Area Errors Between Grid Imagery Boundaries and Vector Actual Boundaries Identifying Waterbodies from Remote Sensing Imagery

In identifying water bodies from remote sensing imagery, a mismatch between grid data boundary and vector boundary has always existed but was seldom studied. Therefore, area errors between grid imagery boundaries and vector real boundaries are the subject of this study. A solution based on the sub-pixel classification method was developed to analyse these errors. A case study from Lake Manasarowar in China showed that the area error proposed in this study is larger than that from different interpretation methods. It was concluded that uncertainties from mixed boundary pixels were greater than that from different methods for identifying lake area using the remote sensing imagery in the study area. Overall, area error analyses for grid imagery boundaries and vector real boundaries are necessary for identifying water bodies from remote sensing imagery. It is also useful for the interpretation of other continuous bodies, such as glaciers.