"llllla/s oJC/aci%gy 25 1997 (- Internati onal G laciological Society

The changing albedo of the Greenland ice sheet: im.plications for clim.ate m.odeling

ANNE W. NOLI:\i, J ULIEN;\fE STROEVE

CoojJerative Illstitlltelor Research in Environmental Seiellce.l. Cllil ' ersi~J I oJColarada, Bauhin CO 80309, US. A.

A BSTR ACT. Although the snow albedo feedback mec hani sm has bee n shown to am­ plify global warming effects in nearl y all models of global climate, it continues to be rep­ resented as a simplisti c pa ra meteri zation. Here, wc demonstrate how changes in snow­ pack energy-ba lance d rive the seasona l flu ctuati ons in sno\\' albedo for the G ree nl a nd ice sh ee t. For a deta il ed, point-based im·esti gation of the relati onship between snow pack energy ba lance a nd albedo, two models arc coupled together; one that caleulates snow g rain-size a nd the other that uses those g rain-size data as input to a radiati \'(:-tra nsf'c r code to obta in spectral a lbedo. These data ind icatc that in thc near-infrared wa\'C lcngths, albedo \'alues drop nearl y 20% during a 10 day pcri od during which grain-sizcs increased dramaticall y. Satellite data were used to map momhly changes in a lbedo o\'C r the enti re G reenland ice sheet during the spring a nd sUl11m er months. These 111 0nthly albedo im ages indicate albedo reducti ons of as much as 80% in coasta l regions. E\'en in areas that ex­ perience lillle or no melt, a lbedo dec reases of 10- 20%, wc re common. From these res ults, it is cl ear that snow a lb cdo parameterizati ons for climate models must incorporate the dynamics of snowpack energy balance.

INTRODUCTION bounda ry condition in GCl\Is, onl y recently has a more phys icall y based treatment of snow a lbedo been added to a Snow is the brightest substa nce of considerable extent on the GC:-'l (.\ la rshall and O glesby, 199+; )'larsha ll and others, surface of our pl anet a nd, because of its hi gh albedo COlll­ 199+), This albedo parameterizati on ta kes into account pared to other natural surfaces, seasona l and perennial changes in snow grain-size, snoll' depth, solar zenith angle, snow cO\'e r pl ays a significant role in the global energy and li ght-absorbing impurities in th e snow. In their applica­ balance (Ba rry, 1985), The connec ti on be tween snow albedo tions of a rea li sti c snow albedo parameteri zati on, ~ I a r s h a ll and climate has long been recogni zed as an importa nt feed­ a nd O glcsby (199+) rera n Nati onal Celller for Atmos pheri c back mecha ni sm (M eehl, 1984; Cess a nd others, 1991), Snow­ Research (NC AR) CCMI simulati ons fo r Antarcti ca and albedo feedback has bee n shown ignificantly to amplify the Northern H emisphere. :\Tot surprisingly, they noted a global warming ( ~ I ee hl and Washington, 1990), The sno\\'­ marked imprO\'e ment in the model's represe ntati on or albedo feedback mechanism is looked upon as a positi\'C changes in seasonal snow cO\'C r. This new parametcrizati on feed back that occurs when warmer temperatures reduce has not ye t been widely incorporated into other G C ~ I s . the snow- covered a rea, revealing a darker substrate a nd I n other work, connecti ons between snowpack phys ical p ro moting increased radi a ti ve heating. properties a nd albedo ha\'e bee n measured and mode led, In this study, we ha\'e revealed an additional aspec t of \\'ith their two-stream radiati\'e-transfer model of snow the snow albedo feedback: that the dynamics of snowpack albedo, Wiscombe and Warren (1980) showed hO\\' an energy balance di reetly influence snow a lbedo over seasonal increase in snow grain-size causes a decrease in snow time-scale a nd over large regions by th e changing size of albedo. J n the visible wa\'elcngths, (0.+ 0.7 {l m ), a single ice snow gra ins in the near-surface layer. Using the G reenland particle transmits nearl y all incidelll radiati on a nd tends to ice sheet as our study area, we have inves ti gated the vari­ scaller mos t of this energy into the forwa rd directi on. Snow abili ty of albedo in detail at a point and spati ally across the is very hi ghl y refl ecti\'e in that pan of the spectrum with ice sheet, and have related these cha nges ro sno\l'pack albedo \'a lues generally ranging fr om 0,80 to 0,98, Ice energy and mass flu xes, becomes moderately absorbing of solar energy in the nea r­ To the human eye, the vast G reenl and ice sheet, measur­ infrared part of the spec trum (0.7- 2.5 {lm ). Figure 1 shows 6 2 ing 1.7 X 10 km , gives the impress ion of homogeneity. that changes in snow grain-size have their greates t influence H owever, large seasonal flu ctuations in surface climate at on a lbedo in the near-infra red, Depending on grain-size these hi gh latitudes create substa nt ial changes in surface a nd wavel ength, infrared snow albedo can range from near energy bala nce (Ste ffen and others, 1995, 1996) Pa rameteri­ zero to 0,8. In Figure 1, the relati onship between spec tral zati ons of snow albedo in mos t of the currelll general circu­ a lbedo and snow grain radius is shO\\'n using two-stream lati on models (G C M s) ha\'e been simpli sti c representations model res ults, Increases in snow gra in-size arc primaril y that vary onl y as functi ons of latitude a nd ai I' temperature, responsibl e fo r lower values of spec trail ), integrated snow Although snow albedo has been identified as a sensitive albedo and because a significant proporti on of the incident

