Thickening of the western part of the Greenland ice sheet

JournalofClaci%gy, T,-ol. 44, .Vo. 148, 1998

Thickening of the western part of the Greenland ice sheet R OBERT

H. THOMAS,I'

BEATA

M.

CSATHO/ SIVAPRA SA D GOGINENI/ KENNETH

C. J

EZEK/

KARL K UIVI NEN:l '2 B yrd -

I Code 1 5,. V.!lSA Headquarters, 300 E Stmt S /I : rI ashillgtoll, DC 20546, CS.A. PoLar Research Center, Th e Ohio State [;niversit)" 1090 Carmack Road. CO/limbus, Ohio '13210, Cs. ... l. 3 PoLar Ice Coring Office, enil'asiLy if. \ ebra;ka- Linco/Il, LiIlCO/II. ,\ ebraska 68588, Cs. .../.

ABSTRACT. NASA's Program for Arctic Reg io n a l C lim a te Assess m e nt (PARCA )

inc ludes measuremel1ls of ice velocity and ice thi ckn ess a long the 2000 m ele\'ation conto ur line in the western part of the icc sh ee t. H ere wc use th ese Illcas u remcnts together with published estim a tes of snow-accumulation rates to infer the mass balance, or rate o f thi cke ning/t hinning, of the ice-sh ee t catchm ent a rea inl and from the ve loc it y trm·e rse. \\'ithin th e acc uracy to which we kn ow sno\\'-accum ul ation rates, the ent ire area is in balance, but localized regions inl a nd from Uperna\' ik Iss trom a ndJakobs h a\'n rsbnc both appear to be thickening by abou t 10 c m a I.

L INTRODUCTION Desp ite seve ral decad es of int ensive Geld im-csti gati o n s, wc still do not kn ow whether the Greenland a nd Antarctic ic e sheets arc increasing or dec reas ing in volume. The answe r to thi s questio n has d ea r reie\'ance to o ur understand ing o f~ a nd ability to predict, sea-level change. The goa l o f~ASA's Program for Arctic R egional C lim a te A ssessment (PARCA ) is to measure a nd understand th e mass ba lance o f th e Greenland ice sheet (Abdalati and o thers, in press ), \\'ith direc t meas urem c- nts of changrs, 0\ '(' 1' time, in ice-surface eleva ti on as the pr i m a ry object i\'C'. Th e program a lso includes "traditi o na l" m ass-balance investigations by co mpa rin g ice discharge w ith tota l snow acc umul at ion . Thi s is the work th at we report he re. Figure I shows t he locati ons of stat ions where ice m ot io n is inferred from repeated global positioning system (GPS ) measurements. Those a long most o r thc \\'es tern part or the ice sh eet were establi shed in 1993 and 1 99 -~ by snowmobilc, a nd remea sured in 1995 and 1996 us ing a Twin Otter a irpl a ne [o r tra nspo rta tion. Thi s part o f the traverse li es 'within th e percolati o n zo ne. "Ve use ice vel oc iti es dcri\'ed from th ese sta ti ons, a nd m eas urements of ice thickness a lo ng th e sam e rou te, to calcu late total ire discharge [or co mpa ri so n wit h total inl a nd snow acc umul a ti on in o rder to infe r thl' rate of thicken ing o r thinning of this part of' the ice sh eet. A lthough th is is a \'olum e-bala nee calc ul a tion, wc use th e te rm "mass ba la nce" elsewhere in thi s paper because thi s is co mmon ly enco unte red in glac io log ica l literature, an d vo lume ba lance can readi ly be con\"e rLed to m ass bala nce with knowledge o r ice density. Each t raverse sta ti on comprises a j ointed a luminu m po le, wh ich is the reference ma rke r fo r GPS meas ure m e nts, a nd a closely ad jacent fl agged bamboo pole to aid re location

Presen t address: EG&G, 900 C lopper Road, Suite 200, Gaithersburg, rfaryland, U.S.A.

/

=

Ice Sheet Local Ice Cap Land \ ® Ice Core Sites + GPS Velocity Traverse

= = ,_ _

50'

-+_20_A~~ 40'

60'

30'

Fig. 1. Green /and mal) showillg low/iolls qjs/a/ions Id/e re ice 1II0/ioll is ill/nredjrom repealed CPS measurements. Contollrs qjsIlOl(' -({(ClllI1l1la /ioll ratesfroll1 Olunllra and Reeh (1991) ill {Ill a I water eqlli1'(l /ent are a/so illcluded. R esu/ts presenter! here flreJor /he bored area. 653

