from a user’s.pdf

A tool for assessing error in digital elevation models from a user’s perspective 1 2 2 A. R. Darnell , N. J. Tate , and C. Brunsdon 1School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ. Tel: +44 (0)1603 591346, Fax: +44 (0)1603 458553 Email: a.darnell@uea.ac.uk 2Dept of Geography, University of Leicester, University Rd, Leicester, LE1 7RH 1. Introduction A Digital Elevation Model (DEM) is a representation of geographic reality. The elevations recorded within DEMs have been shown to contain errors pertaining to sampling, measurement and interpolation (Fisher, 1998). Even a small amount of elevation error can greatly affect derivative products (Holmes et al., 2000). This can potentially have a significant impact on the application of DEMs in Geographical Information Systems (GIS) where first and second order derivatives are considered. DEM vendors generally provide users with a measure of vertical accuracy in the form of the Root Mean Squared Error (RMSE) statistic only. However many papers have reported on the limitations of a single value of accuracy, stressing that DEM error is spatially autocorrelated (e.g. Kyriakidis et al., 1999). Arguably, the best method for error modelling is based on conditional stochastic simulation (Fisher, 1998). Conditioning the simulation model includes sample observations of error and thus allows consideration of spatial autocorrelation. Unfortunately computation is complex. The main aim of this research was to simplify existing procedures to enable the ‘average’ DEM user to perform his/ her assessment on the implic