Downloaded from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. 51 Nolin and Stroeve: Albedo of the Greenland ice sheet

the experiment period was cold but about halfway through, 1.0 '.-...... : .... " __ rod ius= 50 um the snowpack warmed and began lo melt. '\ \ ... \ To examine and quantify the spati al changes in a lbedo 0.8 \., \ , rod ius=500 um " , across the Greenland ice sheet, remote sensing images for --.. .. radius=1000 um i , M ay, June and J uly 1991 were used. Melt on the ice shee t Q) \ 1, / \ usually becomes fairly extensive duringJune, and maximum g 0 .6 \ I I " , E i , values for melt area usua lly occur inJuly (Abdalati and Sref­ u Q) ~ \, fen, 1995; Mote and Anderson, 1995). Although the images 't 0.4 . , "- ~ ,', , and the point-based measurements are not for coincident \./ \ \ .. -\ Solar illumination \ years, similar processes affec t both datasets. 0.2 angle = 30 degree s \ i.. , ... . - ., I Sn owp ack-en e r gy and m ass-ba la n ce model i ' o . o~~~~~~~~~~~~;'~~ __ ~~~~~~ 0.0 0.5 1.0 1.5 2.0 2 .5 SNTHERM, a snowpack- energy and mass-balance model wavelength (microns) OOl'dan, 1991), requires input values of air temperature, relative humidity, wind speed, air press ure, incoming a nd Fig. I. Modeled spectral directional hemispheric riflectance if outgoing shortwave radi ati on, incoming longwave radia­ snow Jor three grain -sizes. Fine -grained new snow is rep ­ ti on, a nd precipitation information at regul a r time-steps resented by the 50 I1m grain radius curve, the 200 flm grain (in this case hourly averaged data). radius curve represents medium-grained snow, and coarse­ For the 1994 ex periment period, snow-accumulation grained melting snow is represented by the 1000 fun Cllrve. data (accumulation rate and grain-size) were obta in ed fr om a combination of synoptic weather observations (every solar radiation is in the near-infrared region, this change in 3- 6 hours), grain-size measurements and additional field albedo has important consequences for surface radiation observations. For the 1995 period, additional snow accum­ balance. A preliminary study by Davis and others (1993) ul ati on information was supplied by a snow-height sensor demonstrated how aging and warming of a snowpack leads mounted on the instrument tower. to an increase in grain-size and an ensuing decrease in near­ The model was initia lized with snowpack physical prop­ infrared albedo. erties including grain-size, temperature and density (from In this paper we investigate these principl es through a field measurements). SNTHERM then generated a time­ combinati on of theoreti cal models, fi eld observati ons and series of snowpack temperature, grain-size and densit y satellite measurements to: profiles corresponding to the time intervals of the input meteorologic data. (I) demonstrate the connection between surface climate, For metamorphism in dry snow, SNTHERM uses a snowpack energy balance, snow grain-size and surface grain-growth algorithm formul ated by Jordan (1991), based albedo; on the previous work of Colbeck (1983): (2) map and quantify the spatia l and temporal variability of Cl 1000 ( albedo on the Greenland iee sheet; ad T)6 laTI at = d D eOs P" 273.15 CkT az (3) rel ate these measurements a nd model results to current where d is snow-grain di ameter (m ), G1 is a grain-growth snow albedo parameterizations in climate models. parameter ( m+ kg- I), Dcos is the effective diffusion coeffi­ cient for water vapor (m2 s \ P" is atmospheric pressure METHODOLOGY (mb), T is snow temperature (K ),CkT is the variation of saturation vapor press ure with temperature (Nm 2 K ) a nd 1'0 examine in detail the relationship between snow pack z is the height (m ) above the snow-ice j nterface. energy balance and albedo, two coupled models arc used. A se parate formulation is used for grain growth in wet One model calculates the energy a nd mass fluxes within snow (Colbeck, 1982) that incorporates the liquid water the snow pack and the second uses radiative-transfer code fr action, f: to compute the spectral albedo from snow optical proper­ ti es. Both models are one-dimensional (I -D ) and are run ~~ = ~2 (f + 0.005) for hourly time-steps. The energy and mass balance model when 0.0 < f < 0.09 and is driven by hourly meteorologic data and calculates, among other things, the grain-size for each snow layer. From this ad G2 at = --;1(0.14) grain-size output, NIie theory is used to compute the optical 2 properties of the snow grains and these a re used as inputlo when f:2: 0.09. The variabl e C2 has units of m S- I and is the radiative-transfer model to obtain spectral albedo at currently set at 4 x 10 I-.') each time-step. These two models are described in greater During the experiment period, 2- 25 May 1994, there detail below. The coupled model was tes ted for two time per­ was no significant snowmelt. M elt occurred only during iods: 2- 25 May 1994 and 15 M ay- 12 June 1995. Measure­ the second half of the 15 M ay- 12 June 1995 experiment ments were made during these time periods at the ETH/ period. CU camp located on the western portion of the Greenland At hourly time-steps, SNTHERM calculates an effec­ ice sheet (located at 69.58° N, 49.30° W ). The 1994 ex peri­ tive sphere size, representative of the panicle-size distrib­ ment period was characteri zed by cold temperatures and ution, for each snow layer in the snowpack. To test that this numerous, small accumulation events during which fin e­ effecti\'e grain-size can be related to a physical quantity such grained new snow was deposited. In 1995, the first part of as mean snow grain-size, measurements of grain-size in the Downloaded52 from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. /VO/ill and Stroeve: Albedo qftlz e Greenland ice sheet

