Kongsvegen and Kronebreen are two tidewater glaciers that converge close to their calving fronts the latter pinning the tongue of the former - within Kongsfjorden, near to Ny Alesund in North-west Spitsbergen (see Figure 46). Kronebreen draws from both Isachsenfonna and Holtedahlfonna in its accumulation area. The Kongsvegen basin has two main tributaries Kongsvegen and Sidevegen with a number of other small basins contributing to these. Whilst both are polythermal glaciers with generally low slope, their flow regimes are markedly different. Kronebreen displays a heavily-crevassed surface, with ice velocities of > 2m per day measured in the central part measured in the field in 1990 (Melvold 1992) and confirmed from SPOT satellite imagery (Lefauconnier et al., 1994). By contrast, Kongsvegen presents a much smoother surface, with velocities in the order of 1-2 cm/day (Melvold, 1992). Kongsvegen is currently in a quiescent stage, and may be building up to a new surge phase. These contrasting glaciers are therefore of interest both glaciologically, and in terms of the utility of InSAR techniques for their monitoring.

Location of Kongsvegen and Kronebreen in NW Spitsbergen. Coverage of ERS track 438, Frame 1989 shown in red box. Extent of ASTER DEM data also shown, as well as GPS stake locations (red) and GLAS elevation locations (blue).

A number of ERS-1/2 Tandem phase (1-day) and ERS-1 3-day repeat pairs of SAR scenes were available to the INTEGRAL project, courtesy of VECTRA project, PI A. Shepherd (SPRI). Scenes highlighted were received and processed prior to creation of this CD other scenes have been ordered, but not yet processed.

Scenes extend over Jan 1992 Aug 1996, with at least one pair for each month of the year except June. Ascending and Descending geometries are covered, allowing for application of DINSAR techniques.

The figure to the left shows the extent of Icesat GLAS laser altimetry data (blue) available for the region around Kongsvegen and Kronebreen. The GLAS sensor has an ~70 m footprint. Whilst there is a temporal mismatch between the ERS SAR (1992 1996) and the GLAS data (200x), we do not anticipate this being a major source of error over the higher reaches of the glacier accumulation areas. This data has been useful in extending GCP coverage more widely over the SAR data than afforded by the limited number of GPS-derived stake locations over Kongsvegen. However, it is clear that data coverage is sporadic, and it may not be possible to rely on this data source for other areas.

Cryosat LRM mode

Cryosat SAR mode

Cryosat SARIN mode

The figure to the right shows Cryosat orbit tracks over the same part of Svalbard including the Kongsvegen and Kronebreen glacier areas over one complete orbit cycle (just over 1 year). CryoSat orbit crossing point density exceeds 100 per 100 x 100 km region at high (>60 degree) latitudes. It can be seen that within one orbit cycle (369 days) of data availability, a moderate network of 3D points could be expected from this area, with good opportunity for assuring data quality at crossovers. Orbit tracks were calculated using ESOV software (http://eop-cfi.esa.int/ESOV): within the area delineated as approximating the Kronebreen and Kongsvegen glacier catchments, some 6969 elevation measurements (3488 Ascending, 3481 Descending) are estimated with a maximum between-track spacing of ~1.5 km.

Estimated Cryosat orbit tracks

CRYMPS simulation for Kongsvegen / Kronebreen

The figure to the left shows results from a run of the CRYMPS simulator over a DEM of the Kongsvegen / Kronebreen area. The image illustrates the location of the centre of the radar altimeter footprint as the track traverses surface topography. Click the image to see an enlarged version in a new window.

The figure to the right illustrates the expected footprint of Cryosat/2 for a track over Kongsvegen / Kronebreen (click image to open larger version in a new window). The hatched line predominately top to bottom is a set of boxes (not scaled) marking the center of the footprint for each along-track step (~250 m) simulated by CRYMPS. The circle shows a nominal Low Resolution Mode footprint of ~ 10 km radius (i.e. when the instrument is operating as a conventional Pulse-limited Radar Altimeter). The line from left to right within this circle shows a SAR-mode footprint strip of ~250 m along-track, within which the nearest-return elevation point will be returned. In SARIN mode, a set of across-track point offsets may be determinable.

