文档介绍：该【Watershed‐Scale Effective Hydraulic Properties of the Continental United States 2021 Arik Tashie 】是由【贾敬】上传分享，文档一共【18】页，该文档可以免费在线阅读，需要了解更多关于【Watershed‐Scale Effective Hydraulic Properties of the Continental United States 2021 Arik Tashie 】的内容，可以使用淘豆网的站内搜索功能，选择自己适合的文档，以下文字是截取该文章内的部分文字，如需要获得完整电子版，请下载此文档到您的设备，方便您编辑和打印。:.
RESEARCH ARTICLEWatershed-Scale Effective Hydraulic Properties of the
United States
Key Points:Arik Tashie1 , Tamlin Pavelsky1 , Lawrence Band2,3 , and Simon Topp1
•Provideempiricalestimatesof
watershed-scaleeffectivehydraulic1DepartmentofGeologicalSciences,UniversityofNorthCarolinaatChapelHill,ChapelHil,NC,USA,2Department
variablesforthecontinentalUnitedofEnvironmentalSciences,UniversityofVirginia,Charlottesville,VA,USA,3DepartmentofEngineeringSystemsand
Statesforthefirsttime
•Watershed-scaleeffectivehydraulicEnvironment,UniversityofVirginia,Charlottesville,VA,USA
conductivitiesareorder(s)of
magnitudehigherthansoil-texture-
basedestimatesofaveragehydraulicAbstractInlandsurfacemodels(LSMs),thehydraulicpropertiesofthesubsurfacearecommonly
conductivitiesestimatedaccordingtothetextureofsoilsattheEarth',
•Maximumdrainablestorage
volumesmaybeusedtorealisticallyfractureflow,heterogeneity,andtheeffectsofvariabledistributionofwaterinthesubsurfaceoneffective
constrainmodelsandreducemodelwatershed-,weempiricallyconstrain
spin-uptimeestimatesofwatershed-scaleeffectivehydraulicconductivities(K)andeffectivedrainableaquiferstorages
(S)ofallreferencewatershedsintheconterminousUnitedStatesforwhichsufficientstreamflowdataare
Supporting Information:available(n = 1,561).Then,weusemachinelearningmethodstomodelthesepropertiesacrosstheentire
(K)(r2 > ;
1% < bias < 9%)andreasonableconfidenceforS(r > ;−70% < bias < −18%).Ourestimatesof
effectiveKare,onaverage,twoordersofmagnitudehigherthancomparablesoil-texture-basedestimates
Correspondence to:ofaverageK,confirmingtheimportanceofsoilstructureandpreferentialflowpathwaysatthewatershed
,
******@
andarespatiallyheterogeneous(5–3,355 mm).BecauseestimatesofSaremuchlowerthantheglobal
Citation:maximumsgenerallyusedinLSMs(.,5,000 mminNoah-MP),theymayservebothtolimitmodel
Tashie,A.,Pavelsky,T.,Band,L.,spin-
&Topp,S.(2021).Watershed-scaleattempttoconstrainestimatesofwatershed-scaleeffectivehydraulicvariablesthatarenecessaryforthe
effectivehydraulicpropertiesoftheimplementationofLSMsfortheentireconterminousUnitedStates.
AdvancesinModelingEarthSystems,
13,:// Language SummaryModelingtheflowofwateroutofthegroundandintostreams
org/,includinghowmuchwater
isavailabletosupportstreamflow(.,thevolumeofmobilegroundwater)andhowquicklythatwater
Received17DEC2020mayflowthroughthematerialswhereitisstored(.,effectivehydraulicconductivity).Currently,most
Accepted11MAY2021
large-
areknowntobepronetoerrorinmanylandscapeswherebedrockfracturesorbiologicalactivity(.,
earthwormtunnelsorrottedtreeroots)arecommon,asthesestructuralfeaturesmayaltertheeffective
,weprovide
empiricalestimatesofthesepropertiesthroughanalysisofhistoricalstreamflowandmakethesedata
availableatthecontinentalscaleforthefirsttime.
1. Introduction
Regionalhydrologicmodelsandlandsurfacemodels(LSMs)arepowerfultoolsforinvestigatingthedistri-
butionofwaterattheEarth',while
modelsofthistypeoftenincorporatemultiplelayersofsoilstructureandtherelativelycomputationally
©'sequationtorepresenttheflowofwaterintheshallowsubsurface,groundwaterand
AdvancesinModelingEarthSystemsbaseflowprocessesremainpoorlyconstrained(Clarket al. 2008, 2015;Fanet al., 2019;Fatichiet al., 2020).
publishedbyWileyPeriodicalsLLConBecausegroundwatersustainshalfoftheglobalstreamflowatannualtimescales(Becket al., 2013),buffers
behalfofAmericanGeophysicalUnion.
