Recommending the IHACRES model for water resources assessment and resolving water conflicts in Afric

JOURNAL OF ARID LAND, 2011. VOL. 3,NO. 1,.40 -48Recommending the IHACRES model for waterresources assessment and resolving waterconflicts in AfricaSamir Mohammad Ali Alredaisy'Department of Geography, Faculty of Education, University of Khartoum, Omdurman 406, SudanAbstract: The International Association of Hydrological Sciences (IAHS) recognized the lack of hydro-logical data as a world-wide problem in 2002 and adopted the Prediction of Ungauged Basins (PUB) as adecadal research agenda during the period of 2003 to 2012. One of the objectives is to further developmethodologies for prediction in ungauged basins and to reduce uncertainties in model prediction. Estimationof stream flows is required for flood control, water quality control, valley habitat assessment and waterbudget of a country. However, the majority of water catchments, streams and valleys are ungauged in mostdeveloping countries. The main objective of this paper is to introduce the IHACRES (lentification of Hy-drographs and Components from Rainfall, Evaporation and Stream) model into Atrican hydrological plan-ning as a methodology for water resources assessment, which in turm can be used to resolve water conflictsbetween communities and countries and to study the climate change issues. This is because the IHACRESmodel is applied for the estimation of flows in ungauged catchments whose physical catchments descriptors(PCDs) can be determined by driving variables (i.e. rainfll and temperature); and also in gauged streamsbut whose gauging stations are no longer operational but historical data are available for model calibration.The model provides a valuable insight into the hydrologic behaviour of the upper water sources for valleys aswell as provides a useful methodology for water resources assessment in situations of scarce financialresources in developing countries. In adition, it requires relatively few parameters in its calibration and hasbeen successful applied in previous regionalization studies. It will also make possible the equitable distri-bution of water resources in intemational basins and rivers' catchments. This paper does not apply themodel anywhere, but recommends it as a methodology for water resources assessment in order to curewater conflicts on the African continent.Keywords: ungauged catchment; water resources assessment; ranfall; runoft; lumped models; Nile Basin; Atricamore, the changes of land use make past stream flow1 Introductionrecords difficult to use for future assessment of waterHydrological data are required for sustainable waterresources. Data's inadequacy occurs not only in un-gauged catchments but also in gauged catchments be-resources planning and management to enable quanti-cause of data's inconsistency, inaccuracy in measure-fication of water quantity and quality (Oyebande, 2001).ments, short duration of recorded data or low networkRiver flow measurements are necessary for water re-density, while some gauging stations are sometimessources planning and conservation, pollution controland for solving many environmental problems. Thelocated in remote areas and become inaccessible duringrequired data can be obtained through measurements atrainy season hindering acquisition of data and main-tenance of equipment. Although the availability ofriver gauging stations. However, most water catch-中国煤化工is restricted tempor-ments in most African countries (Fig. 1) are ungauged,i.e. without adequate recorded data in both quantity andReceivMHCNMHGloi: 103124/51.J.1221.2011.0期Uquality or spatial and temporal distribution and hence●Corresponding author: Samir Mohammad Ali Alredaisy (E-mail: redaisy2@such data may be unavailable when needed. Further- ybo.om)No.1 Samir Mohamnad Ali Aredaisy: Recommending the IHACRES model for water resources asesment and resolving water..41ally and spatially, the needs of data are inevitable.the entire catchment. The models simulate catchmentWater supply, storage, distribution and hydraulic response well although some accuracy may lose as thestructures all require stream flow data, and the needsscale of lumping increases (Kirkby, 1999). The modelsfor stream data flow are increasing due to rising waterrequire less input data, pose lttle computational bur-demand, declining finance, technology and humanden, hence, their use is widespread. Unlike statisticalcapacities for data collection and monitoring.and black-box models, lumped model can not be ap-Hydrological models are substantively a set ofplied to the conditions not reflected in the calibrationmathematical equations based on theoretical principlesdata