Contents lists available at ScienceDirectScience of the Total EnvironmentELSEVIERjournal homepage:www.elsevier.com/locate/scitotenvModeling future water footprint of barley production in Alberta,Canada:CrossMarkImplications for water use and yields to 2064Mohammad Badrul Masud *Tim McAllister b,Marcos R.C.Cordeiro Monireh FaramarziaHIGHLIGHTSGRAPHICAL ABSTRACTA framework is developed to model bar-ley yield,water use and water footprint.Yield and water footprint of barley isassessed under climate change.Rainfed yield is projected to increase andirrigated yield is expected to remain un-changed.Water footprint is projected to decreasein future.Water footprint is adjusted based onwater-stress conditions.Article history:Despite the perception of being one of the most agriculturally productive regions globally.crop production in Alber-Received22Augu过2017ta,a western province of Canada,is strongly dependent on highly variable climate and water resources.We devel-oped agro-hydrological models to assess the water footprint(WF)of barley by simulating future crop yield (Y)andAccepted 1 November 2017Available online 16 November 2017consumptive water use (CWU)within the agricultural region of Alberta.The Soil and Water Assessment Tool(SWAT)was used to develop rainfed and irrigated barley Y simulation models adapted to sixty-seven and elevenEditor:D.Barcelocounties,respectively through extensive calibration,validation,sensitivity,and uncertainty analysis.Eighteendownscaled dimate projections from nine General Circulation Models(GCMs)under the Representative Concentra-tion Pathways 2.6 and 8.5 for the 2040-2064 period were incorporated into the calibrated SWAT model.Based onCrop modelingthe ensemble of GCMs,rainfed barley yield is projected to increase while imigated barley is projected to remain un-changed in Alberta.Results revealed a considerable decrease(maximum 60%)in WF to 2064 relative to the simu-Blue and green water footprintlated baseline 1985-2009 WF.Less water will also be required to produce barley in northern Alberta (rainfedUncertainty predictionbarley)than southern Alberta(imrigated barley)due to reduced water consumption.The modeled WF data adjustedfor waterstress conditions and found a re markable change (increase/decrease)in the irrigated counties.Overall,theresearch framework and the locally adapted regional model results will facilitate the development of future waterpolicies in support of better climate adaptation strategies by providing improved WF projections.2017 Elsevier B.V.All rights reserved.1.IntroductionCanada is home to ~12 million head of cattle with the majority ofCorresponding author.beef production occurring in Alberta(Farm Credit Canada,2012).Alber-ta is globally recognized for its large oil and agricultural exportshttps://doi.org/10.1016/j.scitotemv.2017.11.0040048-9697/0 2017 Elsevier B.V.All rights reserved.MB.Masud et al Science of the Total Ervironment 616-617 (2018)208-222209including beef and live cattle to over 50 countries around the worldMean winter temperatures usually range from-25.1 to -9.6C while(Alberta Cattle Feeders'Association,2017).Beef production in this prov-mean summer temperatures vary between 8.7 and 18.5 C with theince depends largely on annual feed crops such as wheat,canola,andmean annual temperature ranging from 3.6 to 4.4 C (Jiang et al.,barley,with barley being the principal feed.Given that global beef pro-2017).The province has 17 river basins (Fig.1a).Most of the southerduction is expected to increase 74%by 2050(FAO,2017),there will beriver basins are snowmelt dominated in their upstream highland areas,an increased need to produce barley to sustain Alberta beef exportsand glacier melt plays a major role in supplying downstream waterWater use intensity of beef production in Canada indicates that feedneeds in late summer.and pasture production are responsible for the majority of water con-Alberta is the home of 65%irrigationin Canada (Alberta WaterPortalsumption (Legesse et al,2017).However,unreliable precipitation pat-2017).With 6%of Alberta's total water availability (Faramarzi et al.tems,potentially adverse impacts due to climate change,and2017),the southern river basins provide nearly 57%of the irrigationincreasing conflicts among different water users represent serious chal-lenges to the availability of this critical resource for Alberta's beef indus-seeking means of improving water use efficiency,and reducing thetry (Islam and Gan,2014;de Souza et al.,2017).Thus,an assessment ofwater footprint to meet water supply-demand constraints during pe-barley production in the future in light of the uncertainties in waterriods of high water shortages (Faramarziet al.,2017).Irrigation districtsavailability arising from climate change is needed in order to prepareare spread in 11 out of 67 counties (Fig.1c)in the souther part of thethe beef industry for potential issues that can compromise its sustain-province,where there is often not sufficient precipitation and soil mois-ability in Alberta and its contribution to global food security.ture to naturally meet crop requirements.In the central and northemA usable metric for assessing current and future crop water use isareas of the province crop production relies on precipitation.Barley iswater footprint(WF)accounting which offers a quantifiable indicatorone of the most commonly grown crops in the province,and its produc-to measure the volume of consumptive water use (CWU)per unit croption and Y is dependent on the availability of water resources