Thisisanopenaccessarticleunderstreamflowagainstchangesintemperature,nutrients,andprecipitation(PPT)(Ficklinet al., 2016),and
thetermsoftheCreativeCommonssustainsbaseflowandevapotranspiration(ET)duringextendeddryperiods(Yanget al., 2011),modelingit
Attribution-NonCommercialLicense,correctlyisimportantforaccuratelyrepresentingthewatercycleandwaterresources.
whichpermitsuse,distributionand
reproductioninanymedium,provided
theoriginalworkisproperlycitedandInearlyLSMs,baseflowdependedonone-dimensionalfreedrainagebelowathin(2–3 m)soillayer,though
(.,Goodfellowet al., 2014;Winteret al., 1998).
:.
Journal of Advances in Modeling Earth
Thissimplificationresultsinunderestimatesofseasonalstorage,enhanceddrainageduringwetperiods,
andinhibiteddryseasonET(Bakeret al., 2008;Brunkeet al., 2016;Fanet al., 2017;Kuppelet al., 2017;
Miguez-Macho&Fan, 2012a, 2012b;Milly&Shmakin, 2002;).Inanattempttoresolvetheseissues,many
LSMshaveincorporatedaTOPMODELapproachtorunoff(.,Niuet al., 2005orOlesonet al., 2010)or
addeda(linear)groundwaterreservoirbelowthesoilprofile(.,Lianget al., 2003orNiuet al., 2007).
However,thisadditionalmodelcomplexity(NOAA, 2016)hasfailedtogenerateclearimprovementsin
modelperformance(Ganet al., 2019;Yanget al., 2011).
Apotentialconfoundingissueisthattheeffectivehydraulicpropertiesofthesubsurfacecomponentsthat
arecontributingtobaseflowarepoorlyconstrained(Dai,Xin,et al., 2019).Thatis,boththehydraulicprop-
ertiesofthesubsurfaceitself(.,thepotentialdrainableporosityandsaturatedhydraulicconductivityof
eachconstituentsoilgroupandbedrocktypethatunderliesawatershed)andtheeffectivevaluesofthese
hydraulicvariablesatthewatershedscale(whichaccountsforthetime-varyingdistributionofgroundwa-
teramongtheseconstituents),hydraulicpropertiesare
generallycalculatedasafunctionoftheaveragepropertiesoftheoverlyingsoilasestimatedaccordingto
soiltexturalclass,followingGedneyandCox (2003).However,soiltexturalclassesmaybeexpectedtocon-
strainvaluesofhydraulicconductivity(K)onlytowithinarangeofseveralordersofmagnitude(Freeze&
Cherry, 1979;Gleesonet al. 2011;Huscroftet al., 2018;Zhang&Schaap, 2019).Uncertaintiesareknownto
beevengreaterinnon-temperateclimateswheredataaresparser(Henglet al., 2017;Huscroftet al., 2018)
andthereissomedisagreementevenamongcommonpedotransferfunctions(Dai,Shangguan,et al., 2019;
Zhang&Schaap, 2019).Further,soiltextureisknowntobeapoorerpredictorofhydraulicpropertiesthan
othersoilcharacteristicslikehydraulicradius,structure,andsorting(Zhang&Schaap, 2019).Soil-texture-
basedpredictionsmayalsounder-useadditionalinformationavailableinsoilsurveys,anddonotaddress
deepersystemslikesaproliteandbedrock.
Richard'sequationandestimatesofKfromsoiltexturealsoexplicitlyignoremacropores(Beven&Ger-
mann, 2013).EvendirectlabmeasurementsoftheKofasoilsample(onwhichpedotransferfunctionsare
based)maybeorder(s)ofmagnitudelowerthaninsitumeasurementsthatdoincorporatecontributions
frommacropores(Mendozaet al., 2003;Zecharias&Brutsaert, 1988).Macroporeshavebeenshownto
sustaindischargenotonlyintheshallowsubsurface(.,bioturbationorrootrot),butalsoattheinterface
betweenweatheringbedrockandsoils(.,saprolite)(Beven&Germann, 2013;Tromp-vanMeerveld&
McDonnell, 2006a, 2006b).
