This study examined the classification accuracy of specific land use/cover categories from original and derived spaceborne radar measures for a site in Nepal. Six RADARSAT images were compiled and registered to evaluate the effects of incident angle and seasonality on feature delineations. Examination of an initial RADARSAT image for Kathmandu provided poor classification results (56 percent overall accuracy) for the four generalized land covers of urban, residential, agriculture, and grass. Measures of Variance texture were applied to this original RADARSAT data over varied window sizes, improving overall classification accuracy to 75 percent. Dry-season results were generally better than wet-season along with a pattern of improved separability as the incident angle increased from averages of ca. 23.5° to 39.5 to 47°.
No takes yet. Share an insight, caveat, or question.
Barry Haack (2007) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: