Spatial distribution of water supply in the coterminous united ststes - PDF document

SPATIALDISTRIBUTIONOFWATERSUPPLYINTHECOTERMINOUSUNITEDSTATESThomasC.Brown,MichaelT.Hobbins,andJorgeA.RamirezABSTRACT:Availablewatersupplyacrossthecontiguous48stateswasestimatedasprecipitationminusevapotranspirationusingdatafortheperiod1953-1994.PrecipitationestimatesweretakenfromthePara-meter-ElevationRegressionsonIndependentSlopesModel(PRISM).Evapotranspirationwasestimatedusingtwomodels,theAdvection-AriditymodelandtheZhangmodel.Theevapotranspirationmodelswerecali-bratedusingprecipitationandrunoffdatafor655hydrologicallyundisturbedbasins,andthentestedusingestimatesofnaturalrunoffforthe18waterresourceregions(WRR)ofthe48contiguousstates.TheÞnalwatersupplycoveragereßectsamixtureofoutputsfromthetwoevapotranspirationmodels.Political,administrative,andlandcoverboundariesweremappedoverthecoverageofmeanannualwatersupply.Acrosstheentirestudyarea,weÞndthat53%ofthewatersupplyoriginatesonforestedland,whichcoversonly29%ofthesurfacearea,andthat24%originatesonfederallands,including18%onnationalforestsandgrasslandsalone.ForestsandfederallandsareevenmoreimportantintheWest(the11westerncon-tiguousstates),where65%ofthewatersupplyoriginatesonforestedlandand66%onfederallands,withnationalforestsandgrasslandscontributing51%.KEYTERMS:watersupply;evapotranspiration;runoff;forests.) JOURNALOFTHEAMERICANWATERRESOURCESASSOCIATIONVol.44,No.6AMERICANWATERRESOURCESASSOCIATIONDecember2008 haslargelyescapedourgraspuntilitfallsagainelsewhereasprecipitation.WhatremainsisÐuntilitreachestheseaÐavailableforusebyhumansandotherspecies,andinabroadsenseisourfreshwatersupply.Asmuchofourwatersupplyoriginatesinforests,waterrunoffanditsqualityhavelongbeenafocusofforestmanagement.TheÞrstnationalforestpreserveswerespeciÞcallysetasidebyCongressintheOrganicActof1897fortheprotectionofwaterandtimbersupplies.Althoughnationalforestscoveronly8%ofthelandareainthecontiguous48states,theyareofparticularinterestbecausetheyaregenerallylocatedatbasinheadwaters.Variousestimatesofthepropor-tionofthenationÕswatersupplyoriginatingonnationalforests,andonforestsingeneral,havebeensuggestedovertheyears.Forexample,a1970reviewofpubliclandsestimatedthatinthe11contiguouswesternstates61%ofthenaturalrunofforiginatedonfederallands,with54%comingfromthenationalforests(PublicLandLawReviewCommission,1970),andmorerecentlyaForestServicereportestimatedthat14%ofthewatersupplyofthe48statesand33%ofthewatersupplyoftheWestoriginatedonnationalforests(Sedelletal.,2000).Usingnewerdataandmod-els,weaimedtoprovideanimprovedsetofestimates.Theapproachtakenherewastoestimatewatersupplyatitssourceasprecipitationminusevapo-transpirationforeachcellofagridcoveringthecon-tiguous48states(AlaskaandHawaiiwereexcludedbecauseoflackofdata).Overlayingpoliticalandlandownershipboundariesorlandcoverdelineationsthenallowsestimationoftheamountofwateroriginatingwithindistinctlandunits.Wetookourprecipitationestimatesfromawell-establishedmodel.Estimatingevapotranspirationwasmorechallenging,andweexaminedtwoquitediffer-entmodels.FindingthatneithermodelestimatedevapotranspirationwithsufÞcientaccuracyinalllocations,andfocusingclearlyonourverypracticalobjectiveofestimatingmeanannualcontributiontowatersupply,wecalibratedeachmodeltoimprovetheaccuracyoftheestimates,andthenusedoneortheothermodelinspeciÞcbasinswherefurthertestingindicatedthatityieldedthemoreaccurateestimate.Nextweexplainthemethodsinsomedetail.METHODSOurbasicapproachwastoestimatewatersupply)asprecipitation()minusactualevapotranspira-tion()onameanannualbasisateach55-kmgridcellacrosstheconterminousU.S.Havingthesespatiallydistributedestimatesofwatersupplyatitssource,boundarieswerethenoverlaid.Aggregatingestimatesofacrosscellswithinaboundaryindicatestheamountofwatersup-plyoriginatingwithinthedesignatedarea.Thelogicofestimatingmeanannualreliesonassumptionsinvolvingbasinstorage.MorespeciÞcally,whererepresentschangesinstorageandsub-scriptssandgrefertosurfaceandsubsurfacedomains.Assumingastableclimateandanundis-turbedbasin,thelong-termaverageannualnetchangeinoverallbasinmoisturestorageisnegligi-bleandassumedtobe0,sodevolvestoEquationPrecipitationisthemostcommonlymeasuredhydrologicquantity,butbecauseprecipitationisspa-tiallyvariable,especiallyinmountainousregions,modelsareneededforÞllinginbetweenthelocationsofprecipitationgages.Forthispurposeweselectedthemonthlyprecipitationestimatesproducedata5-kmlevelofresolutionbytheParameter-ElevationRegressionsonIndependentSlopesModel(PRISM)etal.,1994).PRISMcombinesclimatologicandtopographicdatatointerpolatebetweenprecipitationobservations.wesetouttoadapttheAdvection-Aridity(AA)model(BrutsaertandStricker,1979)fornation-wideuse.AsubsidiaryaimwastodevelopamodelforestimatingatthemonthlytimescalethatcouldbeappliedoverthecontiguousU.S.,whichcouldthenbeusedinotheranalysesthatarenotatissuehere.TheAAmodelisbasedonthecomplemen-taryrelationshipofactualandpotentialevapotrans-piration,describedbelow,andrelieslargelyonreadilyavailableweatherdata.Unfortunately,theAAmodelwasnotableaccu-ratelytoestimateforsomeareasoftheU.S.,largelybecauseofinsufÞcientdata.InlocationswheretheAAmodelperformedpoorlyweusedanothermodel,oneproposedbyZhangetal.