Upload
others
View
9
Download
0
Embed Size (px)
Citation preview
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 1
The role of the land surface in the climatesystem
By Pedro Viterbo ([email protected])
European Centre for Medium-Range Weather Forecasts, Shinfield Park, Reading RG2 9AX, United Kingdom
Abstract
Therole of the landsurfacein theclimatesystemis illustrated,with a focuson recentexperienceat theEuropeanCentreforMedium-RangeWeatherForecasts(ECMWF). Global energy and water budgetsare discussedand comparedwith theircounterpartsover theocean,highlightingphysicalmechanismsresponsiblefor their differences.Time scalesassociatedwiththeglobalhydrologicalbudgetarepresented.Usingfield dataandmodelresults,soil moistureis shown to beresponsibleformodulatingthesurface-atmosphereinteractionat a continentalscale,on time scalesrangingfrom thediurnal to theseasonal.After a brief review of theimpactof landsurfaceon weather, threeECMWF casestudiesarepresentedwherea morerealisticrepresentationof landsurfacewascrucialfor theperformanceof theforecastsystem.They correspond,respectively, to theroleof soil moisturein determiningthepositionandintensityof theprecipitationmaximumin anextremeeventof mid-latitudessummer, the role of albedoof thesnow in thepresenceof forestsin springandtheeffect of soil waterfreezingasa thermalregulatorof thesurfacein coldclimates.Finally, theevolutionof thesystematicerrorsin theECMWFforecastsof nearsurfacetemperatureandhumidity is presentedover thelasttenyears;aclearsignatureof changesto therepresentationof landsurfaceprocesses (and other physical processes affecting the energy and water fluxes at the surface) can be found on that record.
Table of contents
1 . Introduction
2 . Surface energy and water budget
3 . Time scales and the role of soil moisture
3.1 Global time scales
3.2 The role of soil moisture
3.3 The interplay between the diurnal and the seasonal time scales
3.4 A schematic view of the role of land surface
4 . Impact of land surface on weather: A brief literature survey
5 . Examples from ECMWF recent experience
5.1 Soil moisture
5.2 Boreal forests
5.3 Soil water freezing: A regulator of cold climates
6 . Conclusions
1. INTRODUCTION
Theclimatesystemis influencedby thelandsurfaceatavarietyof timeandspatialscales.Firstly, theatmosphere
The role of the land surface in the climate system
2 Meteorological Training Course Lecture Series
ECMWF, 2002
is in directcontactwith thesurface,whichactsasasourceor sink for theatmosphericenthalpy andmoisture(and
asasinkfor theatmosphericmomentum),via thesurfacesensibleheatflux andevaporation(andthesurfacestress).
Secondly, thesurfaceconditionsactasaregulatorfor importantfeedbackcyclesin theclimatesystem.Thirdly, the
partitioningof netradiationatthesurfaceinto sensibleandlatentheatfluxesdeterminesthesoil wetnessevolution,
whichactsasoneof theforcings– or, at thevery least,amodulator- of low frequency variability. In fact,afterthe
seasurfacetemperature,soil wetnessandsnow massarethemostimportant“memory” mechanismsfor timescales
rangingfrom weeksto seasons.Finally, thesurfaceenergy fluxesdetermineto a largeextent thesurfaceweather
variables,suchasscreen-level temperature,humidity, andwind speedand,to a lesserextent,low-level cloudiness
andprecipitation.Humankindlivesin thelowesttwo-metresof theatmosphereandis directly affectedby theat-
mospheric conditions at the near surface.
Thispaperillustratestheimpactof thelandsurfacein theclimatesystem,with afocusonexamplesbasedonrecent
experienceat theEuropeanCentrefor Medium-RangeWeatherForecasts(ECMWF).Theprimarypurposeof the
paperis to highlight typicalweatherregimes/ecosystemswherethelandsurfaceis of relevanceto theevolutionof
atmosphericvariables,ratherthanpresentinga systematicreview of themultifacetedresearchon surface-atmos-
phere interactions.
Theimportanceof landsurfacein numericalweatherprediction(NWP) stemsfrom a blendof practicalconsider-
ationsandbasicphysicalprinciples.Thosearenotverydifferentfrom thegeneralargumentspresentedin theopen-
ing paragraph,but two key issuesareof paramountimportancefor NWP. Firstandforemost,accurateforecastsof
nearsurfaceweatherparametersarerequestedby theNWPusers’community. Thequalityof suchproductsasthe
diurnal cycle of nearsurfaceair temperatureandhumidity, winds, low level cloudinessandprecipitationis to a
largeextentdeterminedby thephysicalrealismof themodelrepresentationof thesurface-atmosphereinteractions.
Secondly, remotesensingobservationsof theatmospherehaveincreasedimportanceto characterisethestateof the
atmosphereover land.Thesensorsthataresensitive to the lower tropospherecanonly beusedeffectively in the
presence of a good quality background field of skin temperature.
It is importantto emphasisethat,despitethe fact thatdeterministicforecastsof NWP do not extendbeyond two
weeks,themodellandsurfacecontrolsmuchlongertimescales.Thesoil variablesin themodelareinitialisedevery
6 hoursbasedon indirect, imperfectandvery sparseobservations(Douville et al. 2000).Dataassimilationcom-
binesthat informationwith a short-termforecastto getanoptimalestimateof the landsurfaceinitial conditions.
Cycling thecreationof initial conditionsis tantamountto anextendedmodelintegrationnudgedby observations
andextendsthememoryof thesurfacevariablesbeyondtheforecastduration.Therealismof therepresentationof
landsurfaceprocessesis crucialfor handlingcorrectlythememoryproperties,representedby soil moisturein sub-
tropics and mid-latitudes spring and summer and by snow mass in high latitudes and mountainous areas.
Although the importanceof the landsurfacein theclimatesystemhasbeenestablishedwith, amongothers,the
pioneeringwork of Namias(1958),Budyko(1963;1974andreferencestherein),Manabe(1969),Charney(1975)
–seereview by Garratt(1993)- theimpactof landsurfaceonweatherwasonly recentlyrecognised,ascanbeseen
by thescarceliteratureavailable.In Section2 somegeneralremarkson thesurfaceenergy andwaterbudgetwill
bepresented,while Section3 will elaborateon therole of thelandsurface,with emphasison thelong time scales
characteristicof soil water. In Section4 somesurface-atmosphereinteractionmechanismsthatarerelevantfor the
atmosphericsimulationattheregionaltocontinentalscalesarereviewed,basedonnumericalorobservationalstud-
iesthatarerelevantfor NWP. Section5will illustratewhenandwherethelandsurfacecanaffecttheweather, based
on examplesfrom recentexperienceat theEuropeanCentrefor Medium-RangeWeatherForecasts(ECMWF).A
short summary is presented inSection 6.
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 3
2. SURFACE ENERGY AND WATER BUDGET
Fig. 1 representsin a schematicform theenergy andwaterbalanceat thelandsurface.Thesurfacealbedo,con-
trolling thefractionof theincidentshortwaveradiationabsorbedby thesurface,dependsonthesoil andvegetation
typeandstateandon theamountof snow present.Thenetlongwave radiation,LW, dependsalsoon propertiesof
thelandsurface,namelythesurfaceemissivity andthesurfaceskin temperature.Sincethenetradiationflux (the
sumof solarandlongwave radiation)is downward,andbecausethe landsurfacehasa small thermalinertia,the
sumof latentandsensibleheatfluxesmustbeanupwardsflux. Note that thesurfacelatentheatflux, LE, in the
energy budget(left panel)is equalto the latentheat,L, timestheevaporationflux, E, in the waterbudget(right
panel),indicatingthatthewateris availableat thesurfacein acondensedphaseandis passedto theatmospherein
the vapour phase. In that process, the surface undergoes evaporative cooling.
