Mapping wetlands using multi-temporal RADARSAT-1 data and a decision-based classifier

Abstract

The purpose of this study was to determine the suitability of multi-temporal RADARSAT-1 data and a decision classifier for mapping the Lower Islands of the Paraná Delta wetland in Argentina. The information-extraction strategy was based on identification of the interaction mechanisms occurring between the radar signal and the canopy, considering different vegetation phenology and flood conditions. Such information was used in the design of a decision classifier to obtain a land cover map. In addition, results were compared with those obtained from an iterative optimization clustering procedure (ISODATA algorithm). The quality of the maps obtained was assessed by an error matrix evaluation. The decision classifier was able to discriminate among land cover types with an overall accuracy on the order of 85%, and ISODATA had an overall accuracy of 81%. Two of the available scenes were taken during the high flood of the El Niño event of 1998, and the results obtained also show that RADARSAT-1 data are quite effective not only in delineating the inundation area, but also in identifying the flood condition of each of the land cover types considered. Two main conclusions are derived from this research: (1) the need for multi-temporal SAR data acquired under different environmental conditions for mapping wetlands, and (2) the advantages and flexibility of physically based reasoning classifiers for synthetic aperture radar (SAR) data classification.