Downloaded by guest on September 28, 2021 www.pnas.org/cgi/doi/10.1073/pnas.1904242116 J 2018 balance to mass 1972 Sheet from Ice Greenland of years Forty-six Greenland year every loss mass annual an high 1998. maintain since sufficiently to remained equilibrium has above discharge glacier enhanced SMB, 34 and 66 at mm) controlled (9.1 is loss mass The sheet. a Morlighem Mathieu etrmpyeddahg-eouin(5 )ietikesand thickness ice m) (350 high-resolution a yielded velocity map mass- high-resolution A vector a (19–21). with Foundation constrained Moore method Betty conservation and Gordon OIB the from surveys and gravity Opera- and NASA’s (OMG), velocity NASA’s from Greenland surveys Melting ice from bathymetric Ocean 18), of thickness (17, series (OIB) ice IceBridge than revi- time tion in The complete longer gravity. improvements more years with a (12–16), than 20 from longer benefits 1972, years sion 30 in and archive altimetry, with historical Landsat coastal the in Model Elevation changes Digital volume single a ice using (5). (DEM) 1980s quantify Aerial to volume 1900 data. to from satellite ice areas used with The The 1992 were 2002. numbers. before photos before extend large not extend two over does of not SMB method differencing does of the method reconstruction precise i.e., gravity and comprehensive, sheet, ocean requires ice the it the that into is fluxes the method draw- glacier into A this models. flux of numerical mass inform back to (ice important dynamics mass is glacier which surface ocean), and between ablation) other and partitioning of runoff forms minus the one (accumulation i.e., processes only (SMB) loss, control- the balance mass processes is 1992– physical the method period the ling budget time about information the mass provides for or The that 9–11), fluxes 2002–2016. 4, output or 2, versus 2016 (1, input method and budget (8), mass (6, volume gravity ice time-variable in changes 7), using techniques independent three by I north- from are rise (2.0 level sea (4.4 global 1972, extrapolat- west to since of contributions Cumulated series difficulty trends. largest longer-term a the the into to illustrates records due short which 2010–2018 ing summers, in from cold negative switched to of loss 2000–2010 mass 80 in in or acceleration positive 1980s, The the average. since on sixfold decade, or 2010–2018, in 187 Gt/y to 1990–2000, in Gt/y 17 (15%) 51 +47 of balance glaciers of gain mass 110 mass decadal for a from The indirectly directlyswitched fluxes. and ocean 1972 reference D) the from velocity-scaled of into using glaciers (85% D, glaciers 260 discharge, 107 of mass for calculate (SMB) veloc- We balance elevation, 2018. mass surface to 2018) 31, surface thickness, July and review of for ity, using (received survey Sheet 2019 Ice 20, comprehensive Greenland March the approved a of and balance CA, mass Jolla, the La reconstruct Diego, We San California, of University Thiemens, H. Mark by Edited and Denmark; Copenhagen, 1350 Copenhagen, France; Grenoble, 38000 CNRS, eateto at ytmSine nvriyo aiona rie A92617; CA Irvine, California, of University Science, System Earth of Department er ´ otms oteoen(–) h asls a enquantified been has loss has mass (GIS) The Sheet (1–5). Ice ocean the Greenland to the mass decades, lost several last the n ee eetn h asbde ehdt h tr of start the to method budget mass the extend we Here, meMouginot emie ´ ± ± . m reln,wt oa 13.7 total a with Greenland, mm) 0.2 7G/ n18–90 h asls nrae rm41 from increased loss mass The 1980–1990. in Gt/y 17 | ± glaciology . m,suhat(3.0 southeast mm), 0.2 ± %b M 46m) vni er fhigh of years in Even mm). (4.6 SMB by 8% a,b,1,2 | e level sea a rc No Brice , rcRignot Eric , c e rplinLbrtr,Clfri nttt fTcnlg,Psdn,C 91109; CA Pasadena, Technology, of Institute California Laboratory, Propulsion Jet ± | lmt change climate 7G/ n20–00 o286 to 2000–2010, in Gt/y 17 ± el ¨ e 1G/ n17–90t loss a to 1972–1980 in Gt/y 21 ± en Scheuchl Bernd , a,c,1,2 ± . m,adcnrlwest central and mm), 0.3 e %b lce dynamics glacier by 8% nttt o aieadAmshrcRsac teh,UrctUiest,30 AUrct Netherlands Utrecht, TA 3508 University, Utrecht Utrecht, Research Atmospheric and Marine for Institute nesA Bjørk A. Anders , ± | glaciers . mfrteice the for mm 1.1 ± typer Gt/y 6 a n ihe Wood Michael and , ± a,d 20 ± ihe a e Broeke den van Michiel , 1073/pnas.1904242116/-/DCSupplemental. y at online information supporting contains article This (DOIs: NSIDC edu/data/arcticdem/ at available are and elevation surface UCI and the (NSIDC) Center and Data Ice and Snow (DOIs: National at (DOIs: Dash available analysis Dash) are data the (UCI thickness Dash ice perform the Irvine to California, used of boundaries University D1MM37, basin in and deposited been data have velocity Ice deposition: Data 2 1 BY-NC-ND) (CC 4.0 NoDerivatives License distributed under is article access open This Submission.y Direct PNAS a is article This interest.y of conflict no wrote declare E.R. authors and The J.M. J.M., and research; data; analyzed performed M.W. paper. and J.M. y the B.S., research; B.N., M.M., designed R.M., E.R. M.v.d.B., A.A.B., and J.M. contributions: Author ntercn n ertr otiuin fteGSt sea to GIS the of contributions near-term conclude entire rise. and and the level work; recent for prior the region, with estimates by on most the region the compare glacier, until sheet; years by ice history 46 glacier the last which data, the discuss glaciers, over recent methodology; eleva- balance the low outlet mass at present Greenland of melt of We km) ice width 28). 11 of (27, typical of reconstruction tion of instead the fidelity km size, the 1 improves in to match, downscaled spa- in km to improved (5.5 have atmospheric resolution SMB regional reconstruct tial Finally, to used 1972. models glacier since calculate climate surveys, to precision thickness radar with glacier the the of fluxes 1993–2017, series of from time a altimetry date yield laser DEMs the airborne ice at pre-OIB with and elevation combined OIB 30-m- When bed a (26). and and 1980s thickness for 2011–2018, (25), DEM for historic 2007–2008 Minnesota) spacing for of Geospatial (Polar University DEMs DEM Center, WorldView of 30-m-spacing Project series a time Mapping 8-m-spacing use We improve- Ice mapping. major Greenland topography from surface benefit in We ments Greenland. inversion southeast gravity new in with data We combined (24). 3 Version interpolation BedMachine an use of instead principles 23), physical (22, on “BedMachine” based named Greenland, of map elevation bed b owo orsodnemyb drse.Eal [email protected] rjmougino@ or [email protected] Email: addressed. be uci.edu. may correspondence whom work.y To this to equally contributed E.R. and J.M. ntttdsG des Institut ifl ic h 90.Genadhsrie e ee by level years. sea 8 last raised the during has half Greenland 1972, increased since has 1980s. mm loss 13.7 mass the natural The its since 1980s. from the sixfold deviate in to variability dis- of started range ice balance drainage Ice mass 260 glacier over The Greenland comparing snowfall models basins. of climate by the accumulation atmospheric interior regional years with from of ocean 46 the balance past into charge the mass for the Sheet reconstruct We Significance 10.5067/CPRXXK3F39RV y 10.7280/D1GW91, 10.5067/2CIX82HUV88Y, a ocecsd ’nionmn,Universit l’Environnement, de eosciences ´ ). y n h oa esailCne (https://www.pgc.umn. Center Geospatial Polar the and ) 10.7280/D1595V e oanMillan Romain , 10.5067/GDQ0CUCVTE2Q . y d etefrGoeeis nvriyof University GeoGenetics, for Centre raieCmosAttribution-NonCommercial- Commons Creative , 10.7280/D1595V www.pnas.org/lookup/suppl/doi:10. NSLts Articles Latest PNAS a 10.5067/ICESAT/GLAS/DATA209 , and , and , rnbeAlpes, Grenoble e ´ 10.7280/D1NW9K), 10.7280/D11H3X), | 10.7280/ f6 of 1

