Applications: Cryosphere (1)

INTRODUCTION

Mapping the velocity fields of the continental ice sheets and their outlet glaciers is important in order to monitor and model the response of the cryosphere to global climate change.  Since the mid 1990s, space-based SAR data have enabled measurement of ice velocities on a continental scale [1].  Compared to interferometry, Offset Tracking techniques excel in terms of robustness and ease of automation.  Offset Tracking estimates the range and azimuth displacements by determining the position of the peak resulting when cross-correlating two ice patches mapped with a temporal separation, typically between one day and one month.  The term Offset Tracking here refers to a class of methods including (complex or incoherent) speckle tracking [2][3] and feature tracking [4][5][6].  The former does not call for surface features like glacier crevasses, but relies on a stable speckle pattern, which in turn requires coherence to some extent.

PROBLEM

TOPS SAR data offer a wide swath at the expense of a coarser azimuth resolution [8], and the interferometric wideswath (IW) product of the Sentinel-1 SAR [10] supports interferometry, as the bursts are synchronized from pass to pass [9] such that the corresponding data pairs are spatially aligned for stationary scenes.  Between the two data acquisitions, however, a scatterer with an along track velocity component may move from one burst to the neighbouring one.  In the Sentinel-1 GRD product, where the bursts do not overlap [10], this would imply that a patch having a positive Doppler centroid is combined with a patch having a negative Doppler centroid, and since the azimuth steering angle variation of the antenna exceeds the beam width the two Doppler spectra do not overlap, and no cross-correlation peak results.  In this case speckle tracking would fail, whereas feature tracking might still work in presence of surface features.

Fortunately, in the Sentinel-1 SLC product neighbouring bursts overlap [10], so patches from corresponding bursts can be cross-correlated, provided the azimuth displacement of the ice does not exceed the burst overlap.  Hence, when using SLC data, also speckle tracking is expected to work.

The non-zero Doppler centroid rate introduced by the antenna scanning is a challenge for interferometric processing of nonstationary scenes [11].  Offset Tracking is less impacted, as a constant phase difference has no impact, even not on the complex Offset Tracking.  In principle the Doppler centroid rate call for deramping and azimuth common band filtering if speckle tracking is applied to fast moving glaciers [11], but in practice features are more important than speckle when cross-correlating such fast glaciers.

METHODOLOGY

DTU Space has implemented an operational interferometric post processing facility IPP, which has been upgraded with an Offset Tracking capability in the frame of ESA’s Climate Change Initiative [12].  The IPP processor supports the Sentinel-1 SLC product, but currently no Sentinel-1 SLC data pairs from Greenland or Antarctica has been made available.  However, the IPP processor has been tested with RADARSAT-2 data acquired in TOPS mode [13].  An ice velocity map covering the Petermann glacier and the adjacent ice sheet in the accumulation zone has been successfully generated.  Once appropriate Sentinel-1 SLC data pairs become available more results will be generated.  For a range of ice types, the normalized cross-correlations will be computed and compared in statistical terms, thereby testing the hypothesis that

1) in presence of features, a burst in the master data correlates with the corresponding burst in the slave data as well as with the overlapping burst in the slave data

2) in absence of features a burst in the master data correlates with the corresponding burst in the slave data but not with the overlapping burst in the slave data.

If this is verified both GRD data and SLC data can be used to map the velocity of glaciers with surface features, while only SLC data ensure full coverage when the ice does not have any features.  In this case, using GRD data results in gaps with a size corresponding to the ice displacement.

REFERENCES

[1]     E. Rignot, J. Mouginot, B. Scheuchl, “Ice flow of the Antarctic ice sheet”, Science, Vol. 333, p. 1427-1430, 2011.

[2]     A.L. Gray, N. Short, K.E. Mattar and K.C. Jezek, “Velocities and flux of the Filchner ice shelf and its tributaries determined from speckle tracking interferometry”, Canadian Journal of Remote Sensing, Vol. 27, No. 3, 2001.

[3]     I. Joughin, “Ice-sheet velocity mapping: a combined interferometric and speckle-tracking approach”, Annals of Glaciology, Vol. 34, No. 1, p. 195-201, 2002.

[4]     B.K. Lucchitta, C.E. Rosanova, K.F. Mullins, “Velocities of Pine Island glacier, West Antarctica, from ERS-1 SAR images, Annals of Glaciology, Vol. 21, 277-283, 1995.

[5]     R. Michel, E. Rignot, “Flow of Moreno glacier, Argentina, from repeat-pass Shuttle Imaging Radar images: comparison of the phase correlation method with radar interferometry”, Journal of Glaciology, Vol. 45, No. 149, p. 93-100, 1999.