set. This is because the lumping of processes isthat link inputs to outputs and are impotant tools inoften over-simplified. A key factor in the applicationwatershed management. They are used to study com-of lumped models is the stability of the catchmentplex problems and synthesize different kinds of infor-system, stable spatial distribution of precipitation,mation necessary for planning and decision making invegetation and soil characteristics. According to thewater resources management and development (Wool-study of Refsgaard et al. (1996), when the objective ishiser et al, 1982). In stream flow prediction they pro-to simulate time series of stream flow, only lumpedvide the water resources planning and disaster man-conceptual model gives best results compared withagement tool. Models are either lumped or distributedother models.and have been classified into statistical, black box,lumped parameter and physically based models2 Study area(Vertessy et al. 1993). Each model is related to itselfThe African continent (Fig. 1) is a rich mosaic eco-predictive power, utility, accuracy and use.system ranging from the snow and ice field of Kili-Generally, basic models represent producingmanjaro to tropical rainforest to the Saharan desert.mechanisms of individual run-off and spatial varia-Although it has the lowest fossil energy use per capitations in catchment characteristics (e.g. soil propertiescompared with other regions, Africa may be the mostand vegetation cover). Beven (1989) indicated thatvulnerable continent to climate change because wide-models require large amount of input data, which isspread poverty limits country's capabilities to adaptdiffcult to be obtained. The models are poorly used,unless the input data are of very good qualityStrait of_ F(Refsgaard et al., 1992). Statistical models are based(Gibraltar( Meietrancanon regression relationships and have been widely usedin catchment runoff predictions since they are rela-tively simple to be constructed (Stoneman et al, 1989).RiverThe major limitation to these models, however, is that! SenegalNiger RiverRodGulfofthey need to be based on long-term rainfall-unoff re-A BluctBenuecords and that statistical association on catchmentWhitecharacteristics or planned treatments is different(Bosch et al, 1982). Most black-box models arelumped and have good agreement between observed2 Victoriyand predicted daily stream flow using their time seriesangnyikaanalysis function method (Jakeman et al, 1993).These models are incapable of predicting the hydro-Alanticlogical impact of catchment disturbances unlessOeeanFambezi9 Njaestream flow data collected after the disturbance arefirst calibrated.中国煤化工mUnlike statistical and black box models, lumpedmodels assume that system inputs and dynamics areCNMHGrIndianYHOcean(KIveruniform in space and are realized when rainfall-runoffmodels are aggregated spatially and temporally overFig. 1 River basins and lakes in Atrica42JOURNAL OF ARID LANDVol. 3the change. Signs of changing climate in Africa havefor the determination of physical parameters arealready emerged including melting glaciers in theavailable from topographical sheets, soil maps or re-mountains, warming temperatures in drough-pronemote sensing and are derived using GIS. Conceptualareas, and sea-level rise and coral bleaching along theparameters require rainfall, stream flow, and tempera-coastlines. The surface area of the Chad Lake has de-ure data for their calibration, which is achievedcreased from 25,000 km- in 1963 to 1,350 km2 in 2008.through optimization algorithms (Sorooshain et al,Periods of severe drought led to large-scale environ-1995). This involves systematic search for parametermental degradation, population displacement and ur-values that yield computed runoff hydrographs thatbanization.best match observed hydrographs. With model paIn addition, post independence Africa has witnessedrameters determined, the model can be used to im-very violent conflicts over water resources (Smillie,prove data records in gauged catchments, if rainfall2000). In westerm African, sub-region water conflictdata for missing stream flow is available.involved sedentary farmers and mobile pastoralistsIn ungauged catchments, rainfall-runoff models can(Shatima and Tar, 2008). Pastoralists have interactedbe used in data generation after regionalization, whichwith sedentary farmers for millennia, but water scar-involves correlating the conceptual parameters tocity increased tension and conflicts between thesecatchment characteristics from several gauged catch-groups in many parts of the world (Fratkin, 1997).ments (Nathan et al, 1990; Shaw, 1996; Funke et al,Traditionally, resource-based