top 5 cm were made over the course of the 1994 experiment curvature, scanning geometry, orbital errors. Images are period. Grain-size in the top 0- 5 cm of the snowpack was then georegistered to a polar stereographic projection using used for the model calculations. \\'iscombe and Warren a navigation code deYeloped by Baldwin and Emery (1993). (1980) haw shown that albedo is primarily a ffec ted by The AV HRR data are then cali brated using the method rr rain-size in the top few cm so measurements of the sur­ currently employed by the NOAA/NASA Pathfinder pro­ face-l ayer grain-size are su ffi cient. Though onl y the near­ gram. This melhod combines both a relative calibration surface layers haye a direc t affect on snow albedo, the full (using the pre-launch coeffi cients) with aircraft measure­ snowpack is modeled to account properly for heat and mass ments to calibrate channel I and 2 as functions of days ­ transfers that drive these changes in near- surface grain-size. since-launch. The nex t step involves calculating surface albedo from Two-stream r adiative-tran sfer m od el the AVHR R data. For this, the atmospheric correc tion algorithm 6S (Ta nre and others, 1992) was used. I n addi­ To calculate spectral albedo fo r these two experiment peri­ tion, a correcti on for the bidirecti onal renectance of snow ods, onl y snow grain-size in the 0- 5 cm depth is needed. The was made using anisotropi c refl ecta nce factors calculated required optical properties include optical thickness, single from a discrete-ordinates radiative-transfer model (Stroeve scattering albedo and an asymmetry parameter. The single and Nolin, in press). scattering albedo represents the probability that li ght inci­ dent 0 11 a particle will be scattered rather than abso rbed. The res ulting direction of the scattered li ght is described RESULTS by the scattering-phase function of a particle P(e), where e Measured and modeled grain-s ize is the scattering angle. The asymmetry pa rameter, g, is a parameterization of the particle scattering-phase function. 199..J r n this case, because the snowpack was thick enough to be For the period 2- 25 l\ Iay the SNTHER M model generated considered "opti call y-thick" (where the substrate has vir­ the snow surface-l ayer grain-size data that could then be tually no effect on albedo), the optical thickness, a dimen­ compared with measurements of snow grain-size from the sionlcss number, was prescribed to be 2000. Both the single same period. Figure 2 shows good agreement between the scattering albedo and asymmetry parameter are functions mean measured grain-size a nd the modeled grain-size. The of grain-size and these are calcul ated using the refracti\'e irregul a r, sharp d ips in modcled grain-size arc the res ult of indices of ice a nd the effective grain radius from the sno\\'-accumulation epi sodes, mostl y from blowing a nd energy-balance model (\\'iscombe and \ Varren, 1980). Using drifting snow bUl also from prec ipitation. Following an these scattering and absorbi ng parameters, the two-stream accumulation e\'e nt, the surface snow grains showed steady model was used to calculate the spec tral albedo of the snow­ increases in size as metamorphism proceeded in the snow­ pack at each hourly time-step. pack. While the maximum and minimum grain-size meas­ Sn owpack m easurem en ts urements have a wide range (see Table I), the mean grain­ size closely tracks changes in modcled grain-size. The lar­ Snow gra in-size was measured II times oye r the 199+ model gest snow grains renect the effec ls of grain clustering while run period. A gridded card and hand lens were used to the sma ll es t g rain were primaril y broken crys tals. measure the maximum, minimum and mean grain-size for replicate sa mples of snow from the top 5 cm of the snow­ 1995 pack. Six snowpits were exca\'ated to pro\'ide density and Model res ults from the 1995 experimental period show gen­ snow strati graphy information. erall y small changes in grain-size for the first half of the time During the 1994 experiment, snowpack renec ta nce was se ries when air and snow temperatures were cold. Figure 3 measured using a portable field spec trometer that operates shows a dramati c increase in surface-layer grain-size begin- in the 400- 2500 nm spectra l range. These refl ec ta nce values were collected for compa ri son with the albedo estimates cal­ ETH /CU Snow Groin Si ze culated fr om the radiati\'e-transfer model. 50 0~~~~~~~~~~~~~~~~~~~~~ To initia li ze snowpack te mperature in the snowpack 6 m easured gro in size 1994 __ SNTHERM groin si ze energy and mass-balance model, temperature data were 400 acquired from thermistor a nd fin e-wire thermocouples pl aced in the snowpack. Snow grain-size data fi-om snowpit profiles were also used to initiali ze the model.

