Fast retreat of Zachariæ Isstrøm, northeast Greenland

See allHide authors and affiliations

Science  11 Dec 2015:
Vol. 350, Issue 6266, pp. 1357-1361
DOI: 10.1126/science.aac7111

Shrinking shelf and faster flow

Zachariæ Isstrøm, a large glacier in northeast Greenland, began a rapid retreat after detaching from a stabilizing sill in the late 1990s. Mouginot et al. report that between 2002 and 2014, the area covered by the glacier's ice shelf shrank by 95%; since 1999, the glacier's flow rate has nearly doubled; and its acceleration increased threefold in the fall of 2012. These dramatic changes appear to be the result of a combination of warmer air and ocean temperatures and the topography of the ocean floor at the head of the glacier. Rising sea levels should continue to destabilize the marine portion of Zachariæ Isstrøm for decades.

Science, this issue p. 1357


After 8 years of decay of its ice shelf, Zachariæ Isstrøm, a major glacier of northeast Greenland that holds a 0.5-meter sea-level rise equivalent, entered a phase of accelerated retreat in fall 2012. The acceleration rate of its ice velocity tripled, melting of its residual ice shelf and thinning of its grounded portion doubled, and calving is now occurring at its grounding line. Warmer air and ocean temperatures have caused the glacier to detach from a stabilizing sill and retreat rapidly along a downward-sloping, marine-based bed. Its equal-ice-volume neighbor, Nioghalvfjerdsfjorden, is also melting rapidly but retreating slowly along an upward-sloping bed. The destabilization of this marine-based sector will increase sea-level rise from the Greenland Ice Sheet for decades to come.

Zachariæ Isstrøm (ZI) and Nioghalvfjerdsfjorden glacier (NG), in northeast Greenland, drain a sector 198,380 km2 in size, or 12% of the Greenland Ice Sheet (1). These two glaciers together drain the northeast Greenland ice stream, the only large, dynamic feature that extends continuously deep to the ice sheet interior near Greenland’s summit (2). This marine-based sector holds a 1.1-m sea-level rise equivalent (3) (Fig. 1D).

Fig. 1 Ice speed, bed topography, and grounding lines of ZI and NG.

(A) Schematic view with Operation IceBridge and other NASA mission flight tracks in gray, basin boundaries in black, flux gates in thick blue and red (table S2), and the profile used in Fig. 2 in dashed black. (B) Ice surface speed from 2008 to 2009 (1), with velocity profiles used in Fig. 3 in black dots. The profiles in (A) and (B) are different. (C) Differential interferogram showing the tide-induced motion of ZI in December 2014. The inset shows detail about the pattern of tidal flexure at the grounding line. (D) Bed topography above the WGS84 ellipsoid derived from mass conservation on land (3) and gravity data at sea (4). The seafloor bathymetry beneath NG ice shelf (square) is from seismic measurements (6). Inset shows the drainage boundaries of three major marine-based basins in Greenland (3).

We constructed a high-resolution bed topography of both glaciers (Fig. 1) using a mass conservation method over grounded ice (3) and airborne gravity inversion (4) over floating ice. On ZI, we find that the grounding line in year 1996 (5) was positioned 450 m below sea level (bsl), on a previously unknown sill that crosses the entire glacier width. Seaward of the sill, the seafloor drops to 800-m bsl (Fig. 1D). Inland of the sill, the glacier bed remains between 400 and 700 m bsl for 30 km. The bed then rises to reach a ridge at sea level. The ridge is cut across by a 300-m-deep channel that connects with interior regions, where the bed remains 300 m bsl for another 150 km. On NG, the 1996 grounding line was 600 m bsl. We find no sill, and the bed is sloping upward until 45 km inland. Seismic data collected in the 1990s (6) indicate that the ice shelf floats on a 900-m bsl cavity. The seafloor rises to 200 m bsl to the east, where the ice front is anchored by islands and ice rises, and 600 m bsl to the north into Dijmphna Sund.

We use Landsat optical imagery (fig. S1) to document the ice-front positions over the past 40 years. ZI ice shelf was stable until 2002–2003, when a large section broke off (7) and ice debris cleared from Jøkelbugten fjord. The ice front retreated steadily until late 2012, when the northern and southern floating sections became disconnected. In 2013–2014, the ice-front retreat accelerated markedly and the glacier started to calve at its grounding line. By December 2014, the remaining shelf was 52 km2 in size, or 95% smaller than in 2002. Meanwhile, the calving front of NG retreated by only a few km between 2002 and 2012 (7).

We map the glacier grounding lines from 1992 to 2015 (Fig. 1C and figs. S2 and S3) using differential satellite radar interferometry (DInSAR). The grounding line of ZI retreated by 3.5 km at its center between 1996 and 2010, and 3.5 km between 2011 and 2015 (Fig. 1C). The mean rate of grounding-line retreat therefore quadrupled from 230 m/year to 875 m/year before and after 2011. On NG, the grounding line retreated 1 km between 1992 and 2011 and has remained stable since. The DInSAR observations reveal a downward tilting of the ice-front surface of ZI by 75 cm between 16 to 20 December and 20 to 24 December 2014 in a section 1 km wide by 7 km long (Fig. 1C). We attribute this deformation to a buoyancy-driven rotation of the terminus depressed below flotation and facilitated by the propagation of basal crevasses to the water line (8).