conflict has been repre-1999). Therefore, the data to be obtained by rain-sented by the old competition between fammers andfall-runoff models with regionalized parameters arepastoralists over water and land resources (Assal,possible for use. Regionalization of model parameters2006). Sub- Saharan Africa is more vulnerable to wateris the calibration of the model to representativestress than any other regions where about 64% of Af-catchments scattered across the basin and assigningricans rely on the water that is limited and highlythe parameters from these catchments to other catch-variable; croplands inhabit the driest regions of Africaments around them with assumed similar characteris-where some 40% of the irigated land is unsustainable;tics. However, the description of hydrological charac-roughly 25% of Africa's population suffers from waterteristics in terms of physical catchment descriptorsstress; nearly 13% of the population in Africa experi-(PCDs), allows for the estimation of the Unit Hydro-ences drought-related stress once each generationgraph (UH) for any catchment within the region. The(Tatlock, 2006). The conflict in Darfur, Sudan wasresulting relationships are then used to derive modelover water and grazing rights (Schanche, 2007) whereparameters for the ungauged catchments in the samemany sedentary tribes became targets for displacedgeographical and climatic region making it possible togroups from Northerm Darfur and various camel pas-simulate stream flows in ungauged catchments.toralists (Ayoub, 2006).3.2 The HACRES model3 MethodsThe IHACRES model identifies the rainfall-runoffbehavior from data in its parameters at a catchment3.1 Conceptualization of rainfall-runoff relation-scale. To present physical feature, it incorporates theshipconceptualization of the relevant large scale catchmentainfall-runoff models are mathematical expressionsprocess. It is comprised of two modules in series (Fig.of various rainfall-runoff processes. Some models2)，a nonlinear and a linear. The first one is ahave conceptual parameters only, while others havenon-linear loss module that links rainfall and air tem-both conceptual and physical parameters. Although theperature (Rk and Tr) to effective rainfall (U) with pa-majority of these models were developed for humidrameters C T(w) and F (Fig. 3). It uses temperaturetemperate regions, they need to be adapted to localand中国煤化工relative catchmentconditions through calibration and validation beforemoisiYHCNMHGtheproporionofbeing applied locally. This is done by using rainfall,rainfall tnat Decomes ettective rantall. The second is astream flow, and physiographic data. The data neededlinear unit hydrograph (UH) module that links effec-No.1 Samir Mohammad Ali Alredaisy: Recommending the IHACRES model for water resources asessment and resolving watr..43tive rainfall Uk to stream flow Xk with parametersC-Volurme of aT(q)-quick flowconceptualresponse decay timT(q), T(s), and V(s) (Fig. 3). It routes the effectivecatchment wetnessconstant (days)rainfall through any configuration of stores in parallelconstant (days)|andor in series, which is identified from the time se-T(s)-slow flowT(w)-constantresponse decayries of rainfall and stream data but is typically eitherdecaying timeconstant(days)one store only, representing ephemeral streams or par-days)allel two of both slow and quick flows to be repre-sented (Croke et al, 2005). In the HACRES applica-R)- t Non-linerLinear UHtx)moduletion, it was shown that the parameters in the non-linearIoss module(3 parameters)3 parameters)SFTD心module (C, T(w) and F) had significant direct effectsanalogouson the volume and the peak of flow hydrograph, whilethe parameters in the linear module (T(s), T(q) andF-temperatureV(s)-proportiona! volumetricmodulation factorcontribution of slow flow ofV()) had an effect on the peak of flow hydrograph,(Celsius degree-1)streamflowbut not on its volume (Taesombat and Sriwongsitanon,Fig. 3 IHACRES model structure and dynamic response char-2010).acterstics (DRCs). Source: Lttlewoods et al., (1992)The advantage of the spatially *lumped' approach isRainfalll Non-linearffectiveLinear StreamlowTemperature modulerainfallthat it requires only a few (six) parameters, three in thenon-linear module from model rainfall to rainfall ex-cess and three in a linear module from rainfall excessFig.2 Two modules of the IHACRESto stream flow. A model with a few well- defined pa-The IHACRES is a lumped conceptual model,rameters gives better statistical relationships betweenwhich simulates rainfall-runoff response of catchmentsthose parameters dynamic response characteristics,to total stream flow, with calibrated parameters prior(DRCs) and