[6]     R. De Lange, A. Luckman, T. Murray, “Improvement of satellite radar feature tracking for ice velocity derivation by spatial frequency filtering”, IEEE Transactions on Geoscience and Remote Sensing, Vol. 45, p. 2309-2317, 2007.

[7]     R. Bamler, M. Eineder, “Accuracy of differential shift estimation by correltion and split-bandwidth interferometry for wideband and delta-k SAR systems”, IEEE Geoscience and Remote Sensing Letters, Vol. 2, p. 151-155, 2005.

[8]     F. De Zan, A.M. Guarnieri, “TOPSAR: Terrain observation by progressive scans”, IEEE Transactions on Geoscience and Remote Sensing, Vol. 44, No. 9, p. 2352-2360, 2006.

[9]     https://sentinel.esa.int/web/sentinel/user-guides/sentinel-1-sar/acquisition-modes/interferometric-wide-swath

[10]   https://sentinel.esa.int/web/sentinel/user-guides/sentinel-1-sar/product-types-processing-levels/level-1

[11]   R. Scheiber, M. Jager, P. Prats-Iraola, F. De Zan, …, “Speckle tracking and interferometric processing of TerraSAR-X TOPS data for mapping nonstationary scenarios”, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, DOI 10.1109/JSTARS.2014.2360237, 2014.

[12]   http://esa-icesheets-cci.org/?q=products#IV_Description

[13]   D. Geudtner, “Implementation of the TOPS mode on RADARSAT-2 in support of the Copernicus Sentinel-1 mission; RADARSAT-2 TOPS SAR interferometry (InSAR) scene pair data acquisitions”, ESA reference S1-TN-ESA-SY-0452, 2014.

Over the past decade, ice loss from the Greenland Ice Sheet increased significantly as a result of extended surface melting and enhanced ice discharge caused by acceleration of marine-terminating outlet glaciers. Sentinel 1A, launched in April 2014, is the new flagship of SAR satellite missions, enabling comprehensive operational monitoring of the Earth’s surface, including ice sheets and their outlet glaciers. Sentinel-1 SAR supports different acquisition modes, including StripMap and Interferometric Wide Swath (IWS) mode. The latter is planned to be the main acquisition mode over land surfaces. IWS mode applies the novel Terrain Observation by Progressive Scans (TOPS) acquisition technology, providing a spatial resolution of about 3 m and 22 m in slant range and azimuth, respectively. In order to demonstrate the potential of Sentinel-1 for ice sheet observations we derived velocity fields of the outlet glaciers along the Greenland west coast from 12 day repeat pass data of Sentinel 1 data acquired in autumn 2014. We applied an iterative offset tracking algorithm using image cross-correlation enabling to map the full spectrum of flow velocities by using small matching windows. The effect of different Sentinel-1 SAR products, like SLC and GRD-H, as input for the offset tracking is studied for selected outlet glaciers, including Jakobshavn glacier and others. The Sentinel-1 velocity products are compared with ice velocity maps retrieved from repeat pass TerraSAR-X Stripmap mode data in order to assess particular features of the various velocity products. We will present ice velocity maps from different Sentinel-1 products (SLC, GRD-H) and comparisons with TerraSAR-X and TanDEM-X based ice velocity products. Features of the Sentinel-1 ice velocity maps covering the outlet glaciers along the Greenland West Coast of 2500 km length will be high lighted in relation to ice velocity maps available from previous satellite missions.

Ice velocity measurements produced from Interferometric Wide (IW) swath mode Sentinel-1 data are examined to determine how well Europe’s newest Synthetic Aperture Radar (SAR) mission performs in West Antarctica. Since 1992 ice velocity in West Antarctica has sped up markedly, with individual ice streams such as Pine Island Glacier nearly doubling in speed over the 22 year period (Mouginot et al, 2014). Despite a clear long term trend for increasing ice velocity, speed up has not been constant through time and multiple years with no significant change have also been observed (Joughin et al 2010). Ice velocity is an important geophysical parameter that can be used in combination with auxiliary ice thickness and surface mass balance data to measure ice sheet mass balance. Recent studies have shown that ice mass loss from West Antarctica has increased from 31± 25 Gt/yr during the early 1990’s to 137 ± 17 Gt/yr Gt/yr over the past 5 years, an increase of over 70% (Shepherd et al, In Press). It is necessary to make present day measurements of ice velocity to provide an independent means of measuring ice mass loss from one of the most rapidly changing ice sheet regions. Since 1992 ESA SAR satellites such as ERS-1, 2 and ENVISAT have had a pivotal role in measuring ice sheet velocity, with dedicated short temporal repeat campaigns designed to optimise SAR performance over ice. The recent launch of Sentinel-1 in April 2014 brings in a new era of Earth Observation, and the short 12 day repeat period and large scale coverage will be a useful resource for ice sheet monitoring in the future. Here I apply conventional feature tracking and interferometry (InSAR) techniques measure ice velocity from Sentinel-1 data in Antarctica. I will present ice velocity’s measured from tracking features in real valued intensity images produced from both Ground Range Detected (GRD) and Single Look Complex (SLC) data formats, both of which are now freely available from ESA. I will present my first InSAR results from Sentinel-1 data in West Antarctica and I will assess the level of coherence Sentinel-1a achieves over ice with a 12 day repeat, in comparison to ERS with shorter 3 and 6 day repeat periods. I will compare 2014 ice velocity measurements from Sentinel-1 with data from earlier missions including ERS-2 and TerraSAR-X, to extend the record and investigate if West Antarctic ice velocities have continued to increase over time.