J ournal qfClacioLogy of the sta ti o n. Freshl y pl a nted , both po les extend about 2.5 m above the surface, a nd th e ex posed leng th of each pole was m eas ured to prQ\'ide a n estim ate of loca l snow acc umulati on. For the snowmohi le trave rse, one or two stati ons were esta bli shed each work d ay, whereas with the Twin Otter as m a ny as ten stations were establi shed or remeasured in a d ay, w ith brief stops to establi sh a GPS receive r at th e site, and a seco nd \'isit la ter in th e day to retri eve the recei\·e r. Th e GPS receivers acqu ired severa l hours of data at most stati o ns, with a minimum occupation of about 40 min. Comparison of th e repeated GPS locations, separated by a minimum of I yea r, yields estimates of surface ice velocit y with mag nitudes, generally, of a few tens of m a I. iVlost of the GPS so luti ons arc acc urate to better than 20 cm in a ll coordinates, but additiona l errors arc introdueed by tilting of the m a rker poles a nd inadyertent mi slocat ion of the GPS antenn a. H ere, we ass ume the effect of th ese various errors to r es ult in a vcloci t y error of ()

....0

'"

'-..

a

30

Upernavik

Jakobshavn

W

20

1

()

E ...,Cl>

'"

0: bll

.:

'2 .:

:aEo-

'-..

bll

.:

'2 Cl> .;.: ()

:aEo-

0

200

400 600 Distance (D) in km

800

1000

Fig. 3. Rates of ice-Ihickness change ( T ) along pari of the wesl side of Ihe Greenlalld ice sheet. Th e distance, D, is measured perpendicuafr to iceflowJrom the nOTthem end of Ihe ice-motion lraverse. falues ofTJor individual exit gates are shown b..y the vertical bars, and associated errors aTe between ±6 and ±20 cm a I Th egrey band dejJicls values of TcalClllatedJor larger gates, comprising several travem staliol1s, such that the catchm enl area is approximatefy 30000 kill H ere, the lIncertainlj' in T is indicated tv the thickness oflhe grl!)I band. whirh is a smoothed versiolZ ofthe bm'plolthat is It-righted according to catchment area, and conseqllentfy peaks and troughs il1 the two plots do notnecessari{J' malch. C and] indicate sites discussed ill the text. one traverse station a t a tim e, to gi\'e the va lues depicted by the grey band in Fi g ure 3. To reduce errors furth e r, we ca lculated the mass ba lance fo r th e fo ur la rge regions show n in Figure 4, with a reas o f 80000- 100000 km 2 , a nd with di stinc ti ve p atterns of thick-

7'-

o'---------' 100 km

f ig. -1-. FOllr regions. A-D, It'ill! distinctive JJClttems of i(ethickness change shown, in cm a I, b)' th e circled numbers. U and] indicate sites discussed ill the text, and hatching indicates the small area that ajJPears to be rajJidfy thinning.

ness change. H ere, all erro rs a re insignificant compa red to those assoc iated with unce rta inty in the average snow-accumul atio n rates within a region, which we ass ume to be ± 10% . Thi s is equi va le nt to a n error in the ice-thickness change of between ± 4 a nd ± 5 cm a 1 which represe nts about the hi g hest acc uracy to which we can estimate mass balance b y co mparing snow fa ll wit h ice di sch a rge until we have be tte r estimatcs of snow-accumu lation rates. 'L,ke n LOgether. all four o f the regions shown in Fig ure 4 a re in balance, with a n average thickenin g ra te of onl y 1.5 cm a I. Fo rtuitously, thi s \'a lue is almost ide ntica l to that estim a ted far the entire ice shee t from a m odel ing study of its long-te rm beha\'iou r (Hu ybrechts, 1994). The re a re two regio ns of thi c ken ing, Band D , separated by a regio n ofpossible thinning immediately nort h ofJ a kobshav n Isbn'C . Th e most striking aspec t of Figure 3 is the hig h var iabilit y in thic ke ning rates for catc hm ent a reas co rres ponding to individu a l ex it gates. Th e la rge r negative va lu es a re associated w ith \'C ry small catc hment a reas, and associated errors arc la rge. I n par tic u lar, th e \'er y la rge thinning rate of8 0 cm a 1is fo r th e sma ll tri a ng ula r catchm e nt area in th e middl e of th e traverse (Fig. 4), where th cre is a large veloc ity g rad ient ac ross th e gate. Thi s gate straddl es th e ridge se parating th e Upernavik dra in age basin from that o f th e Rink Isbn:c. Consequently, we calc ul a ted ice di scharge through the ga te by assum ing zero velocity at the gate cente r, a pprox im ately a t the cres t of th e ridgc. With thi s ass umption, th e total vo lume of ice discharged was m ore than three times la rge r th an the \'o lum e acc umul ated, but thi s cou ld result from errors in ou r estim a ti on of the catc hme nt a rea. The neighbori ng gate to the so uth has a ve ry la rge catchment a rea, an d has a calc ul ated thi c ke ning rate of 17 c m a 1 Taken together, the catchmen t areas fo r th e two gates have a calcu lated thickening rate of on ly 5 cm a I. Furth e r n o rth, within th e catchmcnt area la belled U in Figure :1, G. H am ilton and 1. M . Whill a ns ( perso na l communication, 1998) measured a local thinning of more than 50cm a 1.\Vca lso infcrthinnin g (of 20± IOcma I) for th e single-gate catc hment area con ta ining this site (F ig. 3). Our results show bo th of the neig hboring single-gate catchmen t