the selected physical catchment descrip-to simulation by comparison with observed streamtors, (PCDs). The parameter T(w) is the rate at whichflow data (Jakeman et al, 1993). It assumes that therethe catchment wetness declines in the absence of rain-is a linear relationship between effective rainfall andfall. Hence a large T(w) gives more weight to the ef-stream flow (effective rainfall Uk for time step k is thatfect of antecedent rainfall on catchment wetness thanpart of rainfall which eventually leaves the catchmentthe smaller one. The value of parameter C (Fig. 3) isas stream flow X}). This assumption allows the appli-set such that the volume of rainfall excess is equal tocation of the Unit Hydrograph (UH) theory whichthe total stream flow over the estimated period. It isconceptualizes the catchment as a configuration ofthe increase in catchment wetness index per unit rain-linear stores acting in series and/or in parllel. Thefall in the absence of any decrease due to evapotran-non-linearity normally observed between rainfall andspiration. The parameter F (Fig. 3) is a temperaturestream flow is therefore accommodated in the module,modulation factor, which determines how T(w) (T)which converts rainfall to effective rainfall (Fig. 3).changes with temperature for a constant Sk. The pa-The simultaneous identification of UH for both highrameter Aq (or a,) describes the rate of decay, orand low flows by the IHACRES model extends theequivalently the time constants T(q) (T(S)) of theutility of the UH approach to include a large portion ofquick (slow) flow hydrograph following a unit input ofthe whole flow regime. The underlying conceptualiza-rainfall excess. The volume V(s) is the proportionaltion is that the catchment wetness varies with antece-volumetric contribution of slow flow to total streamdent rainfall and evapotranspiration. Therefore, aflow. The quantities C, F, T(q), T(w), T(s), V(s) arecatchment wetness index Sk (ideally, 0≤Sk≤1) isDRCs of the catchmpnte (Fio 3)computed at each time step on this basis. The per-Gene中国煤化工he“data basedcentage of rainfall, which becomes effective rainfall inmechan|YHCNMHG02)andhenceisany time step, varies linearly (between 0 and 100%) asable to make efficient use of existing data set. Forthe catchment wetness index Sk (between 0 and 1).calibration, the model requires time series of rainfall,4JOURNAL OF ARID LANDVol. 3stream flow and temperature data. Its parametric effi- jective appraisal can also be carried out.ciency makes it easy to link its DRCs to PCDs making3.3 Flow estimation in ungauged catchmentsit useful in regionalization studies. Despite its struc-tural simplicity, the IHACRES model has been appliedWhen data on stream flow are unavailable to supportsuccessfully to a wide range of catchment types andmodel calibration, other catchments' data become veryfor regionalization studies (Post et al, 1996, 1999,valuable (Kokkonen, 2002). To estimate stream flow2002; Kokkonen et al, 2003). In practice, the selec-from ungauged catchments for existing or forecastedtion of the 'best' IHACRES model is based on per-future conditions, model parameters may be extrapo-formance in both calibration and simulation (ittle-lated from gauged catchments within the same region,a transfer of information known as regionalizationwood et al, 2002).Physical catchment descriptors of slope area, soil(Bloschi et al, 1995). This effective transfer shouldype, elevation, stream density and land cover at aform a relatively homogeneous group (Post et al,1996). According to Bates (1994) a succesful region-catchment scale can be determined using ArcViewalization of rainfall-unoff model depends on accurateGIS and the available physiographic data. Stepwiseestimation of model parameters for the gauged catch-regression can be used to quantify the relationshipments, selection of PCDs with significant influencesbetween DRCs and PCDs and to identify those PCDson catchment response to rainfall, proper identificationwith the strongest statistical relationship with theof homogeneous regions and the degree of correlationDRCs in order to draw in PCDs overlooked in correla-between model parameters and catchment descriptors.tion analysis. The PCDs that change with time allowsRegionalization is an attempt to relate flow character-hydrological response to be affected by changing landistics at gauging stations to physical and climaticuse or climate. An implicit assumption is that the rela-characteristics of their drainage basins (Riggs, 1990).tionships between PCDs and DRCs are constant andIt is the transfer of information in the form of charac-the change in land use or climate will not alter the hy-teristics