As mentionedin theintroduction,thepartitioningof theenergy availableatthesurfaceinto latentandsensibleheat
dependscrucially on thesoil moisture.Vegetatedcoveredsurfaceshave theability to draw waterfrom a depthof
order1m (therootlayer),while for baregroundonly thewaterin thetopfew cmof soil contributesfor evaporation.
Thelatentandsensibleheatfluxes(LE andH, respectively) playadifferentrole for theatmosphere.Sensibleheat
at thebottommeansenergy immediatelyavailableto theatmosphere,andcontributesto theheatingand/ordeep-
eningof theplanetaryboundarylayer(BL), thatshallow portionof theatmospheredirectlyaffectedby thesurface.
Thesurfaceevaporationflux doesnotdirectlyheattheatmosphere,but providesmoistureto theBL or, in thecase
of deepconvection,to thewholetroposphere.In favourableconditions,thatcontributesto precipitationgeneration
mechanisms,with theassociatedreleaseof latentheatinto thewholetroposphere.It is clearthat,whencompared
to thesensibleheatflux, evaporationcanindirectlyleadto averyefficienttransferof energy affectingamuchdeep-
er atmosphericlayer. For anentireatmosphericcolumn,thenet radiative cooling is balancedby energy involved
in phasechangesinsidethecolumn(condensationof watervapourandevaporationof rain) andsensibleheatflux
at the surface. Land surface processes affect directly or indirectly this balance.
Figure 1. Schematicsof surfaceenergy andwaterfluxes.H, LE, LWandSWstandfor surfacesensibleandlatent
heat flux, surface longwave and shortwave radiation flux, respectively; E, P, andY stand for evaporation,
precipitation and runoff. Numbers at the bottom represent averaged values over all land points for the ERA15
reanalysis (1979–1993).
TABLE 1. MEAN SURFACE ENERGY FLUXES IN THE ERA15ATMOSPHERICREANALYSIS
SW LW H LE G0 Bo=H/LE
Land 138 -63 -23 -41 0 0.6
Sea 163 -51 -10 -104 -2 0.1
Surface energy budget
H
LE LW
SW
23 41 63 138 Wm-2
Surface water budget
E
P
Y
0.8 mmd-12.21.4
The role of the land surface in the climate system
4 Meteorological Training Course Lecture Series
ECMWF, 2002
Table1summarisesthesurfaceannualmeanfluxesfor the1979-1993periodcoveredby the15yearEuropeanCen-
trefor Medium-RangeWeatherForecasts(ECMWF)reanalysis(ERA15,Gibsonetal. 1997).Valuespresentedare
globalaveragesover landandseaseparately, in W m-2, anddownwardfluxesarepositive.Thenetheatflux, G0, is
thesumof all thesurfacefluxes.Thecontrastsbetweenlandandseaareclear. Evenfor sucha largetime period,
thenetflux is non-zeroover sea,emphasisingthelarger thermalinertiaof theoceans.Thecontinentshave a fast
responsive surfaceandadjusttheir surfacetemperaturein orderto maintaina zero-heatflux at thesurface,while
theoceanshave a muchlargerthermalinertia,with relatively smallvariationsin surfacetemperatureandflux im-
balancesallowed in muchlongertime scales.The last column,the Bowen ratio, Bo, is the ratio of sensibleand
latentsurfaceheatfluxes.Thelargervaluesoverlandareindicativeof therelativelydifficulty of accessingthewater
at thesurface.Overvegetatedsurfaces,thiscorrespondsto thephysiologicalmechanismscontrollingtranspiration
while over bare ground the water directly accessible for evaporation is limited to the top few soil cm.
Surfaceenergy fluxesfor anentireseason(notshown in thetableabove)still balanceout,indicatingthattheenergy
associatedto theseasonalchangesin soil temperaturearenegligible whencomparedto theindividual surfaceen-
ergy fluxes.However, for thesurfacewaterbalanceon a seasonaltime scale,thestorageterm,or changein total
soil watercontent,canbeof similarmagnitudeto theprecipitationor evaporation,which is self-evidentin any ex-
tendeddroughtperiod.In thenext section,wewill discussthedifferenttimescalesregulatingthesurfaceandreg-
ulated by the surface.
3. TIME SCALES AND THE ROLE OF SOIL MOISTURE
3.1 Global time scales
Figure 2. The global water cycle (fromChahine1992). Units of water in reservoirs: 1015 kg; units of water in
fluxes: 1015 kg yr-1.
Fig. 2 representsthesizeof themoisturereservoirs of theterrestrialatmosphereandthemarineatmosphere(rec-
tanglesin thefigure),theexchangesof moisturebetweenthem,andbetweentheatmosphereandthesurfacebelow
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 5
(arrows in thepicture).Theseasurfaceevaporatesat thepotentialrate,while over landthereareadditionalmech-
anismsthat reducetheevapotranspirationrate:drynessof thesoil or, over vegetatedareas,physiologicalmecha-
nismsthatcanreduceor shuttranspirationfrom theplant leavesandtrunksmakingthewaterfrom theroot zone
effectively unavailablefor theatmosphereabove.Precipitationover landis abouta quarterof thatover sea.Note
thatprecipitationexceedsevaporationover land,while overseathereverseis true.In orderto haveaclosedbudget
for theterrestrialatmosphere,advectionof moistureacrossaverticalwall projectingover thecontinentboundaries
hasto matchthedifferenceprecipitationminusevaporation.Advectionis roughlyhalf of thewaterevaporatedover
land(seePeixotoandOort 1992for estimatesbasedon radiosondeobservations),suggestinganannualrecircula-
tion ratio (ratioof therainfall comingfrom localevaporationover total rainfall rate)of 67%(71/107).To closethe
hydrologicalcycle, theadvectionhasto bematchedby theriver runoff: theglobalaveragedinflux of freshwater
into theoceanis estimatedin this way as36 x 1015 kg yr-1. For continentalareas,annualrunoff, evaporationand
precipitation are approximately in the ratio 1:2:3.
Rigorousformulationsof theatmosphericbranchof thehydrologicalcyclecanbefoundin Peixoto1973andPeix-
otoandOort1983,1992.So-called“aerologicalestimatesof runoff”, basedonmeasurementsof atmosphericwater
vapourtransportanda closureat thesurface,have beenappliedsuccessfullyto basinswith areaslarger than106
km2 (Rasmusson 1967, 1968, 1971;Peixoto 1973).
Therainfall rateandthesizeof thereservoir canbecombinedto giveatimescale4.5/107= 0.042years= 15days:
theterrestrialatmospherewouldbeemptiedby rainfall in 15days.In asimilarway thereservoir wouldbereplen-
ishedby surfaceevapotranspirationin 23 days(4.5/71years)1. The time scalesassociatedto marinerainfall are
only 7.5days,andthecorrespondingvaluefor evaporationis 6.8days.This suggests(a) a morevigoroushydro-
logical cycle over theocean;(b) a landsurfacecontrolover largetime scales(weeksto months),throughtheeva-
potranspirationflux of waterat the surface.The implication of the above on the extendedpredictabilityof the
atmospheredueto exchangesof waterwith thelandsurfacehasbeendiscussedby many authors(see,e.g.,Namias
1958,Mintz 1984,Dümenil and Bengtsson 1993,Dirmeyer and Shukla 1993).
Theaccuracy of thenumbersshown in Fig. 2 varieswidely: seeChahine1992,for thesourcesusedto produce
theseparticularestimates.Any literaturereview showsavery largedispersionin thosenumbers(see,e.g.,Viterbo
1996):globalestimatesof thetotalcolumnwatervapourcanvaryby asmuchas34%,while runoff estimatesdiffer
by 45%.Independentestimatesof precipitationhavesmallerrangesof uncertainty(notwithstandingtheextensive
areasof theplanetwhereobservationsarevery scarce),but director indirectestimatesof evaporationaresubject
to very large uncertainty.