We document 40 years of surface velocity using Landsat and SAR instruments (table S1 and fig. S4). The results show that after 25 years of stability, the speed of ZI increased by 50% from 2000 to 2014, with half of that increase taking place after 2012 (Fig. 2). The glacier sped up 125 m/year every year from 2012 to 2015, or three times as fast as between 2000 and 2012. NG exhibited smaller changes: Its speed increased by 8% between 1976 and 2014, with most of the acceleration occurring after 2006. The glacier accelerations are larger than seasonal variations and extend 80km and 15 km upstream of the 1996 grounding lines of ZI and NG, respectively, indicating that the coastal changes affect a substantial portion of the drainage system (Fig. 2).

Fig. 2 Ice speed of ZI and NG from 1976 to 2015.

(A and B) Ice speed along the profile in Fig. 1B color-coded from blue (1976) to red (2015). Thick vertical lines locate the grounding lines. The inset displays ice speed versus time at the location of the black dashed vertical line.

Repeat measurements of ice thickness (±10-m precision) and surface elevation (±10-cm precision) using radar and light detection and ranging (LIDAR) data (see the supplementary materials) from 1995 to 2014 provide precise information about ice thickness change during the retreat of ZI. About 4.5 km upstream from the 2014 grounding line of ZI, the ice-thinning rate doubled from 2.5 ± 0.1 m/year, consistent with (9), to 5.1 ± 0.3 m/year during 1999 to 2010 and 2010 to 2014, respectively. On the ice shelf, the change in ice thickness is large enough to be directly measured with radar (Fig. 3). After correction for dynamic thinning and changes in surface mass balance (SMB), we find that the ice-shelf thickness at the 1996 and 2011 grounding lines decreased by 161 ± 43 m and 100 ± 50 m, respectively, during 1999 to 2010 and 2010 to 2014, reflecting enhanced bottom melting by the ocean of 14.6 ± 4.1 m/year and 25 ± 12 m/year during those time periods (see the supplementary materials). Application of mass conservation on the ice shelf indicated that in the 1990s, the steady-state bottom melt rates of ZI and NG averaged 8 and 5 m/year, respectively, and reached 25 m/year within 10 km of the grounding lines (5). We conclude that ice-shelf bottom melting doubled in recent years compared with the 1990s and that half of the increase took place between 2010 and 2014.

Fig. 3 Surface elevation and ice thickness of ZI from 1999 to 2014.

(A to C) Radar echograms along the center-line profile in Fig. 1A. Ice surface and bottom (assuming flotation) from LiDAR data, respectively, are white. Bed elevation from radar data is green. (D) Evolution of ZI between 1999 and 2014, with successive ice-front positions color-coded from dark (1999) to light gray (2014), seawater in blue, and bedrock in light brown. Vertical dashed lines locate the grounding lines.

On NG, 3.7 km upstream of the 1996 grounding line, ice thinned 0.9 ± 0.1 m/year and 1.4 ± 0.5 m/year during 1999 to 2012 and 2012 to 2014, respectively. The radar echograms show that 5 km downstream of the grounding line, the ice shelf lost 30% of its total thickness (fig. S5). This corresponds to a bottom melting of 13.3 ± 4 m/year in the past 15 years, or 50% above the melt rate from the 1990s (5). The ice shelf is therefore eroding rapidly from the bottom. We hypothesize that the erosion has not translated into an inland migration of the grounding line and ice-flow acceleration because the bed of NG rises inland and the ice-shelf front did not detach from bay walls, islands, and ice rises (fig. S5).

Combining surface velocity and ice thickness, we calculate the glacier ice discharge from 1976 to 2015 (Fig. 4 and fig. S6). On ZI, the ice flux increased from 10.3 ± 1.2 Gt/year in 1976 to 15.4 ± 1.7 Gt/year in 2015, or 50%. On NG, the ice discharge increased by 8% from 1976 to 2015. Comparing the ice discharge with net accumulation of mass over the drainage basins (fig. S7) using the regional climate model MAR (Modèle Atmosphérique Régional) (10) indicates that ZI was in a state of mass balance until 2003 and is now losing mass at about 5 Gt/year, whereas NG remains close to a state of mass balance (Fig. 4). Our discharge estimates for ZI supersede the overestimates in (7) for the period 1990 to 2012, which employed less reliable ice thickness data (figs. S8 and S9).

Fig. 4 Yearly ice discharge, surface mass balance, and runoff of ZI and NG from 1976 to 2015.

(A and B) Ice flux [D] and SMB from the regional climate model MAR (10) for ZI and NG color-coded blue and orange, respectively. (C) Runoff from MAR (10). (D) Mass evolution of NG, ZI, and both combined. Embedded Image is the mean SMB for the time period 1960–1990.