describing the hydrological data or modelsdrologic response of a catchment or processes gov-from one catchment to another (Bloschi et al., 1995).erming the losses.These characteristics, obtained from maps and weatherThe relationships between DRCs and PCDs can berecords, can be used for ungauged catchments fromderived by using a progression of techniques fromthe derived relationships to estimate their stream flow.Sefton et al. (1998) and Funke et al. (1999) respec-inspection, correlation analysis, stepwise regressionand multiple regression analysis. The derived rela-tively indicated that step-wise regionalization method-tionships should have some meaning in the context ofology is applicable in the situations where the numberthe individual parameter conceptualization and are notof gauged catchments is limited. Most regional predic-simply mathematical abstractions. Stepwise regressiontive studies focus on a certain flow regime. In particu-can be used to identify those PCDs with the strongestlar estimation of flood indices for ungauged catch-ments has received a lot of attention. Vandewiele et al.statistical relationship with DRCs and these can be(1995) reconstructed monthly flows for basin consid-combined in a multiple regression model that ex-ered ungauged, while Post et al. (1996, 1999) andpresses each DRC in terms of PCDs. The validity ofSefton et al. (1998) predicted daily flows by develop-the relationship can be tested by re-calculating DRCsing relationships between the parameters of a dailyfor all calibration catchments and two other catch-time step rainfall-runoff model and PCDs. In suchments that were not used in the derivation of the rela-studies, the models should be parsimonious in order totionships.capture efficiently the hydrological behavior of theUsing regionalized parameters to estimate flows is acatchment. The consequences of over-parameteriz-way of validation of the relationships, and the derivedation中国煤化工re (Pilgrim, 1983,parameters can be used to simulate flows and carry outJakenthe sensitivity analysis in two ungauged catchmentsRe JMHCNMHGdatafrommanyInd comparison between observed and simulatedcatchments in the same region (Kokkonen, 2002), andflows using visual and numerical methods, and an ob-the limited number of gauged catchments is usuallyNo.1 Samir Mohammad Ali Alredaisy: Recommending the IHACRES model for water resources assessment and resolving water..45available, and using large regions increases the num-the gauged valleys as a function of measurable/desir-ber of available gauged catchments but also increasesable PCDs. The assumption is that the resulting pre-the variation of climate and physiography betweendictive equations apply for valleys or streams otherthese catchments leading to introduction of morethan those from the data that are drawn for their de-variables in the regression analysis (Seibert, 1999).velopment. These equations can be validated on twoThe regionalization process is usually accompanied byadditional valleys or streams and then used to deriveloss of accuracy due to optimization error, adjustmentDRCs of two other valleys. Using the derived DRCsof boundary conditions and use of transfer functions.and the available rainfall and temperature, the modelTo determine the overall regionalization efficiency,can be used to simulate annual flows in gauged valleyssimulations should be carried out using the regional-but considered ungauged for the purpose and a com-ized parameters and the efficiency is determined byparison between simulated and observed flows usingcomparing with that of optimization. The transfervisual and oumerical methods.functions should be tested in separate catchments inAn altermative to the regionalization of the parame-the same climatic regions in order to venify their valid-ters of a model is to regionalize aggregate measures ofity.the hydrological response such as water yield and flowThat the model can be calibrated for any period andprobability distribution, and use these methods to con-a 3-year period is sufficient to capture the diversestrain the model parameters. Using the water yieldweather conditions. Selection of a 3-year calibrationfrom similar neighboring catchments has resulted inperiod (Jakeman, 1993) helps to balance problems ofimproved prediction of flows using a regionalizationvariance and bias. Each period can be selected to startapproach. This can be extended through regionaliza-and end on low flows because the model assumes antion of runoff cofficient, enabling prediction across ainitial catchment wetness peak index (Sk) of zero.larger scale. If the model parameters are calibratedModel calibration can be performed on daily rainfallusing the flow duration curve (FDC) rather than thestream flow time series during a common period, fortime series of flow, the concept can be further ex-example the period of 1970 to 1978, for all the numbertended to calibration of the model to the regionalizedof catchments selected. The period of record can beprobability distribution, avoiding the need to deter-divided intnon-overlapping calibration periods.mine the relationships between model parameters andTransfer function parameters can be optimized usingPCDs (Littlewoods et al.,. 