Geoscience and Environment

Using remotely sensed optical, thermal, and radar data for land-use and land-cover classification

Project for the Graduate Certificate in Geo-Spatial Analysis, Department of Earth Science, Emporia Sate University, Emporia, Kansas (December 1, 2003)

Abstract

ASTER data were used to make a provisional map showing land use and land cover for a 120 km2 site in the state of Kedah, Malaysia. The identification of classes was derived from images based on the visible, near-infrared, mid-infrared, and thermal infrared parts of the spectrum. The result was compared with a similar map based on Landsat-ETM+ data. Finally, AIRSAR radar data were used in the field to refine the classification. This study supports research carried out by others suggesting that land-use and land-cover classification may be improved by using radar data as a supplement to optical and thermal data, especially where seasonal cloud cover degrades optical datasets.
The land use and land cover classes identified in this study were: rice paddy (72.2%), roads, villages, bare soil (8.7%), bush and tree-crops (9.7%), open forest (5.4%), dense forest (4.0%). Radar datasets may require special software to access phase relationships and to analyze texture. Projects that rely on fieldwork for verification of remotely sensed data need to provide adequate resources for geo-registration, even where datasets have been previously geo-registered.

(Click image to download report as a PDF file, 3.6 megabytes.)

Return to Menu

The URL of this site is [http://www.geoscience-environment.com]