CRYMPS illustration of Cryosat Footprint

Left: result of merging unwrapped topographic phase from differential processing with mosaiced ASTER DEM simulated phase; Right: result of refinement of composite DEM. Observe detail added in place of too-smooth extrapolated areas.

Interferogram E1-23472/E2-03799 was subtracted from E1-24474/E2-04801 to form a topographic differential interferogram with a perpendicular baseline of ~150m (assuming constant ice velocity between the January and March 1996 acquisition dates).

Much differential topographic phase was unwrappable over glacier areas. However, some areas proved problematic in particular, rock areas, and some glacier margins (see 47). The heavily-crevassed lower part of Kronebreen shows noisy phase patterns and intermittent, generally low coherence, rendering it impervious to our best unwrapping efforts in common with the experience of (e.g.) Eldhuset et al (1996)

The mosaiced ASTER DEM data was geocoded to the SAR geometry and used to simulate topographic phase. The result was then adjusted according to best-fit against unwrapped topographic phase in overlapping areas and a composite topographic phase map was produced, taking unwrapped phase where available, in preference to the ASTER-derived phase. The more complete topographic phase estimate was subtracted from the differential (wrapped) topographic phase, to give an interferogram representing the residual errors from the ASTER-derived estimate. After unwrapping, the result was added to the extended topographic phase estimate to give a corrected topographic phase map. We repeated the same iterative process using the NPI DEM (20m resolution), which is likely to be of a better quality than the ASTER DEM data, and does not require mosaicking of multiple tiles.

Look-direction displacements calculated by subtraction of enhanced NPI DEM-derived topographic phase from shorter-baseline (~35 m) interferogram E1-23472/E2-03799 and unwrapping.

Subtraction of enhanced NPI DEM from the shorter-baseline (~35 m) interferogram E1-23472/E2-03799 and then unwrapping the remaining phase yielded the look-direction displacement map shown left. Repeating this versus the longer-baseline (~180 m) interferogram E1-24474/E2-04801 showed a noisier result from residual topography/baseline error and unwrapping error. Use of the ASTER DEM for the same purposes produced more artifacts. Considering flow units presenting in the look direction, we see that the eastern tributary of Kronebreen flows at circa 20 cm/day across profile A-A, speeding up to ~55-60 cm/day across B-B, after which point the unwrapping fails. Speckle/amplitude tracking trials did not yield useful results beyond this point either. By contrast, the inflow from a more northerly tributary across C-C achieves up to ~7-8 cm/day. From manual comparison with adjacent rock areas, we estimate an error of ~2-3 cm for these displacement values. The south-west flowing Sveabreen shows a look-direction displacement of upto ~45-50 cm/day across profile D-D, rising to 60 cm/day across profile E-E close to the margin. However, it is possible that these latter values are too high by upto ~7-12 cm/day, due to a possible combination of unwrapping / topography / baseline errors.

Full-resolution interferogram over Kronebreen glacier: repeated adaptive filtering clarifies fringe structure considerably, although this risks aliasing effects, and still proves problematic for phase unwrapping

In contrast to other study areas, application of offset tracking techniques using ERS-1/2 SAR data has not proved useful for the fast-moving (> 2 m / day in some parts) Kronebreen glacier. It may be possible to complement InSAR results over slower-moving parts of the glaciers with other sources of displacement data such as ground-based GPS, or offset tracking results from non-SAR (e.g. ASTER visible/near-infra red imagery). We have made some attempt at using full resolution InSAR data, to retrieve displacement values. Whilst there is evidently a pattern of fine fringes discernable to the eye which may be lost through aliasing in processing multi-looked interferograms, the full-resolution data suffers from greater noise and low coherence. The repeated application of adaptive filtering to the full-resolution interferogram provides a visually clearer fringe pattern of higher coherence, although care must be taken to avoid loss of fringes through aliasing. The improvement has not so far been sufficient to allow unwrapping to succeed over the bulk of the lower glacier.