Research Report

The IIMI Water BalanceFramework: A Modelfor Project Level Analysis

C.J. Perry

International Irrigation Management Institute

INTERNATIONAL IRRIGATION MANAGEMENT INSTITUTEP O Box 2075, Colombo, Sri Lanka

Tel (94-1)867404 Fax (94-1) 866854 E-mail IIMI@cgnet.comInternet Home Page http:/ /www.cgiar.org

ISSN: 1026-0862ISBN: 92-9090-331-7

5

2Research Reports

IIMIs mission is to create sustainable increases in the productivity of irrigated agriculturewithin the overall context of water basins and the analysis of water resource systems as awhole. In serving this mission, IIMI concentrates on the integration of policies, technologies,and management systems to achieve workable solutions to real problemspractical, rel-evant results in the field of irrigation and water resources.

The publications in this series cover a wide range of subjectsfrom computer model-ing to experience with water users associationsand vary in content from directly appli-cable research to more basic studies, on which applied work ultimately depends. Some re-search papers are narrowly focused, analytical, and detailed empirical studies; others arewide-ranging and synthetic overviews of generic problems.

Although most of the papers are written by IIMI staff and their collaborators, we wel-come contributions from others. Each paper is reviewed internally, by IIMIs own staff, byIIMIs senior research associates and by other external reviewers. The papers are publishedand distributed both in hard copy and electronically. They may be copied freely and citedwith due acknowledgment.

3Research Report 5

The IIMI Water BalanceFramework: A Modelfor Project Level Analysis

C. J. Perry

International Irrigation Management InstitutePO Box 2075, Colombo, Sri Lanka

4The author C. J. Perry is the Coordinator of Research at the International IrrigationManagement Institute.

Earlier drafts of the framework proposed here benefited from valuable comments and sug-gestions from Jacob Kijne, Jacques Rey, S. G. Naryanamurthy, Harald Frederiksen, BrianAlbinson, Andy Keller, Benoit Lesaffre, S. A. Prathapar, and David Seckler.

Any errors in the analysis remain the responsibility of the author. Suggestions for im-provement and correction are welcome.

Perry, C. J. 1996. The IIMI water balance framework: A model for project level analysis. ResearchReport 5. Colombo, Sri Lanka: International Irrigation Management Institute (IIMI).

/ irrigation management / irrigation programs / surface irrigation / analysis / water use efficiency /water balance / water loss / seepage / groundwater / pumping / computer models /

ISBN: 92-9090-331-7ISSN: 1026-0862

IIMI, 1996. All rights reserved.

Editor: Kingsley Kurukulasuriya; Consultant Editor: Stephen Breth; Artist: D.C. Karunaratne;Typesetter: Kithsiri Jayakody; Editorial/Production Manager: Nimal A. Fernando.

5Contents

Summary 1

Introduction 3

Conceptualization of the IIMI Water Balance Framework 4

Application of the IWBFA Worked Example 5

Calculations 9

Discussion 11

Availability of Data and Related Analytical Issues 12

Literature Cited 13

Annexes

A. Notes on the Use of the Worksheet 14

B. Documentation of Analysis 15

C. The Impact of Watercourse Lining on Cropping Intensity 17

1Summary

The water balance for an irrigation project is a com-plex set of inflows, outflows, consumptive use, andrecycling of water. When changes in cropping pat-terns, management regimes, or infrastructure aremade, they disturb existing balances. The water bal-ance must be fully understood to properly anticipatethe chain of impacts resulting from interventions suchas canal lining.

The IIMI water balance framework (IWBF), whichis in Excel, is an easy-to-use computer model for ana-lyzing the utilization of water from surface irrigationand rainfall within an irrigation project. The frame-work allows explicit definition of losses to seepage,operational losses, and efficiency of field application.Recycling of water through pumping from drains andfrom groundwater is allowed, and the resulting waterbalance is presented, showing flows to groundwater,outfalls from drains, and consumptive use of cropsand other evaporative uses. The model permits asimple validity check of basic assumptions to ensureapproximate internal consistency. Where such consis-tency cannot be demonstrated, either through thismodel or a more complex formulation, the impact ofchanges in management or infrastructure cannot bepredicted.

The data on which the proposed water balancemodel is based are often reasonably well known orassumed to be known. Usually, losses in canals andwatercourses have been measured, and amounts ofrainfall and effective rainfall, water use by crops, andthe volume of pumping from groundwater anddrains are also known within reasonable limits. Ifthese data are not known, and reasonable assump-tions do not allow computation of a credible balance,the first priority in project designor where interven-tions are plannedis additional study.

The model is based on a simple gross water bal-ance. The elements of that balance include the mostcommon set of known or assumed data for an irriga-tion systemcanal inflows; operational, evaporative,and seepage losses; rainfall; crop consumptive use; and

recycling of groundwater and drainage flows. Themodel has been constructed as a workbook (consist-ing of five worksheets) in Microsoft Excel version 5.

Even with sound field data, the IWBF may some-times be inadequate to the type of analysis requiredif precise accounting of the soil moisture status isneeded, if lateral flows are significant, or if waterquality issues are important. But in most cases, thelevel of detail and scope of analysis that the modelprovides will be sufficient to shed light on the inter-actions among the components of the water balancefor example, the relative importance of rainfall, sur-face deliveries and pumped supplies to crop con-sumption. In all cases, this model provides a conve-nient initial analytical approach.

The IWBF accounts for two inflows of water (sur-face-delivered supplies and rainfall), four outflows(crop evapotranspiration, nonbeneficial evaporation/evapotranspiration, drainage runoff, and net flows togroundwater). These elements are interlinked throughseepage from channels and irrigated fields, the dispo-sition of rainfall between runoff, infiltration, andevapotranspiration, and two modes of transfer (pump-ing from groundwater and pumping from drains).