Thegeologicunitsusedinglobalmapsofhydraulicpropertiesbringadditionaluncertainty,astheyoften
explicitlyignorefaultandfracturenetworks(.,Gleesonet al., 2014)despitethesenetworks'enhance-
mentoftheeffectiveporositiesandKoftheirparentrockbyuptoseveralordersofmagnitude(Freeze&
Cherry, 1979).Further,regional(andglobal)mapsofsoilandgeologicunitsareoftenstitchedtogetherfrom
multiplesources,resultinginsharpboundariesinunitclassesatpoliticalboundaries(Dai,Xin,et al., 2019;
Gleesonet al., 2011, 2014;Hartmann&Moosdorf, 2012;Huscroftet al., 2018)andarebasedonsurveysthat
rarelyextendmorethan2mbelowthesurface(Dai,Shangguan,et al., 2019).
Evenwherethethree-dimensional(3D)structureofthesubsurfaceiswellmapped,assessingtheeffective
hydraulicpropertiesofthecomponentsofthesubsurfacethatactivelycontributetostreamflowisprob-
lematic(Binleyet al., 1989).ThesecomplexitiesarecompoundedinLSMsthataggregatethehydraulic
propertiesofthesoilsand/orgeologicmaterialunderlyingawatershedwhetherornotthosematerials
,becausethetime-varyingsourcingofwaterwithinthe3D
structureofawatershedhasamajorimpactonthestabilityofstreamflow(Barnes, 1939),aswellasits
chemistryandbiology(Zhiet al., 2019),researchersaremakingstridestowardincorporatingthesepro-
cessesintomodernLSMs(Fanet al., 2019).Forinstance,includingestimatesofdepthtowatertablein
LSMshelpsmodulatesoilmoisturedistribution,ratesofET,andstreamflowresponse(Koiralaet al., 2019;
Miguez-Machoet al., 2007).Similarly,Brunkeet al. (2016)showedthatincorporatingspatialdifferencesin
depthofunconsolidatedmaterialhasmajorimpactsonpatternsofbaseflow,soilmoisture,andstorage,as
,noregionaldatabasesoftheeffectivehydraulicpropertiesofthesubsurface
-
scaleapplicationrecessionanalysis.
:.
Journal of Advances in Modeling Earth
. Recession Analysis
. Traditional Applications and Assumptions
Theeffectivehydraulicpropertiesoftheaquifer(s)thatactuallysustainbaseflowhavelongbeenestimated
,webrieflysummarizetheapplica-
tionsofandassumptionsunderlyingtraditionalhydrographrecessionanalysis,thoughwereferreadersto
therecessionliteratureforadetailedaccounting(Brutsaert&Nieber, 1977;Harman&Kim, 2019;Kirch-
ner, 2009;Tallaksen, 1995;Tashie,Pavelsky,&Emanuel, 2020;Trochet al., 2013).Wealsooutlinerecent
reinterpretationsofthesemethodsinthefollowingSection ,giveamorethoroughdescriptionofthe
specificmethodsappliedhereinSection 2,anddescribeunderlyinguncertaintiesinSection 4.
ApplyingtheDupuitassumptions(Dupuit, 1863)thatgroundwaterflowishorizontalandneglectingcap-
illarity,theBoussinesqequation(Boussinesq, 1877)isusedtodescribeunitdischarge(q)asafunctionof
effectivesaturatedhydraulicconductivity(K),drainableporosity(f),drainableaquiferthickness(Th),aqui-
ferbreadth(.,widthnormaltoflowdirection;B),slope(i),streamnetworklength(L),andthewatertable
elevation(h),totalsubsurfacedis-
charge(Q)isafunctionoftheaquifer'savailabledrainablestorage(S = Th*f)andthedistributionofwater
 (1977)showedthatifchangeinstreamflow(dQ/dt)isexpressedas
afunctionoftime-independentQaccordingtoapowerlaw:
dQdtaQ/b(1)
thenaisafunctionofwatershedcharacteristics(K,f,Th,B,L,andi)whilebisafunctionofaquifergeom-
−dQ/dt,
itispossibletousehistoricalstreamflowdatatoestimatewatershed-averagehydraulicproperties(.,K,S,
Th,orf)whichareotherwisedifficulttomeasure(Szilagyiet al., 1998).
. Storage-Dependent Recession Behavior
Thoughtheassumptionsunderlyingrecessionanalysisarerestrictive,thistechniquehasrepeatedlybeen
validatedanalytically,usingcomputermodels(.,Pauritschet al., 2015;Rupp&Selker, 2006),table-top
models(Guerinet al., 2014;Luoet al., 2018),tracerexperiments(Winkleret al., 2016),inhighlymo