Many DEMs contain depressions, i.e. enclosed areas of lower elevation than the surrounding area. Regardless of whether these are real features of the landscape or whether they are spurious features arising from measurement error, they are usually considered undesirable elements in a DEM for many hydrological or geomorphological applications, because they hinder continuous surface flows. Consequently, many algorithms have been developed for eliminating DEM depressions, either by filling the low-lying area or by removing terrain that is obstructing the natural flow path. These techniques have the drawback of either artificially creating new elevation or artificially eliminating elevation from the DEM, i.e. the algorithms are not mass-conserving. Some combined algorithms have been developed that try to minimize these volume differences, but currently no algorithm achieves full mass-conservation. This paper presents a new algorithm of removing depressions. The algorithm is, by design, fully massconservative and essentially relies on a redistribution of the DEM elevations in the depressed area and its downstream outflow zone. The algorithm is intuitive and straightforward to implement. For the current research it has been coded as a stand-alone program, but it can easily be implemented in existing terrain analysis software or GIS packages. Tests on a range of DEMs (artificial, LiDAR, SRTM, Aster GDEM2) show that the algorithm works well and does exactly what it is intended to do: remove pits and depressions from a DEM, whilst maintaining the total mass of the terrain. The corrected DEMs also avoid the spurious parallel drainage that often occurs with other depression removal algorithms. Computationally, however, the mass-conserving algorithm is slower than some of the existing non-mass-conserving routines.
|Publication status||Published - 2013|
|Event||8th IAG International Conference on Geomorphology, Paris, France - |
Duration: 27 Aug 2013 → 31 Aug 2013
|Conference||8th IAG International Conference on Geomorphology, Paris, France|
|Period||27/08/13 → 31/08/13|