The user must consider the appropriate level ofdisaggregation, spatially and temporally. Consider-ations include the purpose of the analysis, selection ofphysically appropriate boundaries, and avoidance ofsignificant boundary effects by choosing the rightscale for the analysis.

A single irrigation project may be taken as awhole. But in a large project where deliveries, crop-ping patterns, or groundwater conditions vary, it maybe appropriate to disaggregate the project into distinctareas. Similarly, it is usually appropriate to separateseasons if rainfall and cropping patterns differsharply from one season to the next. The IWBF allowssimultaneous analysis of three agricultural seasons,with different data for each season (except, of course,project size). The model then produces individual sea-sonal analyses as well as summary tables for the year.

2For each season, the model produces a set of threetables. The first traces the flow of water to the field asirrigation water is delivered via canals and water-courses; it also traces associated flows to groundwa-ter and drains. At the field level, rainfall is integratedinto the supplies, together with water pumped fromdrains and groundwater to compute the basic waterbalance. The second table shows the sources, and dis-position of water in both depth terms (as is commonlyused for reporting consumptive use and rainfall) andvolume terms (as is commonly used for reportingflows). The third table summarizes sources and usesof water. A separate set of three tables consolidates theseasonal data on an annual basis.

This report includes documentation of the under-lying formulas in the worksheet on which the analy-sis is based and gives a simple example of employingthe model to examine the impact of an investment inimproved infrastructure.

The analysis provided by the IWBF is of primaryinterest to those involved in designing irrigationprojects, formulating improvements to existing infra-structure, or revising operational rules. Managers ofirrigation projects will also find the analysis useful asa basis for interpreting issues such as water use effi-ciency and identifying the primary causes and effectsof water imbalanceslong-term rises and falls in thewater table.

3Understanding the water balance at projector command level is a prerequisite to ana-lyzing the operation of an irrigation systemand its performance. The relationship be-tween various sources of water (rainfall,canal supplies, and pumping from ground-water and drains), and uses (crop consump-tion, drainage outflows, and rises or falls inthe water table) are complex, and some-times counterintuitive.

Interventions in one aspect of systeminfrastructure or operation will usuallyhave impacts beyond the direct, firstround effect. All too frequently, proposedinterventions that will significantly affectthe components of the water balancelin-ing canals, expanding conjunctive use,changing canal schedules, or introducingdrainage systems or new cropping pat-ternsare considered or implemented with-out analysis of the water balance and thelikely effects of the intervention.

Similarly, irrigation is described as in-efficient at the field or project level, whilevirtually no water leaves the river basinbecause losses in one place are recap-tured, and consumed, downstream. Aproper understanding of the water balance,which identifies sources, uses, and reuses ofwater, will clarify such situations.

Such an analysis is of primary interestto those involved in the design of irrigationprojects or in the formulation of improve-ments to existing infrastructure or opera-tional rules. Managers of irrigation projectswill also find the analysis useful for inter-preting issues such as water use efficiencyand identifying the management or invest-

ment interventions to improve efficiency orinfluence the sustainability of their projectsby controlling undesirable trends in the wa-ter table.

The appropriate degree of sophisticationin formulating a water balance depends onthe purpose of the analysis and on the com-plexity of the system. If the purpose is to de-sign a groundwater control system or a flooddisposal system or to understand the effectsof surface water quality and quantity and soilsalinity and sodicity, detailed and specializedmodeling may be required.

The model proposed here, the IIMI Wa-ter Balance Framework (IWBF), is based ona simple gross water balance. The elementsof that balance include the most commonset of known or assumed data for an irriga-tion systemcanal inflows; operational,evaporative and seepage losses; rainfall;crop consumptive use; and recycling ofgroundwater and drainage flows. Themodel has been constructed in MicrosoftExcel version 5 as a workbook consisting offive worksheets (see Annex A).

The framework presented here is notdesigned to meet specialist needs. Rather itis intended for more general or diagnosticpurposes including:

understanding and quantifying themain factors in the water balance

identifying linkages between sources,uses, and reuses

estimating project water consumptionas a basis for defining actual losses, theefficiency of water use, and the produc-tivity of water

The IIMI Water Balance Framework: A Model for ProjectLevel Analysis

C. J. Perry

Introduction

4 analyzing the potential impact of inter-ventions

This simple gross balance approach alsoprovides insights into related issues such assalinity. Another paper in this series (Kijne1996) addresses the issue of the sustainableirrigated area under known conditions ofwater quality, using information generatedby the IWBF. It is important to note thatwhen the available data and assumptions fora project fail to produce a credible balancewithin the IWBF, understanding of the wa-ter balance is limited, and interventionsshould be delayed until better informationis found. At a minimum, application of thismodel will identify parameters that shouldbe investigated in more detail.

The IWBF differs from CROPWAT(Smith 1992), CRIWAR (Bos, Vos, andFeddes 1996), and the CERES series,1 whichcompute crop water consumption at plant,field, and project levels, as well as appro-

priate irrigation scheduling. CROPWAT andCERES also have yield prediction options.These models produce information that isone element of the project-level water bal-ance, namely, seasonal crop demand.

Other models, such as SHE2 and thevarious HEC models,3 provide analystswith a far more powerful array of toolsthan is available in the IWBF. They addressthe timing of interactions among elementsin the hydrological system at the water ba-sin level, water quality, and many other is-sues. But the data requirements, computa-tional power, and training needs are corre-spondingly more complex. Often, the IWBFwill provide initial insights that the analystmay judge adequate to fulfill the intendedpurpose. Where the IWBF falls short, itspurpose, which is first to encourage ana-lysts to actually attempt to derive waterbalance, will have been fully achieved if theanalyst is persuaded to move on to ahigher-order model.