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES A B C D

NO NO NO NO

NE NE NE NE NW NW NW NW

CW CE CW CE CW CE CW CE

SE SE SW SE SE SW SW SW dH (%) D -D dV (%) M2018-M1972 2018 1972 D2018-D1972 >100 from dH >100 from dV -150 Gt 50 -50 Gt 50 10 Gt/yr 10 Gt/yr -10 Gt 0 0 Speed (km/yr) +10 Gt 300 km -50 1 Gt/yr -50 1 Gt/yr 300 km +50 Gt <0.001 0.1 1>3 <-100 0.1 Gt/yr <-100 0.1 Gt/yr +150 Gt

Fig. 1. (A) Glacier catchments/basins for the GIS and seven regions overlaid on a composite map of ice speed (12). (B–D) For 1972–2018, the percentage (B) thickness change, (C) acceleration in ice flux from each basin, and (D) cumulative loss per basin. The surface area of each circle is proportional to the change in ice discharge caused by a change in (B) thickness or (C) speed; the (blue/red) color indicates the (positive/negative) sign of the change in (B) thickness, (C) speed, and (D) mass.

Results loss of 63 ± 98 Gt or 0.2 ± 0.3 mm eustatic SLR, the smallest We divide Greenland, including its peripheral glaciers and ice contributor in Greenland. 2 caps, into 260 basins (Fig. 1A and Dataset S1; ref. 29) grouped CW holds a 134-cm SLE over an area of 236,648 km drained in seven regions: (i) southwest (SW), (ii) central west (CW), (iii) at 91% by 15 tidewater glaciers (SI Appendix, Fig. S1B). Jakob- northwest (NW), (iv) north (NO), (v) northeast (NE), (vi) central shavn Isbræ controls 45% of D (30.0 ± 7.7 Gt/y in 1975), east (CE), and (vii) southeast (SE). These regions are selected followed by 32% from Store Gletscher (8.6 ± 1.4 Gt/y in 1975) based on ice flow regimes (30), climate (31), and the need to par- and Rink Isbræ (11.0 ± 1.4 Gt/y in 1984). D decreased from tition the ice sheet into zones comparable in size (200,000 km2 to 68 ± 4 Gt/y in 1972–1980 to 65 ± 1 Gt/y in 1990–2000 due to 400,000 km2) and ice production (50 Gt/y to 100 Gt/y, or bil- a slowdown of Jakobshavn Isbræ (33), jumped to 93 ± 8 Gt/y in lion tons per year). Out of the 260 surveyed glaciers, 217 are 2013 due to the speed up of Jakobshavn Isbræ, and decreased marine-terminating, i.e., calving into icebergs and melting in con- to 78 ± 6 Gt/y in 2018. Three other glaciers, Eqip Sermia, tact with ocean waters, and 43 are land-terminating, i.e., with Kangilerngata Sermia, and Sermeq Silarleq, increased their ice zero discharge at the terminus (Dataset S2). The actual number flux by almost 100% from 1984–1998 to 2018. Store Gletscher of land-terminating glaciers is far larger than 43, but we lump and Rink Isbræ fluctuated little. SMB dropped from 62 Gt/y in them into larger units for simplification because we only need 1972–1989 to 38 ± 3 Gt/y in 2010–2018, or 39%. SMB and D the total SMB to conduct the assessment. Peripheral glaciers and were in balance between 1972 and 2000, but combined to form a ice caps are assumed to be in balance at the beginning of our large loss (679 ± 15 Gt) from 2000 to 2018 (Fig. 2). In total, CW lost 738 ± 75 Gt, or 2.0 ± 0.2 mm SLR since 1972 (Fig. 3B). survey, and only SMB processes are considered afterward (32). 2 We calculate the mass balance as SMB over the drainage basin NW holds a 127-cm SLE over 283,654 km drained by 64 minus ice discharge, D, at the glacier grounding line, or at the ice tidewater glaciers (Fig. 1A). D decreased from 90 ± 3 Gt/y front if the glacier does not develop a floating section. Results are in 1972–1980 to 87 ± 1 Gt/y in 1990—2000, and increased to added by regions and for the entire ice sheet (Dataset S2). We 112 ± 1 Gt/y by 2010–2018, or 45% (SI Appendix, Fig. S1C). The calculate mass balance on a decadal time scale to reduce errors. largest changes occurred for Upernavik Isstrøm C (+73 ± 16% For SMB, we use the output products from the recent Regional for 1993–2018), Upernavik Isstrøm N (+141 ± 12% for 1996– Atmospheric Climate Model v2.3p2 downscaled at 1 km (28). 2013), Kakivfaat Sermiat (+136 ± 11% for 1995–2015), Alison We use BedMachine to calculate the eustatic sea level equivalent (+169 ± 10% for 1995–2018), Kjer (+372 ± 30% for 2005– (SLE) of each basin and region. 2018), Steenstrup-Dietrichson (+93 ± 13% for 1985–2012), and SW holds a 74-cm SLE over an area of 216,207 km2 controlled Sverdrup (+112 ± 22% for 1986–2018). A number of glaciers at 72% by land-terminating glaciers. Twelve tidewater glaciers with no speedup were already out of balance in the 1970s: Hayes discharge, on average, 30 ± 5 Gt/y in 1972–2018. Changes in Gletscher M and SS, and Upernavik S. SMB averaged 78 ± 2 D are controlled by Qajuutap Sermia (4.5 ± 1.1 Gt/y in 1987), Gt/y for 1972–1980 but dropped to 45 ± 2 Gt/y in 2010–2018. Ukaasorsuaq (6.4 ± 1.9 Gt/y), and Kangiata Nunaata Sermia NW changed from near balance in the 1970s to a small loss in (6.4 ± 1.9 Gt/y). We find no trend in D during 1972–2018. In con- the 1980s, then equilibrium between 1995 and 2000, before a trast, SMB decreased from 59 ± 4 Gt/y in 1972–1980, 40 ± 4 Gt/y rapid loss from 2000 to present. The cumulative loss is 1,578 ± 56 in 1990–2000 (SI Appendix, Fig. S1A and Dataset S2), and −13 ± Gt, or 4.4 ± 0.2 mm SLR, the largest contributor in Greenland 2 Gt/y in 2010–2018. The total mass balance decreased (Fig. 3C). from +29 ± 6 Gt/y in the 1970s to −43 ± 4 Gt/y in 2010–2018 NO holds a 93-cm SLE over 263,534 km2 drained at 82% by 12 (Fig. 2). Overall, SW gained 426 ± 80 Gt between 1972 and 2001 tidewater glaciers (Fig. 1A). D varied from 23 ± 1 Gt/y in 1972– (Fig. 3A) and lost 486 ± 17 Gt between 2001 and 2018, for a net 1980, to 22 ± 1 Gt/y in 1990–2000, to 24 ± 1 Gt/y in 2010–2018.