Applications: Cryosphere (2)

Ice-sheets in the Arctic region are retreating rapidly since late 1990s. Typical ice loss rates are 0.5 - 1 m/yr for the Canadian Arctic Archipelago, ~ 1 m/yr for the Icelandic ice sheets, and several meters per year for the edge of Greenland ice sheet. Such load decreasing causes measurable (several millimeter per year) deformation of the Earth’s crust from Synthetic Aperture Radar Interferometry (InSAR). Using small baseline time-series analysis, we retrieve the signal after noises such as orbit error, atmospheric delay and DEM error being removed.

We present results from Vatnajokull ice cap, Petermann glacier and Barnes ice cap using ERS, Envisat and TerraSAR-X data. Up to 2 cm/yr relative radar line-of-sight displacement is detected. The pattern of deformation matches the shape of ice sheet very well. The result in Iceland was used to develop a new model for the ice mass balance estimation from 1995 to 2010.  Other applications of this technique include validation of ICESat or GRACE based ice sheet model, Earth’s rheology inversion. Moreover, we find a narrow (~ 1km) uplift zone close to the periglacial area of Petermann glacier which may due to a special rheology under the ice stream.

Matusevich Ice Shelf (MIS), 241 km² in size (1953), the largest floating glacier in the Old World split up into three unequal parts and lost three fifth of its area in late summer 2012. The ice shelf collapse continued in 2013-2014 and exited debates about the emergence of new icebergs threatening the exploration, extraction and transportation of hydrocarbons in the Kara and Barents seas. We located and dated main MIS calving events using optical and radar image time series from LANDSAT, QuickBird, WorldView and Sentinel satellites. Matusevich event icebergs were mostly observed in 2012 with surprisingly few large icebergs found in 2013-2014.

The first MIS elevation model applicable to the computer analysis of ice shelf characteristics and volume changes was generated from the TanDEM-X SAR interferometry data of 05.05.2011 available from the DLR project ID.XTI_GLAC0249 and validated using the concurrent CryoSat-2 interferometric radar and ICESat lidar altimetry data collected within the framework of the MAIRES FP7 research project. The comparison with the cartographic DEM derived from Russian topographic maps 1:100 000 showed that the TDX elevation model is superior in depicting ice surface undulations and compact glaciological features, such as ice fronts, walls, crevasses etc., yet without compromising on positional accuracy. In this model, we detected six elevated and crevassed ice shelf areas alternating with lower areas of thin even ice, apparently of marine origin. MIS top heights of up to 32 m asl are typically observed in front of tributary glaciers that feed into the ice shelf, while extensive areas of thinner ice with medial heights of several meters are to be found along ice-free coasts. This observation sheds some light on spatiotemporal variations of the MIS iceberg production capacity.

The TDX elevation model was further applied to differential interferometric processing of tide-coordinated ERS-1/2 InSAR data of 23/24.09.1995 and Sentinel-1 EW SAR-C data of 9/21.10.2014. The resultant “retrospective” and “prospective” (DIn)SAR models were used for measuring ice flow velocities of tributaries and their inter-annual changes, mapping elevation changes on parent ice caps, reconstructing pre- and post-event ice shelf deformations, estimating surficial strain rates, and detecting the ice shelf grounding line. The boundary between the ice shelf and 7 tributaries was drawn along the intermediate grounding line derived from tide-corrected differential interferograms. A dozen new islands underpinning the ice shelf were first discovered in the InSAR products and, afterwards, verified in high-resolution optical images. We revealed that two out of seven tributaries increased flow speeds by nearly 200 % over the past 20 years, which could not be related directly to the breakup event, however. Besides, we recognized that high strain rate values might reliably indicate areas of further MIS retreat and determined that the present MIS disintegration began no later than mid-August 2011 when the first large portion of the “stressed” south-eastern margin manifested in the strain-rate image product was lost.

