Evaluating digital elevation models for glaciologic applications: An example from Nevado Coropuna, Peruvian Andes
This paper evaluates the suitability of readily available elevation data derived from recent sensors - the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and the Shuttle Radar Topography Mission (SRTM) - for glaciological applications. The study area is Nevado Coropuna (6426 m), situated in Cordillera Ampato of Southern Peru. The glaciated area was 82.6 km 2 in 1962, based on aerial photography. We estimate the glacier area to be ca. 60.8 km 2 in 2000, based on analysis of the ASTER L1B scene. We used two 1:50,000 topographic maps constructed from 1955 aerial photography to create a digital elevation model with 30 m resolution, which we used as a reference dataset. Of the various interpolation techniques examined, the TOPOGRID algorithm was found to be superior to other techniques, and yielded a DEM with a vertical accuracy of ± 14.7 m. The 1955 DEM was compared to the SRTM DEM (2000) and ASTER DEM (2001) on a cell-by-cell basis. Steps included: validating the DEM's against field GPS survey points on rock areas; visualization techniques such as shaded relief and contour maps; quantifying errors (bias) in each DEM; correlating vertical differences between various DEM's with topographic characteristics (elevation, slope and aspect) and subtracting DEM elevations on a cell-by-cell basis. The RMS error of the SRTM DEM with respect to GPS points on non-glaciated areas was 23 m. The ASTER DEM had a RMS error of 61 m with respect to GPS points and displayed 200-300 m horizontal offsets and elevation 'spikes' on the glaciated area when compared to the DEM from topographic data. Cell-by-cell comparison of SRTM and ASTER-derived elevations with topographic data showed ablation at the toes of the glaciers (- 25 m to - 75 m surface lowering) and an apparent thickening at the summits. The mean altitude difference on glaciated area (SRTM minus topographic DEM) was - 5 m, pointing towards a lowering of the glacier surface during the period 1955-2000. Spurious values on the glacier surface in the ASTER DEM affected the analysis and thus prevented us from quantifying the glacier changes based on the ASTER data.