Wavelet-Based Digital Elevation Model Analysis

Estimations of the local roughness of Digital Elevation Models (DEMs) have been used both for a better geological understanding of terrain structures and for a more accurate DEM interpolation leading to higher resolution terrain maps. This work investigates two fractal dimension estimators for the characterization of the roughness of DEMs: the power spectrum estimator and the wavelet-based estimator. The performance of these methods is compared in terms of image segmentation capabilities. Since wavelets and fractals are related by the same multiresolution concept, we expect to have better results using the wavelet analysis method. This expectation is confirmed by experiments on synthetic images and real DEMs: measurements of fractal parameters using wavelet-based methods are more reliable than the same measurements performed using other methods. Finally we discuss some examples in which the fractal analysis of DEMs allows the separation of different roughness classes and reveals artifacts in the computation of elevation data.