Intermediate interferometric models, radar and optical imagery were geocoded and orthorectified using the same TDX DEM and served as basic layers for the output map series and animations of the MIS disintegration in 1931-2014 at 1:100,000 scale (UTM 46N, WGS84), which were included into the “Online atlas of glacier fluctuations” accessible at http://dib.joanneum.at/MAIRES/index.php?page=atlas. Topographic contour lines with 5 m contour interval on the ice shelf surface were generated and ice shelf areas with thin ice were delineated in semi-automatic mode. The MIS present total area was measured as 85 km² (IX, 2014) and the remaining ice shelf volume was given as 7.0 km³, which is twice as small as it was after the previous breakup in 1984-85. Field validation studies including radio-echo sounding were carried out in September 2014.

30-year long records of hydrometeorological and oceanographic data obtained from 4 coastal stations around MIS as well as ancillary oceanographic, bathymetric and gravimetric publications were involved in the analysis of main causes and consequences of the MIS breakup. It was decided that the unfavourable combination of long-term warmth, liquid precipitation and strong winds in summer as well as essentially negative mass balance of both parent ice caps, heterogeneous structure of the ice shelf and, probably, low concentrations of sea ice along the eastern coast of Severnaya Zemlya was the main driving factor for the MIS event in 2011-2014. It was supposed that, under current environmental conditions, MIS will disappear by 2020. Thanks to wide terrestrial coverage of Sentinel data, we were able to compare the rate of MIS breakup with the concurrent ice loss rates on two smaller ice shelves in Severnaya Zemlya (No.19) and Franz Josef Land (No. 8) situated 70 km to the north and 370 km to the west of MIS respectively. We conclude that the rapid collapse of Russian ice shelves is a new sign of climate change in the Eurasian High Arctic and is consistent with the observations made recently in the Canadian Arctic, where large ice shelves demonstrate similar behaviour.

Sparse GPS surveys find that ocean tides modulate the downstream motion of selected Antarctic ice streams, even at positions many tens of kilometers inland from the grounding zone (e.g., Gudmundsson, 2006; 2007).  These surveys suggest that ocean tidal stresses can perturb ice stream motion at distances about an order of magnitude farther inland than tidal flexure of the ice stream alone.   Using Rutford Ice Stream as a real-world comparison, one can show that the assumption that tidal stresses in ice streams propagate elastically large distances inland fails for channelized glaciers due to an intrinsic, exponential decay in the stress caused by resistance at the ice stream margins (Thompson et al., 2014).  This behavior is independent of basal conditions beneath the ice stream and cannot be fit to observations using either elastic or nonlinear viscoelastic ice rheologies without nearly complete decoupling of the ice stream from its lateral margins.  These results suggest that a mechanism external to the ice stream is necessary to explain the tidal modulation of stresses far upstream of the grounding zone for narrow ice streams.  Thompson et al (2014) hypothesize that tidally driven water pressure fluctuations are readily transmitted through the subglacial hydrological system producing time-dependent variability in till strength that could explain transmission of tidal stresses inland of the grounding line.  This conceptual model can reproduce observations from Rutford Ice Stream.  A Coulomb-plastic till has previously been proposed for WAIS (e.g., Kamb, 1991; Tulaczyk et al., 2000).  At the same time, surface velocity observations of glacier flow on Icelandic ice caps using UAVSAR, NASAs airborne InSAR platform, also suggest a strong role for Coulomb-like plastic behavior of the subglacial till.  Such a Coulomb model may be consistent with the inferred conceptual model for Rutford Ice Stream.

To further understand the synoptic-scale response of ice flow to tidal loading in Rutford and other ice streams feeding into the Filchner-Ronne Ice Shelf, we explore the ability of satellite InSAR to resolve temporal variability in 3D vector displacements associated with harmonic tidal forcing at semi-diurnal, diurnal and fortnightly periods.  Relying on the short repeat time observations made possible by the COSMO-SkyMed SAR satellite constellation launched and operated by the Italian Space Agency (ASI), our approach is based on InSAR time series analysis assuming known periods.  This analysis combines both interferometric phase and pixel offset observations from multiple line-of-sight imaging directions for interferometric pairs that span time intervals of a single day to multiple days.   Along with measuring the tidal responses, we explore the ability to use InSAR time series approaches to improve our estimates of glacier surface topography and the spatial complexity in the location of the grounding zone. 