2002).an instrumental variable technique (Jakeman et al,1990), while the parameters of a nonlinear loss mod-4 Discussionule can be selected by a semi automatic parameter.Optimal combination of loss characteristics can be Successful application of the IHACRES model willchosen using objective guidelines based on maximiz-help the African countries to face development prob-ing R' and minimizing average relative parameter er-lems, growing population and demand for water re-ror (ARPE) and bias. The parameters can be consid-sulting from climatic change, and to overcome waterered to represent the dynamic response characteristicsconflicts. The HACRES model has been carried out(DRCs) of the catchments.in humid and arid environments and proved to be suc-Validation can be cited to a specified number of cessful because it requires a few parameters in itsgauged valleys within a chosen area in order to obtaincalibration and regionalization. The study of Jakemana set of dynamic response characteristics (DRCs) de-et al. (1993) confirmed the strong relationship be-scribing their hydrologic behaviour. Physical catch-tween the lumped response parameters of thement descriptors (PCDs) indexing topography, soiIHACRES model and its PCDs. Littlewood (2002)type, climate and land cover can be alotted and linkedproducee and flnw duration curves (FDCs),to hydrological model by overlaying catchmentrespes中国煤化工teled daily stream .boundaries with GIS technique. For example, usingflowYHCNMHGtheRiverTeifitomultiple regression analysis, predictive equations canGlan Teifi in Wales. The model accounts for 88% ofbe developed that predict the calibration parameters of the variance of stream flow over its calibration period46JOURNALOF ARID LANDVol. 3and has the following DRCs: T(w) = 22 days, F =in small to medium size watersheds (Hope et al,2.0/C, C= 69 mm, T(q)= 1.91 days, T(s) = 39.02008). The IHACRES model has been developed indays and V(s)= 0.36 (dimensionless). The FDCs con-the Messara catchment, Crete, with the results show-firm that the model performs well over the 5%- -95%ing that the model is capable of capturing the keyof flow range. The hydrological regime at Glan Teifi is trends over time (Herron and Croke, 2009).characterized by the six DRCs described in Fig. 3 inThe IHACRES model is actually beneficial for wa-our context here. Given similar models for otherter resources assessment in water conflict areas in Af-gauged catchments, statistical relationships can be rica, such as the Nile Basin. The River Nile Basin issought that link DRCs and PCDs. The DRC PCD re-virtually important water catchment in Africa as itlationships can then be applied to ungauged catch-covers an estimated area of 3,254,000 km' represent-ments (Littlewoods et al., 2002). This research hasing 10% of the continent area. This basin support theproved that IHACRES can be effectively used forlives for many centuries, but recent issues on the re-flood estimation and flood routing along the riverquest of source countries for their share, declaration ofcourse.historical right by downstream countries and the par-Dye and Croke (2003) applied the IHACRES modeltial signing of Uganda, Tanzania and Rwanda on ain two South African catchments. In many South Af-separate agreement have shrunk future collaborationrican catchments, the water is an increasingly limitedbetween the Nile Basin Countries. Political conflict onand highly fluctuating resource. Accurate prediction ofwater resources is a facet of future wars on water re-low flows is especially vital if water resource manag-ers are to successfully balance the growing needs ofsources. The Nile Water Treaty signed in the early lastagriculture, industry and rural and urban populations,century is not convincing for many countries in theand to maintain the ecological health of aquatic andNile Basin because of the challenging problems ofriparian ecosystems. The IHACRES shows great popopulation growth, desire for agricultural development,tential in linking proposed land-use change to alteredand so on. However, one of the methods that can in-flow regimes, and efficiently describing the flowcrease knowledge about