Knowledge-based land-cover classification using ERS-1/JERS-1 composites

Land-cover classification based on an ERS-1/JERS-1 composite is explored in the context of eventual world-wide applicability. Since each of these orbiting sensors collect data using different frequencies, polarizations, and look angles, a classification procedure based on data from both sensors should enable a better classification than can either alone, and on a world-wide basis. A simple conceptual model is presented to show how this data relates to the structural features of terrain and vegetation cover. The classification scheme demonstrates an overall accuracy in excess of 90% for five different classes