R em ote-sen s ing-image d ata ana lysis

For the spati al component of this analys is, satellite images of the Greenland ice shee t were acquired from the National Oceanic and Atmospheric Administration 11 Adyanced Very High Resolution R adiometer (NOAA/lIAVHRR ) sensor. For this research, AVHRR channel I (0.58- o 100 200 300 400 500 0.68 J.lm ) and channel 2 (0.725- 1.10 ~lm ) were used. Because May 2 - 25, 1994 ( hours since start of model run) snow types on the G reenl and ice shee t extend O\"Cr tens of km, the AVHRR Global Area Coverage (GAC) data with Fig. 2. SNTHERM modeled snow grain sizeJor 2- 25 1\I1a) a nomina l spatial resolution of + km were used. Each image 1994. Solid fine is the model results and the triangles are is first co-l ocated by correcting for distortions due to Ea rth measured snow gmin -sizeJrom the top 0- 5 cm.

Downloaded from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. 53 Nofin and Stroeve: Albedo ift he Greenland ice sheet

Ta ble 1. ilJeasllred grain -sizes from the snowpack sU1JaceJo r ETH/CU Modeled Albedo May--J une 1994 1. 00 ~_,,_~_c_'='_~ __,,_:-r_~_~_ ~ __,:"",_:'r_r_~_~_ ~__~_~_~_r_~_ ~ __ ~_~_~_~_T_~_~_~_~_~_ ~ _ "..,~~

0.90 llfinimum radius AIean radill s A Iaximum radius '. o pm pm {Lm u .8 0.80 :;;: 100 200 500 ~ u., 0 .70 a. 100 190 550 (/) 50 100 175 ...... 0.70 microns 300 500 0 .60 100 __ 1.03 microns 100 -1-00 1250 1994 100 150 300 _._._._._ broadband albedo

