GEOPHYSICAL
RESEARCH LETTERS, VOL. 22, NO. 16, PAGES 2183-2186, AUGUST
15, 1995
Diagnosisof the recordminimumin Arctic seaice area during
1990 and associated
snow cover extremes
MarkC. Serreze,
James
A. Maslanik,
JeffreyR. Key,andRaymond
F. Kokaly
Cooperative
Institute
forResearch
In Environmental
Sciences,
University
of Colorado,
Boulder
David A. Robinson
Department
of Geography,
Rutgers
University,
Piscataway,
NewJersey
Abstract.
The Arctic sea ice cover exhibited its record mini-
smallestyet recorded.Northern Hemispheresnow extent was
alsoat a recordlow, dominatedby recordminima over Eurasia
from FebruarythroughSeptember[Robinsonet al., 1993]. Fitionsare consistent
with warm,windyconditions
in May and nally,NorthernHemisphere
temperatures
attainedrecordhighs
continuedwarmthin Junepromotingearlymelt andreductions for spring,autumnandthe annualaverage[Jones,1994].Here,
in ice concentration,
followedin Augustby strongcoastalwinds we examinethe 1990 seaice anomalyand arguethat its develforcinga finalbreakup
andretreatof thepackice.Theunusually opmentin partreflectsthis warming,providinga link with the
warmArcticconditions
in 1990 are part of a larger-scale
tem- snow cover extremes.
peratureanomalypattern,linkingthe seaice anomalyto accommum area during 1990, characterizedby extensiveice-free
conditionsduringAugust.alongthe Siberiancoast.Thesereduc-
panyingrecordminima in Eurasiansnowcover.
Observed
Introduction
Monthly-meansea ice areasfor 1990 were computedfrom
daily fields of ice concentrationderived from DefenseMeteorological Satellite Program (DMSP) Special Sensor Microwave/Imager(SSM/I) passivemicrowavedata [Weaver et al.,
1987]. We then calculatedmonthlymeanice areasfor 1980 to
1990 by combiningthe SSM/I (July 1987 onwards)and SMMR
(November 1978 to June 1987) records.Both ice data setswere
Using the combinedpassivemicrowaverecordfrom the Nimbus-5 Electrically ScanningMicrowave Radiometerand the
Nimbus-7 Scanning Multichannel Microwave Radiometer
(SMMR), Parkinson and Cavalieri [1989] concluded that no
overall trend in Northern Hemisphereice extent occurredbetween 1973 and 1987. However, following a reworking of the
SMMR record(1978 to 1987), Gloersenand Campbell [1991]
demonstrated
a small but significantdownwardtrend.Chapman
and Walsh [1993], using a longerrecord(1961 to 1990) based
on the weekly US Navy/lXlOAANationalIce Center(NIC) charts
since1973 (compiledfrom high-resolution
satelliteimagery,aircraft and shipreports,but alsoincludingpassivemicrowavedata
when necessary(C. Bertoia, NIC, pets. comm.)), and regional
data sourcesfor earlier years,confirm a downwardtrend. This
has been accompaniedby upward trends in Northern Hemispheresurfacetemperatures
duringthe past30 yearsover northerr. Eurasia and northwestNorth America during winter and
Ice Conditions
obtained from the National
Snow and Ice Data Center and are
calculatedusing the NASA Team Algorithm with global tie
points[Steffenet al., 1992]. We defineice areafrom the number
of 25 x 25 km pixelsin the entireNorthernHemispheregrid
,
/_
•
''
.'
spring(withcompensating
negative
trendsoversouthern
Green-
Siberia
....
...,.::•i:!•::
•:::!::.:.<'.-.'
.::--'i'"'"'"':¾'i•'•!:•11•::•.::"
:.i"?'::::
"%":'":.:•':.
'"""!i?:'•i'
i '4
L/ 5::..
....
..-....:
.-,.:?-:---
land and the westernsubpolarNorth Atlantic) and reductionsin
NorthernHemispheresnowextent[Robinsonet al., 1993].
