Detecting Landforms Using Quantitative Radar Roughness Characterization and Spectral Mixing Analysis

In Geomorphology the landforms delineation and delimitation are based on traditional techniques which is usually achieved by two different approaches: semantic and geometric. The semantic approach involves interpreting aerial photographs as well as using field knowledge to map the possible landforms extent. The geometric approach involves inferring the probability of extension on related properties through observations of topographic attributes like slope, elevation and curvature. This report, based on a similarity geometric model, uses quantitative roughness characterization, Spectral Mixing Analysis and fuzzy logic to map alluvial fans. The research is applied in the Alashan region in China because the timing of alluvial deposition is tied to land surface instabilities caused by regional climate changes. The main aim of the research is to understand where they form and where they extent in an effort to develop a new approach using the physical response within the SMA, the backscatter roughness parameters and primary attributes (elevation and curvature) derived from the Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM). So that, this study helps to provide a benchmark against which future alluvial fans detection using roughness, SMA and fuzzy logic analysis can be evaluated, meaning that sophisticated coupling of geomorphic and remote sensing processes can be attempted in order to test for feedbacks between geomorphic processes and topography.