Conceptualization of the IIMI Water Balance FrameworkThe IWBF accounts for two inflows of water(canal-delivered supplies and rainfall), fouroutflows (crop evapotranspiration, nonben-eficial evaporation/evapotranspiration,drainage runoff, and net flows to ground-water). These elements are interlinkedthrough seepage from channels and irri-gated fields; the disposition of rainfallamong runoff, infiltration, and evapotrans-piration; and two modes of transfer (pump-ing from groundwater and pumping fromdrains).

In total, there are 11 elements in thesystem, each with potential linkages to theother 10. For example, rainfall on irrigatedland can contribute to crop consumption,

nonbeneficial evapotranspiration (NBET),flows to groundwater, and runoff to drains.The runoff to drains can, in turn, bepumped back for further irrigation, par-tially lost to groundwater, and pumpedagain for irrigation from groundwater. Inthe Excel workbook, the user specifies eachinteraction in the most commonly usedterms (percentage of losses from canals, ef-fective rainfall percentage, etc.) and canthus include or exclude any interaction.

The model evaluates 31 of the possibleinteractions. Figure 1 shows the 31 interac-tions with the source identified in the leftcolumn and the destination or use in therow heading.

1For example, Jones andKinry 1986.2Systme HydrologiqueEuropensee Abbott etal. 1986.3 United States Corps ofEngineers, HaestadMethods Civil Engineer-ing Software, Waterbury,Connecticut, USA.

5FIGURE 1.The 31 components of the IWBF.

Application of the IWBFA Worked ExampleThis section describes the approach themodel takes, allowing potential users tojudge the extent to which it will meet spe-cific needs, and explains how to apply themodel (annex B contains cell-by-cell defini-tions of the formulas in the IWBF).

The user must first decide the appro-priate spatial and temporal parameters.Considerations include the purpose of theanalysis, the selection of physically appro-priate boundaries, and ensuring that thescale chosen for the analysis is consistentwith avoiding significant boundary ef-fects.

A single large irrigation project may betaken as a whole. But in a large projectwhere deliveries, cropping patterns, orgroundwater conditions vary, it may be ap-propriate to disaggregate the project intodistinct areas. Similarly, it is usually appro-priate to separate seasons if rainfall andcropping patterns vary sharply. The IWBFallows simultaneous analysis of three sea-sons, with different data for each season

(except, of course, project size). The modelthen produces individual seasonal analysesas well as summary tables for the year.

For ease of exposition, the followingdescription traces the data and analysis fora single season.

Data are entered by the user into thedata block (fig. 2). The gray cells are thoseinto which data can be entered. All cells ex-cept those into which data are entered arelocked to prevent accidental erasure ormodification. Figure 2 includes a sample setof data, presented as a basis for describingthe analysis.

The data analyzed represent conditionsin the fresh groundwater areas of the Indo-Gangetic plains of northern India for thewinter season. Surface water is adequate foronly about 30 percent of the command, andgroundwater use is significant. The projectarea is 10,000 hectares, with seasonal rain-fall of 100 millimeters and surface suppliesequivalent to 300 millimeters over the entirecommand.

6FIGURE 2.The data worksheet with data entered for one season.

B C D E F G H

2 Title Winter Total

3 Area ha 10,000 10,000 project area

4 Irrigation intensity % 55 55 irrigated cropping intensity

5 Canal inflow 000 m3 30,000 30,000 surface water supply6 Operational losses % 10 10 canal inflows surplused to escapes

7 Canal seepage % 25 25 canal inflow lost to seepage in canals

8 Watercourse seepage % 27 27 inflows lost to seepage in watercourses9 Field efficiency (surface) % 70 70 field deliveries from watercourses used by crop

10 Irrigation losses to runoff % 10 10 field losses going to drainage

11 Drain seepage % 10 10 drain flows lost to seepage12 Losses to NBET % 30 30 losses (except runoff) going to NBET

13 Rainfall mm 100 100 rainfall

14 Effective rain (irrigated) % 70 70 rainfall used by crop15 Effective rain (unirrigated) % 50 50 rainfall on unirrigated area to evapotranspiration

16 Rain to runoff % 20 20 noneffective rainfall going to drains

17 Pump recovery (groundwater) % 110 110 flows to groundwater recovered through pumping18 Pump recovery (drains) % 10 10 flows to drains recovered through pumping19 Field efficiency (pump) % 80 80 pumped field deliveries used by crop

In defining the data, the common ter-minology for losses is used, partly be-cause these are terms and data with whichfield practitioners are familiar. The impor-tant ongoing redefinition of these terms, ei-ther as components of effective efficiency(Keller and Keller 1995) or as consumed,recoverable and nonrecoverable fractions(Willardson, Allen, and Frederiksen 1994)helps to clarify the interactions specifiedhere, and in turn, the analysis produced bythe IWBF provides a clear basis for identi-fying the components of these redefinedconcepts of efficiency. In the descriptionsthat follow, traditional terminology is used.Thus canal seepage is described as a loss,although the model allows recapture ofsuch losses through groundwater pumping.The data may be summarized as follows:

Cell D2 TitleUsed to designate the project, season,or other information.

Cell D3 AreaThe physical command area of theproject (or subproject)the maximum

area that could be irrigated in a seasonin the absence of constraints on wateror other inputs.

Cell D4 Irrigation intensityThe proportion of the area that is irri-gated from surface water or groundwa-ter, or both, in the period under analy-sis. The irrigated area is equal to thearea multiplied by irrigation intensity.Taking the data in figure 2 as an ex-ample, the irrigated crop area for theseason would be 5,500 hectares (10,000* 55%). No differentiation should bemade between area irrigated from sur-face water and area irrigated throughpumping from groundwater or drains.

