Applications of SHEBA/FIRE data to evaluation of snow/ice albedo parameterizations

JOURNAL OF GEOPHYSICAL

RESEARCH, VOL. 106, NO. D14, PAGES 15,345-15,355, JULY 27, 2001

Applications of SHEBA/FIRE data to evaluation of snow/ice albedo parameterizations J. A. Curry and J. L. Schramm Programin Atmospheric andOceanicSciences, Department of Aerospace Engineering Sciences, Universityof Colorado, Boulder

D. K. Perovich

u.s. Army Cold RegionsResearch and Engineering Laboratory,, Hanover,New Hampshire J. O. Pinto Programin AtmosphericandOceanicSciences,Departmentof AerospaceEngineering Sciences,Universityof Colorado, Boulder

Abstract. Climatemodelsusea wide variety.ofparameterizations for surfacealbedosof the ice-covered ocean.Theserangefromsimplebroadband albedoparameterizations thatdistinguish amongsnow-covered andbareiceto moresophisticated parameterizations thatincludedependence on iceandsnowdepth,solarzenithangle,andspectralresolution.Severalsophisticated parameterizations havealsobeendevelopedfor thermodynamic seaicemodelsthatadditionallyinclude dependence on iceandsnowage,andmelt pondcharacteristics. Observations obtainedin the Arctic Oceanduring1997-1998in conjunction with the SurfaceHeatBudgetof the ArcticOcean (SHEBA) andFIRE ArcticCloudsExperimentprovidea uniquedatasetagainstwhichto evaluate parameterizations of seaicesurfacealbedo.We applyeightdifferentsurfacealbedoparameterizationsto the SHEBA/FIRE datasetandevaluatethe parameterized albedosagainstthe observedalbedos.Resultsshowthattheseparameterizations yield verydifferentrepresentations of theannual cycleof seaice albedo.The importance of detailsandfunctionalrelationships of thealbedoparameterizations is assessed by incorporating intoa single-column seaicemodeltwo differentalbedoparameterizations, onecomplexandonesimple,thathavethesameannuallyaveragedsurface albedo.Thebaselineseaicecharacteristics andstrength of theice-albedo feedbackarecompared for the simulations of the different surface albedos.

1.

associated with a substantial

Introduction The correct determination of sea ice albedo in climate models

is essentialfor proper treatmentof the ice-albedo feedbackand related cryospheric processesin climate models. The icealbedo feedback mechanism is described as follows:

As tem-

peratures increase.the extent of snow and ice is reduced, decreasingthe surfacealbedo and increasingthe amountof sunlight that is absorbedby the Earth-atmosphere system. Conversely,a temperaturedecreasewill increasethe surfacealbedo and thusreintbrcethe cooling. Curry et al. [1995] furtherconsideredthe impact of internal sea ice processes(e.g., formation of melt ponds. aging of sea ice)on the ice-albedo t•edback mechanism.The ice-albedofeedbackhas been a particular sub-

retreat of sea ice, in some cases

the summertimearctic sea ice completely disappears.The predictedwarming in the high latitudeshas been at least partly attributed to the ice-albedo

feedback mechanism.

The albedo of a surface is defined as the ratio of the solar en-

ergy reflectedby the surtb. ce to that incidentuponit. The spectrally integratedvalue is a function of surfacecharacteristics, solar zenith angle, and atmosphericproperties, including cloudiness.The opticalpropertiesof a snowor ice cover which determineits albedo are the coefficientsof spectralabsorption and scatteringand the associatedphasefunctionof single scattering [e.g., Maykut et al., 1992]. Snow and ice are transparent in the visible regionand moderatelyabsorptivein the near infrared. As described by Grenfell and Maykut [1977] and