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1904242116 Mouginot et al. Downloaded by guest on September 28, 2021 Downloaded by guest on September 28, 2021 ognte al. et Mouginot oteaslt antd ftecag nSBo D. or proportional SMB is in circle change the D the of and size of gain), The magnitude for gain). absolute for blue the blue loss, to loss, for for (red (red tone tone dark light in in or SMB gigatons, are in Shown 1972–2018 2010–2018. period time the for 10 regions by dD, discharge, ice 2. Fig. 12 AB CD 9019,(C 1980–1990, (B) 1972–1980, (A) kg; E attoigo h asls ewe nmle nSB SB and dSMB, SMB, in anomalies between loss mass the of Partitioning 0021,ad(E and 2000–2010, (D) 1990–2000, ) ) 9219 oprdwt 9221,bttegairvariations glacier in the data years velocity but fewer 30 1992–2018, have to with We 20 20%. compared than by 1972–1992 more time by in quality in records and prior extends assessment Our Discussion 4,976 lost sheet ice the 13.7 1972, or Since Gt, 2). (Fig. 2010–2018 in 187 1990s, the 50 in lost Gt/y and 1972–1980, in Gt/y 48 a and 286 to 555 than at 456 higher averaged peak Gt D 15 1972–2000, and In equilibrium 2017. above in well is D because Gt, 55 449 at equilibrium above 145 at minimum a and 2018, 410 to S1H Fig. Appendix, has but 1990s the before balance 1,089 near lost was SE 2010–2018. in Gt/y 131 Kangerlua. from Anorituup Fjord, decreased and Umiivik SMB Fjord, S, Tingmiarmiut and Bernstoff, C P. Bugt A. Køge glaciers Helheimgletscher well-known by less driven was and 2000–2004 in increase the rapid The has SE front. 158 ice to the Gravimet- at fjords, (136 D (39). the D calculating largest in firn of bathymetry the problem OMG high the in with of solved water combined area (21), liquid an surveys with in ric snowfall speed to of and due assess thickness rates to glacier difficult been in has uncertainties budget mass Its glaciers. water SMB glaciers. 508 SLR. totals two mm 1972 0.3 to since one loss mass of The speedup 2010–2018. the 73 to from due decreased 1996 in Gt/y 8 20 increased D 87 and 1990–2000, in Gt/y 75 from moderately increased Kangerlussuaq D from is D (8.2 largest The 1A). (21.1 (Fig. glaciers water 532 is loss 1.5 cumulative The 2010–2018. in 22 from from dropped increased SMB D Overall, 38). 31 (37, up builds 36.2 surge next 1.2 the as 20.6 with 1972, glacier in surge-type charge a 36), is (35, 10 Storstrømmen 2012 after from up D sped doubling Isstrøm Zachariæ period. entire the 1 SLE) (26-cm SLE), (Fig. Storstrømmen (60-cm and SLE), Nioghalfjerfjorden (56-cm Isstrøm mostly Zachariae basins, tidewater 14 by 474 is loss mass cumulative 1.3 The or 2010–2018. in 10%. Gt/y 25.4 by 2 D from increased it decreased and when (34) SMB 2010, events until calving stable major experienced relatively remained Petermann 2018. hreicesdfo 4.8 from increased (1.7 charge Ostenfeld and 1985), (10.5 (4.8 Humboldt Petermann from is discharge largest The reln’ M vrgd422 averaged SMB Greenland’s Ehlsa5-mSEoe 165,349 over SLE 55-cm a holds SE 218,628 over SLE 72-cm a holds CE 425,250 over SLE 180-cm a holds NE ± ± ± ± tyi 9020,adu o35 to up and 1990–2000, in Gt/y 1 ± . mSLR. mm 0.1 ± .DfrNohljrjre a nrae y1%over 10% by increased has Nioghalfjerfjorden for D A). ± ± . ty,adDuar-esn(8.9 Daugaard-Jensen and Gt/y), 0.7 tyi 9218,t 40 to 1972–1980, in Gt/y 1 ± . tyi 94,UnmdDcpinøC n CS and CN ø Deception Unnamed 1984), in Gt/y 2.0 . mSR h ags osprui discharge. unit per loss largest the SLR, mm 0.1 7G/ nte18sad19s 251 1990s, and 1980s the in Gt/y 17 ± 0G/ n21–08det n18 an to due 2010–2018 in Gt/y 20 tyi 0021 n 160 and 2000–2010 in Gt/y 2 ± ± %dces nSB h c he and47 gained sheet ice The SMB. in decrease 9% ± 1G ic 92 r3.0 or 1972, since Gt 91 . mSLR. mm 1.1 ± 2G/ n21.I oa,tems osincreased loss mass the total, In 2018. in Gt/y 12 ± tyi 9218)i reln.Dincreased D Greenland. in 1972–1980) in Gt/y 6 ± tyi 05 10 2005, in Gt/y 9 .I erae rm506 from decreased It ). ± . tyi 95,Rdr(3.0 Ryder 1975), in Gt/y 1.4 ± ± ± . tyi 9018 o60 to 1960–1989 in Gt/y 1.5 ± . tyi 99 n tyatr1990 after Gt/y 0 and 1979, in Gt/y 3.4 tyi 92t 18 to 1972 in Gt/y 1 7G/ nte20s n 286 and 2000s, the in Gt/y 17 ± ± . tyi 9119 o13 to 1961–1990 in Gt/y 1.3 ± ± 5G,btteiesetsills 105 lost still sheet ice the but Gt, 55 ± ± ± tyi 0021.I igeyears, single In 2010–2018. in Gt/y 2 . tyi 92t 6.4 to 1972 in Gt/y 1.1 . tyi 9018 o111 to 1960–1989 in Gt/y 1.6 . tyi 94.Hmod dis- Humboldt 1984). in Gt/y 0.2 5G/ n21.I 08 M was SMB 2018, In 2012. in Gt/y 55 . tyi 9218 to 1972–1980 in Gt/y 1.5 ± ± ± 7G/ nte18s 41 1980s, the in Gt/y 17 tyi 9218 o78 to 1972–1980 in Gt/y 5 ± tyi 9019,bc to back 1980–1990, in Gt/y 1 ± ± 0G/ n16–99(SI 1961–1989 in Gt/y 10 ± . mSLR. mm 0.3 km km tyi 91 n 7 and 1991, in Gt/y 7 ± NSLts Articles Latest PNAS ± km ty erblne to balance, near Gt/y, 1 ± ± 2 tyi 2010–2018. in Gt/y 2 2 tyi 2010–2018. in Gt/y 1 ± 2G ic 92 or 1972, since Gt 52 2 ± ± rie y5 tide- 59 by drained 8G/ nte1970s the in Gt/y 18 rie y4 tide- 41 by drained ± otolda 87% at controlled ± . tyi 1980s). in Gt/y 0.7 %ices nD in increase 1% ± . tyi 1975), in Gt/y 1.1 0G/ n2010– in Gt/y 20 ± tyi 2018. in Gt/y 2 3G,o 1.4 or Gt, 83 ± . tydis- Gt/y 1.0 ± ± . tyin Gt/y 0.2 tyfor Gt/y 3 ± ± ± tyin Gt/y 1 0Gt/y 20 | 0Gt, 30 Gt/y 2 ± −5 ± ± f6 of 3 ± ± 400 21 17 ± ± ± ± 3 2