657

J ournal qfGlaciology areas to be thickening, a nd wc origin a lly ass umed thi s la rge \·ar iability over sma ll di ta nces to be caused by errors in our reco nstruction of the nowlin es. However, th e reaso na ble ag ree ment with th e single-point estim ate of H amilton a nd Whill a ns suggests that th ere m ay be a hig h degree of spa ti a l \·a ri a bility in ice-thickening rates in thi s region. If" th ere is, then highl y locali zed meas urements of ice-thickness cha nge should be interpreted with great ca re. Errors a re significa ntly less for results frOI11 the approxima tel y 30000 km 2 catchment a reas, d epicted by the g rey ba nd in Figure 3. H ere, the dominant features are the hi gh rates of thickening close to Uperna\'ik Isstrom (14 ± 6 cm a I) a ndJa kobshav n Isbra:: (10 ± 6 cm a I). Th ese thickening r a tes a re equi va lent to abo ut 35% (Up ernav ik ) and 20 % Ua kobsha\'n ) of th e snow-accumulati on rates meraged over th e catchment a reas. Evidence fro m recent cores (perso nal communicati on from R. C. Bales a nd others, 1998) indi cates tha t Ohmura a nd Reeh's acc umul ation r ates are approxim atel y 25 % too la rge a t a site upstream of Uperna\' ik Isst ro m (U in Fig. 4) but ve ry e10se to obsen ·ed \·alues a t a site sli ghtl y north of th e J a kobsha" n drain age basin U in Fig. 4). Thus, a lthough th e single-ga te thi ckening ra te a t U p ern avik in Figure 3 is probably too hig h, we believe o ur error assumpti ons to be reasonabl e, a nd th at ac tu al th ickening rates, ave raged over 30000 km 2 a reas, li e within th e sha ded band shown in Fig ure 3. Our results a re broadl y consistent with those from o ther studi es. Weidick (1991) presented evidence for retreat o f m a ny of the outl et glaciers in so uthern Gree nl and, fro m the mid-19th century until th e 1950s, since wh en there has bee n a readva nce. Of pa rticul a r relevance to our slUdy is th e genera l readvance of the ice sheet so uth ofJakobsh av n Isbra:: (o ur region D ), with co ntinuo us retrea t immedi a tel y to th e north (our regio n C ). Zwally (1989) a nd Davis a nd othe rs (1998) interpreted satellite rada r a ltimetr y d ata to co nclude th at th e ice sheet, south of 72 N, thickened betvveen 1978 a nd 1986- 88. Z wa ll y's resu lts show significa nt thickening for most of thi s area, but r ecent a nalysis ta king acco unt of improved satellite orbits, etc. , indicates ave r age thickening rates of abo ut 6 cm a I, or a bout 25% o f the ea rli er estimate'S (perso nal communi cati on from H.]. Z wa lly, 1997). Davis and others' results indicate an ave r age thickening rate of 1 m between 1980 a nd 1993 at locations approxim a tely 300 km so uth of th e region (D in Fi g. 4) where \Ve infer thi ckening 01' 6 cm a I. Th ere arc mi xed res ults from th e soca lled EGIG lin e, approx im ately a long th e ma rgin between our regions C a nd D, where repeated le\"Cling indicated thickening of about Im between 1959 a nd 1968 (Sec kcl, 1977), but thinning by abo ut 3.5 m between 1968 and 1992 (K ock, 1993). H owever, it is difficult to assess the rel iabilit y of th e EGIG res ults. It is importa nt to no te that these va rio us m ass-bala nce estim ates refer to different time peri o d s, with our va lues indicative of trend s over the pas t few decades. D a ta collected during 1997 will permit us to co mplete

our mass-bala nce stud y a round the entire trave rse shown in Figure I. Th en, during 1998 a nd 1999, repeat a irborne surveys of a ll the 1993- 94 lase r-altimeter sur vey li nes will re\·ea l a ll a reas of signifi ca11l thickening or thinning over the ice shee t; and other PARC A im"Cstigati ons should help identify th e causes of these cha nges. Changes d etec ted by the airborne lase r sur veys will refer to th e 5 year period 1993- 94 to 1998- 99. Longer-term trends wi II be measured by ~ASA's G eoscience Lase r A ltimeter Sys teITl (GLAS ), to be launched in 2001. This w i II acc urately measure ice-sheet ele\·ati ons a t latitudes up to 86°, covering th e entire Greenland ice sh eet a nd most o f Antarctica. GLAS data will extend th e time seri es of Greenl a nd aircra ft m eas urements, and wi ll provide a first opportunity to meas ure ice-thickn ess changes over la rge areas in Anta rctica.

