Digital Elevation Models (DEMs) provide a model of the continuous representation of the earth¿s elevation surface. This form of spatial data contains errors from the true elevation values. The nature and extent of these errors are often unknown and not readily available to users of spatial data. Topographic attributes are frequently derived directly from DEMs and DEM errors propagate to derived parameters. Normally, DEM accuracy is quantified using the Root Mean Square Error (RMSE) that assumes DEM errors are random. However assessment of DEM uncertainty requires more information on the spatial structure of the errors. The aim of this study was to provide an example of simple indicators of the performance of a given dataset for hydrological networks derivation and to model the errors propagation in such process. Three DEMs of different origin and cell resolution were compared in an area of about 120 km2 in the Rhein basin. SRTM dataset with resolution of 100m, DEM dataset mosaic from various sources with a resolution of 60m and ASTER derived dataset with a resolution of 30 m were used. All datasets were resampled to a common cell-size of 30m. The sinks were removed prior to the derivation of hydrological parameters. The river network was then extracted for different flow accumulation thresholds. The Strahler order and the contributing area network were calculated. Significant differences were found among the considered DEMs, showing the impact of the original resolution. The DEM with lower original resolution extracted river network with lower Strahler order. The DEM with the highest original resolution showed a higher contributing area for lower Strahler order. The propagation of errors was estimated using a Monte Carlo simulation including the spatial variability of DEM errors.

POGGIO Laura;  SOILLE Pierre; 

2009-02-05

University of Natural Resources and Applied Life Sciences (BOKU)

JRC48774

http://www.ecsss.net/web/MDB/media_folder/39_abstracts.pdf,    https://publications.jrc.ec.europa.eu/repository/handle/JRC48774,