As model predictionsindicatethat the effectsof increasing
greenhousegas concentrationswill be strongestin the Arctic
[e.g., Mitchell et al., 1990], thesetrendsare of interest.Groisman et al. [1994] suggestthat the snowcoverretreat can be relatedto an increasein snowcoverradiativefeedback,accounting
for part of the increasesin spring temperatures.Within this
context,the year 1990 is particularlynoteworthy.Basedon the
Chapmanand Walsh [1993] studyand our analysisof passive
microwavedata through1993, the sea ice area in 1990 is the
;.:-'
.........
s25¬
I
Figure 1. August1990 ice conditions.
Medium shadingindicatesthe 1990 ice area,dark shadingareasnormallyice covered
Copyright1995 by the AmericanGeophysicalUnion.
but ice free in 1990. The inset shows the absolute minimum ex-
Paper number95GL02068
tent (mediumshading)and the areaice free only in 1990 (dark
shading).Coastalpixels have beeneliminatedfor clarity.The
numbersindicategeographic
regions:1) ChukchiSea, 2) East
SiberianSea,3) LaptevSea,4) Kara Sea,5) BarentsSea.
0094-8534/95/95 GL-02068503.00
2183
2184
SERREZE
ET AL.'
ARCTIC
SEA ICE
(commonto bothdata sets)havingat leasta 15% ice concentration. Earlier studies[e.g., Parkinsonand Cavalieri, 1989] employeda subsetof the grid. Estimatesof the 1990 areaanomalies were then obtained.The l 1-year period was selectedas
Arctic Oceantemperature
dataare availablefor thesameyears.
Ice areasduringMay andJune1990 are 6% and 10% below
average,respectively.
By July,ice areadecreases
to 16% below
normal,largelydueto openwateralongSiberiaandin the Bar-
entsSea.Theanomaly
grows
to21%forAugust
(2.2x 106km2
fromthemeanof 10.5x 106km2),almost
entirely
duetoice-free
conditionsalong Siberia (Figure 1). Essentiallyno areas in
August 1990 show above-average
ice cover.ll•ese conditions
persistinto September
(19% belownormal).Anomalies
recalculated after eliminatingpixels adjacentto land areas(whichmay
containfalse concentrationestimates)are similar.
The accuracy
of total ice concentrations
derivedfrompassive
microwavedata is generallywithin 5% in high concentration
regionsduringthe cold months[e.g., Steffenand Maslanik,
1988]. However, errors increaseto 10 to 15% during summer Figure2. Meantemperature
field(solidlines)forMay to June
melt conditionsandlocallyhigherdueto meltpondsand flooded 1990and anomalies
(positive:dottedlines;negative:
dash-dot
ice. Errorsalsoincreasewhere thin ice is present.Someerrors lines), both in Kelvins.
arealsoexpected
in ouranalysis
dueto differentsensor
characteristicsof SMMR and SSM/I. Althoughanalysisof the SSM/I
and SMMR overlapperiodindicatesthat the latter problemis
minorfor our simpleanalysisof ice area,the effectsof summer
melt are causefor greaterconcern.However,ice marginsdepicted in the SSM/I and NIC analysesshowcloseagreement.
Althoughtheseare not entirelyindependent
data sets,DMSP
shortwaveimages(0.6 km resolution)for earlySeptember1990
reconfirmthe ice marginpositions.
As basingthe anomalieson the 11-yearmean maskslarge
interannualvariabilityin the Siberiansector,we alsocompiled
an absoluteminimum field of mean Augustice area by treating
anypixel with< 15% ice concentration
in anyyearfrom1980to
1990 as open water. We then identified thosepixels ice-free
only in 1990. This open water area (Figure 1) representsa reduction of 30% in the Siberian
sector from 140øE to 160øW
north to approximately80øN. The concentration
in this area is
70% averagedover 1980 to 1989 but only3% in 1990.This differenceis too large to be due to errorsin the concentration
retrievalsandfurtherpointsto the significance
of theanomaly.
betweendayswith gooddata. We stressthat our focusis primarilyon the Siberiancoast,wherethe fieldsare stronglyinfluencedby plentifulstationdataandfew buoysaredeployed.