50 100 250 o 100 200 300 400 500 50 100 300 May 2 -- 25. 1994 (hours since start of model run) 50 150 350 100 180 400 75 125 350 Fig. 4. Two-stream modeLed spectral albedo Jar three wave­ lengths Jor 2- 25 May 199 J.. Top, second and bollom curves are Jor wavelengths q! 0.46 pili, 0.70 {L nl and 1.03{un, res­ pectivel)!. The second-:from -bollom cu rve re/Jrese nts measll red ETH / CU Modeled Snow Groin Size broadball d aLbedo Jor the same period. 500~~~~~~~~~~~~~~~~~

1995 modcled albedo values. Albedo increascs and dec rcases at 1.03 {Lm correspond to accumulation e\" cnts from both pre­ -;;;- 400 c cipitation a nd bl owing/drifling snow. These new snow eu grain-sizes are typi call y \'e ry fin e, with g rain radii of about E 100 pill. A warming trend from 12-16 M ay created a 55°/., ~ 300 increase (fr om 250 to 400 J.lm ) in grain-size a nd res ulted in '6 o er a 13% dec rease (.75 to .65) in albedo in the near-infra red. c '0 Only sma ll changes in albedo a re evident for 0.70 {UTI and l5 virtuall y no changes occurred for 0.46 pm.

100~~~~~~~~~~~~~~~~~~~ 19.95 o 100 200 300 400 500 600 For the 1995 model period, thc spec tral albedo data closely, May 15 -- 25. 1995 (hours since start of model run) but im'erscl y, mirror the grain-size data with a strong decrease in albedo beginning with the onse t of melt. This change is most strongly seen in the 1.03 {Lm data in Figure 5 Fig. 3. SNTHERM modeled sno wgrain-si::.eJor 15 M a..v- 12 where lhere is a nearl y 20% relative difference in albedo J une 1995. Grain -sizes show a large increase starting about31 comparcd to the Sla rt of the model run. As expected, albedo JIlay when th e snowpack is warming and melt is beginning.

ning on about 31 !\1ay when the effec ti\'e grain radius ETH / CU Modeled Albedo increases from about 100 pm to nearl y 500 lLm by the end 1.00~~~~~~~~~~~~~~~~~~~~ ------of the period. This increase in g rain-size coincides with the ------onse t of melt (Steffen and others, 1996). Subsequent accum­

ulati on event were sma ll and, with the warm snowpack, 0.90 these smaller surface grains were quickl y melted or sub­ 0 0.46 microns limed away, revealing the older a nd larger grains beneath. u., D :;;: It appears that once melt begins, such sma ll accumulation 0 .70 microns 0 .80 events have little effect on grain-size. "§ U., 1.03 microns a. Measured and rnodeled albedo (/) 0.70 1994 Figure 4 shows a time seri es of modelcd spec tral albedo 1995 \'alues for three wavelengths (0.40 {Im , 0.70 {Im and 1.03 pm) 0 .60 for the 1994· experimenl period. Broadband albedo meas­ 0 100 200 300 400 500 600 May 15 -- 25. 1995 (hours since start of model run) urements also shown in this fi gure closel y correspond to trends and magnitudes of the modeled values. These broad­ band albedo data were calculated using a pair of pyran­ Fig.5. Two-stream modeled spectral albedo Jor three -wave­ omcters, one upward a nd one downward looking; their lengths Jor 15 M a..y--12 June 1995. Top, middle, and bollom speetral range is 0.3- 3.0 pm. These measured values corres­ curves aTeJor wavelengths qfO. 46 {tm, 0.70 {tin and 1.03 {t11l, pond well wilh both the trends and magniludes of lhe res/lee t ivety.

Downloaded54 ·from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. .\ oLiIl alld StlOeve: Albedo qfthe Green/and ire sheet

changes are negligible at 0.46 pm even during melt; and portions of the ice sheet but even the northeastern regions albedo dec reases a re moderate at 0.70 fliT! . show decreases or as much as 10- 20% . Changes in surface albedo are even more pronounced in the Channel 2 data with May-to-July decreases of as much Albedo from AVHRR as 50% (see Figure 7). Using the formulati on dC\'elopecl by Stroe\'C (1996) the broaclband albedo \\'as calculated for Of signifi cance are the large variations in surface a lbedo lVIay a ndJuly and the per cent change in broadba nd a lbedo from AVHRR channels I a nd 2. In a time se ri es or monthly is shown in Figure 8. Although th e a lbedo remains relati\'c1 y averaged albedo data from M ay through July for AVHRR consta nt during the summer months O\'e r the celllra l por­ channels I and 2, it is cl ear that spatial and temporal tion of the ice sheet, there appears to be a sli ght increase in changes in the albedo of the Greenland ice sheet are sub­ albedo due to new snowfall. Large a reas of the ice sheet stantial. The Channel I images in Figure 6 show the largest show changes in the 10- 20% range and in the abl ati on zone decreases (40% ) in albedo a long the so uthern a nd western where snow is melting away to reveal the ice substrate, the

