A new method to quantify the accuracy of classification and spatial delineation in land cover maps

Abstract

Land cover maps provide spatial information on the physical cover of the earth. With increasing strain on natural resources, these maps enable knowledge-based nature management. Such maps represent a generalization of nature, and their accuracy is often unknown. There are two main sources that generate inaccuracies in land cover mapping. Classification inaccuracies and spatial delineations inaccuracies. In classification inaccuracies, mappers delineate roughly the same, but assign different ecosystem units. In spatial delineation inaccuracies, mappers assign the same ecosystem unit, but delineate the polygon borders different. The main objective of this thesis is to investigate these main sources of inaccuracies separately, further focusing on ecosystem unit characteristics and biome complexity. To investigate biome complexity, four different biomes were mapped and compared.

Land cover maps provide spatial information on the physical cover of the earth. With increasing strain on natural resources, these maps enable knowledge-based nature management. Such maps represent a generalization of nature, and their accuracy is often unknown. There are two main sources that generate inaccuracies in land cover mapping. Classification inaccuracies and spatial delineations inaccuracies. In classification inaccuracies, mappers delineate roughly the same, but assign different ecosystem units. In spatial delineation inaccuracies, mappers assign the same ecosystem unit, but delineate the polygon borders different. The main objective of this thesis is to investigate these main sources of inaccuracies separately, further focusing on ecosystem unit characteristics and biome complexity. To investigate biome complexity, four different biomes were mapped and compared.