Figure2 showsthe meantemperatureand anomalyfieldsfor
the two-monthperiodMay to June1990. Positiveanomalies
are
foundover all of the Arctic Ocean,with all areassouthof an approximateline from the LaptevSea, acrossthe ArcticOceanto
Alaskashowinganomalies
between2.5 to 3.0øC.Thispatternis
mostsimilar to that for May. During this month,anomaliesincreaseto 5øC alongthe 180ø meridianfrom 70 to 75øN. June
anomaliesarelargestalongthe Siberiancoast,lyingbetween2.0
to 3.0øC in the Laptev Sea. Temperaturesalong the Siberian
coastare slightlybelowand abovenormalfor July and August,
respectively
(0.5 to 1.0øC).Surfacetemperature
reportsfromthe
Historical Arctic Rawinsonde Archive (HARA) [Kahl et al.,
1992] confirmthesepatternsfor coastalstations,but with typically larger anomalies.Inferredmelt onsetis at least ten days
Atmosphericand Radiative Forcings
The ice anomaly(Figure 1) appearsto haveits originsin the
unusu:filly
warm
conditions
during
spring
andearly
summer.
To
illustratethis link, daily averagesof 12-hourlysurfacetemperatures from the International Arctic Buoy Program (IABP)
[Colony and Rigor, 1993] were used to computefields of
monthly-mean
temperatureandthe estimateddateof snow-melt
onsetfor 1980 to 1990. Resultsfor 1990 were thenexpressedas
anomalies.Following Serrezeet al. [1993], melt onsetat each
grid locationwasdefinedas the first day of the first seven-day
run of daily meantemperatures
of at least0øC.
The IABP fields are compiledfrom coastalstationdata and
temperatures
measured
by driftingbuoys.Despitelikelypositive
biasesin the buoytemperatures
related to insulationby snow
coverduring
winterandradiative
heating
in sununer
[Munoz
andMartia' 1995], thesefieldsrepresentthe bestavailableArctic Oceandata source.Employingdaily averagesreducesthese
problems.Data errors(< 5% of all data) were flaggedthrough Figure 3. Mean SLP field for May, 1990 (solid lines) and
limitschecks,
withupperandlowertemperature
bounds
vary'raganomalies(positive:dottedlines;negative:dash-dotlines),both
by monthand region, and re-filled throughlinear interpolation in mb.
SERREZE
ET AL.: ARCTIC
SEA ICE
2185
dominatedby an anomalousanticyclonecentered north of
Alaska.More typically,the AugustSLP distributionover this
area exhibitsa weak closedlow [Serrezeet al., 1989]. The isobarsindicatestrongeasterlywinds over the Chukchiand East
Siberianseas,wherethe ice retreatbetweenJuly and August
was especiallypronounced.As observedin the NIC charts,this
final retreatwaslargelyaccomplished
duringthe third weekof
August,whenthe anticyclonic
regimewas bestdeveloped.Dur-
ing summer,the ice motiontendsto be to the fight of the
geostrophic
wind, implyinga polewardcomponentof the ice
motionawayfrom thecoast.
Estimatednetradiationfluxesfor August1990alongthe Si-
berian
coast
are> 20W m'2(30%)above
average
overanarea
greater
than4.0x 105km2,primarily
duetotheextensive
open
waterandlow concentration
ice. Nevertheless,
positiveanomaliesagaincharacterize
essentially
theentireArcticOcean.