3.2 The role of soil moisture
In orderto illustratetheroleof soil moisturein shapingtheinteractionsurface-atmosphere,wewill usemodeldata
over theRed-ArkansasRiverbasin,asub-basinof theMississippiRiverbasin.Datais basedon theECMWFrea-
nalysisERA15(Gibsonetal. 1997)usedalsoin Section2.Bettsetal. (1998,1999)havestudiedtheECMWFmod-
el energy andwaterbudgetover theMississippiRiver basin,asdescribedby short-termforecasts.Basin-average
datawasusedfor nineyears,1985–93andanalysedatdifferenttemporalscales;wewill concentratein thissectio
on summertime 5-day average ECMWF data.
1. Theglobalview ontheatmosphericwaterbudgetpresentedin thetext canberegionalised,allowing, undercertainrestrictiveassumptions,an estimate of how much moisture that precipitates comes out from local evaporation versus horizontal transport. For a discussion on the“intensityof thehydrologicalcycle”, i.e.,thetimescalesassociatedto theemptyingandrepleneshingof theatmosphericwaterreservoir atdif-ferent locations on the globe, seeTrenberth (1999).
The role of the land surface in the climate system
6 Meteorological Training Course Lecture Series
ECMWF, 2002
Figure 3. (a) Scatterplot of 5-day average evaporative fraction over warm soils (0-7 cm layer soil temperature >
296K) against0-7cmsoil water(SW1). (b) As Fig 3(a),but for pressureheightto thelifting condensationlevel
(PLCL) (from Bettset al. 1998).
We will illustrateherethat theECMWF datashows a similar couplingin themidsummeron the5-daytimescale
betweensoil water, evaporationandlow-level thermodynamicsastheFirstInternationalSatelliteLandSurfaceCli-
matologyProjectField Experiment(FIFE) data2 (BettsandBall 1998)show on thediurnaltime scale.We define
a 5-day mean evaporative fraction as
.
whereH andLE arethe5-dayaveragesof thesurfacesensibleandlatentheatfluxes.Fig. 3 (a) shows thestrong
couplingbetweenEF andthe top layermodelsoil water, SW1.Thedataplottedareall the5-dayvalues(1985-
1993) for which the 0-7 cm mean soil temperatures are > 296 K, representative of the warm months.
Fig.3 (b) showsasimilarcouplingof PLCL (thepressureheightof thelifting condensationlevel,determiningcloud
base)to SW1for thesamedata.Themodelresistanceto evaporationbetweenthesaturatedinteriorof a “leaf” and
thesurroundingair is dependenton soil water, andthis vegetationresistanceis thereforeonekey factorin deter-
miningtheequilibriumsaturationlevel difference,PLCL, in thesaturationpressurebudgetof theBL (BettsandBall
1998).
Wehaveshown 5-dayaverages,but thepatternsandslopesin Figs.3 (a)and(b) aresimilar (but shiftedslightly to
higherPLCL andlowerEF) if the12-hdaytimeaverageareusedinstead.Notethatthelower limit in Fig. 3 (b) (cor-
respondingto very wet soils) is neartheoceanicequilibriumof hPa (e.g.,BettsandRidgway 1989).
The oceanicsurfaceboundaryis saturated,andhasno additionalresistanceof evaporationcorrespondingto the
vegetativeresistanceover land.Figs.3 (a)and(b) areparticularlysignificantbecauseneitherof theserelationships
of EF andPLCL on soil water depend strongly on soil temperature at this warm temperatures.
2. FIFEwasanexperimentmeasuringthedifferentcomponentsof thesurfaceenergy andwaterbudgetof a15x 15km2 of naturalgrasslandlocated in Kansas, US, and heavily instrumented over the period 1987-1989.
(a) (b)
��� � �� � �+--------------------=
�LCL 60≈
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 7
3.3 The interplay between the diurnal and the seasonal time scales
Figure 4. (a) Daytime diurnal cycle of potential temperature (q) and mixing ratio (q) at 2 m from 1145 to 2345
UTC for monthlydry-daycomposites(FIFEaveragesfor 1987);(b) (q,q) plot of surfacedatafor selected28days
from July and August 1987, composited by soil moisture, showing the dependence of mean diurnal cycle on
surface evaporation (fromBettset al. 1996).
Fig. 4 (a) (from Bettset al. 1996)shows thedaytimediurnalcycle of theFIFE 2-m thermodynamicdatafor the
predominantlysunny anddry daysfrom May to October1987.Theaxesarepotentialtemperature(θ) andmixing
ratio (q). This (θ, q) plot canberegardedastheheatandmoisturebudgeton orthogonalaxes(Betts1992).There
are19, 21, 25, 22, 23, and22 datesin eachaveragefrom May to October. Theselectioncriteriawerenear-noon
surfacenetradiationaboveathreshold(whichwas450W m-2 in midsummer, falling to 300W m-2 in October)and
nosignificantdaytimerainfall. Herewecanseethediurnalandseasonalcycletogether. Thepointsareplottedhour-
ly, startingat 1145UTC, shortlyaftersunrisein midsummer. Theseasonalriseandfall of meantemperatureand
mixing ratio canbeseen:July is thewarmestmonth.Octoberis noticeablydrier, afterthevegetationhasdiedand
evaporationis low. Saturationpressurelinesof 970and800hPa areshown dashed.Thesurfacepressureis near
970hPa.It canbeseenthatat themorningminimumtemperature,the2 m air is about30hPa from saturation,ex-
ceptin October, whenit is moreunsaturated.Thediurnalrangeof mixing ratio q is relatively small in all months.
Thereis generallya riseof q in themorning,whentheBL is shallow andcappedby relatively moistair from the
BL of theprecedingday, anda fall in theafternoon,asthegrowing BL entrainsdrier air from higherlevels.May
shows no afternoonfall of q, probablybecauseof thehighersoil moistureandevaporation.May andJunedo not
reachas low afternoonsaturationpressuresas the later monthsof July, August,September, andOctober. This
meansa lower meanlifting condensationlevel (seealsoprevioussection)or cloudbasein spring.Probablythis
reflectsaseasonaldryingof thesurface,althoughchangesin upperair thermodynamicstructuremaybeinvolved.
It is clearthattheafternoonmaximumof equivalentpotentialtemperatureθe is controlledmostlyby theseasonal
shift. Theisoplethsof θe =310,330,350K areshown dotted.Theriseof θe from morningminimumto afternoon
maximum is around 14 K in all months.
Thesumof surfacesensibleandlatentheatfluxesis asurfacesourcefor increasingθe (e.g.,BettsandBall 1998).
This surfaceθe is proportionalto thesumof H+LE, andit is not affectedby theBowenratio. It is entrainmentof
low θeair from above theBL, togetherwith thedeepeningof theBL, thatreducetheBL θe rise,andsofeedback
on boththeshallow andevenmoreimportantlyon precipitatingconvection.Thusoneof theimportantaspectsof
theBL evolutionover landis how largeentrainmentatBL top is. ThedaytimeBL over landis primarily thermally
generated(in strongwinds,shearplaysa role),andthuslinkedto thesurfacevirtual heatflux (which over landis
usuallydominatedby thesensibleheatflux). Henceif thesurfaceBowenratio is large,althoughthesurfaceθeflux
(a)(b)
The role of the land surface in the climate system
8 Meteorological Training Course Lecture Series
ECMWF, 2002
maybeunchanged,thelargeH flux drivesmoreentrainment,producesa deeperBL, andthediurnal riseof θe is
reduced.Fig. 4 (b) shows how this diurnal cycle over land dependson soil moistureand,asa consequence,the
surface evaporation.
