Assessing uncertainties associated with digital elevation models for object based landslide delination

Abstract

Digital elevation models (DEMs) are representations of topography with inherent errors that constitute uncertainty. DEMs data are often used in object based analyses without quantifying the effects of these errors. The main objective of this research is to establish a semi-automated object-based image analysis (OBIA) methodology for modelling uncertainty associated with DEMs when applied for locating landslides. In order to assess the uncertainty of DEMs, the Monte Carlo Simulation methodology was employed for evaluation of the effects of uncertainty on elevation and derived topographic parameters. The effect of DEM error is investigated, using stochastic conditional simulation to generate multiple equally likely representations of an actual terrain surface. Accordingly, distributional measures including accuracy surfaces, spatial autocorrelation indices, and variograms were also employed to quantify the magnitude and spatial pattern of the uncertainty. The semi-automated object based rule-sets for landslide delineation were developed based on two approaches a): without uncertainty assessing of DEMs and b): under applying uncertainty assessing of DEMs to examine the probable and possible uncertainties in delineating the landslides and measuring the improved accuracy after minimizing these associated uncertainties. The results of two approaches were validated using a landslide inventory database and very accurate GPS dataset. Results indicated very significant improvement in accuracy of results (> 28 %) when employing DEMs under uncertainty assessment. This research demonstrates how application of this methodology can address DEMs uncertainty, contributing to more responsible use of elevation and derived topographic parameters, and ultimately results obtained from their use