ACKNOWLEDGEMENTS \ Ve th ank M :. Fa hnestock for hi help in remeasuring some of the traverse stations in 1995, Vv. Krabill a nd ]. Sonntag for processing th e GPS data, H o ng Xu for assistance with the mass-ba la nce computation, a nd P. Hu ybrechts for prm·iding res ults o f his model simula tions of the ice sh eet. \Ve are particul a rl y indebted to the pi lot, ]. Finnboga son, a nd crew orthe Gree n la ndairTwin Otter used in 1995 and 1996 to re\·isit th e traverse stati ons. Their skill, pati ence a nd hard wo rk ma d e thi s proj ect feasible. We thank R. Bindsc hadlcr and D. M acAyeal for helpful comments tha t significantl y impro\·ed the text. This proj ect IS supported by NASA's Polar R esea rch Program. Thi s is Byrd Pol a r Research Ce nter co ntributi on No. 11 16.

REFERENCES Abda lati , W., R . C. Bales and R. H . T homas. In press. Prog ram of Arctic Regiona l C li mate Assessment ( PA RCA ): ill\"Cstigations of the climate and stat e o f ba lance of the Gree nl a nd ice shee t. :i rctic Res. Cs. Ali en. c. . S. Gogineni, B. Wohl etz, K. J ezek and T. C hu a h. 1997. Airbo rne radio ec ho sounding of outlet glac iers in Greenla nd. 1nl. J Rell/ote Sel/ .ling. 18 (1 +), 3103-3107. Dm·is, C. H ., C. A. Kl ue,·e r a nd B.]. H ai nes. 1998. Ele,·at ion change of the southern Gree nl and ice sheet. Science, 279 (5359), 2086 2088. Ekholm. S. 1996. A full cove rage, hi g h resolution topographic model of Green la nd com puted from a varie t y of d igita l elevati on d ata . J Geo/Jhp . Res., 101 ( BIO), 21,961 21,972. Goginen i, S. , T. C huah, C. Ali en, K. J ezek and R. K. ]\l oorc. 1998. An impr(}\"Cd coherent radar de pth so un de r. J Glllriol. , 44 (14-8), 659 669. Huybrechts, P. 199+. The present evo lu tion of the G ree nl a nd ice sheet: an assess ment by model ling. Global and Planelal)' Change, 9 (1 2), 39- 51. Huybrechts, P. 1996. Basa l tem perat ure cond it ions of the Green land ice sheet d u r in g t he glacia l cycles. Ann. Glaciol., 23, 226 236. Kock, H. 1993. Height determ inatio n a long the EG IG line a nd in the GR I P area. Granlands Geoloaiske Cndtrsogelse. Open File 93-5, 68- 70. Krabili, \ \ '., R. T homas, K. J ezek, K . Kui,·inr n and S. M an izade. 1995. Gree nl and ice sheet th ick ness ch a nges measured by laser a ltime try. Geo /!hp. Res. Lell., 22 (1 7), 234-1- 234-4-. Ohmura, A. a nd N. Reeh. 199 1. New prec ipitation a nd acc um ulation maps for Gree n la nd. J Glaciol. , 37 (125), 1+0 14-8. Scckel, H . 1977. H6henanderu nge n im griinlii ndischen Inl a ndeis zwischcn 1959 undl 968. Muld. Gronl., 187 (4-). " ·eidick, A. 199 1. Present·day ex pansion of the sout hern part of the inland ice. Gmn/ands Geologiske Cndersagelse. Rapporl 152, 73 79. " ·h ili alls, 1. ]\ l. a nd C. J. ,·an der Veen. 1993. l\ew and im p ro,·ed determi nations or ve loc it y 0(" Ice Strea m s B a nd C, West Anta rcti ca. J Glaciol. , 39 (133), +83 -190. Z,,·ally. H .]. 1989. Growth of Gree nl a nd ice shee t: interpretat ion. Srifllfe, 246 (+937), 1589-1591.

illS received 20 Oc/ober 1997 and accefJled ill rellisedjorm 12 J une 1998 658