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES Sea level rise equivalent (mm) altimetry because of the difficulty for radar altimeters to sam- AB-1 ple coastal sectors with steep slopes. With the Gravity Recovery 0 ±

Gt) and Climate Experiment (GRACE), the loss of 280 58 Gt/y 3 1 for 2003–2013, with an acceleration of 254 ± 12 Gt/y per decade (8), matches our 274 ± 17 Gt/y estimate, with an acceleration of 2 249 ± 54 Gt/y per decade. For 1992–2011, the 142 ± 49 Gt/y 3 mass loss from multiple methods (41) is within errors of our Southwest (SW) Central West (CW) 144 ± 13 Gt/y estimate. Few estimates exist before the 2000s (42– 4 45), because of mission duration (GRACE launched in 2002) or Sea level rise equivalent (mm) limitations over steep slopes (radar altimeters). C D -1 Our assessment indicates with certainty that the ice sheet was 0

Gt) Mass (10 near balance (−13 ± 14 Gt/y) in 1972–1990. Within that time 3 1 period, SMB was above balance in 1972–1980 and slightly below balance in 1980–1990, whereas D was slightly above equilibrium 2 the entire time. This result contradicts an earlier study that con- 3 cluded the ice sheet was losing mass the entire time (5) based on Northwest (NW) North (NO) differences in ice elevation between two epochs 83 years apart, 4 with no data in the ice sheet interior. We posit that balance Sea level rise equivalent (mm) conditions in 1972–1990 were compensated by periods of high EF-1 loss, e.g., in the 1920s through 1940s (46), so that the mass loss