Discussion
Figure 4. Surfacealbedoanomalyfield for June 1990. Heavy
(light) stipplingdenotesareaswith a negativealbedoanomalyof
at least0.10 (0.05).
early along most of the coastfrom the Laptev Sea eastwardto
Alaska (late May) and over two weeks early in the northern
LaptevSea (early June),but alsoearly over the remainderof the
ice cover.The temperatureanomalyfields also arguefor rapid
melt duringJune,especiallyin the LaptevSea.
The May 1990 mean sea level pressure(SLP) and anomaly
field, calculatedfrom National MeteorologicalCenter (NMC)
data (1980 to 1990) which includepressuredata from the drifting buoys(Figure 3) is highly unusual,dominatedby a low of
996 mb (an anomalyof-20 rob) centeredat about80øN, 100øE.
As inferredfrom the isobars,strongsoutherlywindsare found
alongthecoastfrom theLaptevSeaeastwardto northof Alaska.
For example,surfacewinds at stationBukhta Tiski (71.6øN,
128.2øE)
average
5.2m s4 forthemonth,
overtwicethe1980to
We conclude
thattheice anomalynorthof Siberiawassetup
by unusuallywarm andwindyconditions
duringMay and continuedwarmthin June,promoting
earlymelt andlocalbreakup
of theicecover,with consequent
enhancement
of the surfacenet
radiationflux hastening
melt throughJuly.Althoughthe anomaly was alreadylargeduringJuly,we suspectthat the extreme
ice-freeconditions
observed
for Augustmaynothavedeveloped
withoutthe additionalcirculation"boost"in the form of strong
coastalwinds,forcinga final breakupandpolewardadvectionof
the ice cover.
Other possiblecontributingfactors,suchas anomalousocean
heatfluxes,remainto be addressed.
It is alsonotclearwhy large
negativesea ice anomaliesdid not developnorth of Alaska,
whichexperienced
a similarwindregimeaswell asconsistently
positivetemperature
anomalies.Perhapsthisreflectstheparticularly earlyalbedoreductions
overthe Siberiansector(Figure4).
The locationof the ice anomalyalsocontrasts
with the general
pictureprovidedby ChapmanandWalsh [1993] that overall,sea
ice reductionshave been largestover the Atlantic sector.As
suggested
from their analysis,this may reflectextensiveseaice
over thisregiondm'ingthe late 1960s.
1990mean.Althoughstrongsoutherly
windsareconsistent
with
thetemperature
anomaliesandearlymelt, theymayalsoexplain
the localdevelopment
duringthismonth(asdepictedin theNIC
andSSM/I analysis)of areasof openwater and low concentraAs discussed, 1990 also has the smallest snow extent on rectionice (20 to 30%) in theLaptevandEastSiberianseas.
ord for the continental
NorthernHemisphere,
dominated
by recSchweigerand Key [1994] used the InternationalSatellite
CloudClimatology
Project(ISCCP)C2 data set, a compilation
of monthlysatellite-derived
cloudstatistics,
surfacereflectivity,
atmospherictemperature, and other variables [Rossow and
Schiffer,1991], to providea griddeddata set of surface.
broad-
ß""? '
band albedo and radiative fluxes for the Arctic Ocean for 1984
to 1990.Althoughrecognizing
uncertainties
dueto problems
in
cloud/snow
discriminationand modelingassumptions,
surface
albedosfor June 1990 alongthe Laptevand East Siberiansea
coasts are shown as 0.10 to 0.30 below the 1984 to 1990 means
(Figure
4). Netradiation
fluxesaretypically
10to30W m':and
locally50 W m': abovemeansof 80 to 120W m':.TheJune
ISCCP cloud cover north of Siberia is 5 to 15% above normal.
ß
."" •
--'4I
:.;.....'
.."".
-..•:;:- ..
.•..' ,! "....•. .........
,
'...........
;..".......' •....
;" •
3¾"
'
::
......
.'./.
::I:'7'.'.:.
. "...
?' ..
As cloudforcingis negativeat this time [Schweiger
and Key,
1994], this would reducethe surfacenet radiation.Therefore,
the net radiation anomaliesreflect the reducedalbedo,due both
to earlymelt andexpansion
of theopen-waterfeaturesfh'stseen
in May. Smallernegativealbedoanomaliescharacterizethe remainderof the ArcticOcean.Julyshowsa similarpattern.
As discussed,
the ice anomalycontinues
to growrapidlyduring August.Figure 5 showsthat the AugustSLP patternis
"'/'""'••I:...;
'"'?'
..'
'..,':........
Figure5. SameasFigure3, butforAugust1990.
2186
SERREZE ET AL.: ARCTIC
SEA ICE
Gloersen,P. andW.J. Campbell,Recentvariationsin ArcticandAntarctic sea-icecovers,Nature, 352, 33-36, 1991.
Groisman,P.A., T.R. Karl, R.W. Knight and G.L. Stenchinkov,
Changesof snow cover,temperatureand radiativeheat balance
over the Northern Hemisphere,J. Climate, 7(11), 1633-1656,
1994.
Groisman,P.A., V.V. Koknaeva,'T.A. Belokrylova,and T.R. Karl,
Area-average
precipitation
overthe mainpart of the formerUSSR,
in: Trends '93. A Compendiumof Data on Global Change(T.A.
Boden, D.P Kaiser, R.J. Sepanski and F.W. Stoss, eds.),
ONRL/CDIAC-65, CarbonDioxide InformationAnalysisCenter,
OakRidge NationalLaboratory,pp.906-910, 1993.
Jones,P.D., Hemisphericsurfaceair temperaturevariations:A reanalysisand an updateto 1993, J. Climate,7(11), 1794-1802,
1994.
Kahl, J.D., M.C. Serreze,S. Shiotani,S.M. Skony,andR.C. Schnell,
In-situ meteorological
soundingarchivesfor Arctic studies,Bull.
Amer. Meteorol. Soc., 73(11 ), 1924-1830, 1992.
Figure6. Departures
of April(top)andMay (bottom)
snow
cover over Eurasia for 1990. Snow cover in the shaded(open)
areaswasat least20% below(above)normalin 1990.
ordminimaoverEurasiafrom FebruarythroughSeptember
(but
with the winter 1989/90 averagebeingnearnormal)[Robinson
et al., 1993]. Interestingly,1990 precipitationover mostof the
former Soviet Union was exceededonly by two years (1957,
1958) since1891 [Groisman
et al., 1993].Figure6 showsthe
snowcoveranomalyfield for April andMay 1990 with respect
to 1980 to 1990 meansanalyzedfrom NOAA showcharts[see
Robinsonet al., 1993].Snowcoveris markedlyreducedoverthe
westernand easternportionsof the continent.The patternfor
March is similar.Maps from the ClimateDiagnostics
Bulletin
[NOAA, 1990] for all threemonthsshowsurfacetemperatures
north of 40øN in the upper 90th percentileof the normal
(Gaussian)
distribution
fit to the 1951to 1980baseperioddata
for western and eastern Eurasia as well as northwest North
America.All regionsnorth of 60øN lie in at least the 70th to
Mitchell, J.F.B., S. Manabe, V. Meleshko,and T. Tokioka,Equilibriumdimatechangeandits implications
for the future,in: Climate
Change:The IPCC Assessment
(J.T. Houghton,G.J. Jenkinsand
J.J.Ephraums,
eds.),CambridgeUniversityPress,pp.135-165plus
figures,1990.
Munoz,E.A. and S. Martin, GfiddedArcticsurfacetemperature
data
sets:a comparison
with Sovietice stationandlandtemperatures
for
1987-88,J. Climate(in press).
NOAA, ClimateDiagnosticsBulletin,March-June1990, 1990.
Parkinson, C.L. and D.J. Cavaliefi, Arctic sea ice 1973-1987: Sea-
sonal, regional and interannual variability, J. Geophys.Res.,
94(C10), 14,499-14,523, 1989.
Robinson,D.A., K.F. Dewey, and R.R. Heim, Global snow cover
monitoring:an update,Bull. Amer. Meteorol. Soc., 74(9), 16891696, 1993.
Rossow,W.B. and R.A. Schiffer,ISCCP clouddata products,Bull.
Amer. Meteorol. Soc., 72, 2-20, 1991.
Schweiger,A.J. and J. Key, Arctic Oceanradiationfluxesand cloud
forcingbasedin the ISCCP C2 clouddata set,J. Appl. Meteorol.,
33(8), 948-963, 1994.
Serreze,M.C., R.G. Barry, and A.S. McLaren,Seasonalvariationsin
sea ice motion and effects on sea ice concentration in the Canada
Basin,J. Geophys.Res.,94(C8), 10,955-10,970,1989.
Serreze,M.C., J.A. Maslanik, G.R. Scharfen,R.G. Barry, and D.A.
Robinson, Interannual variations in snow melt over Arctic sea ice
90th percentile,the exceptions
beingnearor slightlybelowavand relationships
to atmospheric
forcings,Ann. Glaciol.,17, 327eragetemperatures
over the CanadianArctic, and in May, over
331, 1993.
central Eurasia.The IABP data reveal positivetemperature Steffen,K. and J.A. Maslanik, Comparisonof Nimbus7 Scanning
anomalies
overthe ArcticOceanin May andJune,but theApril
Multichannd
Microwave Radiometer radiance and sea ice concendataalsorevealslargepositiveanomalies,
similarto May.
tration with Landsatimageryfor the North Water area of Baff'm
The unusuallywarm ArcticOceanconditions
in springand
Bay,J. Geophys.Res.,93, 10,769-10,781,1988.
earlysummerassociated
with the seaice anomalyhenceappear Steffen,K., J. Key, D.J. Cavaliefi,J. Comiso,P. Gloersen,K. St. Germain,and I. Rubenstein,
The estimationof geophysical
parameters
to be the high-latitudesignalof a muchlarger-scale
temperature
usingpassivemicrowavealgorithms,in: MicrowaveRemoteSensanomalypattern.Althoughthe straightforward
interpretation
is
ing of SeaIce (F.D. Carsey,ed.), Geophys.Monogr.Ser.,Vol. 68,
that the snowcoverextremesrepresenta response
to thislarge-
pp. 201-231,AGU, Washington,D.C., 1992.
scaleforcing,the issueof whetherthe reducedsnowextentitself
Weaver, R., C. Morris and R.G. Barry, Passivemicrowavedata for
providedfor a feedbackaccentuating
the temperatureanomalies
snowand ice research:plannedproductsfrom the DMSP SSM/I
[cf. Groismanet al., 1994]meritsfurtherinvestigation.
Acknowledgments:This studywassupported
by NSF grantsOPP9321547, ATM-9315351, ATM-9314721, SBR-9320786, NASA
grantsNAGW-3568,NAGW-2407andtheEOSprogram.
References
Chapman,W.L. and J.E. Walsh,Recentvariationsof seaice and air
temperaturein high latitudes,Bull. Amer. Meteorol. Soc., 74, 33-
system,EOS Trans. Amer. Geophys. Union, 68(39), 776-777,
1987.
M.C. Serrezo,J.A. Maslanik, J.R. Key and RaymondF. Kokaly,
CIRES, CampusBox 449, Universityof Colorado,Boul&r, Colorado
80309-0449.(e-mail:serreze@kryos.
colorado.edu)
David A. Robinson,Departmentof Geography,
RutgersUniversity,
Piscataway,
New Jersey08854. (email:drobins@gandalf.
rutgers.edu)
47, 1993.
Colony,R.L. andI.G. Rigor,International
ArcticBuoyProgramData
Reportfor 1 January1992-31 December1992, Tech.Mem. APL-
(ReceivedFebruary14, 1995; revisedJune16, 1995; accepted
June
UWTM 29-93,APPL,Univ.WA, Seattle,WA, 215 pp,1993.
19, 1995.)