Application of DEM data to Landsat image classification: Evaluation in a tropical wet-dry landscape of Thailand

Integration of ancillary data in digital image classification has been show n to improve land-use/land-cover discrimination and classification accuracy . Studies demonstrating such techniques in the context of the tropical land scape are lacking. Cloud related problems limiting data acquisition during the crop season are frequently encountered in the humid tropics. Generally cloud-free data are obtained during the dry season. The reason for carrying out this study was to explore and evaluate the use of ancillary data, such as elevation, in Landsat thematic mapper classification to produce a land- use/land-cover map of the Sakae Krang watershed of Thailand. Altogether, 12 feature sets containing the TM original bands, ratios, normalized differen tial vegetation index, principal components, and a digital elevation model were tested using unsupervised ISODATA clustering. Incorporation of elevati on data was found to improve land-cover discrimination with relatively high er classification accuracy (77.5 percent) compared to TM data alone (65.3 p ercent). Further improvement in the classification accuracy (84.3 percent) was obtained when using elevation data under a supervised technique. The st udy also indicated that the classification results can be further improved by incorporating other geomorphometric variables, such as slope and soil mo isture regime.