Suitability of DEMs derived from SAR interferometry and ASTER stereoscopy for hydrological applications using GIS. A case study of Al-Jafer basin, Jordan

Digital elevation models (DEMs) provide an essential quantitative environmental variable for a vast amount of the research published in remote sensing, GIS and physical geography. Traditionally, DEMs have been derived from ground surveys and photogrammetric techniques, and from topographic maps using contour data and spot heights. Satellite remote sensing now provides the most accurate digital elevation datasets with worldwide coverage by means of optical, radar or laser sensors. The aim of this study is to evaluate the accuracy and reliability of DEMs generated from InSAR (ERS-1/2) and ASTER data over a sparsely vegetated drainage system in central Jordan. DEMs of the study area were generated by each of these systems and an accuracy assessment was carried out through verification of the DEM based on the characteristics of the terrain against a number of independent check points collected using differential GPS data and references to generate a DEM from a topographic map (scale 1:50,000). The accuracy of independent check points used in this study is less than 1 m, which is more accurate than remote sensing techniques. The accuracy of the DEMs derived from InSAR and ASTER are represented by their RMSE, which were ± 6.95 m and 13.28 m respectively. The increase in errors in high elevation areas for ASTER DEM was higher than in InSAR DEM. The effect of these errors is investigated using stochastic conditional simulation to generate multiple equal likelihood representations of an actual terrain surface. The propagation of data uncertainty to the elevation derivatives, and the impact on the extracted surface flow are assessed. The results suggest that an elevation error propagates to flow accumulation especially in low slope areas. The effects of DEM resolution on a set of topographic parameters, including slope, accumulated flow area, flow length and catchment area, are examined.