Cell D5 Canal inflowSurface water delivery at the canal headfor the period of analysis (season, year,etc.) in thousands of cubic meters. Notethat surface deliveries may be pumpedfrom a river or other source. However,elsewhere in the model, pumped sup-plies refer to internal recycling of flowsfrom groundwater and drains.

7Cell D6 Operational lossesThe proportion of canal inflow that is re-leased through escapes,4 and hence todrains, expressed as a percentage of ca-nal inflow. These losses are assumed tooccur at the canal level, not in water-courses. If such losses also occur at thewatercourse level, they should be in-cluded in the estimate of operationallosses for canals. Losses at the water-course level may be recaptured throughpumping from drains.

Cell D7 Canal seepageThe proportion of canal inflow that goesto seepage. Such losses may be recov-ered through pumping from groundwa-ter.

Cell D8 Watercourse seepageThe proportion of watercourse inflowthat is lost to seepage. Such losses maybe recovered through pumping fromgroundwater.

Cell D9 Field efficiency (surface)The proportion of water arriving at thefield from canals that is used in evapo-transpiration in the course of thegrowth of the crop. Conventionally, thisvalue includes evaporation of moisturefrom the wetted field surface.

Cell D10 Irrigation losses to runoffThe proportion of field losses that goesto drains. This value is different fromrainfall-related runoff (Cell D16), andwill depend on the irrigation technol-ogy. It may often be zero in water-shortcommands with less than 100 percentirrigation intensity.

Cell D11 Drain seepageThe proportion of flows in drains lostto seepage. This value may be differentfrom the value used for canal or water-course seepage. Drains are usually low-lying and often in areas with a rela-tively high water table, tending to re-

duce seepage. Nonbeneficial evapo-transpiration from rainfall onunirrigated land is accounted for sepa-rately (see below).

Cell D12 Losses to NBETThe proportion of seepage and fieldlosses that is evaporated by weeds ortrees or directly from the surface, ex-cluding those evaporation losses thatare conventionally accounted (in pro-grams such as CROPWAT) as part ofcrop demand. Residual losses go togroundwater (surface runoff is alreadyaccounted for through D10).

Cell D13 RainfallTotal depth of rain during the analysisperiod.

Cell D14 Effective rain (irrigated)The percentage of rain falling on irri-gated land that is used for crop transpi-ration. That portion of the rainfall thatis not used by the crop goes to runoff,NBET (from weeds, trees along canals,or evaporation directly from the sur-face), or groundwater in accordancewith specified ratios.

Cell D15 Effective rain (unirrigated)The proportion of rain falling onunirrigated land that is lost throughevapotranspiration. This value may bedifferent from the corresponding valuefor the irrigated area. If no rain-fedcropping is practiced, this rainfall willbe entirely nonbeneficial (which is howit is accounted for in the model).

Cell D16 Rain to runoffThe proportion of rainfall that is notused by the irrigated crop and goes tosurface drainage as runoff.

Cell D17 Pump recovery (groundwater)The proportion of surface and rainfalllosses to groundwater that is recoveredthrough pumping. If the value in cell

4This occurs when, forexample, rainfall causesa sharp fall in demandfor irrigation water.

8Cell D19 Field efficiency (pump)The proportion of deliveries frompumping of groundwater and drainpumping that is used by the irrigatedcrop. The residual is lost to NBET(from weeds, trees along canals, orevaporation directly from the surface)or to groundwater in accordance withspecified ratios. These losses may differfrom the corresponding value for sur-face deliveries because of the bettercontrol of the latter.

Figure 3 shows one of the three sea-sonal worksheets containing the water bal-ance values computed on the basis of thedata shown in the data block (fig 2). Figure4 is a schematic flow diagram of the resultsshown in figure 3.

D17 is set to 100 percent, the watertable will remain at a constant level.Setting it at less than 100 percent re-sults in a drainable surplus to ground-water and a rising water table. The es-timated value will depend upon waterquality, installed capacity, and aquiferconditions. If the volume of pumping isknown (from survey data or estimates),the value given in D17 can be adjustedso that the computed volume matchesthe known volume.

Cell D18 Pump recovery (drains)The proportion of surface and rainfallrunoff to drains that is recoveredthrough pumping. If the volume ofsuch pumping is known (from surveydata or estimates), the value in D18 canbe adjusted so that the computed vol-ume matches the known volume.

FIGURE 3.Portion of a worksheet showing the water balance computed from data in the data block worksheet.

Season: Winter Surface Rainfall PumpingTable 1: Water balance water Irrigated Unirrigated Total Groundwater Drains

000 m3

25 CANAL INFLOW 30,000

26 Operational losses 3,000

27 Drain outfall 2,700

28 NBET 90

29 To groundwater 210

30

31 Seepage 7,500

32 NBET 2,250

33 To groundwater 5,250

34

35 WATERCOURSE INFLOW 19,500

36 Seepage 5,265

37 NBET 1,580

38 To groundwater 3,686

39

40 FIELD DELIVERY 14,235 5,500 4,500 10,000 18,819 413

41 Crop use 9,965 3,850 3,850 15,055 331

42 Losses 4,271 1,650 6,150 3,764 83

43 Drain outfall 384 297 405 702 339 7

44 NBET 1,166 406 2,250 2,656 1,028 2345 To groundwater 2,720 947 1,845 2,792 2,398 53

9FIGURE 4.Schematic flow diagram of the water balance shown in figure 3.

