SPATIAL MODELING AND UNCERTAINTY ASSESSMENT OF FINE SCALE SURFACE PROCESSES BASED ON COARSE TERRAIN ELEVATION DATA

Frequent acquisition of high-resolution digital elevation models (HR-DEMs) over large areas is expensive and difficult. Satellite-derived low-resolution digital elevation models (LR-DEMs) provide extensive coverage of Earth’s surface but at coarser spatial and temporal resolutions. Although useful for large scale problems, LR-DEMs are not suitable for modeling hydrologic and geomorphic processes at scales smaller than their spatial resolution. In this work, we present a multiple-point geostatistical approach for downscaling a target LR-DEM based on available high-resolution training data and recurrent high-resolution remote sensing images. The method aims at generating several equiprobable HR-DEMs conditioned to a given target LR-DEM by borrowing small scale topographic patterns from an analogue containing data at both coarse and fine scales. An application of the methodology is demonstrated by using an ensemble of simulated HR-DEMs as input to a flow-routing algorithm. The proposed framework enables a probabilistic assessment of the spatial structures generated by natural phenomena operating at scales finer than the available terrain elevation measurements. A case study in the Swiss Alps is provided to illustrate the methodology.