Fig. 6. AVHRR Channel J surfacealbedoJor A1ay, J une andJuly 1991. Albedo decreasesG7"e evidelllJro mJll nf loJ uly especially ill coastal regions.

Fig. 7 AVHRR Channel 2 sll1j ace albedoJor M ay, June an dJuDI1991. Substantial albedo decreases occur as snowpack wa rming and melt OCCllr on the ice sheet.

Downloaded from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. 55 Nolin and Stro eve: Albedo ifthe Greenland ice sheet

Fig. 8. Relative decrease in broadband albedo between May and June. Note the decreased albedo along coastal regions where extensive melt is occurring, and in the northeast, where low accumulation rates allow suiface grains to continue to grow. Negative values correspond to those pixels showing slight increases in albedo resultingfrom new snow deposition in southern portions qf the ice sheet.

change in broad band albedo is even larger, in some places CONCLUSIONS reaching 80%. It is interesting to note that the 17% AVHRR-derived relative change in albedo for the ETH/ The results of this work demonstrate an important point: CU camp is nearly identical to the 18 % decrease calculated that there does not need to be any reduction in snow- and by the coupled model. Although these represent different ice-covered area for there to be a substantial decrease in time periods, the warming and melting of the snowpack is albedo over the course of a season. Changes in snowpack being demonstrated by both types of data. energy balance can strongly and rapidly affect albedo The large albedo changes that have been shown here through grain growth especiall y during periods of snow­ would not be accounted for in GCMs using the simple para­ melt. Rapid flu ctu ations in surface-layer grain-size and meterizations for snow albedo. For instance, the snow albedo result from even sm a ll accumulation events on the albedo values of Meehl and Washington (1990) are fixed at ice sheet. However, when the snowpack is near O°C, these 85 % for cold snow and 60% for warm snow over the short­ small grains are rapidly sublimed or melted away making wave region (0- 0.911m ). Yet, AVHRR Channel 1 albedos for this a relatively short-term change. May 1991 range from 60- 96% and much of the ice sheet has Use of AVHRR data was successful for identifying both an albedo of>90%. And longwave albedo has been shown spatial and temporal albedo changes. Remote-sensing data to be more highly varying than shortwave, both in the such as these can be used to monitor changes in visible, near­ coupled-model results and the remote-sensing data, but infrared, and broad band albedo for the Greenland ice sheet. Meehl and Washington (1991) promote a fix ed value of If accurate predictions of climate change are to be made, 55% regardless of snow conditions. Only an albedo para­ seasonal albedo variations need to be accounted for in meterization such as that proposed by Marshall and Ogles­ GCMs. To date, only one physically based albedo para­ by (1994) can encompass the changes demonstrated here. meterization takes into account the changes in snow grain-

Downloaded56 from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. ,\'OLin and S/meve: Albedo of the Greenland ice sheet