Gt) 0

3 varied significantly between the two epochs. Secondly, the same 1 study suggested that the contribution from glacier dynamics to the mass loss was constant over the entire period 1900–2016. Our Mass (10 2 annual to subannual time series show that D increased continu- 3 ously during 1972–2018 and overall has had a larger impact on Northeast (NE) Central East (CE) mass balance than SMB. SMB domination of the mass balance 4

Sea level rise equivalent (mm) is only recent. Over the past 46 years, ice dynamics contributed G H 66 ± 8% of the mass loss versus 34 ± 8% for SMB. Hence, ice 0 dynamics, i.e., changes in glacier flow, play a major role in the Gt) Mass (10

3 mass budget. In 2000–2012, half of the cumulative loss was from four gla- 5 ciers, (i) Jakobshavn Isbræ, (ii) Kangerlussuaq, (iii) Køge Bugt,

Mass (10 and (iv) Ikertivaq S (4), but, between 1972 and 2018, Ikertivaq S 10 gained 26 ± 15 Gt. Over 46 years, the conclusions are differ- SouthEast (SE) Greenland Ice Sheet (GIS) ent. The largest losses are from (i) Jakobshavn Isbræ (327 ± 40 Gt), (ii) Steenstrup-Dietrichson in NW (219 ± 11 Gt), (iii) Year Year Kangerlussuaq in CE (158 ± 51 Gt), (iv) Humboldt in NO (152 ± Fig. 3. Cumulative anomalies in SMB (blue), discharge (D, red), and mass 7 Gt), (v) Midg˚ardgletscher in SE (138 ± 5 Gt), and (vi) Køge (M, purple) in gigatons (gigaton = 1012 kg) for the time period 1972–2018 Bugt C in SE (119 ± 37 Gt), hence highlighting glaciers that for the seven regions of Greenland and the entire ice sheet component: (A) are seldom mentioned in the literature. Steenstrup-Dietrichson, SW, (B) CW, (C) NW, (D) NO, (E) NE, (F) DE, (G) SE, and (H) GIS. Humboldt, and Midg˚ardgletscher contributed to the mass loss during the entire period, versus only after year 2000 for Jakobshavn, Kangerlussuaq, and Helheimgletscher. This result in speed are also less in 1972–1992 (11, 18, 40). Our long time illustrates the risk of summarizing the ice sheet loss on the basis record makes it possible to examine decadal estimates of mass of the fate of a few glaciers. balance instead of yearly estimates, which reduces our errors Some glaciers gained mass during the survey: Saqqap, by a factor of 3. We constrain 85% of D with precision thick- Majorqaq, and Russell glaciers in SW gained 282 ± 18 Gt in ness and 15% from velocity-scaled reference fluxes (see Materials 1972–2006. This mass gain is consistent with the glacier advance and Methods). For precision thickness, we use BedMachine at in SW in the 1970s to 1980s (47, 48, 49) and satellite-derived ice the ice front of glaciers contributing to 47% of D, contrary sheet growth (50, 51), which improves confidence in the SMB to ref. 11, who use it for all glaciers. For the glaciers that we reconstruction. Conversely, a few large glaciers did not undergo exclude, the uncertainty in BedMachine exceeds our require- large dynamic changes (<10%): Rink Isbræ, Hayes N and NN, ments (±100 m) and yields errors in flux up to 100%. For the Petermann, Nioghalfjerdfjorden, and Daugaard-Jensen glaciers. remaining 38% of D, we use gravity-derived thickness (13%) and We attribute their stability to the bed configuration. Daugaard- direct radar-derived thickness upstream of ice fronts (25%). The Jensen calves on a stabilizing ridge. Nioghalfjerdfjorden retreats gravity-derived thickness used for SE (21) changes its balance along a prograde bed. Rink and Petermann are protected by a flux from 47.2 Gt/y to 64.2 Gt/y, or +37%, thereby enhancing the buttressing ice shelf. Petermann and Nioghalfjerdfjorden sped role of SE in the total budget. For the glaciers using a velocity- 10% since 2010 and 2006, respectively, following a weakening of scaled reference flux, a 10% uncertainty in reference flux yields their buttressing ice shelves (34, 36). a 1.5% error in total mass balance, or 4 Gt/y, which is negligible. In terms of partitioning between discharge and SMB (Fig. 2), Our study also uses systematic corrections for ice thickness. The the loss from D anomalies (deviations in D from a reference uncertainty in ice thickness correction is less than 1%. Without state) decreased from 47 ± 19 Gt/y in the 1970s to 41 ± 8 Gt/y it, D would be 10% too high in 2018, or 55 Gt/y. in the 1990s before reaching 127 ± 9 Gt/y in 2010–2018, for a Radar altimetry indicates a loss of 269 ± 51 Gt/y in 2011– cumulative 3,312 ± 124 Gt since 1972, or 9.1 ± 0.3 mm SLE 2014 (7), and laser altimetry indicates a loss 243 ± 18 Gt/y for (Fig. 3H). SMB anomalies (deviations in SMB from a refer- 2003–2009 (6). We find 323 ± 28 Gt/y and 220 ± 21 Gt/y, respec- ence state) were positive in 1970s (+95 ± 20 Gt/y) and near tively, for the same periods; that is, our estimates agree within equilibrium in the 1980s and 1990s (−1 ± 17 Gt/y and 0 ± errors, especially laser altimetry. The loss is lower with radar 17 Gt/y, respectively) before becoming negative in 2000–2018

