An object-based method to map wetland using RADARSAT-1 and Landsat ETM images: test case on two sites in Quebec, Canada

The Canadian Wildlife Service, Quebec region, of Environment Canada tested a multiscale object-based classification method on two test sites using satellite images to map wetlands in the context of the Canadian Wetland Inventory (CWI). The objective of this study was to assess the method most adapted for the Canadian inventory program to map five wetland classes (bog, fen, swamp, marsh, and shallow water), for a minimal geographical unit of 1 ha, from RADARSAT-1 and Landsat-7 enhanced thematic mapper (ETM) data. The top-down object-based classification selected was based on the Canadian Wetland Classification System and identifies quickly and precisely the ecologically meaningful polygons of wetlands. Validation was done on two levels: (i) between “other” versus “wetland” and (ii) between each class of wetland. Global accuracy values for the first level are greater than 80% for both test sites and about 76% and 67%, respectively, for the two sites for the second level. This approach is well adapted to wetland mapping on both the thematic and the spatial level.