(2001).TheZhangmodelemploysaverydifferenttheoreti-calapproachfromthatofthecomplementaryrela-tionshipandestimatesonameanannualbasisonly.ThelossofseasonalitywiththeZhangmodelwasnotashortcominginthecurrentcontext,asourobjectivehereistoestimatemeanannualwatersupply.Bothmodelsaredescribedbrießybelow,asaremethodsformodelcalibrationandtesting.ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1475JAWRA TheAAModelforEstimatingETInhydrologyithasbeencustomarytomodelasafunctionofpotentialevapotranspiration(andsoilmoisture,whereistheevapotrans-pirationthatwouldtheoreticallyoccurwerewaternotlimiting.Inthisapproach,isconsideredanindependentclimaticvariable,constrainingtheamountofbutremainingunaffectedbytheamountofandthuslargelyconstantoverrangesofactualmoistureavailabilityatagivenlocation.However,asproposedbyBouchet(1963),isnotunaffectedby.Rather,whenlackofsoilmoisturekeepsbelowthelevelthatavailableenergywouldsupport(i.e.,),thesurplusenergyheatsanddriesthesurroundingair,raisingabovewhatitwouldbeifmoremoisturewereavailableandweregreater.Whenmoistureisnotlimiting,equalsataratereferredtoaswetenvironmentevapotranspiration().However,asBouchetproposed,atregionalscalesandawayfromsharpenvironmentaldiscontinuities,ifmoistureislimitingandtheavailableenergyisconstant,energynotusedcausestoriseabovebytheamountfallsbelowitFigure1illustratesthecomplementaryrelationship.Complementaryrelationshipmodelsrelylargelyonatmosphericobservations.Themodelsavoidtheneedfordataonsoilmoisture,stomatalresistanceproper-tiesofthevegetation,orothersurfaceariditymea-suresbecauseverticalgradientsoftemperatureandhumidityintheatmosphericboundarylayerreßectthepresenceofmoistureavailabilityatthesurface.Modelsbasedonthecomplementaryrelationshiphypothesishavebeensuccessfullyusedtomakepre-dictionsofregionalatdifferenttemporalscales,therebyprovidingindirectevidenceofitsvalidity(BrutsaertandStricker,1979;Morton,1983;Hobbinsetal.,2001a;Szilagyi,2001).Furthermore,thecomplementaryrelationshiphasrecentlybeendemonstratedobservationallywithdatafromasetofsitesacrossthefreeze-freezoneoftheU.S.usingannualpanevaporationmeasurestoestimatedenotedhereas,andawaterbalanceapproachtoestimate,denotedhereas(Hobbinsetal.2004b).Thewaterbalanceapproachsimplycomputesasprecipitation()minusbasinwatersupplyestimatedasstreamßow()fromthewatershedssurroundingthepansPlotsofannualmatchedcloselythelinesforinFigure1(Hobbinsetal.2004b;Ramirezetal.,2005).Severalcomplementaryrelationshipmodelsarenowavailable,themostwidelyknownofwhicharetheComplementaryRelationshipArealEvapotrans-piration(CRAE)model(Morton,1983)andtheAAmodel(BrutsaertandStricker,1979).Hobbinsetal.(2001b)comparedthesetwomodelsforabilitytoestimateacrosstheU.S.andselectedtheAAmodelasthemostpromisingapproachforsuchanapplication.Inthismodel,iscalculatedbasedonderivationsoftheconceptofequilibriumevapotranspi-rationunderconditionsofminimaladvection,ÞrstproposedbyPriestleyandTaylor(1972),andcalculatedbycombininginformationfromtheenergybudgetandwatervaportransferinthePenmanequation(Penman,1948)[seeHobbinsetal.fordetailsonthemodiÞedAAmodelthatweemployed].DatafortheAAModel.TheAAmodelofrequiresdataontemperature,humidity,solarradiation,windspeed,albedo,elevation,andlatitude.Mostoftheavailableclimatologicaldataonthesevariablesarepointvaluescollectedatweatherstations.Lackingmoresophisticatedmodels,likethatavailableforprecipitation,forgeneratingspatiallydistributedestimatesof(i.e.,estimatesforeachgridcell),weusedspatialinterpolationtechniquesontheinputvariablesandthenappliedtheAAmodelateachcell(wecouldnotinterpolatefromstationestimatesofbecausetheinputdatastationnetworkswerenotcoincident).Krigingwastheprioripreferenceforspatialinterpolationoftheclimatologicalinputswhosestationnetworkswouldsupporttheinherentsemivariogramestimationprocedure(OliverandWebster,1990).Otherwise,aninverse-distance-weightedschemewasused. FIGURE1.SchematicRepresentationoftheComplementaryRelationship.JAWRA1476OURNALOFTHE Thepointdataweregatheredfromvarioussourcessoastocreatemonthlytimeseriesfortheperiod1953-1994.Thetimeperiodwasconstrainedateitherendbytheavailabilityofsolarradiationdata.ThedatausedfortheAAmodelareasfollows:1.Temperaturewasestimatedasthemeanoftheaveragemonthlymaximumandaveragemonthlyminimumtemperaturesdrawnfrom9,271sta-tionsintheÔÔNCDCSummaryoftheDayÕÕ(EarthInfo,1998a;NCDC,2004b)database.2.Humiditydata,intheformofdew-pointtemper-atures,weredrawnfrom324stationsintheÔÔSolarandMeteorologicalSurfaceObservationNetwork(SAMSON)ÕÕ(NCDC,1993)andÔÔNCDCSurfaceAirwaysÕÕ(EarthInfo,1998b;NCDC,2004a)databases.3.Solarradiationwasderivedfromthesumofdif-fuseradiationanddirectradiationcorrectedfortheeffectsoflocalslopeandaspect(Hobbinsetal.,2004a).Solarradiationdataweredrawnfrom258stationsintheSOLMET(NCDC,2001),SAMSON(NCDC,1993),andÔÔNCDCAirwaysSolarRadiationÕÕ(EarthInfo,1998b;NCDC,2004a)databases.4.Windspeeddataweredrawnfrom568stationsintheSAMSON(NCDC,1993)andÔÔNCDCSur-faceAirwaysÕÕ(EarthInfo,1998b;NCDC,2004a)databases.5.AveragemonthlyalbedosurfaceswerebasedonAdvancedVeryHighResolutionRadiometer(AVHRR)datafromGutman(1988).TheAVHRR-derivedalbedoestimateshaveanorigi-nalspatialresolutionofabout15km.6.Elevationdatawereextractedfroma30-arc-secondDEM.Unfortunately,thedensityofstationsislowerintheinteriorWest,especiallyforhumidity,solarradia-tion,andwindspeed,which,giventheWestÕsextremetopographicheterogeneity,ispreciselytheregionwhereapplicationofacomplementaryrela-tionshipmodelwouldrequirethedensestsetofdatapoints.TheimpactofthisdifÞcultyisseenbelow.ModelTestingandCalibration.TheAAmodelwastestedbycomparingmeanannualfromthemodel()towaterbalance-derivedestimatesofactualevapotranspiration()computedasinEqua-tion(4)for655minimallyimpactedbasinsincludedintheHydro-ClimaticDataNetwork(HCDN)(SlackandLandwehr,1992;HydrosphereDataProducts,1996)(Figure2).StreamßowdatawereextractedforU.S.GeologicalSurvey(USGS)gagesattheoutletsofthe655basins.Precipitationinthebasinswasesti-matedusingthePRISMmodelasdescribedabove.Relatingrevealedconsiderablevarianceintheestimatesof=0.76,=655),andtendedtounderestimate,especiallyinwetterregions.Becauseofthisproblem,andinlightofourobjectiveofaccuratelyestimating,itwasdecidedtouseatthetest-basinstocalibratethemodel.CalibrationoftheAAmodelfollowedadetailedprocesswherecoefÞcientsofthewindfunctionofthemodel(Hobbinsetal.,2001a)werealteredateachofasubsetofthe655test-basinssoastominimizethedifferencesbetweenmonthly.Thentheresultingmodelsforthetest-basins,orinsomecasescombinationsofthosemodels,wereappliedtosur-roundingwatershedstoÞllintherestofthestudyarea,producingcalibratedestimatesof(seeetal.,2001afordetails).Resultingestimatesofwatersupply()arethenequaltoAplot(notreproducedhere)offromtheregionallycalibratedAAmodelagainstforthetest-basinsshowsthatthepointsareclusteredaboutthe1:1line,withrelativelylittlescatter(=0.88,=655),suggestingthatthe-basedcalibrationoftheAAmodelpredictsfairlywell,withnobiasovertherangeofobservedAsanothermeasureoftheperformanceofthecali-bratedmodel,awaterbalanceclosureerrorwascom-putedforeachtest-basinasfollows: P42i¼1P12j¼1ETaði;jÞETAAaði;jÞ42i¼1P12j¼1Pði;jÞ100ð5Þ FIGURE2.Locationsofthe655Test-BasinsandWaterResourceRegionsintheConterminousU.S.(seeTable1forthekeytotheWRRs).ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1477JAWRA expressesthecumulativeerrorasapercentageofcumulativeprecipitation(refertoyearandmonth,respectively).Acrossthe655test-basinstheis0.15%(median=0.12%,standarddevia-tion=4.9%,minimum=47%,maximum=31%),indicatingaslightoverallunder-estimationofEighty-Þvepercentofthearewithintherangefrom5%to+5%ofmeanannual.Possibleexplanationsofnonzeroare:(1)violationsoftheassumptionsinherentinEquation(3)usedtodevelopmates,perhapsthroughunaccountedground-waterpumping,surface-waterdiversions,ornetground-waterßowoutofthebasin;(2)violationsoftheassumptionofstationarityinclimatologicalforcing;and(3)errorsinducedbyspatialinterpolationoftheclimaticvariables.ModelPerformanceattheWRRLevel.paringmodeledfromthetest-basinsdoesnotconstituteanindependenttestbecausethetest-basinswereusedforcalibratingthemodel.Dataforwaterresourceregions(WRR)(Figure2)allowamoreindependenttest(althoughnotawhollyindependenttest,asthetest-basinscover21%ofthecontiguous48states).TheWRRsweredeÞnedbytheU.S.WaterResourcesCouncil(1978)forthepurposeoflarge-scaleassessmentandplanning.Toassistsuchplan-ning,theUSGSestimatedannualoutßowfromtheWRRsusingstreamßowgagerecordsfortheperiod1951-1983(Graczyketal.,1986).FromtheUSGSestimates,wesubtractedinßowstoWRRs8and15fromupstreambasins,andinßowstoWRRs9and17fromCanada,soastoisolatetheamountofßoworiginatingwithintheU.S.portionsoftherespec-tiveWRRs.WeaveragedtheannualestimatesofoutßowsfortheWRRs()andthenestimatedmeanannualvirginßow()originatingintheWRRsaswhereisgroundwaterdepletion,isreservoirevaporation,isconsumptiveusefromdiversions,representsnetimportstothebasin.weretakenfromFoxworthyandMoody(1986)andwastakenfromPetsch(1985)andMootyandJeffcoat(1986)(becausePetschandMootyandJeffcoatestimatedtransfersfor1973-1982,themeanannualtransferswerenecessarilyassumedtoapplytotheentireperiodofinterest,1953-1983).TheresultingestimatesofmeanannualarelistedinTable1.Estimatesofmeanannualwatersupplyoriginat-ingintheWRRsovertheperiod1953-1983werecom-putedfromthedistributedmonthlyestimatesofbysummingacrossmonthsandacrossthegridcellswithinaWRR(Table1).Figure3presentsthecomparisonofthetwosetsofestimatesofmeanannual(listedintheÞrsttwocolumnsofnumbersinTable1).Theoverall TABLE1.MeanAnnualWaterSupply()OriginatinginWRRsforthePeriod1953-1983,EstimatedFromUSGSGageDataandasUsingtheTwoCalibratedModels.)