We use satellite observations to document rapid acceleration and ice loss from a formerly slow-flowing, marine-based sector of Austfonna, the largest ice cap in the Eurasian Arctic. During the past two decades, the sector ice discharge has increased 45-fold, the velocity regime has switched from predominantly slow (~ 101 m/yr) to fast (~ 103 m/yr) flow, and rates of ice thinning have exceeded 25 m/yr. At the time of widespread dynamic activation, parts of the terminus may have been near floatation. Subsequently, the imbalance has propagated 50 km inland to within 8 km of the ice cap summit. Our observations demonstrate the ability of slow-flowing ice to mobilize and quickly transmit the dynamic imbalance inland; a process that we show has initiated rapid ice loss to the ocean and redistribution of ice mass to locations more susceptible to melt, yet which remains poorly understood.

We employ InSAR and Landsat time series of the northeast sector of Greenland, in combination with ice thickness and surface mass balance data, to determine its state of mass balance from the 1970s to present. Ice velocity is calculated using Landsat, ERS-1/2, Envisat, RADARSAT-1, RADARSAT-2, ALOS PALSAR, TerraSAR-X, TanDEM-X, and Sentinel-1a. Ice thickness combines information from airborne radio echo sounding data, ice motion vector (to apply mass conservation in between sounding lines) and airborne gravity. We also reconstruct the sea floor bathymetry in front of the glaciers using NASA's Operation IceBridge (OIB) airborne gravity. Surface mass balance is from the University of Utretch's RACMO2. The northeast sector of Greenland drains into three major outlet glaciers: 79north, Zachariae Isstrom and Storstrommen. Starting around 2004, the ice shelf in front of Zachariae Isstrom broke off, which resulted in a progressive but moderate speed up of the glacier. Between 1996 and 2011, ERS-1/2 3day repeat indicate that the grounding line retreated by several km but the glacier did not speed up in a major way. Around 2012, however, the glacier speed up became more significant, suggesting that this part of Greenland is now in a stage of progressive collapse since this is a marine-based sector. Records of ocean temperature do corroborate these glacier observations, i.e. ocean temperature has been increasing over that time period and ice shelf melt rates must have increased, triggering the retreat. We conclude on the contribution to sea level rise from north-east Greenland and its possible evolution in the coming decades and how future observations from Sentinel-1a/b will help us observe and understand this evolution, in particular with the goal to make numerical simulations of ice sheet evolution more realistic. 

We combine measurements of ice velocity from Landsat feature tracking and satellite radar interferometry from ERS-1/2, RADARSAT-1/2, ALOS PALSAR, TanDEM-X and Sentinel-1a, with ice thickness from existing compilations to document 41 years of mass flux from the Amundsen Sea Embayment (ASE) of West Antarctica. We also measure the grounding line retreat of glaciers draining into ASE using ERS-1/2 satellite radar differential interferometry from 1992 to 2011. The total ice discharge has increased by 77% since 1973. Half of the increase occurred between 2003 and 2009. Ice speeds of Pine Island Glacier stabilized between 2009 and 2013, following a decade of rapid acceleration. Flow speeds across Thwaites Glacier increased rapidly after 2006, following a decade of near-stability, leading to a 33% increase in flux between 2006 and 2013. Haynes, Smith, Pope, and Kohler Glaciers all accelerated during the entire study period. Associated with the flux and speed increases, the grounding lines have retreated 31 km on Pine Island, 14 km on Thwaites, 10 km on Hayne and 35 km on Smith/Kohler glaciers. We conclude on the possible development of a marine ice sheet instability in this part of Antarctica and how future observations from Sentinel-1a/b are crucial to follow the rapid evolution of ASE in the coming decades.