Land cover discrimination in a wetland environment using TopSAR data

Abstract

A long-term monitoring system based on remote sensing techniques is currently being developed for the Mary River Catchment in the Northern Territory, Australia. Two of the major environmental issues in this area are saltwater intrusion and weed infestation. An intensive ground based monitoring program provides in situ data for sites in a variety of wetland habitats. TopSAR data have been acquired for the catchment during the PACRIM 2000 campaign to investigate the potential of SAR data for the long term monitoring of this environment. In this paper, the results of a preliminary investigation of the use of TopSAR data in a wetland environment are presented. A land cover classification was derived using a previously developed methodology. This involves the correction of TopSAR data for variation in incidence angle and a maximum likelihood classification based on training areas. The separability of the major land cover classes in this environment, especially weed species was assessed against available ground data with encouraging results. Additionally, the digital elevation data derived from the C-band interferometry are evaluated for their relative accuracy in this environment in comparison to survey data.