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1904242116 Mouginot et al. Downloaded by guest on September 28, 2021 Downloaded by guest on September 28, 2021 r orcin tteflxgt,Di ui ioeesprya steinte- the is year per kilometers velocity cubic ice in of D no rule) apply gate, (trapezoidal We flux gral speed. the ice At depth-averaged the correction. equals firn speed ice Appendix surface (SI is, D that flux, ice calculate to Discharge. Ice Methods and Materials the manyfold for discharge. of potential ice the because in and level rise, increase sea level above ice sea of to reserve to Greenland large importance of whole part greatest a northern future, of the as the in become In sheet changes 1972–1990. ice mass period the the time expect that we the find over balance also near all We includes was glaciers. that the balance large mass illustrating the of the hence series of time in previously, extensive role an reported of stronger than value a loss played mass have total NE) (NO, C. Greenland and Bugt SE, Køge North glaciers, and NW, Several Steenstrup-Dietrichson, from glaciers. Humboldt, is tidewater including by loss controlled mass are largest which The mass CW, of in 1980s. increase the sixfold reconstruction a from and loss 46-years record from entire the dominance a over a discharge reveals ice present that Greenland ice we across elevation, changes glacier SMB, surface thickness, and ice velocity, of records improved Using Conclusions weakened are shelves ice rele- the greatest change. as climate of change by level therefore sea is future and decades to largest NO coming vance of the the evolution in tap The glaciers would Greenland. the NE in which if cm) counterparts, exists (273 NW SLE D fast as potential and in flowing SE increase start and their large shelves in as a ice those buttressing for with their potential lose compared glaciers the low NW; are speeds or glacier SE the but 2018), in ueahg (25.9 D high a duce eini qal attoe ewe n M tpresent that at in SMB loss and The D between importance. partitioned greatest (62 equally of is are region sectors NE and deceleration. of sign no with 18 NW in increased especially D budget, mass where sheet ice the dominate to continue ognte al. et Mouginot their of fate (64 the dynamics 86 by ice controlled i.e., be glaciers, will tidewater NW and CW, SE, years. even 20 whereas sheet, last ice the in the role pri- of larger of loss a mass remain played SMB the flow though driving glacier 1998. in in since importance changes year mary every that equi- mass implies above lost well this has is sheet Second, D ice the because fact, SMB, mass In high loses This librium. sheet of because 1970s. ice years loss, the in the 2018, even mass in e.g., First, the conditions implications. on several equilibrium has impact above result in larger that already decrease a however, 38% was had note, a it D versus to is, time, important that that is during SMB; 18% It by decades. increased (55 two D dominated last (66 SMB years. the loss 46 in mass last the the during in 10%) role stronger a 1,670 4.6 of or loss 1972, cumulative a for respectively), 2018, (−99 .Ka A ta.(05 reln c he asblne review. A balance: mass sheet ice Greenland (2015) al. et Accelera- SA, (2011) Khan J Lenaerts 3. A, Monaghan M, Broeke sheet den ice van Greenland I, the Velicogna of E, balance Mass Rignot (2008) E 2. Hanna E, Burgess JE, Box E, Rignot 1. ntrso ogtr otiuint e ee ie h NO the rise, level sea to contribution long-term of terms In nteftr,S ilrmi otoldb M processes, SMB by controlled remain will SW future, the In ino h otiuino h reln n naci c hest e ee rise. level sea 78:046801. to sheets ice Antarctic and Greenland Lett the Res of Geophys contribution the of tion 2007. to 1958 from ± ± ± % epciey.W xetta W E n Wwill CW and SE, NW, that expect We respectively). 4%, 1 n 60 and 11% 7G/ and Gt/y 17 ± ecmieietikes(92)adievlct (13–16) velocity ice and (19–22) thickness ice combine We . mSE ec,gairdnmc a played has dynamics glacier Hence, SLE. mm 1.0 38:1–5. epy e Lett Res Geophys ± ± 5,rsetvl) h lcesd o pro- not do glaciers The respectively). 15%, −160 ± . tyad39.5 and Gt/y 1.9 . typrdcd rm19 o2018, to 1998 from decade per Gt/y 0.1 ± 35:1–5. 0G/ o 0021 n 2010– and 2000–2010 for Gt/y 20 .W suen nenldeformation; internal no assume We ). v i ntedrcinnra otegate the to normal direction the in ± ± ± % hnSB(34 SMB than 8%) . ty respectively, Gt/y, 2.7 2,65 12%, ± 7 tsince Gt 379 ± ± e rgPhys Prog Rep 4,and 14%, % only 5%) ± σ error, associated The thickness ice times i.S2 Fig. Appendix, (SI D in and S2 SMB in errors the combines dM/dt in Loss. error Mass the so that glaciers, assume individual We between estimate. is independent available D increas- are closest error the D relative at 100% from a in extended or year assume errors per estimate, we is 2.5% interpolated, available D error is at closest relative If D ing the the When series. that beyond years. assume time year 14 we after the per