A total of 28 daysfrom July andAugust1987duringFIFE,which wereaffectedlittle by precipitationor cold air
advection,werecompositedby soil moisture(SM: measuredgravimetricallyin thetop10cm).Thepointsarehour-
ly valuesfrom 1115UTC (nearsunrise)to 2315UTC. Thedry soil composite(SM= 13%,for whichthemeasured
meansurfaceBowenratioatnoonwas0.8)reachesahigherafternoonmaximum K (theθe isoplethsare
shown dotted).In contrast,thewetsoil composite(SM = 23.4%),for whichBowenratioatnoonwas0.4),reaches
a muchcoolerafternoonθ maximum,but a muchwetterq value,sothat theafternoon K. Someof this
shift of θe is associatedwith theshift of theentirediurnalpathof higherq with highersoil moisture,but abouthalf
is theresultof reducedentrainmentof dry low θe into theBL. Overwetsoils,H is muchreducedandtheBL deep-
enslessrapidly. For all the threecomposites,thesurfaceavailablefluxes(net radiationminusgroundheatflux)
werenearlyidentical,sothatthesurfaceθe fluxesweresimilar. This local feedbackbetweensoil moisture,evap-
orationandafternoonθe equilibriumprobablyproduceson largespatialscalesa positive feedbackbetweensoil
moistureandprecipitation,which hasbeenthesubjectof muchresearch(see,for example,Brubakeret al. 1993,
Trenberth1999).Theanalogyover thetropicaloceansis thelink betweenBL andseasurfacetemperature(SST),
whichinfluencestheprevalenceof deepconvectionoverwarmwater. Overlandvariationsin soil moisturecanlead
to aslarge variationsin BL θe aslarger asthoseproducedby several degreesof changein SST. On continental
scales,highersoil moistureandhigherevaporationoverlandwouldleadto ahigherafternoonθemaximumrelative
to thesurroundingoceanandshiftmoreof theglobalprecipitationoverthecontinents(Bettsetal. 1996).Thisfeed-
backhasbeenseenin globalmodels(Mintz 1984).Ontheregionalscale,theMississippifloodcasestudy, present-
edin Section5.1below, suggestedthatthemultipleBLs overtheMidwesternUnitedStatescontrolledthelocation
of precipitation, rather than this mechanism.
3.4 A schematic view of the role of land surface
To concludethis section,a schematicdescriptionof theinteractionsbetweenthesurfaceandtheatmospherewill
bepresented.Inspiredby anearly, muchmorecomplex diagramby Horton(1931),Dooge(1992)(seealsoKuhnel
et al. (1991))summarisedthe interactionbetweenthe landsurfaceandtheatmospherein thepicturereproduced
(with adaptations)in Fig.5 . Thediagramillustratesthebehaviour of thesoil andtheatmospherewithin acomplete
cycle composedof a wet periodfollowedby a dry period.Let usstartjust aftera long episodeof rainfall, point A
in thefigure.Thesoil wateris availablein abundancein theroot layer3 andits evolution is goingto bedetermined
by evaporation.While thesoil hasplentyof water, therateof evaporationis controlledby thenear-surfaceatmos-
phericmoisture:the regime is controlledby theatmosphereandtheevaporationis at thepotentialrate.Below a
certainlevel of soil moisture(pointB in thepicture),physiologicalmechanismswill limit thesupplyof waterfrom
theroot layerinto theatmosphere,andtheevaporationwill dropbelow its maximumvalue(potentialevaporation,
Epot). The regime is undera vegetation(soil) control.Whenprecipitationstarts(pointsC) it will meeta soil dry
enoughduringtheinitial stages,sothatinfiltration (I f, thatpartof waterthatfallsasprecipitationandis effectively
collectedby thesoil for futureuse)will equalprecipitation.Theevolutionof waterin thesoil is oncemoreatmos-
phericcontrolled,via therateof precipitation.Beyonda certainvalue(point D), thesoil doesnot have theability
to soakall precipitation,someof it goesinto runoff. This lastphaseis again soil controlled;thestateof thesoil
determines the rate of infiltration.
Landsurfaceparametrizationshave to representcorrectlythesurfacefluxesandtheevolution of soil moisturein
3. For the purpose of this discussion, field capacity may be considered as the threshold beyond which there is a minimum canopy resistanceto evaporation.
θe 352≈
θe 361≈
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 9
all four phasesof thecycle,andto switchfrom theatmosphericcontrolinto thesoil controlregime.Theevolution
of soil moisturewill determinewhenpoint D will occur, andtheevaporationformulationwill determinepoint B.
Thecrucialareas,from thepointof view of theatmosphere,arethequadrantsBC andCD.Duringspringandsum-
mer(wheretheatmosphericdemandcanbevery large),thesystemremainsmuchlongerin thehalf-circumference
BCD than in the opposite part of the cycle.
Figure 5. The hydrological rosette: Schematic depiction of the interaction between the soil hydrology and the
atmosphere (adapted fromDooge (1992)). Epot and If represent potential evaporation and infiltration, respectively.
4. IMPACT OF LAND SURFACE ON WEATHER: A BRIEF LITERATURE SURVEY
Bettsetal. (1996)review theimpactof landsurfacein thecontext of globalnumericalweatherprediction:Diurnal
andseasonalfeedbackloopsarediscussedaswell asfeedbackloopscontrollingtheBL evolution.We will high-
light heretypical mechanismscontrolling the interactionbetweenlandsurfaceandtheatmosphere,over theUS,
Europe, and the tropics.
Earlymodellingefforts (BenjaminandCarlson1986;Laniccietal. 1987)haveshown thesensitivity of precipita-
tion in theUSGreatPlainsto evaporationupstream,in theMexicanPlateau.Thecharacteristicstormenvironment,
leadingto heavy precipitationovertheMidwest,involvesthebreakdown of acappinginversionformedby anover-
lying pre-existing boundarylayer from theMexicanplateau,which overliesthecool moistBL originatingin the
Gulf of Mexico. This complex patternof differentialadvectionimpactson thestrengthof thecappinginversion,
andthestrengthof the inversionis controlledby evaporationupstreamof theprecipitationarea.Lower valuesof
evaporationleadto a strongercappinginversion,andthelow level flow from theGulf will not breakthroughthe
inversionuntil muchfurthernorth.Thelocationandextentof theheavy precipitationassociatedwith theJuly1993
USfloodswasfoundto behighly sensitiveto thecorrectrepresentationof thesemechanismsin theECMWFmodel
(Beljaarset al. 1996;Viterbo and Betts 1999a;Section 5.1).
Spatialgradientsof soil moisturecanalsoenhancethedifferentialheatingmaintainingandreinforcingthesurface
front in apre-stormenvironmentandintensifyingthethermallydirect(ageostrophic)circulation(ChangandWet-
zel 1991;FastandMcCorcle1991).A similar mechanismis active in a cold front associatedwith a severesquall
line developing explosively (Kochet al. 1997).
The role of the land surface in the climate system
10 Meteorological Training Course Lecture Series
ECMWF, 2002
Theseasonalityof leafareaindex impactsonthesystematicerrorsof USMidwestlower tropospherictemperature
in summer(Xue et al. 1996;Yang et al. 1994).At a smallerscale,therearemany observationalandmodelling
studiesdemonstratingthe importanceof mesoscalefluxesandsmaller-scaleheterogeneity, speciallyat low wind
speeds.