CalculationsCanal inflows are traced through canals andwatercourses to the field. The losses speci-fied in the data block are accounted for ateach stage. In the example (fig. 3), opera-tional losses to drains (cell D26) are 10 per-cent (cell D6) of canal inflows.5 Seepagelosses from drains (D11) are 10 percent, sothat the net flow in the drain is 2,700,000cubic meters (D27). Of the 300,000 cubicmeters of seepage losses from drains, 30percent (D12) goes to NBET (D28), with theresidual going to groundwater.

Canals lose 25 percent of inflows toseepage (from cell D7), which is further al-located between NBET and water going togroundwater. The residual canal flow(30,000,000 3,000,000 7,500,000 =19,500,000 m3) arrives at the watercourse

level, and further seepage losses within thewatercourse are computed and allocated toarrive at the final delivery of surface sup-plies to the field level (D40).

At the field level, the calculations be-come more complex because supplies to thefield come from watercourses (D40), as out-lined above, rainfall (E40 and F40), andpumping from groundwater and drains(H40 and I40, respectively).

Rainfall on irrigated land contributes tocrop use (cell E41, based on cell D14), theresidual going to runoff (E43) and NBET(E44) in accordance with the ratios specifiedin the data block (D16 and D12, respec-tively), with the residual going to ground-water (E45).

5 Cells numbered from 2to 19 are shown in thetable in figure 2, andcells numbered from 25to 45 are shown in thetable in figure 3.

10

Rain falling on unirrigated land does notcontribute directly to irrigated crop use.6 Itgoes to NBET (F44, based on the proportionspecified in D12), to drainage (F43, based onD16), or to groundwater (F45).

Pumping allows transfer of water fromdrains and groundwater to irrigated crop-ping in accordance with the recovery per-centages specified in the data block foreach. The amount of water available for thisis computed from the various relevantsourcesfor example, water going togroundwater is the total of those compo-nents of canal (D33), drain (D29), and wa-tercourse (D38) seepage that go to ground-water, plus the calculated proportions offield losses (D45) and rainfall infiltration(G45) going to groundwater.

Calculations of field delivery fromgroundwater and drainage pumping (H40,I40) are the sum of a converging serieswater is recovered from groundwater andapplied to the field, and losses from this ir-rigation application further contribute togroundwater.

The water balance table (fig. 3) is fur-ther disaggregated in figure 5, which showssources and disposition of water, calculatedseparately for the irrigated and unirrigatedareas. The data are presented both in vol-ume measures and in depth measures be-cause analysts tend to think in terms of vol-umes for deliveries of water and drainageoutflows and in terms of depths for con-sumption and rainfall.

Allocation of rainfall between irrigatedand unirrigated areas is based on the pro-

FIGURE 5.Portion of a worksheet showing sources and allocation of water.

B C D E F G H I

Season: WinterTable 2: Sources and allocation Command Irrigated Unirrigated Command Irrigated Unirrigated

by area 000 m3 mm

50 SOURCES51 Diversion 30,000 22,906 7,094 300 416 15852 Rainfall 10,000 5,500 4,500 100 100 10053 Total 40,000 28,406 11,594 400 516 25854 ALLOCATION55 Net drain outfall 3,719 3756 From surface 3,365 3457 From rain 768 858 Drainage pumping -413 -45960 Crop use 29,200 29,200 292 53161 From surface 9,965 9,965 100 18162 From rain 3,850 3,850 39 7063 From groundwater pumping 15,055 15,055 151 27464 From drainage pumping 331 331 3 66566 Nonbeneficial ET 8,791 4,778 4,014 88 87 8967 From surface 5,085 3,322 1,764 51 60 3968 From rain 2,656 406 2,250 27 7 5069 From groundwater pumping 1,028 1,028 10 1970 From drainage pumping 23 23 0 07172 To groundwater -1,711 -7,671 5,960 -17 -139 13273 From surface 13,850 9,734 4,115 119 141 9174 From rain 3,259 1,414 1,845 28 17 4175 Groundwater pumping -18,819 -18,819 -164 -29976 Leaching fraction at field (%) 19 16

6But it may of course beutilized by unirrigatedcrops through the coeffi-cient for effective rainfall(unirrigated), specifiedin the data block (fig. 2).

11

portion of the area irrigated. In the case ofdiversions, the allocation of water is basedon the assumption that canals and water-courses run through all areas, so that seep-age losses are distributed in proportion tothe irrigated and unirrigated areas. Calcula-tions related to field losses and pumped de-liveries are confined to the irrigated area.

The leaching fraction (D76 and E76 infig. 5) is defined as the proportion of totalwater entering the soil profile that goes togroundwater. The calculation is made at thefield level, ignoring seepage from canals,watercourses and drains, which is localizedand does not affect the salt balance in the

root zone. Although the model includes noanalysis of water quality (which is funda-mental to determining an appropriate valuefor the learning fraction) users can see thesource of each component of recharge andof field application (from rain, surface, orpumping). Thus they can calculate, to theextent that data are available, the likelyquality of the irrigation water applied andhence estimate the adequacy of the com-puted leaching fraction.

The table shown in figure 6 summa-rizes this information to provide a quickreference of what water goes where.

FIGURE 6.Portion of a worksheet giving a summary breakdown.

B C D E F G H I

Season: Winter Crop Consumptive Use

Table 3: Summary Direct Pumped Total Drain NBET Ground-water

81 Diversion (000 m3) 9,965 12,457 22,421 3,028 5,935 -1,38582 Rain (000 m3) 3,850 2,929 6,779 691 2,856 -32683 Total 13,815 15,386 29,200 3,719 8,791 -1,71184 Diversion (%) 25 31 56 8 15 -385 Rain (%) 10 7 17 2 7 -186 Total 35 38 73 9 22 -4

DiscussionThe information presented in figures 3 and5 can be compared with other calculationsand information available about conditionsin the field (for example, whether the watertable is rising, whether drains are frequentlyfull, etc.). Particular points of reference are:

Calculated crop consumption (fig. 5,cell H60, 531 mm) should be comparedwith calculated consumptive use froma program such as CROPWAT.