the actual amounts of watercharacteristics within catchments. Jakeman et al. (1990)within the Nile Basin is the application of theand Jakeman and Homberger (1993) studied to deter-IHACRES model. This will make database for equita-mine whether uncertainty in daily river flow predic-ble distribution of water resources within this basintions using the IHACRES model in small to moderateand also, elsewhere for whole Africa, since the major-size watersheds (50 400 km2) in southerm Califormiaity of catchments and basins are ungauged.would increase if URD gridded rainfall data were usedin place of point rain gauge data to calibrate the model.5 ConclusionThis investigation was a part of a model regionaliza-tion project funded by NASA, Land Cover/Land UseThe general conclusions of this paper are as follows:(1) The IHACRES model requires a few parametersChange (LCLUC) program (Hope et al, 2005). TheIHACRES model was selected for this project becausein its calibration and was scessfully applied in manyit has a . parsimonious, lumped-conceptual structureregionalization studies.that avoids uncertainties associated with over parame-(2) The IHACRES model is beneficial at the com-terization and is well suited to parameter regionaliza-munity and country level for water resources assess-tion studies. Unlike distributed hydrologic models,ment.lumped models such as the IHACRES do not require(3) The IHACRES model can provide a solution forinformation on the spatial distribution of rainfallcurrent water problems in Africa and provide databasewhich would be problematic using URD gridded datafor future water resources planning and management.中国煤化工ReferencesHCNMHGAssal M A M.2006. Sudan: identity and conflict over natural resources.Ayoub M. 2006. Land and conflict in Sudan. [2010-11-08].Development, 49: 101-105.htp:/:/www.r.or/our work/accord/sudan/Land conflict.php.No.1 Samir Mohammad Ali Alredaisy: Recommending the IHACRES model for water resources asessment and resolving water..47Bates B C.1994. Regionalization of bydrological data, a review. In:Kokkonen T s.2002. Rinall-runoff modeling comparison of modelinga report as part of project D2. Regionalization and scaling ofstrategies with focus on ungauged predictions and model integration.hydrological data. Report 94/5, CRC for Catchment Hydrology,Ph D Thesis, Hisiki University of Technology, Finland.Linsley R K, Crawford N H. 1966. Digital simulation in hydrology,Beven K J. 1989. Changing ideas in hydrology: the case of physicallyStanford watershed model IV, Technical report 39, Dep, ofCivil Eng,modeling. Joumal of Hydrology, 105: 157-172.Stanford University, Stanford, CA.Bosch J M, Hewlett J M.1982. A review of catchmnent experimentsto Litlewood G Jakeman AJ, Croke B. et al. 2007. Unit hydrographdetermine the efets of vegetation changes on water yield andcharaterization of flow regimes leading to stream flow estimation inevapotranspiration. Joumal of Hydrology, 55: 3- -23.ungauged catchoments (egionalizatin), l: Predictions in UngaugedBoschi G Sivaplan M.1995. Scale isues in hydrological modeling: aBasins: PUB Kick-of, Prodeedings of the PUB Kick off Meingreview. Hydrological Processes, 9: 251-290.Held in Brasilia, 20 -22 November, 2002. Sheffield: IAHS Publica-Croke B W, Anderson F, Jackeman A J, e1 al. 2005. Re design of thetion, 309.IHACRES ranall-runoff model. Engineers Australia. 29th Hydrol-Litlewood I G 2002. Improved unit hydrograph characterisation of theogy and Water Resources Symposium, 21-23 February, Camberra,daily flow rgime (including low flows) for the River Teif, Wales:Australia.towords better rafll-steamflow models for regionalization. Hy-Dye P J, Croke B F W.2003. Evaluation of stream flow predictions bydrology and Earth System Sciences, 6(5): 899 -911.the IHACRES rainfall runoff model in wo South African catchments. Nathan R J, McMahon T A.1990. Identification of homogeneous他-Environmental Modeling & Software, 18: 705 -712.gions for purposes of regionalizaion. Journal of Hydrology. 121:Frakin, Elliot. 