OverEurope,RowntreeandBolton(1983)demonstratedtheroleof localandnon-localresponseof medium-range
rainfall forecaststo anomaliesin theinitial soil. Themechanismsrelevantto this soil moisture-precipitationfeed-
backwerescrutinisedin Schäretal. (1999)in astudydemonstratingtheimpactof idealised(andlarge)anomalies
of soil wateron theEuropeansummercirculation.Unlike thedesert-albedofeedbackhypothesis(Charney1975),
theEuropeansoil-precipitationfeedbackis notof alarge-scaledynamicalnature,i.e.,it is notassociatedto changes
in large-scaleflow. Precipitationrecycling hasalsoa small role in Europe.Threemainfeedbackloopshave been
identified.Wet soils,associatedwith low Bowenratios,leadto thebuild-up of a shallow BL, concentratingmoist
entropy atlow levelsandgiving highervaluesof convectiveavailablepotentialenergy (CAPE).Additionally, lower
Bowenratiosleadto higherrelativehumidity, loweringthelevel of freeconvection.Finally, apositive feedbackof
radiativeorigin, with increasedcloudcover, but largernetradiativeflux, leadsto largermoistentropy andconvec-
tive instability.
All theexamplesabove referto extratropicsspringandsummerexamples,in snow freesituations.Section5.2be-
low presentsspringexamplesin thepresenceof snow. Thereis little ornoimpactof landsurfaceontheatmospheric
circulation in winter (see Giorgi (1990) andSection 5.3 below).
Despiteanextensive list of publicationsontheroleof land-surfacein thetropicalclimate,thereis scantyevidence
of its impact for short-andmedium-rangeforecasts.A notableexceptionis the work of Walker andRowntree
(1977)whodemonstratedtheroleof enhancedsoil moisturegradientsontheshort-range(1-2daysahead)forecast
of the generation of easterly waves, using a simplified model over West Africa.
5. EXAMPLES FROM ECMWF RECENT EXPERIENCE
5.1 Soil moisture
July 1993showedanomalouslyhigh precipitationover theCentralUSA, with exceptionalfloodingof theMissis-
sippi(Changnon1996).Duringthismonth,thenew versionof theECMWFmodel(CY48)andthethenoperational
version(CY47)4, wererunningin parallelat full resolution(spectraltruncationT213,grid-pointspacing~ 60km),
includingdataassimilation.Beljaarset al. (1996)comparedtheperformanceof the two schemes,looking at the
averageof all one-,two- andthree-dayforecastsverifying between9 and25July. While thedayoneprecipitation
of thetwosystemswereverysimilar, andsimilarto theobservedprecipitation,theforecastsatday3weremarkedly
different.In thenew system,the locationandintensityof themaximumprecipitationwassimilar to theobserva-
tions (40 N, 95 W), while theold systemhadlessthenhalf theprecipitationamountin theareaof theobserved
maximumandhadaspuriousmaximumof precipitationdisplaced800km NE. In theold system,therewasagrad-
ual reductionin precipitationfrom day1 to day3, while thenew systemwasableto bettermaintaintheintensity.
However, evaporationat theareaof maximumprecipitationwassimilar for theold andnew system,andin both
systemstherewasno evidenceof forecastspindown, stronglysuggestingthat the local evaporationwasnot re-
4. ThemaindifferencesbetweenCY47andCY48rely onthesurfaceandboundarylayerprocessesparametrization.Thesoil modelin CY48has 4 layers, with no heat flux and free drainage bottom boundary condition, with soil properties (heat and water conductivities and diffusivi-ties) dependent in a non-linear way on soil moisture (Clapp and Hornberger 1978). CY47 has 2 layers plus 1 climate layer underneath, withconstant water soil properties. CY48 has a smaller roughness length for heat than for momentum, while in CY47 they are identical. For moredetails seeViterbo and Beljaars (1995).
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 11
sponsiblefor thedifferencesin precipitation.It turnsout that themaximumof theevaporationdifferencewaslo-
catedover the Mexican Plateau,1000 km SW of the precipitationmaximum,two to threedaysupstreamas
suggestedby backwardstrajectoriesendingupat750hPa,40N, 95W. Themeanthermodynamicprofiles,similar
for day1 forecasts,werevery differentfor day3 forecasts.Theold modelshoweda too strongcappinginversion
above theBL, with air too warmandtoo dry andmuchlower valuesof CAPE.It is clearthat thedifferentialad-
vectionmechanismcharacteristicof theUS Monsoonwasresponsiblefor thedifferencesin precipitation.When
comparedto thenew model,thesoil ontheMexicanPlateauhadmuchlowervaluesof soil moisturein CY47,giv-
ing amuchreducedevaporation,which in turnproducedawarmanddry air massthatcappedtheBL downstream,
inhibiting convection.In CY47, thesoil modelvalueswerestronglyforcedto anerroneous,too dry, climatology:
In suchadatadensearea,atmosphericprofileswereinitiatedto correctvalues,but duringtheforecastthey slowly
felt the influenceof theerroneoussoil moisturevalues.In CY48 thesoil moisturevalueswereinitialisedto field
capacityatthebeginningof July, consistentwith valuesof Juneprecipitationin theareamuchabovenormal.There
wasnoforcingto climatologyin CY48(ViterboandBeljaars1995)andthemodelwascapableof maintaininghigh
valuesof moisturethroughoutJuly. Monthly integrationsperformedwith CY47 andCY48 suggestedthe impor-
tanceof thememoryassociatedto idealisedsoil moistureanomaliesin theinitial conditions(Beljaarsetal. 1996).
The monthly precipitation fields with CY48 compared much better to observations than those of CY47.
Figure 6. Profiles(so-calledtefigrams)of temperature(solid line) anddewpoint (dashedline) with CY47(a)and
CY48(b) of theaveragesverifying from 9 to 25Julyat theforecastrange78hours.Themodellocationis 40N 95
W. (c) shows theaverageverifying analysis.Theshadingindicatestheareawhereaparcellifted from thesurface
hasalowertemperaturethanthesurroundingair, i.e.,theshadedsurfaceareais ameasurefor thestabilityaparcel
has to overcome before convection can occur (fromBeljaars and Viterbo 1999).
CY48surfacemodelranwith predictedsoil moisturethroughoutthefirst half of 1994.It wasclearthataverylarge
near-surfacewarmanddry biasdevelopedovertheNH continentalareasattheendof springandbeginningof sum-
mer. A schemeto initialisesoil waterbasedontheshort-termforecasterrorsof near-surfaceatmospherichumidity
wasdevelopedto overcomethatproblem(Viterbo1996).In orderto testthenew scheme,threecompletedataas-
The role of the land surface in the climate system
12 Meteorological Training Course Lecture Series
ECMWF, 2002
similation-forecastexperimentswereranat T213for themonthof June1994:(a) Control(CY48,no assimilation
of soil moisture);(b) As control,but usingtheinitialisedsoil moisturevalues,and;(c) As in (b), but usingaprog-
nosticcloudschemewith muchmorerealisticcloudcover over land(Tiedtke1993).Thenear-surfacewarmand
dry bias,reducedfrom Controlto theinitialisedsoil waterexperiment,in responseto awettersoil. A lower tropo-
sphericwarmbiasdevelopedin theControlmodelandwasgreatlyreducedwheninitialisationof soil wateris used.
Bothexperimentshadtoolittle cloudcoverover landwith toolargesurfaceshortwaveradiativefluxes,but thewet-
ter soil conditionsof experiment(b) managedto maintainevaporationin thefaceof excessive netradiationat the
surface.Thethird simulationdisplayedevensmallerbiases,associatedwith alarger, morerealisticcloudcoverand
smaller radiative biases.
Thealgorithmto initialise soil wateris successfulin controllingmodeldrifts but dampenstheseasonalcycle and
interannualvariationsof evaporationandsoil moisture.Viterbo andBetts(1999a)revisitedtheJuly 1993simula-
tion, usingtheinitial soil wateralgorithmandtheprognosticcloudscheme.Thenew systemgivespoorerresults
for precipitationthanBeljaarset al. (1996).AlthoughmuchbetterthanCY47, thereis a suggestionof northward
displacementandreductionin theprecipitationmaximumin day2 forecasts.It appearsthattheinitialisationof soil
moistureat field capacityat thebeginningof July in Beljaarset al. wascrucial to obtaina goodsimulationof the
excessive rainfall events.