size that affect albedo (Marshall a nd Oglesby, 1994). Models Da\'is, R. E. , A. \\'. '1oli n, R.J ordan a ndJ Dozier. 1993. TO\\'ards predicting such as SNTHERM, that characterize the dynamics of lemporal changes oflhe spectral signature of snow in \"i sibl e and near­ infrared w3\·elenglhs . .-111 11. Glacial., 17. 1+3 1+8. snowpack energy balanc e, can be used further to improve J ordan, R. 1991. r\ ollc-dimcnsional lcmperature model for a snow eO\'er: this parameteri zation by more accurately representing technical doeumentalion for Si\"TH ER~ l. 89 . CRREL Spec. Rep. 91-16. grain growth for a wide range of conditions. ~I ars h all. S. and R.J O glesby. 1994. An imprO\'Cd snow hydrology for GCl\ [s. Part I: Snoll' cO\'er fraction, albedo. grain size. and age. Climale Dp.. 10 I 21,21- 37. ACKNOWLEDGEMENTS l\ larshal l, S .. J O. Roads and G. G lalzmaier. 1994. Snoll' hydrology in a general circulation mode\. J CLimate, 7 (8), 1251 1269. This research was supported by "'ASA Polar Program grant :\ Ieehl. G. A. 198+. ;\lodeling lhe Earth's c1imale. Climatic G!wllge, 6 (1), 259- 286. ~kehl, G. A. and \\'. 1\\. Washington. 1990. CO climate sensiti\'il), and NAWG-2l58 and the NSF Division of Polar Programs grant 2 sno\\"- sca-ice- a lbedo paramelerizalions in an atmospheric GCl\1 9423530. We are grateful for the efforts ofW. Abdalati who coupled to a mixed-layer ocean. CLimalic Change. 16(3). 283- 306. made snow temperature measurements for 1995 and the :-- Iote, T. L. and .\l. R. Anderson. 1995. \ a rialions in snowpack melt on lhe helpful co nversations with K. Steffen, both of CIRES, Uni­ Greenland ice shecl based on passi\'e-micro\\,a\T measurements. J CLa ­ ciol., 41 137, 51- 60. versity of Colorado. SldTen, K ., \\'. Abdalali. J Stroe\"c, ~ I. Slobcr. ,\ . :\olin andJ Kc),. 1995. Assessment q{ climate zlariabili~J' rif the Greelllalld ice sheet: integration qf ill situ alld satellite data. Boulder, CO. Lni\'ersity of Colorado. Cooperali"e REFERENCES Institule for Research in EIl\'ironmental Sciences (C l RES). Progress Report 1\ '.) Abdala li , \\'. a nd K. SlcfTcn. 1995. Passi\·c microwa\"C-deri\'cd snow mell StcfTen, K . \ \'. Abdalati,J. Slroe\·e. l\ \. Slobcr a ndJ. Key. 1996. AssessllZml cif regions on lhe Greenland ice sheet. Ceoph}S. Res. Lell .. 22(7),787- 790. climale mriabililJ cif Ihe Greelllalld ice sheel: integralion cif ill silu and salel/ile Baldwin, D. G. and \\'.J. Emery. 1993. A syslematized approach lO A\,HRR data. Boulder. CO. Uni\Trsil), of Colorado. Cooperative Inslitute for image na\·igalion ..- / 1111. Glaciol., 17, +1+-+20. Research in EIl\'ironmcnlal Sc iences (C TRES). ( Progress Report \ ~ ) Barry. R. G. 1985. The cryosphere and c1imalic change. 111 l\ lacCrackcn. Stroe\·e. J 1996. Radiation climatology of the Greenland icc sheel. ( Ph.D. :\ I. C. and F ;\1. LUlher, eds. Detectillg the climatic wcts '!! ill creasing carboll lhesis, Uni\'ersit)' of Colorado, Boulder.) diolide. Wash ington, DC, U.S. Departmelll of Energy, 109- 1+8. (Report Stroe\'e, J. a nd A. Nolin. [n press. Comparison of AVHRR-deri\"Cd and in DOE/ER-0237.) situ surface albedo O\'Cr the Greenland ice sheel. Remole Sensing Em'iron. Cess, R. D. alld 31 others. 1991. Illlerprelarion of snow- climate feedback as Tanre, D. , B. N. Holben and YJ. Kaufman. 1992. Almospheric correeli on produced by 17 general circulat ion models. ScifllCf, 253(5022), 888- 892. algorilhm lor NOAA- AV H RR producls: lheory a nd appli cation. IEEE Col beck, S. C. 1982. An o\'CJ"\"iew of seasonal snow melamorphism. Rfl'. 1/mlS. Ceosei. Remote Sensillg. GE-30 '2). 231 250. C,0/)19'5. S;Jnl"f Php .. 20 1\, +5 61. \\'iscolllbe, \\'. J a nd S. G. \\·an·en. 1980. A model for the spectral albedo of Col beck, S. C. 1983. Theory of mClamorphism of dry snow. J (;eopl,)'s. Res., sno\\·. I. Pure snow. J Atlllos. Sei., 37(12), 2712 2733. 88 (C9), 5+75- .''>+82.

Downloaded from https://www.cambridge.org/core. 25 Sep 2021 at 00:09:24, subject to the Cambridge Core terms of use. 57