1997. Pastoralism: govemance and development isues.217- -238.Annual Review of Anthropology, 26: 235- 261.Oyebande L.2001. Water problems in Africa. How can science Help?Funke R Schuman A H, Schulz G A.1999 Regionalization of paHydrological Science Journal, 46 (6): 947- 961.rameters in mesoscale hydrological models. ln: Kirkby MJ SchroderPerrin C, Michel C, Andereassian V. 2001. Does a large number ofU. Rcgionalization in Hydrology. Proceedings of A Conference Heldparameters enbance model performance? Comparative assessment ofat Braunschweig. March 1997. Sheffield: IAHS Publication. No.254,common catchment model structures. Jourmal of Hydrology, 242:171-179.275- -301.Herron N F, Croke B F W.2009. IHACRES-3S - A 3-store formulationPilgrim D H.1983. Some problems in tansfering hydrological rela-for modeling groundwater-surface water interactions. In: Andersseotionships between small and large drainage basins and between re-R s, Braddock R D, Newham TH 18t World IMACS Congress andgions. Journal of Hydrology, 65: 49-72.MODSIM Intermational Congres on Modeling and Simulation.Post D A, Jakeman A J.199. Predicting the daily stream flow of un-Moudelling and Simulation Society of Australia and New Zealandgauged catchments in S.E Australia by regionalizing the paranetersand Interational Associaion for Mathematics and Computer inof a lumped conceptual riafl-runoff model. Ecological Modeling,Simulation Cairns Australia, 13- -17 July.123: 91-104.Hope A, Decker J, Jankowski P. 2008. Uility of gridded rainfall forPost D A, Croke B F. 2002. Predictig hydrologic response fromIHACRES daily river flow predictions in Southern Califomia wa-physio-climatic atributes: an application to ungaugedtersheds. Joumal of the American Water Resources Association.sub-catchments of the Burdekin river, North Queenlands. ln: Pro-44(2): 428-435.cedings Intermational Environmental Modcling and Software Soci-Jakeman A J, Litlewoods A I, Whitehead P G 1990. Computation ofety Conference, 2427 June. Lugano, Switzerland, 1: 363- 374.instantaneous unit hydrograph and idenifiable component flows withPost D A. Jakeman A J.1996. Relationships between catchment arib-application to two small upland catchments. Journal of Hydrology,utes and hydrological response characteristics in small Australian117:275- -300.catchments. Hydrological Processes, 10: 877- -892.Jakeman A J, Homberger G M.1993. How much complexity is war-Refsgard J C, Seth S M, BathurstJ C, et al. 1992. Application of theranted in ranall-runoff model!? Water Resources Research, 29(8):SHE model to caltchments in India, part 1, general resuls. Journal of2637- -2649.中国煤化工Kirkby MJ.199 Traslating models from hlslpe (I ha) to catchmeatRetsgvalidatio and inter com-(1000 km) scale. ln: Kirkby M J, Schroder U. Regionalization inMHC N M H Gnolels. Waer Reourcs .Hydrology. Proceedings of A Conference Held at Braunschweig,Research, 32(7): 2189- 2202.March 1977. Sefeld: IAHS Pubication. No. 254, 1-12.Riggs H C.1990. Estimating flow carateristis at ungauged sites. In:18JOURNAL OF ARID LANDVol. 3Regionalization in Hydrology. Intermational Association of Hydro-Stoneman G L, Schofield N J.1989. Silviculture for water production inlogical Sciences, Washington DC. 1990. Seffeld: IAHS Publication,Jarrot forest of Westem Australia -a0 evaluation. Forest Ecology andNo. 191: 159-169.Management. 27: 273- _293.Schanche D. 2007. Scarce resources, ethnic strife fuel Darfur conflict.Taesombat w, Sriwongsitanon N. 2010. Flood investigation in the upper[2007-10-29]. h:/://w/.wg/e/elt/s/story/story.php? sto-Ping River Basin using mathematical models. Kasetsart Journalryld=6425093(Natural Science), 44: 152- -66.Sefon C E, Howarth S M.1998. Relationships between dynamic地Tatlock c W.2006. Water stress in Sub-Saharan Africa, council onsponse charateristics and physical ceatchment descriptors of catch-foreign relations. [2010-11-10]. hp://www.f.or/publication/ments in England and Wales. Journal of Hydrology, 211: 1-16.11240/water stess in sub Saharan africa html.Seibert J.1999. Regionalization of parameters for a conceptual rain-United Nations Developmeat Program, Sudan 2010. Reduction of reall-unoff model. Agriculure and Forest Meteorology, 98 99:source based conflicts armong pastoralists and farmers. (2010-11-10].179-293.h:///wwd.ud.op/prpjpcscp3.tmo.pShatima A. Tar 0.2008. FarmerVandewiele G L, Elias A. 19959. Monthly water balance of ungaugedcaused and consequences. Information, Society and Justice, 1(2): ，catchments obtained by geographical regionalization. Journal of163-184Hydrology 170: 277-291.Shaw E M.1994. Hydrology in Practice. London: Taylor & Francis.Woolhiser D A. Brakensick D L.1982. Hydrologic system synthesis. In:399- 421.Haan C T. Johnson H P, Brakensiek D L. Hydrologic Models ofSmillie I, Gberie L, Hazleton R.2000. The Heart of the Matter- SierraSmall Watersheds. American Society of Agricultural Engineers,Leone, Diamonds and Human Security. Otawa: Partnership AfricaASAE Monograph. No. 1-15.Canada.Young P C.2002. Data-based mechanistic and top- down modeling. In:Sorooshain s, Gupa V K.1995. Model calibration. In: Singh V P.Proceedings Intemational Environmental Modeling and SofwareComputer Models of Watershed Hydrology. Chapter 2. Colo: WaterSociery Conference, 24- -27 June 2002. Lugano, Switzerland, I:Resources Publications Highlands Ranch. 23- -68.363- 374.中国煤化工MYHCNMHG