5.2 Boreal forests
Surfacealbedois theprimeregulatorof thenetenergy availableat thesurface.Thealbedoof snow-freelandsur-
facesrangesfrom valuesof 0.1 in foreststo valuesof 0.35over deserts.For areasseasonallycoveredwith snow,
thatrangecanextendupto 0.85.BettsandBall (1997)analysedtheannualcycleof albedoin theBOREASexper-
iment,performedin Canadaduring1994,focussingon thesnow season,comparingseveralmeasurementsiteslo-
catedovergrass,aspenandconiferous.Representativevaluesfor daily averagedalbedoof snow-coveredgrasssites
are0.75,while correspondingvaluesfor theaspenandconifersitesare0.21and0.13,respectively, with valuesas
highas0.4,oneto two daysaftersnowfall. Theloweralbedoof theborealforestsin thepresenceof snow corrob-
oratesdatafrom otherobservationalstudiesandthefew attemptsof makingahemispheric-satellitebasedestimate
of albedo.
TheECMWF modelversionof 1994,at thetimeof theBOREASexperiment,treatedthealbedoof snow covered
areaswith no regard to the landcover: beyonda critical valuefor snow depth,thealbedoof snow coveredareas
wasrarelyoutsidethe0.7-0.8range.As aresult,whencomparedto experimentalresults,netradiationwastoolow
andnear-surfaceair temperaturesweretoocold.A modificationto theschemewasdesignedsuchasthealbedoof
snow-coveredsurfacestendedto theasymptoticvalueof 0.2 in thepresenceof forestsand0.7otherwise(Viterbo
andBetts1999b).Themodifiedschemehasmuchreducedbiasesin temperatureandradiationin thehighlatitudes.
Thecold biasin temperaturein thecontrolschemeextendedin thevertical to thewhole troposphere,increasing
with forecastrangeandaffectingmostcontinentalareas.Thebottompanelof Fig.7 showstheday5 forecasterror
of 850hPa temperature,averagedfor MarchandApril in 1996.Thevery high albedoinducescoolingerrorsex-
ceeding–3K in NorthAmericaand–7K in Asia.Thetoppanelshowsthecorrespondingfigurefor 1997.In spring
1997,with thenew snow albedoscheme,thecoldbiasover theborealforesthasbeenalmosteliminated.Thenew
schemealsoimproved thequality of themedium-rangeforecasts,asevidencedby betterscoresof 500hPa geo-
potential fields.
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 13
Figure 7. Comparison of the average 5-day forecast temperature errors at 850 hPa, for the ECMWF operational
model during March-April 1996 (bottom) and 1997 (top) (fromViterbo and Betts 1999b).
Theforecastresultsabove corroboratethestudyof ThomasandRowntree(1992)on therole of theborealforests
in conditioningtheclimateat high latitudes.Thespringmonthsof two five-yearexperiments,thefirst with a (re-
alistic) snow albedoandthesecondwith thehigh latitudeforestsremovedarecompared.Thelatterexperimentis
colderthantheformerin thecontinentalareasnorthof 50N. PielkeandVidale(1995)suggestedthattheboundary
betweentundraandborealforestsis a regionof enhancedhorizontaltemperaturegradients,actingasapre-condi-
tioner for baroclinicinstability and“locking” the climatologicalpositionof the polar front. In analysingfurther
refinementsto theECMWFsnow model,vandenHurk etal. (2000)show that(a)simulatingtheborealforestcon-
trol on evaporationin spring(reducedtranspirationfrom frozensoils) and(b) increasingthe runoff over frozen
soils, improves the agreement of model results with observations.
5.3 Soil water freezing: A regulator of cold climates
TheoperationalECMWF forecastsfor thewintersof 1993-94throughto 1995-96showeda tendency to produce
a cold biasover continentalareasin winter. This error wasparticularlysevereduring the winter of 1995-96for
Scandinavia, a yearcharacterisedby reducedsnow amountsand,consequently, larger thermalcouplingbetween
thesoil andtheatmosphereabove. Viterbo et al. (1999)diagnosedtwo mainproblemscontributing to thaterror.
Firstly, theenergy involved in phasechangesin thesoil wasnot taken into account.Whenpositive temperatures
approach0 C, asubstantialamountof theexternalcoolingdemand(i.e., infraredcooling)is usedto freezethesoil
The role of the land surface in the climate system
14 Meteorological Training Course Lecture Series
ECMWF, 2002
water, therebydecreasingtherateof soil cooling;a similar effect occursin melting.Soil waterphasetransitions
actasa thermalbarrier, increasingthesoil inertiaat temperaturescloseto 0 C. Secondly, thedownwardsensible
heatflux, prevailing in winter conditions,wastoo small, leadingthe modelinto a positive feedbackloop where
coolingreducedtheheatflux, andmadethesoil evencolder. Modelchangesweredesignedto incorporatethemiss-
ingphysicalmechanisms.Separateseasonalintegrationswith bothmodelschemesrevealedagreatlyalleviatedsoil
andnear-surfaceatmosphericcoolingdrift: Screen-level temperaturesreducedfrom –10C to closeto zero.Despite
theconsiderablewarmingin themodelsoil andnear-surfacewinter climatein continentalareas,therewasnegli-
gible impacton thefreeatmospheretemperatureandtheatmosphericflow. In winter, stablesituations,theatmos-
phereis decoupledfrom thesurfaceandchangesat thesurfacedo not propagateupwards,unlessthey affect the
momentum budget.
6. CONCLUSIONS
Theexamplespresentedabovehaveshown thatlandsurfacecanhaveasignificantimpactontheatmosphereat the
synoptic/continentalscalewhenit affectsthepartitioningof thesurfaceenergy into sensible/latentheat,via thesoil
water. Thiseffect canbelocalor non-local.On theotherhand,landsurfacehasasignificantimpacton theatmos-
phereat thesynoptic/continentalscalewhenit affectsthenetenergy at thesurface,e.g.,changeof thealbedoin
snow-coveredareasin spring.However, in winter, staticallystable,conditionsthelandsurfaceis decoupledfrom
theatmosphere:Largevariationsin surfacetemperatureaffect only thelowesthundredmetresof theatmosphere
and do not have a significant impact on the circulation.
RecentexperienceatECMWF, summarisedabove,showsthatforecastssystemsaresensitiveto misrepresentations
of longertimescalesin theland-surface/atmosphereinteraction.Progresscanonly bemadewith sustainedefforts
on validationof themodelcomponents.Dataassimilation/forecastsystemsareideal toolsto validateparameteri-
sationsbecauseof their constantconfrontationwith observations(in a “perfect synoptics”scenario)anda very
largecommunityof usersrequiringaccuratediurnalcycleof weatherparameters.In thiscontext, GEWEX(Global
Energy andWaterExperiment)ContinentalScaleExperimentsandtheCoordinatedEnhancedObservationPeriod
plannedfor 2002will play a prominentrole. If thesurfaceandupper-air observationsarrive at theNWP centres
they canbeassimilated.Analysisandshorttermforecastsfrom thosecentresarethebesthopeof having acoherent
picture of surface and atmosphere variables and of the surface fluxes.
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 15
Figure 8. Historyof monthlybiases(thick lines)andstandarddeviations(thin lines)with respectto observations
of thedaytime(72-hour:redline) andnight-time(60hour:blueline) operationaltwo-metretemperatureforecasts,
averaged for all available surface stations in the European area (30º N to 72º N and 22º W to 42º E).