%DifferenceFromUSGSUSGSAAModelZhangModelAAModelZhangModel1NewEngland10692912Mid-Atlantic1321201153South-Atlantic-Gulf2982732814GreatLakes10796845Ohio1921821796Tennessee606261537UpperMississippi1061041058LowerMississippi10112013018289Souris-Red-Rainy91110301610Missouri901209333311Arkansas-White-Red8799801312Texas-Gulf51424713RioGrande6216259214UpperColorado1847201641515LowerColorado44459232116GreatBasin1131141722517PaciÞcNorthwest32332029718California11210310248states1,8111,8851,7214Note:Bm,billioncubicmeters.JAWRA1478OURNALOFTHE relationshipbetweenthetwosetsisverygood=0.97,=18).However,therearelargediffer-encesforthefourdry,southwesternbasins(WRRs13-16),wheretheAAmodelappearstounderestimate.ThiscomparisonisnotdeÞnitivebecausetheUSGSestimatesarethemselvessubjecttosubstantialerror.However,therelativelackofweatherstationdataintheSouthwestfortheAAmodel,andthescarcityoftest-basinsformodelcalibrationinrange-landanddesertareas(Figure2),suggestthat,intheseareas,improvementsinestimatesofmodel-areneeded.TheZhangModelofMeanAnnualETTheproblemwiththeAAmodelintheSouthwestledustoconsiderthenewlyavailableZhangmodeletal.,2001)forimprovingtheestimates.TheZhangmodelemploysadifferenttheoreticalapproachfromthatoftheAAmodel,andhasamoremodestgoalthanourimplementationoftheAAmodelinthatitestimatesonlymeanannualThemodelusesmeanannualestimatesofalongwithaparameterreßectingthedominantvegetationtypeTheZhangmodelofmeanannual 1þwg1þwgþ whererepresentsarelativemeasureofwateravail-abletoplantsand.FollowingZhangetal.(2001),wesetto2.0forforestcoverand0.5other-wise.WecomputedusingthePriestley-Taylorequation,asinourimplementationoftheAAmodel.MeanannualwasagainestimatedusingPRISM.Vegetationcoverwastakenfromthenationallandcoverdatabase(U.S.GeologicalSurvey,1992),withforestcovercomposedofcoverclasses41(deciduousforest),42(evergreenforest),and43(mixedforest).Estimateswereproducedforeachgridcell,bothforthefullstudyperiod(1953-1994)(Table2)andformatchingtheUSGSdatafortheWRRs(1953-1983). FIGURE3.MeanAnnualWaterSupplyforthe18WRRsoftheConterminousU.S.forthePeriod1953-1983EstimatedFromUSGSData()andastheTwoEvapotranspirationModels( TABLE2.ComparisonofMeanAnnualEvapotranspiration()FromtheMixedModelWithMeanAnnualPrecipitation()andResultingWaterSupply()fortheWRRs,forthePeriod1953-1994.Volume(BmQETPET1NewEngland17076940.810.452Mid-Atlantic2891641251.320.573South-Atlantic-Gulf8866122752.230.694GreatLakes2511521001.530.605Ohio4662801871.500.606Tennessee14582631.300.567UpperMississippi4113021092.780.748LowerMississippi3612361251.880.659Souris-Red-Rainy8070107.150.8810Missouri690591995.990.8611Arkansas-White-Red498409894.610.8212Texas-Gulf364314506.300.8613RioGrande124118717.720.9514UpperColorado11190214.340.8115LowerColorado117111619.380.9516GreatBasin11197147.080.8817PaciÞcNorthwest5882853030.940.4818California239144951.510.6048states5,9134,1441,7692.340.70ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1479JAWRA ModelTestingandCalibration.TheZhangmodelwastestedinasimilarfashiontotheAAmodel,bycomparingwithbasin-derivedesti-matescomputedasinEquation(4)()forthe655test-basins.Comparisonoffromthetwosourcesrevealedconsiderablevarianceintheestimates=0.44,=655).Becauseofthisvariation,modelestimatesofwereadjusted,basedonmeanannualfromthetest-basinsfor1953-1994.Thecalibrationinvolvedseparatingthetest-basinsinto14groupsbasedoninspectionoftherelationshipof,witheachgrouprepresentingasetofUSGS4-digitbasins.Thenwithineachgroup,test-basindatawereusedtocomputecoefÞcientsforadjustingtomorecloselypredictwithinthegroup.Theseadjustmentsforeachgroupwereappliedthroughoutthesetoffour-digitbasinsrepresentedbythegroup.DetailsonthecalibrationareavailableinthediscussionpaperÔÔThesourceofwatersupplyintheUnitedStatesÕÕ(availableathttp://www.fs.fed.us/rm/value/discpapers.html).Whentestedagainstthecalibratingdata()forthe655test-basins,thecalibratedZhangperformedreasonablywell.Plottingagainstthatthepointsareclusteredalongthe1:1linewithnobias,butthescatter(=0.73,=655)isgreaterthanwiththeAAmodel.Intermsofclosureerror(computedasinEquation(5)withreplacing),acrossthe655test-basinsmean0.22%(median=0.08%,standarddeviation=8.1%,minimum=50%,maximum=36%).Only63%ofarewithintherangefrom5%to+5%ofmean,whichagainshowslessprecisionthantheAAmodel.ModelPerformanceattheWRRLevel.withtheAAmodel,dataattheWRRlevelallowedarelativelyindependenttestofthecalibratedZhangmodel.Estimatesofmeanannualwatersupplyorigi-natingintheWRRswerecomputedbysummingacrossthegridcellswithinaWRR.Table1presentstheresultsofthisprocedure,andFigure3showsthecomparisonofmeanannual.Theoverallrelationshipbetweenthetwosetsisverygood(=0.98,=18).Importantly,forthefourproblematicsouthwesternbasins(WRRs13-16),aremuchclosertothanareThecalibratedZhangmodelalsosubstantiallyout-performsthecalibratedAAmodelinWRRs9-12.ModelSelectionAsseeninTable1andFigure3,formostWRRstheestimatefromatleastoneevapotranspirationmodelisclosetotheUSGSestimate.BecausetheAAmodelproducesgenerallysmallerclosureerrorsthantheZhangmodel,theAAmodelistobepreferred,allelseequal.AmongWRRs1-8,17,and18theAAmodeleitherproducestheestimatethatisclosertotheUSGSestimateorifnotthenproducesanesti-matethatisveryclosetotheZhangmodelestimate.However,forWRRs9-16theZhangmodelproducesanestimatethatissubstantiallyclosertotheUSGSestimatethanistheAAmodelestimate.Interest-ingly,theseeightWRRsareamongthedriest,allwithmeanannualprecipitationestimatesoflessthan700mm,andhavesubstantialnonforest(largelyrangeland)cover.Asmentionedearlier,datafortheAAmodelandtest-basinsforitscalibrationarerela-tivelysparseonrangelands.Themodelsdifferconsid-erablyintheirestimatesforrangelandsbutnotforothercovertypes.ApparentlytheZhangmodel,whichattemptsonlytoestimatemeanannual,islesssensitivetodatadensityandlessdependentoncalibration,andthusbetterabletoestimatemeanwhendataforestimationandcalibrationaresparse.Insum,weusedtheZhangmodelforWRRs9-16andtheAAmodelelsewhere.ThiscombinationofthetwomodelsiscalledtheMixedmodel.Wehastentoadd,however,thatourwork,althoughadequateforourpurposes,doesnotprovideadeÞnitivecompari-sonofthemodels.AdministrativeBoundariesandCoverTypeAsasourceofwatersupply,federalownershipwasdistinguishedfromnonfederal(stateandpri-vate)ownership,andÞvecategoriesoffederalown-ership(ForestService,ParkService,BureauofLandManagement,BureauofIndianAffairs,andother)weretracked.ThefederalboundariesweretakenfromtheFederalLandsoftheU.S.database(U.S.GeologicalSurvey,2004).Becauseofthemanysmallunits,thiscoveragewasmodeledata100-mresolution.Thenationallandcoverdatabase(U.S.GeologicalSurvey,1992)wasusedtodistinguishamonglandcovertypes.Fivecoverclasseswereformedfromtheoriginal21landcoverclassesasfollows:forest(coverclasses41,42,and43),rangeland(classes51and71),waterwetland(classes11,12,91,and92),agriculture(classes61,81,82,83,and84),anddevelopedclasses(classes21,22,23,31,32,33,and85).OfÞcialboundariesofsomestatesextendintomajorwaterbodies,suchastheGreatLakes(e.g.,Wisconsin)ormajorbaysandestuaries(e.g.,Wash-ington).InthesecasesweclippedstateboundariesatthewaterÕsedge.JAWRA1480OURNALOFTHE RESULTSDistributedmeanannualovertheperiod1953-1994,atthe5-kmcelllevel,fromtheMixedmodelisshowninFigure4.Intheenergy-limitedeasternstates,theseestimatesshowdecreasingwithenergysupplyasonemovesnorthward,whereasinthemoisture-limitedwesternstatesdecreaseswithasonemoveswestwarduntiltheWestCoastregionisreached,andalsoincreasingwithelevationasdoesFigure5showsthedistributionofmeanannualwatersupply,,computedasinEquation(1)usingestimatesfromtheMixedmodel.AgeneralcomparisonofFigures4and5showsthat,asexpected,usuallyexceeds.IntheEast(WRRs1-8),theoverallratioofis1.8,andalongtheWestCoast(WRRs17and18)theoverallratiois1.1,whereasinthedrierWest(WRRs9-16)theover-allratiois6.1.ContrarytothegeneralÞnding,insomeareas,mostnotablyinportionsofWRRs1and17,butalsoinpartsofWRRs2,5,and18.OvertheentireconterminousU.S.,meanannualareestimatedtobe538and230mmrespectively.AcrosstheWRRs,meanannualrangesfrom271mmyearinWRR16to910mmyearinWRRs3and8,whereasrangesfrom16mmyearinWRR15to595mmyearinWRR6(Figure6).Figure6arrangestheWRRsinorderofincreasingmeanannual,andshowssev-eralinterestingÞndings.First,forthesixWRRswiththelowestnearlymatchessothatismini-mal.Second,asexpected,WRRswithmoretendtohavemorewith=18).Third,intworelativelymoistandcoolWRRs(1and17)exceeds.Fourth,although FIGURE4.MeanAnnualFromtheMixedModelforthePeriod1953-1994. FIGURE5.MeanAnnualWaterSupply()UsingFromtheMixedModel,forthePeriod1953-1994.ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1481JAWRA risessubstantiallyasonemovesfromthedryinter-mountainWesttotheEastandSouth,risingtoover1,300mmyearinthethreesoutheasternWRRs,alsorises,especiallyinthesoutheasternWRRs,sobarelyreaches600mmyeareveninthemostmoistWRRs.Twenty-fourpercentofthenationÕs(48contiguousstatesonly)watersupply(429billionmperyear)originatesonfederalland(Table3).EighteenpercentofthenationÕswatersupplyoriginatesonlandsofthenationalforestsystemalone,althoughthoselandsoccupyonlyabout11%ofthesurfacearea.For-estServiceandParkServicelandsyieldonaverage384mmyear,whiletheotherfederallandsyieldanaverageofonly58mmyearandstateandprivatelandsyieldanaverageof238mmBecausefederallandsareconcentratedinthewes-ternportionofthecountry,thecontributionsoffed-erallandstowatersupplyaregreatestintheWest.WestoftheMississippiRiver,42%ofthewatersup-plyoriginatesonfederalland,with32%onForestServicelandalone,comparedwith9%and7%,respectively,eastoftheMississippi.Or,dividingthecountryintoÞvesections,weseethatinthewesternsection,comprisedofthe11westernstates,66%ofthewatersupplyoriginatesonfederallands,with51%onForestServicelands(althoughForestServicelandsoccupyonlyabout21%ofthesurfacearea)(Table4).TheseestimatesfortheWestaresurpris-inglyclosetothoseoftheFederalLandLawReviewCommissionin1970,whichestimatedthat61%camefromfederallands,with54%fromthenationalfor-ests(U.S.PublicLandLawReviewCommission,1970).DetailedtablesofresultsforWRRsandstatesarepresentedintheAppendix.TheroleoffederallandsdiffersdramaticallyacrosstheWRRs.WRRs1,2,3,5,7,8,and12allreceivelessthan10%oftheirwatersupplyfromfederallands,whereasWRRs14-16receiveatleast80%fromfederallands(TableA1).Theroleoffederallandsalsodiffersgreatlyamongthestates,with24statesreceivinglessthan10%,andsevenstatesreceivingmorethan75%(Arizona,Colorado,Idaho,Montana,Nevada,Utah,andWyo-ming),oftheirwatersupplyfromfederallands(TableA2).Shiftingtocovertype,weÞndthat53%ofthenationÕswatersupplyoriginatesonforestland,26%onagriculturalland,and8%onrangeland(Table5).Forestcontributes65,56,and68%ofthewatersup-plyintheWest,South,andEastsectionsoftheU.S.,respectively(Table6).AgriculturallandismostimportantinthePlainsandMidwest,contributing49and58%ofthewatersupply,respectively.TheroleofforestsvariessubstantiallyacrosstheWRRs,withWRRs9and10receivinglessthan20%oftheirwatersupplyfromforests(TableA3),whereasWRRs1,6,and17receiveatleast70%.Acrossthestates,theroleofforestsvariesaswell,withÞvestatesreceivinglessthan10%oftheirwater