Effects of Various Factors on the Accuracy of DEMs: An Intensive Experimental Investigation

Abstract

A series of tests on the accuracy of DEMs is described. The effects of four factors on DEM accuracy have been tested, i.e., the accuracy, the density of source data, the characteristics of the terrain su$ace, and the modeling approaches. A large area covered by two 1 :I 0,000-scale maps was selected for testing. The terrain types range from flat to hilly to mountainous. Various sources (photographs and contour maps) were used and various measurement methods were employed. From test results, the following has been found: (1) source data measured manually with analytical plotters are the most reliable because automated measurement using image matching techniques could generate systematic errors; (2) the accuracy of DEMS decreases (or the RMsE becomes larger) with an increase in sampling interval, and the relationship is linear; (3) the inclusion of feature points and break lines improves the accuracy of DEMS significantly, especially when the terrain is rough; (4) generally speaking, the accuracy of DEMS decreases with an increase in relief;. however, this is not always the case, and the best results may be obtained in hilly areas; and (5) direct modeling from originally measured data to form a triangular network will yield better results than indirect modeling using a random-to-grid interpolation to form a grid network. The difference could be significant if the terrain is rough. From these conclusions, some advice on DEM production could be made as follows: (1) when using automated photogrammetric systems for data acquisition, editing by experienced operators should be considered; (2) in hilly areas, photogrammetric contouring can be the most efficient method for DEM data acquisition if analytical plotters are used; (3) feature points should always be measured and kept in order to reduce the data volume while retaining the fidelity of the DEM; and (4) when the terrain su$ace is rough, triangulationbased methods are recommended.