Premium
Removal of artifact depressions from digital elevation models: towards a minimum impact approach
Author(s) -
Lindsay John B.,
Creed Irena F.
Publication year - 2005
Publication title -
hydrological processes
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.222
H-Index - 161
eISSN - 1099-1085
pISSN - 0885-6087
DOI - 10.1002/hyp.5835
Subject(s) - digital elevation model , terrain , elevation (ballistics) , artifact (error) , geology , hydrology (agriculture) , drainage , environmental science , cartography , remote sensing , computer science , mathematics , artificial intelligence , geotechnical engineering , geography , geometry , ecology , biology
Artifact depressions in digital elevation models (DEMs) interrupt flow paths and alter drainage directions. Techniques for removing depressions should enforce continuous flow paths in a way that requires the least modification of the DEM. Impacts on the spatial and statistical distributions of elevation and its derivatives were assessed for four methods of removing depressions: (1) filling; (2) breaching; (3) a combination of filling and breaching, with breaching constrained to a maximum of two grid cells; (4) a combination of filling and breaching based on an impact reduction approach (IRA). The IRA removes each depression using either filling or breaching, depending on which method has the least impact, in terms of the number of modified cells and the mean absolute difference in the DEM. Analysis of a LiDAR DEM of a landscape on the Canadian Shield showed significant differences in the impacts among the four depression removal methods. Depression filling, a removal method that is widely implemented in geographical information system software, was found to impact terrain attributes most severely. Constrained breaching, which relies heavily on filling for larger depressions, also performed poorly. Both depression breaching and the IRA impacted spatial and statistical distributions of terrain attributes less than depression filling and constrained breaching. The most sensitive landscapes to depression removal were those that contained large (i.e. >10%) flat areas, because of the occurrence of relatively large depressions in these areas. Copyright © 2005 John Wiley & Sons, Ltd.