FIGURE6.MeanAnnualDepth(mm)byWRR,WithFromtheMixedModel,forthePeriod1953-1994. TABLE3.WaterSupplyoftheConterminousU.S.byLandOwnershipType,forthePeriod1953-1994.YearPercentNFS32618386BLM26137NPS362364BIA16162Otherfederal241141Stateandprivate1,33976238Total1,768100229Note:NFS,NationalForestSystemoftheU.S.ForestService;BLM,U.S.BureauofLandManagement;NPS,NationalParkSer-vice;BIA,BureauofIndianAffairs,U.S.DepartmentofInterior. TABLE4.WaterSupplyoftheConterminousU.S.byLandOwnershipTypeandU.S.Region,forthePeriod1953-1994(BmWestPlainsMidwestSouthEastTotalNFS245.23.418.651.96.6325.7BLM26.00.00.00.00.026.0NPS31.70.20.53.50.436.3BIA11.71.72.10.30.216.0Otherfederal5.22.22.213.31.624.4Stateandprivate163.6109.4278.3548.6239.31339.2Total483.4116.9301.7617.6248.11767.7Notes:West:AZ,CA,CO,ID,MT,NM,NV,OR,UT,WA,WY.Plains:KS,ND,NE,OK,SD,TX.Midwest:IA,IL,IN,MI,MN,MO,OH,WI.South:AL,AR,FL,GA,KY,LA,MS,NC,SC,TN,VA,WV.East:CT,DC,DE,MA,MD,ME,NH,NJ,NY,PA,RI,VT.SeeTable2foragencynames.JAWRA1482OURNALOFTHE supplyfromforests(TableA4)(Iowa,Kansas,NorthDakota,Nebraska,SouthDakota)andsixstatesreceivingmorethan70%(Maine,NewHampshire,Oregon,Vermont,Washington,andWestVirginia).Theseresultsreßectourbasicunderstandingofrespondstoavailablemoistureandenergy.Indryenvironments,itistheavailabilityofwaterthatconstrains.Aswateravailabilityincreasesconvergetowardinverywetenvironments.Hereitistheavailabilityofenergythatdetermines,andinsuchenergy-limitedenvironments,increasingwillnotchangewillincrease(assumingstoragechangesremainnegligible).Increasingenergyinanenergy-limitedenvironmentwilldriveupwardsandconse-quentlydrivedownwards.Overall,theWRRsspanmuchofthecontinuumbetweentheextremesofwaterandenergylimitations.Themostextremesitu-ationsarefoundinportionsofWRRsorduringcer-tainseasons.Forexample,duringwinterintheOlympicRainforestofWRR17,theavailabilityofenergylimitsregardlessoftheabundanceofandthedriestpartsoftheBasinandRangeregionofWRR16arewater-limitedinJuly,whensolittlefallsthatittypicallyevaporatesbeforeitcanwetthesoillongenoughforanothereventtogenerateThereisgrowingevidencethattemperature,pre-cipitation,andothervariablesaffectingwatersupplyintheU.S.arechanginginresponsetonaturalandanthropogenicforcings(Seageretal.,2007;Barnettetal.,2008;KrakauerandFung,2008).Thesechangessuggestthatourestimatesofmeanannualwatersupplymayalreadybeslightlyoutofdate,andthat,ifthesehydro-meteorologicaltrendspersist,theestimatesofwatersupplyshould,atsomepointinthefuture,berevisitedinlightofnewdata.SUMMARYWehaveestimatedwatersupplyasprecipitationminusevapotranspirationateach5-kmby5-kmcellacrosstheconterminousU.S.Weusedtwoevapo-transpirationmodels(theAAandZhangmodels)andcomparedourresultswithindependentestimatesfortheWRRs.Basedonthiscomparison,weselectedresultsfromtheAAmodelfor10oftheWRRsandfromtheZhangmodelintheothereightWRRs.Overlayinglandownershipandlandcovertypeboundariesonthewatersupplycoverageallowedestimationofthequantityofwateroriginatingindif-ferentlandareasofinterest.ThisprocedurehasresultedinwhatarelikelythemostaccurateestimatesavailablesofarofoverallU.S.watersupply.Theestimatesshowthatforestsarethesourceof53%oftheU.S.watersupply,andoffullytwo-thirdsofthewatersupplyintheWestandSouth.Further,nationalforestsandgrasslandsarethesourceof18%ofthewatersupplynationwideand51%ofthewatersupplyintheWest,withallotherfederallandscontributinganother6%and15%,respectively.Becauseforestsarealsogenerallythesourceofthehighestqualityrunoff(Brown&Binkley,1994),itisnotanexaggerationtosaythatforestsplayanextremelyimportantroleintheprovisionofwaterintheU.S.Itmayalsobesaidthatweareindebtedtothoseofourforbearswhosetasidesomuchforestandotherlandinthepublicdomain,therebyhelpingtomaintainthequalityofmuchofourwaterrunoff,especiallyintheWest.Asprivatelandscontinuetobedeveloped,publicandotherprotectedlandswillgrowinimportanceassourcesofhighqualityrunoff. TABLE5.WaterSupplyoftheConterminousU.S.byLandCoverClass,forthePeriod1953-1994.YearPercentForest93153417Rangeland146855Agriculture46126225wetland1378307Other935284Total1,768100229 TABLE6.WaterSupplyoftheConterminousU.S.byLandCoverTypeandU.S.Region,forthePeriod1953-1994(BmWestPlainsMidwestSouthEastTotalForest313.019.983.0346.2169.2931.2Rangeland111.625.83.84.40.3145.9Agriculture23.657.3173.8163.642.9461.2wetland7.28.929.673.417.5136.5Other28.05.111.530.118.292.9Total483.4116.9301.7617.6248.11,767.7Note:ForregionaldeÞnitions,seeTable4.ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1483JAWRA APPENDIXThefollowingfourtablesofmeanannualwatersupply()estimatesarebasedontheMixedmodelfortheperiod1953-1994.TABLEA1.WaterSupply(Mmyear)oftheConterminousU.S.byLandOwnership.WRRNFSBLMNPSBIAOtherFederalState&PrivateTotal13,8890807944389,14893,64024,70307380993118,101124,534315,90104541376,369251,861274,72148,717013496069989,01099,520514,9140556682,048168,966186,551612,35101,8821701,93946,65762,99972,3600150592771104,932108,80485,49905512,468117,344125,36791,2880109757737,5339,7601017,7911,9314,5373,10472170,55098,635118,999772608672,00776,57488,783122,2962138387646,60549,920133,98226332283752,0096,6431414,6522,225185311513,38320,806153,037476119949371,1205,738168,1212,77284331182,60513,73417154,59112,95420,7486,7243,024105,113303,1541842,6015,3136,0669781,71238,53695,207Total325,69126,01336,32616,01824,4231,340,0451,768,516Note:Mm,millioncubicmeters.TABLEA2.WaterSupply(Mmyear)oftheConterminousU.S.byStateandLandOwnership.StateNFSBLMNPSBIAOtherFederalState&PrivateTotalAL2,852045090266,61570,414AR6,170012801,26351,32358,884AZ2,51721393949359024,709CA43,3175,0965,8789781,56836,21093,047CO15,3841,509478107755,00622,559CT0000107,6837,693DC006037281DE0000391,9501,989FL2,14903031031,98239,93044,467GA5,32905301,67652,87059,928IA001415331,14131,299ID41,3723,4981311,75529714,01161,064IL1,45100027238,34740,070IN1,00509037830,34531,736KS0079221816,14716,464KY4,1830204087444,12349,384LA1,71303811,37952,48555,616MA2022014211,92212,088MD008001469,3939,619ME1990657918450,34350,870MI6,56907628939239,56046,886MN2,50501731,14516423,83927,826MO3,4900130045846,64250,720MS5,303073582859,06065,232MT29,8051,0844,0572,85523410,58448,620NC8,82501,0311701,18043,12354,329ND841378513,8714,088NE47081555111,23911,500JAWRA1484OURNALOFTHE TABLEA2.(Continued)StateNFSBLMNPSBIAOtherFederalState&PrivateTotalNH2,7150003511,93214,682NJ009302409,3439,676NM2,46828611365702,2875,486NV2,1592,69871462435815,798NY3001516442064,44165,070OH1,432050013235,13036,744OK7140786486621,31823,769OR43,0169,2124741,04930243,72797,779PA1,783084024559,24961,362RI0000101,8281,838SC2,034034062125,63428,323SD146212502264,3285,015TN2,94901,03601,86455,96061,809TX2,373015731,00052,49656,029UT6,90380134133272,0409,936VA5,3980517056731,67338,156VT1,9150009711,15013,162WA46,9504215,9632,8912,22745,080113,151WI2,17708566822033,25736,407WV4,9900132013625,82431,081WY11,2701,5784,5425391273,16821,224Total325,69326,01936,34216,01924,4281,339,1801,767,681TABLEA3.WaterSupply(Mmyear)oftheConterminousU.S.byWRRandLandCoverClass.WRRForestRangelandAgricultureWaterWetlandOtherTotal171,0883215,3059,2037,72393,640278,900029,1016,8389,695124,5343155,6003,93357,78239,19518,212274,721443,9471,15434,92415,5993,89699,5205106,02515969,7273,8646,776186,551644,728014,0281,9252,31862,999725,0601,70969,1398,5094,388108,804842,28426253,40725,6693,744125,36791,0084344,2843,8971379,7601017,55827,93446,0433,1753,92598,6351133,55210,89535,7575,8432,73788,7831212,1798,22020,1615,7233,63749,920133,0333,43819161376,643149,02010,36624224193620,806152,3133,361116475,738164,3998,45536321530113,73417226,85837,10113,4715,38020,345303,1541856,44628,3895,9357173,71995,207Total933,998146,131459,699136,01692,6731,768,516TABLEA4.WaterSupply(Mmyear)oftheConterminousU.S.byStateandLandCoverClass.StateForestRangelandAgricultureWaterWetlandOtherTotalAL47,321014,5455,5383,01070,414AR27,2456924,1486,1191,30358,884AZ1,9272,73455374,709CA53,57428,5876,6647843,43893,047CO9,45311,3015082241,07322,559CT4,46607667711,6917,693DC180006381DE52909763771061,989FL15,0943,9477,12412,8225,48144,467ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1485JAWRA TABLEA4.(Continued)StateForestRangelandAgricultureWaterWetlandOtherTotalGA35,95511613,2526,1334,47259,928IA2,3931,77225,32786394531,299ID38,18215,7563,5068632,75861,064IL5,69132629,8642,0052,18440,070IN6,44213522,8879991,27331,736KS3785,2839,83457939016,464KY29,720016,7281,6131,32349,384LA18,21727817,18217,5062,43455,616MA7,375166431,5082,54612,088MD4,08703,7329718299,619ME40,5692912,4115,3182,28150,870MI19,7761,00414,7239,7011,68146,886MN6,34620312,4777,99580527,826MO14,84220931,8522,2741,54250,720MS32,5121021,8858,5062,32065,232MT27,80515,1672,5095462,59348,620NC34,087011,5406,1222,58054,329ND99512,899200304,088NE163,7697,17836617111,500NH12,07207681,10773514,682NJ4,21401,8521,3832,2269,676NM2,5372,8267411385,486NV9984,58234351495,798NY43,133014,2403,9443,75465,070OH12,748820,6911,0822,21436,744OK6,3885,8879,4291,36170423,769OR78,2937,8996,6799773,93197,779PA41,310015,7071,1073,23761,362RI1,1980532423451,838SC15,67106,3364,2752,04128,323SD519233,736266405,015TN38,059018,3922,6632,69561,809TX13,0238,94524,2026,1013,75856,029UT4,2615,0611781113259,936VA25,74308,8171,8761,71938,156VT10,203131,79977337413,162WA87,8857,6953,0322,89811,641113,151WI14,76015515,9444,65988936,407WV26,54803,63919570031,081WY8,07710,0093917302,01721,224Total931,202145,925461,156136,50492,8931,767,681LITERATURECITEDBarnett,T.P.,D.W.Pierce,H.G.Hidalgo,C.BonÞls,B.D.Santer,T.Das,G.Bala,A.W.Wood,T.Nozawa,A.A.Mirin,D.R.CayanandM.D.Dettinger,2008.Human-InducedChangesintheHydrologyoftheWesternUnitedStates.Science319(5866):Bouchet,R.J.,1963.EvapotranspirationReelleEvapotranspirationPotentielle,SigniÞcationClimatique.InternationalAssociationofScientiÞcHydrology,Berkeley,California.Brown,T.C.andD.Binkley,1994.EffectofManagementonWaterQualityinNorthAmericanForests.GeneralTechnicalReportRM-248.RockyMountainForestandRangeExperimentSta-tion,FortCollins,Colorado.Brutsaert,W.andH.Stricker,1979.AnAdvection-AridityApproachtoEstimateActualRegionalEvapotranspiration.WaterResourcesResearch15(2):443-450.Daly,C.,R.P.Neilson,andD.L.Phillips,1994.AStatistical-TopographicModelforMappingClimatologicalPrecipitationoverMountainousTerrain.JournalofAppliedMeteorology33(2):140-158.EarthInfo,1998a.NCDCSummaryoftheDay[TD-3200ComputerFile].EarthInfo,Inc.,Boulder,Colorado.EarthInfo,1998b.NCDCSurfaceAirways[TD-3280ComputerFile].EarthInfo,Inc.,Boulder,Colorado.Foxworthy,B.L.andD.W.Moody,1986.NationalPerspectiveonSurface-WaterResources.NationalWaterSummary1985ÐHydrologicEventsandSurface-WaterResources,WaterSupplyPaper2300,D.W.Moody,E.B.Chase,andD.A.Aronson(Editors).U.S.GeologicalSurvey,Washington,D.C.,pp.51-68.Graczyk,D.J.,W.R.Krug,andW.A.Gebert,1986.AHistoryofAnnualStreamßowsfromthe21Water-ResourceRegionsintheUnitedStatesandPuertoRico,1951-83.Open-FileReport86-128.U.S.GeologicalSurvey,Madison,Wisconsin.JAWRA1486OURNALOFTHE