A Digital Elevation Model of West Antarctica from MODIS and ICESat: Method, Accuracy, and Applications

An image enhancement approach was used to develop a new digital elevation map of West Antarctica, combining multiple MODIS images and ICESat laser altimetry profile data. The method combines the wide image coverage of MODIS, and its high radiometric sensitivity (which equates to high sunward slope sensitivity), with the high precision and accuracy of ICESat track elevation data. By using pixel brightnesses for image data where ICESat tracks provide an accurate slope, we calibrated brightness-to-slope for several MODIS images of the West Antarctic. Using the calibrations, we then created, first, a slope map of the entire ice sheet surface from the image data, and then integrated this to yield a complete DEM for the region. ICESat (as of September 2006) has acquired a series of eight near-repeat tracks over the Antarctic during the period September 2003 to June 2006, covering the continent to 86 deg S. ICESat data are aquired as a series of spot elevations, averaging a ~60m diameter surface region every ~172m. Using the multiple-track 'ribbons' of elevation data, two of us (DY and JZ) were able to generate along- and across-track surface slope fields for the Antarctic continent. However, ICESat track paths have spacings wide enough (2 km at 85 deg; 20 - 50 km at 75deg) that some surface ice dynamical features (e.g. flowlines, undluations, ice rises) are missed by the slope and track data. We combined the restricted-coverage but high-accuracy ICESat data with cloud-cleared MODIS band 1 data from the 2003-2004 austral summer, used in generating the Mosaic of Antarctica, MOA, surface morphology image map. Past analyses of the slope-brightness relationship for MODIS have shown ice surface slope precisions of +/- 0.00015. Multiple images can improve the single-scene precision (e.g. the effective radiometric resolution) and spatial resolution (nominally 250m for a single scene). ICESat spot elevation have nominal precisions of ~5 cm under ideal conditions, although thin-cloud effects and mis-location errors can magnify these. A suite of applications for an enhanced DEM are identified and explored. A full representation of the WAIS undulation field permits a better investigation of the relationship between accumulation and topography, and surface temperature and topography. Further, addressing the shape modifications introduced by the variations in accumulation across undulations is a necessary prerequisite before inverting surface topography for bed elevation. Lastly, surface topography and detailed bed topography are both required for inferring sub-ice-sheet hydrostatic pressure. We will discuss these potential applications.