An importantfeedbackloop in thesurface-atmosphereinteractionis thelink betweentheanomaliesof soil mois-
tureandprecipitation.Its correctrepresentationin themodelrelieson the interplayof severalparameterisations:
evaporationandsoil moisture,boundarylayer processes,cloudsandconvection.Most NWP modelshave their
poorestscoresonprecipitationin thetropicsandspring/summerextratropicsandimprovementsontheunderstand-
ing of thesoil moisture-convectioninteractionwill mostlikely greatlyalleviatethatdeficiency. Validationof model
resultsagainstresultsfrom theLargeScaleBiosphere-AtmosphereExperimentin Amazonia(LBA) (seeforthcom-
ing SpecialIssueof J. Geophys.Res.2001)observationsof thediurnalcycleof precipitationandatmosphericther-
mal and humidity profiles over the Amazon River Basin will probably be a key for progress in that area.
Probablythebestsummaryof theimpactof landsurfaceonweathercanbeshown onFig.8 , displayingthehistory
of ECMWF operationalshort-rangeforecasterrorsof 2 m temperatureover Europeasa time-seriesof monthly
averages.Theseerrorsshow alargeannualcycle,aredifferentfor nightandday(72and60hourforecastsverifying
at12and00UTC, respectively), andhavearich historyof themany modelchangesthatweremadeovertheyears.
We will discuss only the model changes made from 1993 onwards.
In August1993,asurfaceschemewith aclimatologicaldeep-soilboundaryconditionfor temperatureandmoisture
wasreplacedby thefree-runningfour-layerscheme(ViterboandBeljaars1995),but theimpactisnotveryobvious.
The summerdaytimebiasof August1993wassmallerthanthat of the previous yearbut, at that time, the soil
schemehasbeenrunningfreely for only two-months(includingtheJuly paralleltestdescribedearlier).Thenext
summershowedapronouncedwarmbiasrelatedto agradualdryingoutof thesoil whichwasreducedin July1994
by resettingthesoil moistureto field capacityover vegetatedareas.A simplesoil moistureanalysisschemewas
introducedin December1994(Viterbo1996)with aclearbeneficialimpacton thedaytimebiasfor summer1995.
Thenight-timetemperatureshave beenbiasedcold for many years,relatedto anoverly largeamplitudeof thedi-
The role of the land surface in the climate system
16 Meteorological Training Course Lecture Series
ECMWF, 2002
urnalcycle.Thewinterof 1995/1996wasparticularlybad,mainlybecausetheEuropeanareawasblockedfor most
of thewinter with easterlywindsandvery cold temperatures,althoughchangesto thecloudschemeor theoro-
graphicdragmighthavehadanegativeimpactonnight-timetemperatures.It is interestingthatthereductionof the
daytimebiasactuallyincreasedthenight-timebiasby displacingtheentirediurnalcycle to coldertemperatures.
Soil freezingandincreasedboundarylayerdiffusionin stablelayers,introducedin September1996,improvedthe
monthlyerrorstatisticsconsiderably. Thewintertimebiaswaslargelyeliminatedandtheamplitudeof thediurnal
cycle wasdown to a reasonablelevel. Thesnow albedoreductiondescribedin SectionBorealforestswasintro-
ducedin December1996,but its impactis notclearoverEurope,dueto therelatively smallareacoveredby snow
andtheoverallmagnitudeof theerrorslinkedto theexcessivesoil cooling,correctedthreemonthsearlier. Finally,
a muchmoreselective way of initialising soil moisture(Douville et al. 2000),introducedin April 1999,might be
responsiblefor aslightreductionof thestandarddeviationof temperatureerrorsin thatyear. A new surfacescheme
wasintroducedin June2000(vandenHurk etal. 2000),but it is tooearlyto assessits operationalperformance.It
is fair to point out that thestatisticspresentedin Figure8 areaveragedover a monthandover a largearea.The
errorsonaday-to-daybasiscanstill belarge,but arelesssystematicandareoftenrelatedto errorsin theforecasted
clouds or the presence of snow.
ACKNOWLDEGMENTS
Thisnotesummariseswork doneatECMWFduringthelasttenyears,in closecollaborationwith many colleagues,
in particularAnton Beljaarsand Alan Betts. Furthermore,the different contributions of Martin Miller, Jean-
FrançoisMahfouf, JoãoTeixeira,andEric Woodaregratefullyacknowledged.Part of this work will alsoappear
as a book chapter to be published by Springer-Verlag in 2002 (Viterbo and Beljaars 2002).
REFERENCES
Beljaars, A.C.M., P. Viterbo,M. Miller, andA.K. Betts,1996:Theanomalousrainfall over theUSA duringJuly
1993: Sensitivity to land surface parametrization and soil moisture anomalies.Mon. Wea. Rev., 124, 362-383.
Beljaars, A.C.M., andP. Viterbo,1999:Soil moisture-precipitationinteraction:Experiencewith two landsurface
schemesin theECMWF model.Global energy andwatercycles.K. Browning andR. Gurney, Eds.,Cambridge
University Press, Cambridge, 223-233.
Benjamin, S.G.,andT.N. Carlson,1986:Someeffectsof surfaceheatingandtopography on the regionalsevere
storm environment. Part I: Three-dimensional simulations.Mon. Wea. Rev., 114, 307-329.
Betts, A.K., 1992: FIFE atmospheric boundary layer methods.J. Geophys. Res., 97D, 18,523-18,531.
Betts, A.K., and J.H. Ball, 1997: Albedo over the boreal forest.J. Geophys. Res., 102D, 28,901-28,909.
Betts, A.K., andJ.H.Ball, 1998:FIFEsurfaceclimateandsite-averagedatasets1987-1989. J. Atnos.Sci., 55, 1091-
1108.
Betts, A.K., J.H.Ball, A.C.M. Beljaars,M.J.Miller andP. Viterbo,1996:Thelandsurface-atmosphereinteraction:
A review based on observational and global modeling perspectives.J. Geophys. Res., 101D, 7209-7225.
Betts, A.K., J.H.Ball andP. Viterbo,1999:Basin-scalewaterandenergy budgetsfor theMississippifrom theEC-
MWF reanalysis.J. Geophys. Res., 104D, 19,293-19,306.
Betts, A.K., andW.L. Ridgway, 1989:Climatic equilibriumof theatmosphericconvective boundarylayerover a
tropical ocean.J. Atmos. Sci., 46, 2621-2641.
Betts, A.K., P. Viterbo,andE.F. Wood,1998:Surfaceenergy andwaterbalancesfor theArkansas-Redriverbasin
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 17
from the ECMWF reanalysis.J. Climate, 11, 2881-2897.
Brubaker, K.L., D. Entekhabi,andP.S.Eagleson,1993:Estimationof continentalprecipitationrecycling. J. Cli-
mate, 6, 1077-1089.
Budyko, M.I., 1963:Atlasof theheatbalanceof theEarth (in Russian).GlobnaiaGeofiz.Observ., Moscow, 69pp.
Budyko, M.I., 1974:Climate and Life. Academic Press, 510 pp.
Chahine, M.T., 1992: The hydrological cycle and its influence on climate.Nature, 359, 373-380.
Chang, J.-Y., andP.J.Wetzel,1991:Effectof spatialvariationsof soil moistureandvegetationon theevolutionof
a prestorm environment: A numerical case study. Mon. Wea. Rev., 119, 1368-1390.
Changnon, S.A., (Ed.), 1996:The great flood of 1993. Westview Press, xii+324 pp.
Charney, J.G.,1975:Dynamicsof desertsanddroughtsover theSahel.Quart.J. Roy. Meteor. Soc., 101, 193-202.
Clapp, R.B.,andG.M. Hornberger, 1978:Empiricalequationsfor somesoil hydraulicproperties.WaterResources
Res., 14, 601-604.
Dirmeyer, P. andJ. Shukla,1993:Observationandmodellingstudiesof the influenceof soil moistureanomalies
on atmospheric circulation.Prediction of interannual climate variations. J. Shukla, Ed., Springer, Berlin, 1-23.