ject of discussionin the contextof greenhousewarming [e.g., Grenfell and Perovich [1984] from field observations,snow Spehnan and Ylanabe. 1984; Dickinson et al., 1987; Wash- opticalpropertiesdependon grain size and shape,depth of the ington and Meehl. 1986; Ingram et al., 1989]. Atmospheric snow layer and optical properties of the underlying surface, models with doubled CO: concentrations have found that the surfaceroughness,liquid water content, and any impurities. warming is considerablyamplified in the Arctic [e.g., (IPCC), Seaice optical propertiesdependon ice thickness,brine and 1990]. Most model projectionsof amplifiedpolar warmingare air bubbles in the ice, and surface conditions, such as an ice Copyright 2001by theAmericanGeophysical Union. Papernumber2000JD900311. 0148-0227/01/2000JD900311 $09.00

cruston the surface,frost flowers on young ice, or melt ponds [Perovich,1996]. The albedoof d• snow has a specularcomponentand hencedependson solar zenith angle. The effectof clouds on incomingsolar radiation is to changethe spectral distributionand to alter the zenith angle of the radiation inci15,345

15,346

CURRY ET AL.'

SHEBA ALBEDO PARAMETERIZATION

dent on the surface. The net effect of overcastskieson the spec- bedoparameterization may be used,but if the modelis productrally averagedalbedo of snow and ice surfacesis to increase inga seaice field that is substantiallyin error,then the simulated albedo will be in error.

the albedo.

The SurfaceHeat Budget of the Arctic Ocean (SHEBA) A diversityof ice and snow albedo schemes are usedin climatemodelsand uncoupled sea ice models (for a review, see [Perovichet al., 1999] and FIRE Arctic Clouds Experiment Barry, [1996], also section3 of this paper). Most parameteriza- [Curt3,,et al.. 2000] observationaldataset providesan excelof sea ice altionsare very,simple,dependingon surfftcetype (ice. snow,or lent opportunityto evaluateparameterizations open water) and surfacetemperature.A few parameterizationsbedo. This experiment,describedin section2, providescomsurfacealbedomeasurements of a multiyearice floe include snow depth and ice thickness and considerationof prehensive melting snow. Fewer still include spectraldependence (visi- over an entire annual cycle. In addition to surfacealbedo ble versus near infrared) and zenith angle dependence. These

measurements, observationswere made of all the parametersre-

prescribed albedosare usuallybasedon limiteddatafromsite- quiredas inputvaluesto the diverseseaice albedoparameterspecificcasestudiesor are subjectively chosen. In somein- izations. Furthermore,a completesetof observationsof surface stances,modelerstune their albedo formulationsto give a good controlsimulation. For example..•lanabe and Stouffer [1980] used unrealistically low albedosto offsetthe model'sneglect of oceantransports.Somemodelersuseartificiallyhigh melting surtb. ce albedosto prevent ice thicknesseswhich are too small

fluxes, sea ice mass balance. and ice floe dynamics was ob-

models,a summertimesurt3.cealbedo that is too high can give

the simulations of the different surface albedos.

tained. which

can be used to force and evaluate a one-

dimensional sea ice model.

In thispaperwe usethe SHEBAdatato evaluatea varietyof seaice albedoparameterizations thatareusedin climatemodels [e.g.,Hibler. 1980]. Suchtuning can have undesirableconse- and uncoupledsea ice models. The seasonalcycleof the alis evaluatedspecificallyfor the winter quences. In uncoupledmodels,tuning of the sea ice albedo bedoparameterizations may concealseriouserrorsin the model formulation. In cou- periodof freshand dry snoxv.the !atespringperiodof snowpled models.tuning of the albedohas considerableeffectson melt. the summermelt season,and the autumnalperiod of fleezof detailsand thnctionalrelationshipsof model climate sensitivity[e.g., IYashingtonand Meehl, 1986]. ing. The importance For climate simulations, it is not sufficient to have a correct the albedoparameterizations is assessed by incorporatinginto surface albedo and sea ice mass balance t•)r the control simulaa single-columnseaice modeltwo differentalbedoparametertion, but the correctphysical dependenciesmust be included izations.onecomplexandonesimple,whichhavethe sameanso that the sea ice albedo feedback mechanism and its interacnuallyaveragedsurfacealbedo.The baselineseaice characterarecompared for tion with other feedbacksis correctly simulated. In coupled isticsand strengthof the ice-albedotEedback the wrong sign of the cloud-radiative lbrcing [Cttr.rvet al.. 1996].