Calculated value for effective rainfall(fig. 5, D62 and G62) should also beconsistent with results from CROPWAT.

Net flows to groundwater (fig. 5, D72and G72) should be consistent with ob-servations of the water tablethese re-sults show net withdrawals of 124mm/season for the irrigated area (H72);net recharge of 132 mm/season for theunirrigated area (I72) and near balance(17 mm overdraft, G72) for the entirecommand.

If field observations deviate substan-tially from results from the model, furtherrefinement of the input data is needed. Themodel also provides a convenient frame-

12

work for testing the sensitivity of results tochanges in assumptions and for identifyingthe most sensitive linkages.

The calculated crop consumption of 531millimeters is consistent with climatic datafor a crop season of 130 to 140 dayspanevaporation varies from 3 to 5 mm/dayduring the winter months. The calculatedgroundwater balance for the area, where itis known that the water table is stable ordeclining, is also reasonable.

The summary table in figure 6 showsthe relative importance of various water

sources in crop consumptive use. Rainfall,for example, is much less important thanpumped water, and almost 10 percent ofavailable water is lost to surface runoff(comparing cell G83 with total availability,cell D53, fig. 5), indicating the scope for ex-panded use of this sourceperhapsthrough artificial recharge to replenishgroundwater. Recycled surface deliveries(E81) contribute significantly more to cropconsumption than direct deliveries (D81).

Availability of Data and Related Analytical Issues

An approximate estimate of the water bal-ance is a powerful tool. The data requiredfor the analysis presented above are close tothe minimum set that can be expected toprovide meaningful resultsindeed thiswas an objective in designing the IWBF.Nevertheless, several items may prove dif-ficult to obtain or estimate. In particular, theproportions of rainfall going to consump-tive use, runoff, and groundwater are diffi-cult to assess with precision, and the extentof groundwater pumping is often poorlydocumented. Even in areas where powersupplies to wells are metered, the dynamicinteraction between pumping, local con-ing of the water table, and the relationshipbetween power consumption and the vol-ume of water pumped is complex. Simi-larly, where hours of operation are mea-sured, the rate of delivery may vary signifi-cantly over time. In such cases, the solutionis generally to be found through successiveiterations around reasonable estimates. Ifusers know how many wells there are in anarea and have indications of pump size dis-tribution and usage, then they can estimatethe reasonable range of pumping and its re-lationship to canal deliveries.

Because consumptive use of the crop isthe value analysts tend to spend most time

calculating, it is the value we often knowbest. And we know it cannot be greaterthan seasonal evapotranspiration nor muchless than 60 percent of seasonal evapotrans-piration if the crop matures. Analysts alsohave a reasonable fix on effective rainfall,which cannot exceed crop evapotranspira-tion and must, together with the quantity ofwater delivered to the field through irriga-tion, equal or exceed the value for evapo-transpiration.

With this basic data defined withinknown ranges, coming to a realistic waterbalance is not an insurmountable task, eventhough some of the initial estimates may belittle more than informed guesses. The verynature of the system and the fact that it is abalance analysts are seeking ensure that themodel has to close on a consistent andreasonable set of data. Where known datacannot be brought together within thisframework, it is clear that some of the datawill need to be reassessed. The power ofthe model is that it identifies these issues,points to the areas where uncertainties arereal and important, and sometimes simplyforces the analyst to move to a more so-phisticated model. Such results are far pref-erable to continued intervention and invest-ment without knowledge.

13

A number of issues will always needcareful, location-specific, and perhaps sea-son-specific, attention. Canal losses are agood example: It is assumed in the modelthat losses are a (specified) proportion offlows. Often, losses may be more dependenton the duration of flow than the rate, andlosses will be a far higher proportion oflow flows (for example while deliveries aremade for domestic uses). Such attention todetail is important for assessing efficiencyof deliveries at different times of the year.In such cases, the loss percentage can bevaried by season and even selected toachieve a specific volume of losses.

A further difficult issue is assigningrunoff to irrigation or rainfall (in otherwords, does one assume that it rained just

after an irrigation, and thus the rain goes tothe drains, that it rained during an irriga-tion, so that a mixture goes to the drains, orthat it rained just before an irrigation, sothat irrigation water goes to the drains?).One view7 is that since the irrigation wateris the additional input, incremental flows todrains should be charged to irrigation(in)efficiency. The model as formulated fol-lows the middle course described above,which may fail the test of intellectual rigor,but has the advantage of recognizing thelikely physical composition of the water(since over time, it will sometimes be rain,sometimes irrigation, and sometimes a mix-ture that goes to drains). This then allowsconsideration of the quality of water goingto the water table and to drains.

Literature Cited

Abbott, M. B., J.C. Bathurst, J.A. Cunge, P.E. O Connell, and J. Rasmussen. 1986. An introduction to theEuropean hydrological systemSystme Hydrologique Europen, SHE, 2: Structure of a physically-based, distributed modelling system. J. Hydrol., 87: 61-77.

Bos, M. G., J. Vos, and R. A. Feddes. 1996. CRIWAR 2.0: A simulation model on crop irrigation water requirements.Publication 46. Wageningen, Netherlands: International Land Reclamation Institute.

Jones, C. A., and J. R. Kinry, eds. 1986. CERES-Maize: A simulation model of maize growth and development.College Station, Texas, USA: Texas A&M University Press.