Gutman,G.,1988.ASimpleMethodforEstimatingMonthlyMeanAlbedofromAVHRRData.JournalofAppliedMeteorology27(9):973-988.Hobbins,M.T.,J.A.Ramirez,andT.C.Brown,2001a.TheComple-mentaryRelationshipinEstimationofRegionalEvapotranspi-ration:AnEnhancedAdvection-AridityModel.WaterResourcesResearch37(5):1389-1403.Hobbins,M.T.,J.A.Ramirez,andT.C.Brown,2004a.DevelopingaLong-Term,High-Resolution,Continental-Scale,SpatiallyDis-tributedTime-SeriesofTopographicallyCorrectedSolarRadia-tion.ProceedingsoftheTwenty-fourthAnnualAmericanGeophysicalUnionHydrologyDays,ColoradoStateUniversity,FortCollins,Colorado.Hobbins,M.T.,J.A.Ramirez,andT.C.Brown,2004b.TrendsinPanEvaporationandActualEvaporationAcrosstheContermi-nousU.S.:ParadoxicalorComplementary?GeophysicalResearchLetters31(13):L13503.Hobbins,M.T.,J.A.Ramirez,T.C.Brown,andL.H.J.M.Claessens,2001b.TheComplementaryRelationshipinEstimationofRegio-nalEvapotranspiration:TheComplementaryRelationshipArealEvapotranspirationandAdvection-AridityModels.WaterResourcesResearch37(5):1367-1387.HydrosphereDataProducts,1996.HydrodataUSGSDailyValues,Volume8.0.HydrosphereDataProducts,Inc.,Boulder,Colo-Krakauer,N.Y.andI.Fung,2008.IsStreamßowIncreasing?TrendsintheCoterminousUnitedStates.HydrologyandEarthSystemSciencesDiscussions5(2):785-810.Mooty,W.B.andH.H.Jeffcoat,1986.InventoryofInterbasinTransfersofWaterintheEasternUnitedStates.Open-FileReport86-148.U.S.GeologicalSurvey,Tuscaloosa,Alabama.Morton,F.I.,1983.OperationalEstimatesofArealEvapotranspira-tionandTheirSigniÞcancetotheScienceandPracticeofHydrology.JournalofHydrology66:1-76.NationalClimaticDataCenter(NCDC),1993.SolarandMeteoro-logicalSurfaceObservationNetwork1961-1990(CD-ROM),Ver-sion1.0.NationalClimaticDataCenter,Asheville,NorthNationalClimaticDataCenter(NCDC),2001.Solmet[TD9724ComputerFile]SolarandSurfaceObservations,Daily1952-1976,SolmetVolume1-UserÕsManualTD-9724.NationalCli-maticDataCenter,Boulder,Colorado.NationalClimaticDataCenter(NCDC),2004a.SurfaceAirways,TD3280.NationalClimaticDataCenter,Asheville,NorthCaro-NationalClimaticDataCenter(NCDC),2004b.SurfaceLandDailyCooperativeSummaryoftheDay,TD-3200.NationalClimaticDataCenter.,Asheville,NorthCarolina.Oliver,M.A.andR.Webster,1990.Kriging:AMethodofInterpolationforGeographicalInformationSystems.InternationalJournalofGeographicalInformationScience4(3):313-332.Penman,H.L.,1948.NaturalEvaporationFromOpenWater,BareSoilandGrass.ProceedingsoftheRoyalSocietyofLondon,Petsch,H.E.Jr.,1985.InventoryofInterbasinTransfersofWaterintheWesternConterminousUnitedStates.Open-FileReport85-166.U.S.GeologicalSurvey,Lakewood,Colorado.Priestley,C.H.B.andR.J.Taylor,1972.OntheAssessmentofSur-faceHeatFluxandEvaporationUsingLarge-ScaleParameters.MonthlyWeatherReview100:81-92.PublicLandLawReviewCommission,1970.OneThirdoftheNationÕsLand:AReporttothePresidentandtotheCongressbythePublicLandLawReviewCommission.PublicLandLawReviewCommission,Washington,D.C.Ramirez,J.A.,M.T.Hobbins,andT.C.Brown,2005.ObservationalEvidenceoftheComplementaryRelationshipinRegionalEvap-orationLendsStrongSupportforBouchetÕsHypothesis.Geo-physicalResearchLetters32(15):L15401.Seager,R.,M.F.Ting,I.Held,Y.Kushnir,J.Lu,G.Vecchi,H-P.Huang,N.Harnik,A.Leetmaa,N-C.Lau,C.Li,J.VelezandN.Naik,2007.ModelProjectionsofanImminentTransitiontoaMoreAridClimateinSouthwesternNorthAmerica.Science316(5828):1181-1184.Sedell,J.,M.Sharpe,D.D.Apple,M.Copenhagen,andM.Furniss,2000.WaterandtheForestService.U.S.ForestService,Wash-ington,D.C.Slack,J.R.andJ.M.Landwehr,1992.Hydro-ClimaticDataNet-work(HCDN):AU.S.GeologicalSurveyStreamßowDataSetfortheUnitedStatesfortheStudyofClimateVariations,1874-1988.Open-FileReport92-129.U.S.GeologicSurvey,Reston,Szilagyi,J.,2001.OnBouchetÕsComplementaryHypothesis.Jour-nalofHydrology246(1-4):155-158.U.S.GeologicalSurvey,1992.NationalLandCoverDataSet(Updated).U.S.DepartmentofInterior,Reston,Virginia(http://landcover.usgs.gov/natlandcover.asp).U.S.GeologicalSurvey,2004.FederalLandsoftheUnitedStates:NationalAtlasoftheUnitedStates.200412.U.S.DepartmentofInterior,Reston,Virginia(http://nationalatlas.gov/atlasftp.U.S.WaterResourcesCouncil,1978.TheNationÕsWaterResources1975-2000.GovernmentPrintingOfÞce,Washington,D.C.Zhang,L.,W.R.Dawes,andG.R.Walker,2001.ResponseofMeanAnnualEvapotranspirationtoVegetationChangesatCatch-mentScale.WaterResourcesResearch37(3):701-708.ISTRIBUTIONOFATERUPPLYINTHEOURNALOFTHE1487JAWRA