and influ-yield (Y).The WF of crops can be quantified as being composed ofenced by other climate and phenological factors.For instance,rainfedblue,green,and grey forms(Hoekstra et al.,2011).The blue WF ofbarley had an average annual yield of 2.5 t/ha,dropped to its lowestcrops is based on freshwater consumption (e.g.,lakes,river,and aqui-level(1.5t/ha)in 2002,due to drought,generating the lowest yield dur-fers),while green WFis based on effective precipitation that is consumeding the 25-year period (1983-2007).While irrigated yield sustainedin the form of soil moisture.The grey WF is quantified based on thearound 4.5 t/ha with sufficient irrigation to compensate climate varia-amount of freshwater required to assimilate pollutants to meet specificwater quality standards.Application of the water footprint concept hasbeen particularly challenging at large-scale studies (Marano and Filippi,2015:Ma and Ma,2017:Shrestha et al.,2017a),due to uncertainties inrepresenting soil-plant-water-atmosphere relationships at the regionalThe SWAT model divides each river basin into sub-basins based onscale,where changes in water consumption and crop yields are influ-topography and subsequently into Hydrologic Response Unit(HRU)ac-enced by both natural and anthropogenic factors.In this context,futurecording to the soil,land use,and slope characteristics.The plant growthprojections of crop production in relation to climate change involve in-component of SWAT,which is a simplified version of the Erosion Produc-teractive and dynamic soil-plant-atmosphere processes that are assessedtivity Impact Calculator (EPIC;ref.Williams,1995),is capable of simulat-with integrated system models (Ahuja et al.,2007).ing a wide range of crops,grassland,and pasture.In the SWAT model,One example of such models is the Soil and Water Assessment Toolcrop biomass development (above-and-under ground)is simulated(SWAT),a process-based,time-continuous,bio-physical model (Arnddon a daily time-step based on light interception and conversion of COet al.,1998)that simulates hydrology and soil-water dynamics at a dailyto biomass.Actual crop yield is then calculated as a product of actualtime step (Vigiak et al.,2015).It is widely used to simulate the impactsabove ground biomass and the actual harvest index.Actual harvestof climate,water,and agricultural management practices on hydrology,index is calculated,on a daily basis,as a fraction of above ground plantvegetation growth,and related bio-physical processes in small and largedry biomass removed as dry economic yield.A plant is assumed tobasins(Tuo et al.,2016).SWAThas been used ina number of large and re-start growing once the temperature exceeds its base temperature (T).gional scale studies (Abbaspour et al.,2015:Vigiak et al.,2015;PalazzoliThe portion of the mean daily temperature exceeding Th will contributeet al.,2015:Liu et al.,2017;Faramarziet al.,2017:Shrestha et al,2017a)to growth over the growing period.If the temperature falls below ThIt is hypothesized that barley production and its water use in Albertathen the plant is assumed to enter dormancy.The actual crop water up-can satisfactorily be projected by considering the key influential factorstake is simulatedina daily time step.It is based on soil waterdynamics inand adapting to the regional/local conditions using the SWAT model.different soil layers and crop potential evapotranspiration (PET).ThereThe main objective of this study is to assess the WF of barley as the prin-are three different methods (i.e.Penman-Monteith,Priestley-Taylor,cipal feed crop used in beef cattle production.The specific objectives are:and Hargreaves)available in the model to calculate PET,with the1)to set up a high-resolution crop model of Alberta for rainfed and iri-Penman-Monteith being the most comprehensive as it considers variousgated barley by utilizing available local agro-hydrologic data and inforclimatic factors(Allen et al.,1989).The Penman-Monteith approach wasmation;2)to adjust physical parameters of the rainfed and irrigatedconsidered in this study as the base to simulate CWUmodels to local conditions by calibration and validation of annual cropIn this study,four and five management practices were selected foryields;3)to project future climate change impacts on Y,CWU,andrainfed and irrigated condition,respectively.The practices wereblue and green WF of rainfed and irrigated barley:and to address theploughing,fertilizer application,irigation,planting operation and har-most influential bio-physical factors affecting the likely changes;and 4)vest and kill operations.Two options are available for application of irri-to adjust the results to local water stress conditionsgation water and timing of fertilizer application:user specified andautomatic.In the automatic option,an irrigation event is triggered by2.Methods and datawater stress threshold and fertilizer is applied based on nutrients stressfactor.The total amount of fertilizer applied during the growing season2.1.Study areais specified by the user.More details are given by Neitsch et al.(2011).Alberta is a semi-arid western province in Canada with an area span-2.3.Input data,model development and parameterizationning 661,000 km(Fig.1a,b).It has a highly variable climate with meanannual precipitation ranging from 280 mminthe south to 1000 mmIn this study,ArcSWAT 2012 (Rev.632)was used to set up the model.at the higher elevations of the Rocky Mountains(Masud et al.,2015).A hydrology model of the province,developed and calibrated in an