Quantifying Uncertainty of Digital Elevation Models Derived from Topographic Maps

This paper explores a methodology for quantifying the uncertainty of DEMs created by digitising topographic maps. The origins of uncertainty in DEM production were identified and examined. The uncertainty of DEM data was quantified by computing a vector total of Root Mean Square Error (RMSE) from the source map, sampling and measurement errors, and the interpolation process. Distributional measures including accuracy surfaces, spatial autocorrelation indices, and variograms were also employed to quantify the magnitude and spatial pattern of the uncertainty. The test for this methodology utilises a portion of a 1:24 000 topographic map centred on Stone Mountain in northeastern Georgia, USA. Five DEMs, constructed with different interpolation algorithms, are found to have the total RMSE ranging from 4.39 to 9.82 meters, and a highly concentrated pattern of uncertainty in rugged terrain. This study suggests that the RMSE provides only a general indicator of DEM uncertainty. Detailed studies should use distributional measures to understand how the uncertainty varies over a surface.