More sophisticatedsurt3.cealbedo schemesare available 2. Data [e.g..Schramm et al., 1997: Flato and Brown, 1996]. The The data used to evaluate the surthcealbedo parameterizaSchrammetal. schemehas synthesized a variety of observations are obtained from the SurfaceHeat Budget of the Arctic tional and modelingresultsinto a surfacealbedoparameterizaOcean(SHEBA) project [Perovich et al., 1999] and the FIRE tion that considers both the spectral variation in albedo and Arctic Clouds Experiment[Curry et al., 2000]. The SHEBA its dependence on solarzenithangle. Five surfacetypesare inobservationswere madeduring the period October 30, 1997 cluded,manyof whichcan be presentin a single grid cell at a through October 10, 998. The Canadian Coastguard ice giventime step: new snow,meltingsnow,bareice,meltwater breakerDes Groseilliers was deployed in a multiyearice floe ponds,andopenwater,for eachof thefour intervalsin the solar at 75ø16.3'N, 142ø 41.2' W. Over the course of the field spectrum.Becausethisschemedepends on sur/hcefeaturesthat study,the SHEBA ice campdrifted considerablynorthwestare not easyto simulate(e.g., melt ponds)and cloud properties, ward, reaching80•øN162øWby the end of the experiment.The errorsin othercomponents of the modelmay lead to a lessrealFIRE Arctic CloudsExperimentincludedaircraftoverflightsof istic control simulation than a cruder albedo parameterization the SHEBA ice camp duringApril 8 throughJuly 28, 1998. scheme.However, it is preciselythese functionalalbedo deA variety of measurements of the atmosphere, sea ice, and pendencies that are necessary to simulatethecorrectice albedo upper oceanwere made during SHEBA and the FIRE Arctic feedback. CloudsExperiment[Perovichet al.. 1999; Curry et al., 2000]. Climatemodelershavejustified using simple parameterizaHere we focusspecificallyon observationsof surfacealbedo tions based on the lack of observations against which to and also the meteorologicaland surthceparametersthat are evaluate them. Some limited comparisonsof modeled versus usedas inputsto surfacealbedoparameterizations for multiyear observed surfacealbedos have been conductedusing satellite seaice: surfaceskin temperature.surfaceair temperature,snow estimates[e.g.,Rossand }F'alsh,1987].Severalsatelliteanalydepth,ice thickness,melt pondfraction,and melt ponddepth. sesof arctic surfacealbedohave been prepared[e.g., Rossowet al., 1989;Lindsay and Rothrock. 1994]. Uncertaintiesin the satellite-derived

surt3.ce albedos are associated with

cloud

2.1.

Instrumentation

Measurements were made of wavelength-integrated and thresholding,aerosols,ozone, and water vapor. Since these albedo. using Kipp and Zonenradianalysesare madefor only clear sky pixels,the albedovalues spectralvalues of surt3.ce are biasedlow relative to all-sky values. Although these sat- ometersthat integrateover wavelengthfrom300 to 2800 nm. ellite-derivedsurfacealbedo data setshave some significant er- Thesevaluesare accurateto within 0.01. Spectralalbedosfrom

rors,theyarestillof someusein evaluatingthe performance of 400 to 2000 nmwerealso measured usinga SpectronEngiclimatemodels. However, comparingmodeledto satellite- neeringSE-590 spectroradiometer. Albedo measurements were derived fields is not sufficientto evaluatethe sur•3.cealbedo madeat leastweeklyfrom April to Octoberandevery other day

parameterization. For example, a pertEctly accurate seaice al- l¾omJune to August. Measurements weremadeevery2.5 m

CURRY ET AL.'