The MAR reconstruction shows that the mean surface runoff tripled from 0.6 to 1.8 Gt/year, respectively, during 1960 to 1990 and 2002 to 2014 (Fig. 4C) as a result of warmer air temperatures. Higher melting thins ice from the top and contributes to grounding-line retreat as floating ice achieves hydrostatic equilibrium farther upstream. Meltwater ponding on the ice shelf likely contributed to its break-up via hydrofracturing (11). Warmer air temperatures melted the ice mélange that keeps ice floes glued together in the fjord (7). Enhanced glacier runoff increased subglacial freshwater discharge at the grounding line, which drives a stronger thermohaline circulation at the ice underside and increases the rate of melt by the ocean (12).

Ocean in situ measurements over the period 1997 to 2010 show an increase of +1°C in mean temperature of the warm, salty subtropical-origin Atlantic Water (AW) advected from the North Atlantic toward the Arctic Ocean via the West Spitsbergen Current (WSC) (13). Although resolving the transport pathways and water-mass transformation of AW in the Northeast Atlantic and East Arctic Ocean is an area of active research (14, 15), it is known that some fraction of these warm northward-flowing waters recirculates in the northern Greenland Sea and in the southern Nansen Basin to join the southward-flowing East Greenland Current (EGC) (16). Ocean temperatures observed from moored instruments spanning Fram Strait at 78°50′N from 1997 to the present (17) show that temperature anomalies in the northward-flowing WSC also appear in the southward-flowing EGC (18). Although high-resolution (2 to 4 km) ocean simulations show that ocean temperature anomalies on the EGC propagate from the continental shelf break into Belgica Trough to within 50 km of ZI (19), the seafloor bathymetry in these critical last 50 km is not known well enough to simulate ocean circulation close to the glacier. Yet observations from 1996–1997 (6) and 2009 (20) reveal the presence of warm AW at the mouth of the NG ice-shelf cavity. A 1°C increase in AW would have increased bottom melting by 10 (21) to 20 m/year (22), which is within the range of our observations. We conclude that ocean warming most probably played a major role in triggering the glacier retreat, more important than the sea-ice concentration decrease (7). Oceanographic observations near ZI are critically needed to address the effect of thermal ocean forcing on the glacier evolution in more detail.

ZI has now transformed into a tidewater glacier calving along an ice cliff as a result of warmer air and ocean temperatures. The mass loss is driven by the increase in ice discharge rather than a change in SMB (Fig. 4). The glacier detached from a stabilizing sill and retreated into a retrograde basin 700 m bsl. Tidewater glaciers are known to retreat rapidly along retrograde beds until the bed rises again (23). We project that ZI may continue retreating rapidly for another 20 to 30 years. Its ice front will progressively widen from 19 km at present to 50 km about 30 km upstream, thereby increasing ice discharge. The height of the calving cliff will increase from its current 75 m to enhance the risk of ice fracture (11). With the formation of a calving cliff, the ocean-induced melt rates will increase considerably because buoyant meltwater plumes rise faster along a vertical face than along a near-horizontal ice-shelf bottom (5, 12). Beyond 30 km, the retreat will be slowed down by a rising bed topography, but submarine channels will maintain the contact with the ocean into the deep interior.

The ZI/NG sector is one of three major marine-based basins in Greenland (fig. S10), along with Jakobshavn Isbræ (JI) and Petermann (PG)–Humboldt glaciers, each holding a 0.6-m sea-level equivalent. JI started a rapid retreat (18 km from 2001 to 2015) after the collapse of its ice shelf and has undergone massive calving events since 2010 (24) (fig. S11). The central channel of the PG ice shelf lost 250 m of ice from 2002 to 2010, and the ice front retreated 33 km from 2010 to 2012 (25). The NG ice shelf will become vulnerable to breakup in the near future if thinning continues. These observations combined suggest that all three major marine-based basins are undergoing substantial changes at present. JI and ZI have already transitioned to a tidewater glacier regime, with increased calf-ice production and ice melting by the ocean. The retreat of these marine-based sectors is likely to increase sea-level rise from Greenland for decades to come.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S11

Tables S1 and S2

References (2643)

References and Notes

  1. T. Haran, J. Bohlander, T. Scambos, T. Painter, M. Fahnestock, Modis Mosaic of Greenland (MOG) Image Map (2013).
  2. Acknowledgments: This work was performed under NASA grants NNX13AI84A (E.R.), NNX14AB93G (E.R.), NNX13AD53A (J.P.), and NNX15AD55G (M.M.), and NSF grant ANT-0424589 (J.P.). The work of I.F., A.K., and E.R. was carried at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA. We gratefully acknowledge European Space Agency, Canadian Space Agency, Japan Aerospace Exploration Agency, Agenzia Spaziale Italiana, National Aeronautics and Space Administration, and Deutsches Zentrum für Luft- und Raumfahrt e.V. for providing SAR data and Polar Space Task Group for coordination of SAR acquisitions.
View Abstract

Navigate This Article