Keller, J., and A. Keller. 1995. Effective irrigation efficiency. Arlington, Virginia, USA: Winrock International.Kijne, Jacob. 1996. Water and salinity balances for irrigated agriculture in Pakistan. IIMI Research Report 6.

Colombo: International Irrigation Management Institute.Smith, M. 1992. CROPWAT: A computer program for irrigation planning and management. FAO Irrigation and

Drainage Paper 49. Rome: FAO.Willardson, L. S., R. G. Allen, and H. D. Frederiksen. 1994. Universal fractions and the elimination of irriga-

tion efficiencies. Paper presented at the 13th Technical Conference, U.S. Committee on Irrigation andDrainage at Denver, Colorado, October 1994.

7Harald Frederiksenpersonal communica-tion.

14

GeneralIt is assumed here that the user is fa-

miliar with or has access to the manual forMicrosoft Excel. Words in italics in this an-nex are explained in the Excel manuals. Themodel is available in formats for Excel ver-sion 5 and does not readily translate intoother software application because exten-sive use has been made of multipleworksheets in a single workbooka featurethat does not translate. The workbook occu-pies about 100 kilobytes of disk space.Where large numbers of situations are to beanalyzed, it may be worth creating linkeddata sheets, which would each occupy onlya few kilobytes, referencing the data sheetfrom the data cells in the main sheet. Theworkbook consists of five worksheetsData,Season 1, Season 2, Season 3, and Total.

Data Entry and ProtectionData may only be entered in the Data

worksheet. The workbook is protected so thatthe user can only enter information into theshaded data cells. However, there is no pass-word, so the user can readily unprotect theworksheets to see and modify formulas, butit is strongly recommended that this not be

done in normal use. It is easy to accidentallyerase or modify a cell, generating spuriousresults. On the other hand, password protec-tion has been deliberately excluded to allowusers to modify the worksheets for specificuses where the present formulation is unsuit-able, to create linked worksheets, as sug-gested above to conserve disk space, or tomake use of the logic of this model for otherpurposes. It is recommended that any alter-native or additional calculations be done bycreating additional worksheets within thepresent workbook (or separate, linkedworksheets) so that the integrity of thepresent model is preserved.

Linkage to External CalculationsIt may often be convenient to use this

framework as part of a larger water balance(for example where there are significant ad-ditional components in the water balance,such as deliveries and withdrawals fornonagricultural uses). It is relatively easyfor an experienced user of Excel to linksuch flows into the model, generating theassociated impacts on groundwater, canallosses, etc., while maintaining the integrityof the underlying analytical framework.

ANNEX A

Notes on the Use of the IWBF

15

The analysis is documented by definingrange names that are applied to the cells inthe worksheet, corresponding to the namesin the data block. Thus, for example, wher-ever cell D7 is referenced, this is replacedby Canal_seepage. Formulas thus are of theform:

Canal_inflow* (1-Canal_seepage-Opera-tional_losses).

This translates to=$D$5*(1$D$7$D$6)in the underlying worksheet. The disk ver-sion of the worksheet is in the conventionalformat, making it considerably smaller andfaster.

ANNEX B

Documentation of the Model

B C D E F G H I

Season Surface Rainfall Pumping

Table 1: Water balance water Irrigated Unirrigated Total

000 m3 Groundwater Drains

25 Canal inflow Canal_inflow

26 Operational losses Operational_losses * Canal_inflow

27 Drain outfall D26 * (1 - Drain_seepage)

28 NBET (D26 - D27) * (Losses_to_NBET)

29 To groundwater (D26 - D27) * (1 - Losses_to_NBET)

30

31 Seepage (Canal_inflow * Canal_seepage)

32 NBET (Canal_inflow * Canal_seepage)* Losses_to_NBET

33 To groundwater Canal_seepage * Canal_inflow(1 - Losses_to_NBET)

34

35 Watercourse inflow Canal_inflow* (1 - Canal-seepage -Operational_losses)

36 Seepage D35 * Watercourse_seepage

37 NBET D35 * Watercourse_seepage *

38 To groundwater D35 * Watercourse_seepage *(1 - Losses_to_NBET)

39

40 Field delivery D35 * (1 - Watercourse_seepage) Area * Rainfall * Area * Rainfall (1- Area * Rainfall/100 D1 * Pump_recovery__groundwater* F1 *Pump_recovery__drains+I1*Irrigation_intensity/100 Irrigation_intensity/100 H1 *Pump_recovery__groundwater* Pump_recovery__drains * (D1*

(D1 *Pump_recovery__groundwater + Pump_recovery__groundwater + F1*F1 *Pump_recovery__drains)/(1-)H1* Pump_recovery__drains)/(1-(H1*Pump_recovery__groundwater + I1* Pump_recovery__groundwater + I1*Pump_recovery__drains)) Pump_recovery__drains))

41 Crop use D40 * Field_efficiency__surface G40 * Irrigation_intensity * G40 * Irrigation_intensity Field-efficiency__pump * H40 Field-efficiency__pump * 140Effective_rain__irrigated *Effective_rain__irrigated

42 Losses D40 * (1 - Field_efficiency__surface) E40 _ E41 G40 - G41 H40 - H41 140 - 141

43 Drain outfall D40 * (1 - Field_efficiency__surface)* E40 * (1 - F40* (1 - Effective_rain_unirrigated) E43 + F43 H40* (1 - Field_efficiency__pump)* 140* (1 - Field_efficiency__Pump)*Irrigation_losses_to-runoff * (1 - Effective_rain__irrigated)* *Rain_to_runoff* (1 - Drain__seepage) Irrigation_losses_to_runoff * (1 - Irrigation_losses_to_runoff * (1 -Drain_seepage) Rain_to_runoff * (1 - Drain_seepage) Drain_seepage)

Drain_seepage)

44 NBET (D40 - D41 - D43) * Losses_to_NBET (E40 - E41 - E43) * F40 * (Effective_rain__unirrigated) E44 + F44 (H40 - H41 - H43) * Losses_to_NBET (140 - 141 - 143) * Losses_to_NBETLosses_to_NBET

45 To groundwater (D40 - D41 - D43) * (1 - (E40 - E41 - E43) * (1 - F40 - F43 - F44 E45 + F45 (H40 - H41 - H43 - H44) 140 - 141 - 144 - 143Losses_to_NBET) Losses_to_NBET)

FIGURE B.1.Formulas in the worksheet.

