Status of land cover classification accuracy assessment

Abstract

The production of thematic maps, such as those depicting land cover,using an image classification is one of the most commonapplications of remote sensing. Considerable research has been directedat the various components of the mapping process, including theassessment of accuracy. This paper briefly reviews the backgroundand methods of classification accuracy assessment that are commonlyused and recommended in the research literature. It is, however, evidentthat the research community does not universally adopt theapproaches that are often recommended to it, perhaps a reflectionof the problems associated with accuracy assessment, and typicallyfails toachieve the accuracy targets commonly specified. The community oftentends to use, unquestioningly, techniques based on the confusionmatrix for which the correct application and interpretation requiresthe satisfaction of often untenable assumptions (e.g., perfect coregistrationof data sets) and the provision of rarely conveyed information (e.g.,sampling design for ground data acquisition). Eight broad problemareasthat currently limit the ability to appropriately assess, document,and use the accuracy of thematic maps derived from remote sensingareexplored. The implications of these problems are that it is unlikelythat a single standardized method of accuracy assessment and reportingcan be identified, but some possible directions for future researchthat may facilitate accuracy assessment are highlighted.