Dooge, J.,1992:Sensitivity of runoff to climatechange:A Hortonianapproach. Bull. Amer. Meteor. Soc., 73, 2013-
2024.
Douville, H., P. Viterbo,J.-F. Mahouf,andA.C.M. Beljaars,2000:Evaluationof theoptimuminterpolationandthe
nudging techniques for soil moisture analysis using FIFE data. Mon. Wea. Rev., 128, 1733-1756.
Dümenil, L., andL. Bengtsson,1993:Observationandmodellingstudiesof theinfluenceof landsurfaceanomalies
onatmosphericcirculation(futuredirections).Predictionof interannualclimatevariations. J.Shukla,Ed.,Spring-
er, Berlin, 25-47.
Fast, J.D.,andM. D. McCorcle,1991:Theeffectof heterogeneoussoilmoistureonasummerbarocliniccirculation
in the Central United States.Mon. Wea. Rev., 119, 2140-2167.
Garratt, J.,1993:Sensitivity of climatesimulationsto landsurfaceandatmosphericboundary-layertreatments—
A Review. J. Climate, 6, 419-449.
Gibson, J.K.,P. Kallberg, S.Uppala,A. Hernandez,A. Nomura,E. Serrano,1997:ERA description.ECMWFRe-
Analysis Project Report Series, 1, 72 pp.
Giorgi, F., 1990:Sensitivity of wintertimeprecipitationandsoil hydrology simulationover the westernUnited
States to lower boundary specification.Atmosp.-Ocean, 28, 1-23.
Horton, R.E.,1931:Thefield,scopeandstatusof thescienceof hydrology. Trans.Amer. Geophys.Union, 12, 189-
202.
Koch, S.E.,A. Aksakal,J.T. McQueen,1997:Theinfluenceof mesoscalehumidity andevapotranspirationfields
on a model forecast of a cold-frontal squall line.Mon. Wea. Rev., 125, 384-409.
Kuhnel, V., J.C.I.Dooge,J.P.J.O’Kane,andR.J.Romanowicz, 1991:Partialanalysisappliedto scaleproblemsin
surfacemoisturefluxes.Land Surface-Atmosphere interactionsfor climatemodeling. Observation,modelsand
analysis. E.F. Wood, Ed., Kluwer, Dordrecht, 221-247.
Lanicci, J.M., T.N. Carlson,andT.T. Warner, 1987:Sensitivity of the GreatPlainssevere-stormenvironmentto
soil-moisture distribution.Mon. Wea. Rev., 115, 2660-2673.
The role of the land surface in the climate system
18 Meteorological Training Course Lecture Series
ECMWF, 2002
Manabe, S., 1969:Climateandthe oceancirculation.1. The atmosphericcirculationandthe hydrology of the
earth’s surface.Mon. Wea. Rev., 97, 739-774.
Mintz, Y., 1984.Thesensitivity of numericallysimulatedclimatesto land-surfaceboundaryconditions.GlobalCli-
mate. J. Houghton, Ed., Cambridge University Press, 79-105.
Namias, J.,1958:Persistenceof mid-troposphericcirculationsbetweenadjacentmonthsandseasons.Theatmos-
phere and sea in motion (Rossby memorial volume). B. Bolin, Ed., Rockefeller Institute Press, 240-248.
Peixoto, J.P., 1973:Atmosphericvapourflux computationsfor hydrologicalpurposes.WMO Publ.357,World Me-
teorological Organization, Geneva, Switzerland, 83 pp.
Peixoto, J.P., andA. Oort,1983:Theatmosphericbranchof thehydrologicalcycle andclimate.Variationsof the
global water budget. Street-Perrott, A., M. Beran, and R. Ratcliffe, Eds., Reidel, London, 5-65.
Peixoto, J.P., and A. Oort, 1992:Physics of climate. American Institute of Physics, New York, 520 pp.
Pielke, R.A., andP.L. Vidale,1995:Theborealforestandthepolarfront. J. Geophys.Res., 100D, 25,755-25,758.
Rasmusson, E.M., 1967:Atmosphericwatervaportransportandthewaterbalanceof NorthAmerica.Part I: Char-
acteristics of the water vapor flux field.Mon. Wea. Rev., 95, 403-426.
Rasmusson, E.M., 1968: Atmosphericwater vapor transportand the water balanceof North America.Part II:
Large-scale water balance investigation.Mon. Wea. Rev., 96, 720-734.
Rasmusson, E.M., 1971:A studyof thehydrologyof easternNorthAmericausingatamosphericwatervapordata.
Mon. Wea. Rev., 99, 119-135.
Rowntree, P.R.,andJ.A.Bolton,1983:Simulationof theatmosphericresponseto soil moistureanomaliesoverEu-
rope.Quart. J. Roy. Meteor. Soc., 109, 501-526.
Schär, C.,D. Lhthi, U. Beyerle,andE.Heise,1999:Thesoil-precipitationfeedback:A processstudywith aregion-
al climate model.J. Climate, 12, 722-741.
Thomas, G., and P.R. Rowntree, 1992: The boreal forests and climate.Quart. J. Roy. Meteor. Soc., 118, 469-497.
Tiedtke, M., 1993: Representation of clouds in large-scale models.Mon. Wea. Rev., 121, 3040-3061.
Trenberth, 1999:Atmosphericmoisturerecycling: Roleof advectionandlocal evaporation.J. Climate, 12, 1368-
1381.
vandenHurk, B.J.J.M.,P. Viterbo,A.C.M. Beljaars,andA.K. Betts,2000:Offline validationof theERA40surface
scheme. ECMWF Tech. Mem. 295, 42 pp.
Viterbo, P., 1996:Therepresentationof surfaceprocessesin generalcirculationmodels.Ph.D.Thesis,University
of Lisbon, 201 pp. Available from the author, ECMWF, Shinfield Park, Reading RG2 9AX, England.
Viterbo, P., andA.C.M. Beljaars,1995:An improvedlandsurfaceparametrizationschemein theECMWF model
and its validation.J. Climate, 8, 2716-2748.
Viterbo, P., andA.C.M. Beljaars,2002:Impactof landsurfaceonweather. Vegetation,water, humansandthecli-
mate:A new perspectiveonan interactivesystem.M. Claussen,P. Dirmeyer, J.H.C.Gash,P. Kabat,M. Meybeck.
R. Pielke, Sr., and C. Vörösmarty, Eds. Springer-Verlag, to appear.
Viterbo, P., A.C.M. Beljaars,J.-F. Mahfouf,andJ.Teixeira,1999:Therepresentationof soil moisturefreezingand
its impact on the stable boundary layer. Quart. J. Roy. Meteor. Soc., 125, 2401-2426.
Viterbo, P., andA.K. Betts,1999a:Theimpactof theECMWF reanalysissoil wateron forecastsof theJuly 1993
The role of the land surface in the climate system
Meteorological Training Course Lecture Series
ECMWF, 2002 19
Mississippi flood.J. Geophys. Res., 104D, 19,361-19,366.
Viterbo, P., andA.K. Betts,1999b:Theforecastimpactof thealbedoof theborealforestsin thepresenceof snow.
J. Geophys. Res., 104D, 27,803-27,810.
Walker, J., andP.R. Rowntree,1977:The effect of soil moistureon circulationandrainfall in a tropical model.
Quart. J. Roy. Meteor. Soc., 103, 29-46.
Xue, Y., M.J.Fennessy, andP.J.Sellers,1996:Impactof vegetationpropertiesonU.S.Summerweatherprediction.
J. Geophys. Res., 101D, 7419-7430.
Yang, R.,M.J.Fennessy, andJ.Shukla,1994:Theinfluenceof initial soilwetnessonmedium-rangesurfaceweath-
er forecasts.Mon. Wea. Rev., 122, 471-485.