16

17

FIGURE C.1.Water balance based on a watercourse seepage of 5 percent and irrigation intensity of 55 per-cent.

Season: WinterTable 2: Sources and allocation Command Irrigated Unirrigated Command Irrigated Unirrigated

by area 000 m3 mm

SOURCESDiversion 30,000 25,066 4,934 300 440 115Rainfall 10,000 5,700 4,300 100 100 100

Total 40,000 30,766 9,234 400 540 215ALLOCATION

Net drain outfall 3,771 38From surface 3,430 34From rain 760 8Drainage pumping -419 -4

Crop use 30,066 30,066 301 527From surface 12,968 130 228From rain 3,990 40 70From groundwater pumping 12,773 12,773 128 224From drainage pumping 335 335 3 6

NBET 7,615 4,333 3,282 76 76 76From surface 4,150 3,018 1,132 41 53 26From rain 2,571 421 2,150 26 7 50From groundwater pumping 872 872 9 15From drainage pumping 23 23 0 0

To Groundwater -1,451 -5,856 4,404 -15 -103 102From surface 11,311 8,670 2,641 97 124 61From rain 3,203 1,440 1,763 27 17 41Groundwater pumping -15,966 -139 -244

Leaching fracion at field (%) 19 17

8This is achieved mostconveniently using thegoal seek function in Ex-cel. This allows the userto specify that the cellcontaining cropping in-tensity should be set to avalue that results in cropconsumptive use againequaling 531 mm.

ANNEX C

The Impact of Watercourse Lining on Cropping Intensity

The following example demonstrates asimple application of the model, startingfrom the data already presented in the bodyof this report (fig. 2, 3, and 6). The analysisfocuses on a single change to the data in-putreduction in seepage losses from wa-tercourses as a result of lining.

The originally assumed watercourseseepage loss is reduced from 27 percent (fig.2) to 5 percent. The impact of this change,among other things, is to increase the com-puted value of consumptive use to 545 mil-limeters, compared with 531 millimeters inthe original calculation (fig. 5). Clearly, ifthe original 531 millimeters is consistentwith computed crop needs, there would bean opportunity to increase the cropped area

after lining the watercourses. The potentialimpact can readily be assessed by experi-menting with different values of irrigationintensity.8 Increasing this value from 55 per-cent to 56 percent reduces consumptive useto 536 millimeters, and a further increase ofirrigation intensity to 57 percent reducesconsumptive use to 527 millimeters (fig.C.1). We can therefore estimate that the in-crease in irrigation intensity permitted bythe lining program is less than 4 percent .That is a small increase given the very sub-stantial reduction in losses from water-course seepagefrom 27 percent to only 5percent. The explanation, of course, lies ingroundwater use, which captures most ofthe losses for reuse.

18

FIGURE C.2.Summary results based on a watercourse seepage of 5 percent and irrigation intensity of 55percent.

Season: Winter Crop Consumptive Use

Table 3: Summary Direct Pumped Total Drain NBET Ground-water

Diversion (000 m3) 12,968 10,228 23,196 3,087 4,847 -1,131Rain (000 m3) 3,850 2,929 6,779 691 2,857 -326

Total 16,818 13,158 29,975 3,778 7,704 -1,457Diversion (%) 32 26 58 8 12 -3Rain (%) 10 7 17 2 7 -1

Total 42 33 75 9 19 -4

The most important change from re-ducing watercourse seepage to 5 percent isthe saving in pumpingfrom 15,386,000cubic meters pumped in the first case (fig.6) to 13,158,000 cubic meters after lining(fig. C.2). Watercourse lining has a minimalimpact on cropped area, but saves about 17percent in power. This solution is, of

course, strongly related to the exploitationof the aquifer. It would be quite different ifthere were no potential to exploit losses togroundwater. This exercise demonstratesthe relative simplicity of identifying andquantifying such important relationshipsonce a reasonable approximation of the wa-ter balance has been assembled.

19

20

Research Reports

1. The New Era of Water Resources Management: From Dry to Wet Water Savings. David Seckler,1996.

2. Alternative Approaches to Cost Sharing for Water Service to Agriculture in Egypt. C.J. Perry, 1996.

3. Integrated Water Resource Systems: Theory and Policy Implications. Andrew Keller, Jack Keller,and David Seckler, 1996.

4. Results of Management Turnover in Two Irrigation Districts in Colombia. Douglas L. Vermillionand Carlos Graces-Restrepo, 1996.

5. The IIMI Water Balance Framework: A Model for Project Level Analysis. C. J. Perry, 1996.

Research Report

The IIMI Water BalanceFramework: A Modelfor Project Level Analysis

C.J. Perry

International Irrigation Management Institute

INTERNATIONAL IRRIGATION MANAGEMENT INSTITUTEP O Box 2075, Colombo, Sri Lanka

Tel (94-1)867404 Fax (94-1) 866854 E-mail IIMI@cgnet.comInternet Home Page http:/ /www.cgiar.org

ISSN: 1026-0862ISBN: 92-9090-331-7

5