Columbia Glacier is a grounded tidewater glacier located on the south coast of Alaska. It has lost half of its volume during 1957-2007, more rapidly after 1980. It is now split into two branches, known as Main/East and West branch due to the dramatic retreat of >23 km and calving of iceberg from its terminus in past few decades. In Alaska, a majority of the mass loss from glaciers is due to retreat , increased ice flow and calving icebergs into tidewater and lacustrine environments.

We use time series of high-resolution TanDEM-X stripmap satellite imagery during 2011-2014. The active image of the bistatic TanDEM-X acquisitions, acquired over 11 or 22 day repeat intervals, are utilized to derive surface velocity fields using SAR intensity offset tracking. Due to the short temporal baselines, the precise orbit control and the high-resolution of the data, the accuracies of the velocity products are high. We observe a pronounce seasonal signal in flow velocities close to the glacier front of East/Main branch of Columbia Glacier. Maximum values at the glacier (Main branch) front reach up to 14.43 m/day, recorded in May, 2012, 11.7 m/day in June, 2013. Minimum velocities at the glacier front are generally observed in September-October with lowest values below 2 m/day in October 2012. Months in between those dates show corresponding increase or decrease resulting a kind of sinusoidal annual course of the surface velocity at the glacier front. The seasonal signal is consistently decreasing with the distance from the glacier front. At a distance of 17.5 km from the ice front, velocities are reduced to 4-5 m/day and almost no seasonal variability can be observed. We attribute these temporal and spatial variability to changes in the basal hydrology and lubrification of the glacier bed. A fully developed basal drainage system leads to decrease the speeds to their minimum while in early winter, due to closure of the basal drainage system leads to increase the speeds. We also analyze the variation in conjunction with the prevailing meteorological conditions as well as changes in calving front position in order to exclude other potential influencing factors. In a second step, we also exploit TanDEM-X data to generate time series of digital elevation models (DEMs). The multi-temporal DEMs are used to estimate the difference in surface elevation and respective ice thickness changes. All TanDEM-X DEMs are tied with one TDX DEM. The global offsets between the DEMs are estimated over ice free moraines and rocky areas. The quality of the TanDEM-X DEMs on snow and ice covered areas are further assessed by a comparison to laser scanning data from NASA Icebridge campaigns in 08/2011 and 03/2012. The time wise closest TanDEM- X DEMs were compared to the Icebridge tracks from winter and summer surveys in order to judge errors resulting from the radar penetration of the x/band radar signal into snow, ice and firn. We can identify clear dramatic signals leading to elevation differences partly caused by varying penetration depth in summer and winter. Retreat rates of the glacier front are derived manually by digitizing the terminus position. This leads us to estimate the temporal volume change of the glacier.

While the glaciers outside of Greenland and Antarctica are a small percentage of the surface area of global land ice (about 4%), their contribution to sea level rise between 2003-2009 has been about equal to all of Greenland and Antarctica.  For the rest of the 21st century, the contribution of these glaciers and ice caps (GICs) outside Greenland and Antarctica to sea level rise will be significant (perhaps 0.1 to 0.2 m out of their total potential contribution of 0.4 m).  Further,  GICs are a water resource and hazard to nearby populations.  To better undestand the changes occurring at GICs, we have focused multi-instrument observations at two of the most rapidly changing GICs (Alaska and Patagonia) as well as in the Russian Arctic, where the GICs were thought to be changing less.  We derive ice speeds using pixel-tracking on radar and optical imagery, with additional information from InSAR.   Thinning rates over the last few decades are derived by regressing stacked elevations from multiple Digital Elevations Models (DEMs) sourced from ASTER and Worldview stereo-imagery and cartographically derived DEMs. DEMs are calibrated and co-registered using ICESat returns over bedrock.  In this presentation, we will compare and contrast the effectiveness of the different measurement techniques and the observed glacier behavior in the different geographic and climatic settings.  While we see rapid changes at several glaciers, one of the  outliers is Inostrantseva Glacier in Novaya Zemlya of the Russian Arctic that has accelerated by 200 to 300% and retreated by more than 3 km since 2006. 

During the round table, seed questions proposed by the chairs will be discussed with the audience.