Geometric Accuracy Assessment of Classified Land Use/Land Cover Changes

European initiatives to harmonize geo-data and the emergence of object-based image analysis techniques in remote sensing lead to increased demands regarding the quality assessment of thematic classification results. While common metrics for the thematic accuracy assessment have already been established for decades, there is a deficit in generally accepted geometric accuracy metrics enabling the assessment of two-dimensional thematic objects. In this study, geometric accuracy metrics are introduced which base on differences in area and position between classified and reference objects. They are exemplarily calculated on classified and thematically verified agricultural fields in a German test site. We demonstrate how the metrics 1.) can be used for both the assessment of the total dataset and of single objects as well as 2.) can be coupled with thematic accuracy assessment results of a change detection analysis.