Use of preclassification image masking to improve the accuracy of wetland mapping undertaken in support of statewide land cover classification
The state of Wisconsin is engaged in a statewide project to map land cover using Thematic Mapper (TM) satellite imagery. This paper describes research conducted to develop and test the methodology proposed for the wetlands classification aspects of this project. Two areas with very different wetland flora were chosen as study sites. Wetland maps created from aerial photography were digitized and used to mask uplands from the satellite data to produce an image containing only wetlands. These data were classified using an unsupervised routine and accuracy was assessed utilizing aerial photography as reference data. Another classification and accuracy assessment was performed using TM imagery containing wetlands and uplands together, in order to quantify the improvement associated with pre-classification masking. This paper represents a progress report of results to date.