series, 5% of time the by beginning of increases or beginning or end end the the at if stant flux. speed, on reference for the errors factor independent scale and the is flux reference as in scaling error the systematic error, the associated of The tion previously. established changes in are correction SMB for variance used is area unexplained D and the and SMB variables, mean these between between fit (R least-squares SMB and linear correlation 1990s the significant before D a regres- of linear values reveals The GRA D. and initial MC, GAT, and the between SMB sion mean between we relationship glaciers, the in other error the For D. of D 85% use control glaciers 107 These introduce respectively. would D average) in on error 1972–2018 that 0.7% in in error a 18% 2% by A increased time. (D that value at equilibrium for initial in SMB was average discharge the the that adding assume by data SMB velocity (or earliest 1961–1989 the reference with a gate to measure- flux converted D the is sounder gate flux, radar flux most-retreated front the direct the at ice With of flux position. measured km the front 1 (GAT), ments ice within or flux line the grounding calculate we (GRA), inversion position. front den- ice ice or an using line mass kg/m into converted 917.2 is of flux volume sity The speed. in error the is e ne ainlSineFudto P wrs1461 599,and 1559691, 1043681, Cen- Awards Geospatial OPP Polar Foundation post-IPY,1542736. Science the and National by and IPY under provided ter thickness during were Group efforts ice DEMs Task acquisition of Space Worldview Polar data respectively. use and satellite the Group Task coordinating for Land- Space the for the missions for thank IceSAT Survey We Geological and data. US elevation OIB and NASA’s NASA data, data, Agency SAR sat Space of German use and the Japanese (ASI), for (CSA), Agency (DLR) acknowledge Agency Space Italian Space We (JAXA), Canadian CF17-0529). Agency (ESA), Space (Grant Agency Space Foundation European Carlsberg the sup- Research the acknowledges A.A.B. Scientific from Center. for Science port System Organization Earth Netherlands Netherlands the the Recherche and of la Program from de Polar funding acknowledge the Nationale B.N. L’Agence and Sci- M.v.d.B. French ANR-15-IDEX-02). (Grant Cryosphere National and (ANR) Centre Propulsion NASA (CNES) French the the Jet Spatiales from Technology’s d’Etudes with funding acknowledges of A J.M. Program. NNX13AI84 Institute ence Contract California under at Laboratory and Irvine, ifornia, the 362 over ACKNOWLEDGMENTS. using spread rise level naturally sea is eustatic SLE. SMB to mm converted in 1 is = anomaly Gt gigatons) The (in proxy. Mass to sheet. a distance ice as and thickness) front times SMB ice (speed density – the flux (D between D ratio the in using (SMB anomalies SMB between in anomalies partitioning and the compute we basin, .Kede K ta.(05 pta n eprldsrbto fmass of ice distribution Greenland of pattern temporal complex reveals and altimetry Laser (2014) Spatial al. et BM, (2015) Csatho 6. al. sheet. ice et Greenland the for KK, budget mass Kjeldsen improved An 5. (2014) al. et EM, Enderlin 4. D .W siaed/tyal ewe 92ad21 Fg ) o each For 2). (Fig. 2018 and 1972 between yearly dM/dt estimate We ). o er ihn eoiydt,Di ieryitroae,o etcon- kept or interpolated, linearly is D data, velocity no with years For D scaling by calculated is D Annual gravimetric and (MC) conservation mass from sets data gridded With osfo h reln c he ic D1900. AD since Sheet Ice dynamics. sheet Greenland the 400. from loss Lett Res Geophys = q P ref P i qa oSMB to equal v σ i σ D 2 asblne Md,i M iu c icag .Teerrin error The D. discharge ice minus SMB is dM/dt, balance, Mass H . i + σ D q ref rcNt cdSci Acad Natl Proc ref = 3 41:866–872. lxgtsaepae ttems-erae grounding most-retreated the at placed are gates Flux . si e.2 aey ecretteflxotie at obtained flux the correct we Namely, 2. ref. in as , P ref eue6,1,ad2 u ae o C R,adGAT, and GRA, MC, for gates flux 20 and 16, 61, use We . H α i ewe h u aeadteiefot hti,we is, that front; ice the and gate flux the between ) (H v pe n c hcns r ape vr 0 m. 200 every sampled are thickness ice and Speed . σ ref α hswr a efre tteUiest fCal- of University the at performed was work This D i h σ h ro nDcmie h ro nSMB in error the combines D in error The . stesmo ytmtcadrno rosas errors random and systematic of sum the is , σ v i D ) ref ref 2 α where , % ealdrslsfrec lce,gate, glacier, each for results Detailed ±7%. h + M) h nml nDi pedspatially spread is D in anomaly The SMB). – stesaefco o hcns,adD and thickness, for factor scale the is 111:18478–18483. q 2 (α 0.93; = σ v H D i IAppendix SI ref ref steerri c hcns and thickness ice in error the is σ α ihtesedadthickness and speed the with i.S2 Fig. Appendix , SI h ) NSLts Articles Latest PNAS 2 + (α . h D σ ref D stecombina- the is , σ Nature α v ) 2 and where , .Uiga Using ). | Dataset 528:396– ref f6 of 5 ref and ref σ α ref v is v i )

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES 7. McMillan M, et al. (2016) A high-resolution record of Greenland mass balance. 28. Noel¨ B, et al. (2018) Modelling the climate and surface mass balance of polar ice Geophys Res Lett 43:7002–7010. sheets using RACMO2–Part 1: Greenland (1958–2016). Cryosphere, 12:811–831. 8. Velicogna I, Sutterley TC, van den Broeke MR (2014) Regional acceleration in ice mass 29. Mouginot J, Rignot E (2019) Data from “Glacier catchments/basins for the Greenland loss from Greenland and Antarctica using GRACE time-variable gravity data. Geophys Ice Sheet.” UC Irvine Dash. Available at https://doi.org/10.7280/D1WT11. Deposited Res Lett 41:8130–8137. March 29, 2019. 9. Howat I, Joughin I, Scambos T (2007) Rapid changes in ice discharge from Greenland 30. Rignot E, Mouginot J (2012) Ice flow in Greenland for the International Polar Year outlet glaciers. Science 315:1559–1561. 2008-2009. Geophys Res Lett 39:L11501. 10. Howat IM, et al. (2011) Mass balance of Greenland’s three largest outlet glaciers, 31. Weidick A (1984) Studies of glacier behaviour and glacier mass balance in Greenland: 2000-2010. Geophys Res Lett 38:L12501. A review. Geografiska Annaler Ser A Phys Geogr 66:183–195. 11. King MD, et al. (2018) Seasonal to decadal variability in ice discharge from the 32. Noel¨ B, et al. (2017) A tipping point in refreezing accelerates mass loss of Greenland’s Greenland ice sheet. Cryosphere 12:3813–3825. glaciers and ice caps. Nat Comm 8:14730. 12. Mouginot J, Rignot E, Scheuchl B, Millan R (2017) Comprehensive annual ice sheet 33. Joughin I, Abdalati W, Fahnestock M (2004) Large fluctuations in speed on velocity mapping using Landsat-8, Sentinel-1, and RADARSAT-2 data. Remote Sensing Greenland’s Jakobshavn Isbrae glacier. Nature 432:608–610. 9:364. 34. Nick F, et al. (2012) The response of Petermann Glacier, Greenland, to large calving 13. Mouginot J, Rignot E, Millan R, Wood M (2018) Data from “Annual ice velocity of events, and its future stability in the context of atmospheric and oceanic warming. J the Greenland Ice Sheet (1972-1990).” UC Irvine Dash. Available at https://doi.org/ Glaciol 58:229–239. 10.7280/D1MM37. Deposited January 19, 2019. 35. Khan SA, et al. (2014) Sustained mass loss of the northeast Greenland ice sheet 14. Mouginot J, Rignot E, Scheuchl B, Millan R, Wood M (2018) Data from “Annual ice triggered by regional warming. Nat Clim Change 4:292–299. velocity of the Greenland Ice Sheet (1991-2000).” UC Irvine Dash. Available at https:// 36. Mouginot J, et al. (2015) Fast retreat of Zachariae Isstrøm, northeast Greenland. doi.org/10.7280/D1GW91. Deposited January 19, 2019. Science 350:1357–1361. 15. Mouginot J, Rignot E, Scheuchl B, Wood M, Millan R (2019) Data from “Annual ice 37. Reeh N, Boggild CE, Oerter H (1994) Surge of Storstrømmen, a large outlet glacier velocity of the Greenland Ice Sheet (2001-2010).” UC Irvine Dash. Available at https:// from the inland ice of north-east Greenland. Gronl¨ Geol Undersogelse¨ Rapp 162:201– doi.org/10.7280/D1595V. Deposited March 29, 2019. 209. 16. Mouginot J, Rignot E, Scheuchl B, Wood M, Millan R (2019) Data from “Annual ice 38. Mouginot J, Bjørk AA, Millan R, Scheuchl B, Rignot E (2018) Insights on the surge velocity of the Greenland Ice Sheet (2010-2017).” UC Irvine Dash. Available at https:// behavior of Storstrømmen and L. Bistrup Brae, Northeast Greenland, over the last doi.org/10.7280/D11H3X. Deposited March 29, 2019. century. Geophys Res Lett 45:11197–11205. 17. Gogineni S, et al. (2001) Coherent radar ice thickness measurements over the 39. Forster RR, et al. (2014) Extensive liquid meltwater storage in firn within the Greenland ice sheet. J Geophys Res 106:33761–33772. Greenland ice sheet. Nat Geosci 7:95–98. 18. Paden J, et al. (2018) IceBridge MCoRDS L2 Ice Thickness, Version 1(NASA Natl Snow 40. Rignot E, Kanagaratnam P (2006) Changes in the velocity structure of the Greenland Ice Data Cent, Boulder, CO). ice sheet. Science 311:986–990. 19. An L, et al. (2017) Bed elevation of Jakobshavn Isbræ, west Greenland, from high- 41. Shepherd A, et al. (2012) A reconciled estimate of ice-sheet mass balance. Science resolution airborne gravity and other data. Geophys Res Lett 44:3728–3736. 338:1183–1189. 20. An L, Rignot E, Mouginot J, Millan R (2018) A century of stability of Avannarleq and 42. Krabill W, et al. (2004) Greenland ice sheet: Increased coastal thinning. Geophys Res Kujalleq glaciers, west Greenland, explained using high-resolution airborne gravity Lett 31:L24402. and other data. Geophys Res Lett 45:3156–3163. 43. Thomas R, Frederick E, Krabill W, Manizade S, Martin C (2006) Progressive increase in 21. Millan R, et al. (2018) Vulnerability of southeast Greenland glaciers to warm Atlantic ice loss from Greenland. Geophys Res Lett 33:L10503. water from Operation IceBridge and Ocean Melting Greenland data. Geophys Res 44. Zwally HJ, et al. (2005) Mass changes of the Greenland and Antarctic ice sheets and Lett 45:2688–2696. shelves and contributions to sea-level rise: 1992-2002. J Glaciol 51:509–527. 22. Morlighem M, Rignot E, Willis J (2016) Improving bed topography mapping of Green- 45. Zwally HJ, et al. (2011) Greenland ice sheet mass balance: Distribution of increased land glaciers using NASA’s Oceans Melting Greenland (OMG) data. Oceanography mass loss with climate warming; 2003–07 versus 1992–2002. J Glaciol 57:88–102. 29:62–71. 46. Fettweis X, Hanna E, Gallee H, Huybrechts P, Erpicum M (2008) Estimation of the 23. Morlighem M, et al. (2017) BedMachine v3: Complete bed topography and ocean Greenland ice sheet surface mass balance for the 20th and 21st centuries. Cryosphere bathymetry mapping of Greenland from multibeam echo sounding combined with 2:117–129. mass conservation. Geophys Res Lett 44:11051–11061. 47. Bjørk AA, et al. (2012) An aerial view of 80 years of climate-related glacier 24. Bamber JL, et al. (2013) A new bed elevation dataset for Greenland. Cryosphere fluctuations in southeast Greenland. Nat Geosci 5:427–432. 7:499–510. 48. Weidick A (1991) Present-day expansion of the southern part of the inland ice. Rapp 25. Howat IM, Negrete A, Smith BE (2014) The Greenland Ice Mapping Project (GIMP) Geol Surv Greenland 152:73–79. land classification and surface elevation data sets. Cryosphere 8:1509–1518. 49. Weidick A (1994) Historical fluctuations of calving glaciers in south and west 26. Korsgaard NJ, et al. (2016) Digital elevation model and orthophotographs of Greenland. Rapp Geol Surv Greenland 161:73–79. Greenland based on aerial photographs from 1978-1987. Scientific Data 3: 50. Zwally HJ (1989) Growth of Greenland ice sheet: Interpretation. Science 246:1589– 160032. 1591. 27. Noel¨ B, et al. (2016) A daily, 1 km resolution data set of downscaled Greenland ice 51. Davis CH, et al. (2001) Elevation change of the southern Greenland ice sheet from sheet surface mass balance (1958–2015). Cryosphere 10:2361–2377. 1978 to 1988: Interpretation. J Geophys Res Atmospheres 106:33743–33754.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1904242116 Mouginot et al. Downloaded by guest on September 28, 2021