SHEBA ALBEDO PARAMETERIZATION

05/09

06/17

15,347

1'0 t.

0.8

0.6 f' 02

04/01

04/20

05/29

07/06

07/25

08/14

09/02

09/21

10/11

Figure 1. Observations of surfacealbedo obtained at the SHEBA ice station. Circles, surfacemeasurements along the 500-m albedo line: pluses.values obtainedfrom the C130 aircraft.

along a 200-m-long albedo line that encompassed different 2.2. Description of Observations snowand ice conditionsover the heterogenous sea ice cover of The annualcycle of surfacealbedo is shownin Figure 1. the multiyearice floe. The values fbr each day were used to Observations are presentedfrom surt•.cemeasurements obtained computean areally averagedalbedo. In addition to the surface-based measurements,observations of surl•.ce albedo were made t¾om the NCAR C-130 research air-

fromthe albedoline and fromthe C-130 aircrafl;the slightly lower values obtained from the C-130 aircraft reflect contribu-

•,..,,+• ,,• v,,a:,,•:,,., • ........... e...... ,r• ^•e• [Pop tionsf¾omthin ice and leads. During April and May, when the et al.. 2000]. which includes uplooking and downlooking surt•.ceconsistsof dry snow.surthcealbedoaveraged0.84. The multichannelradiometersthat have a hemisphericfield of view melt season lasted a total of 80 days t¾omlate May to midand seven channels t¾om the ultraviolet

to the near infrared

(0.2-3.9_m). The aircraRobservations covereda regionof 20 x 20 km, centeredon the SHEBA ice camp. Ice thickness was measured using ablation stakes and thickness gauges at more than 100 different locations on the ice floe. The gaugeswere manuallyreadever>,other day from Juneto Augustand approximatelyweekly the remainderof the year.Verticalprofilesof ice temperatures were automaticallyrecorded every hour throughout the experimentat seven sites [Perovichet al., 1999]. Snowdepthwas measuredevery 1-5 m along the main 500-m survey line either manually using a graduatedski pole or a magnaprobe[Perovich et al., 1999]. Manual measurements were rounded off to the nearest centime-

ter. and magnaprobevalues were accurateto within 0.5 cm. Snow depths xvereaveragedalong the survey line. Measurementswere made every 1-2 weeks from October to May and every other day from Juneto August.

August. During the first two weeks of June when most of the snowmeltoccurred,the average surtS. ce albedo was 0.77. Melt

ponds tbrmedin early June and continued to develop into August. The surfacealbedo decreasedthroughout the summer

melt seasonand reacheda minimumvalue of 0.38 on August 12. Beginningon August 12. the melt pondsbeganto freeze, causingthe albedoto increase. Snow beganto accumulateon the ice in late August and by mid-Septemberthe snow cover was roughly 10 cm deep. The albedo began to increase,reaching its winter-springvalue by the end of September. The annual cycle of surf•.ceconditions at SHEBA are describedby Perovich et al. [1999]. The thickness of the unde-

formedmultiyearice at deployment was 1.7-2.0m. The ice grew 75 cm duringwinter and melted 110 cm duringthe summermelt season. Snow depth increased quickly in the fall and more

slowlythroughthe winter, reachingan averagedepth of 34 cm in the spring. The first raint•11of the seasonon May 29 her-

Meltpondcharacteristics (pondfraction andponddepth) alded the onset of

snowmelt.

