Rule-Based Landform Classification by Combining Multi-Spectral/Temporal Satellite Data and the SRTM DEM

  • Ho L. T. K.
  • Yamaguchi Y.
  • Umitsu M.


This study developed a rule-based method for generating a landform classification map of an alluvial plain for further assessment of flood susceptibility. Thresholds of the Modified Normalized Difference Water Index (MNDWI) and land cover characteristics were extracted from Landsat and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data. Local relief, average elevation, and channel features were calculated using a Shuttle Radar Topographic Mission Digital Elevation Model (SRTM DEM). Then, relative position indices of polygons were used to classify small-scale landform objects. These inputs were combined based on a rule. The landform classification map by this method was compared to a manual map by visual interpretation. Significant consistence of dominated landform categories between the two maps demonstrates the effectiveness of the rule-base method despite the limitation on spatial resolution of medium-resolution data for detecting some small features. Moreover, the remarkable merits of the rule-based method in comparison with the manual method are its relative time savings, objectivity, and ease of editing.