, The Netherlands, 167pp.

Groundwater Cell, 2002. Groundwater Atlas of Sirsa District.Department of Agriculture, Sirsa (Haryana), India.

Hellegers, P., Perry, C.J., 2003. Water as an economic good inirrigated agriculture – theory and practice. Research Report3.04.12. Agricultural Economics Research Institute, The Hague,The Netherlands, 152pp.

Jacobs, C., and de Jong, J., 1997. Constraints and opportunities forimplementation of improved irrigation management in a water-logged area – the case of Hisar District. MSc thesis, WageningenUniversity and University of Twente, The Netherlands, 88pp.

Jensen, M.E., 1996. Irrigated agriculture at the crossroads. In:Pereria, L.S., Feddes, R.A., Gilley, J.R., Lesaffre, B. (Eds.),Sustainability of Irrigated Agriculture. Kluwer Academic Pub-lishers, The Netherlands, pp. 19–33.

Kroes, J.G., van Dam, J.C. (Eds.), 2003. Reference manual SWAPversion 3.03. Alterra Green World Research, Alterra Report 773,ISSN 1566-7197. Wageningen University and Research Centre,Wageningen, The Netherlands, 211pp.

Malik, R.S., Kumar, R., Dabas, D.S., Dhindwal, A.S., Singh, S.,Singh, U., Singh, D., Mal, J., Singh, R., Bessembinder, J.J.E.,2003. Measurement program and description database. In: VanDam, J.C. and Malik, R.S. (Eds.), 2003. Water productivity ofirrigated crops in Sirsa district, India. Integration of remotesensing, crop and soil models and geographical informationsystems. WATPRO Final Report, Including CD-ROM. ISBN 90-6464-864-6, pp. 29–39.

Molden, D., 1997. Accounting for water use and productivity. SWIMPaper 1. International Irrigation Management Institute,Colombo, Sri Lanka. 16pp.

Molden, D., Murray-Rust, H., Sakthivadivel, R., Makin, I., 2003. Awater-productivity framework for understanding and action. In:Kijne, J.W., Barker, R., Molden, D. (Eds.), Water Productivity inAgriculture. Limits and Opportunities for Improvement. CABIPublishing, Wallingford, UK, pp. 1–18.

Navalawala, B.N., 1999. Water resources development and manage-ment. Yojana July (1999), Publications Division, Ministry ofInformation and Broadcasting, Govt. of India, NewDelhi, pp. 4–9.

Singh, R., 2005. Water productivity analysis from field to regionalscale: integration of crop and soil modelling, remote sensing andgeographical information. PhD thesis, ISBN 90-8504-155-4.Wageningen University, The Netherlands, 147pp.

Singh, R., Kroes, J.G., van Dam, J.C., Feddes, R.A., 2006.Distributed ecohydrological modelling to evaluate irrigationsystem performance in Sirsa district, India. I. Current watermanagement and its productivity. J. Hydrol. (this issue),doi:10.1016/j.jhydrol.2006.03.037.