Most of the snow cover melted in

snowdriftslasting were measuredalong the albedo line. During July, statistics justa few weeks,witha few small,scattered on melt pond •¾actionwere also obtained t¾omthe C-130 video until mid-July. There were occasionalsnow flurries during camera[Tschudi et al.. 2000]. The surface-based melt pond summer,but it was not until late August that snow beganto analysiscoveredthe 200-m-longline. while the aircraftanaly- accumulate on the surl•.ce. Melt ponds formedin Juneas the snow melted,then deepsiscoveredanareaof 20 x 20 kin. During the later part of July ened and grew in horizontal extentduring Junethrough midthe surface-based melt pondtYactionexceededthe aircraflvalue, mostprobablya resultof small-scalevariabilityin pond cover- August. Along the albedo line. ponds covered 38% of the surt•.cearea during the peak of the melt seasonand reachedan age. Surfaceskin temperaturewas measuredusing a broadband Eppley radiometerand assuminga surt•.ceemissivity of 0.99 [Claffeyet al., 1999]. Measurements weremadeevery10 s, and hourlyaveragedvaluesare used in this study. Comparisonof severaldifferentmethodsof determiningsurfaceskin temperature suggestsan uncertaintyin the surfacetemperatureof about

averagedepthof 39 cm.

ments are accurate to within about 0.1 øC.

July 17.

Surfaceskin and air temperaturesare shown in Figure 2. When comparedwith climatology,SHEBA was relatively cool during winter and warm during spring. The onset of summer

melt(May 29)is reflectedby surfacetemperature reaching0øC. During the summermelt period the surfacetemperatureoscilIøC. Air temperature wasmeasured usingVaisalatemperaturelatesslightlyaroundthe melting point but, accountingfbr the sensorsat various levels in the atmosphere;the value used errorof themeasurements, appears to remainat the meltingtemherehas been interpolatedto 10 m. Air temperaturemeasure- peratureduringthis period exceptfor during a clearperiod on

15,348

CURRY

ET AL.'

SHEBA ALBEDO

PARAMETERIZATION

05/09

05/29

06/17

07/25

1.0 o

0.8

•. 0.6 •

0.4

o3 0.2 0.0

04/01

04/20

07/06

08/14

09/02

09/21

10/11

1.0 0.8

0.6 0.4

I

0.2

GFDL (b)

04/01

04/20

05/09

05/29

06/17

07/06

07/25

08/14

09/02

09/21

10/11

1.0

0.8

0.6 •

0.4

co0.2 0.0

04/01

•

,

SHCE

!

04/20

,

I

05/09

,

I

05/29

,

I

06/17

,

I

07/06

,

I

07/25

,

I

08/14

,

I

09/02

,

I

09/21

(c) ,

10/11

Plate 1. Comparisonof parameterizedversusobservedsurfacealbedos. Parameterizedalbedo values were determinedusing SHEBA observationsas input. Solid black line representsthe observedalbedo and colored lines the parameterizedalbedos. (a) NCAR CCM3 [Briegleb and Bromwich, 1998; PW79 Parkinson and lPbshington, 1979;Hibler. 1980], (b)UKMO [Ingramet al., 1989: Rossand Walsh, 1987], GFDL [Manabe et al., 1992], (c)Flato and Brown [ 1996] and Schrammet al [1997].

CURRY ET AL.:

SHEBA ALBEDO PARAMETERIZATION

15,349

270 '•-

• 265;: •

'•

'

1

260

•

I::::l.

'

E 255

'

!,-

-

250

-: r

"

245:

!

04/21

'

05/11

'

05/31

'

,

'

•

06/20 07/10

,

07/30

!

08/19

•

09/08

,

•

09/28

Figure2. Observations of surface temperature obtained fi'omthemeterological towerat theSHEBAice station. Solidcuve,surfaceskintemperature: dashed. surfaceair temperature (10 m).

3.

Albedo

Parameterizations

0.80 =io.6•-o.o3h

In this sectionwe focuson the albedosobservedalong the 200-m survey line over multiyear sea ice, during the period April to October. The albedoparameterizations chosenfor this

[0.65

0.65

studyaim to represent the spectrumofparameterizationsused in generalcirculation models(GCMs), regional modelsof the Arctic basin.and uncoupledsea ice models.

I0.45 +0.04T a [0.45

if T.,.< 268 K

if 268 K < Ts lm} ifh < lm

where

o•i

if Ts > Tm

o½* = O• i +0.025(T m- Ts) if(T.,- 10K)