Large scale forest classification for topographic models

Task of this investigation is to investigate in how far high resolution satellite data can be utilized to support the establishment of the Austrian Topographic Model. The model aims at representing the surface of the entire Austria un-generalised in vector format. The investigation concentrates on the classification of the “object type” forest. For this purpose KFA-1000 data acquired during the AUSTROMIR mission were used. The test site covers an alpine area and thus is characterised by great height differences, steep slopes and heterogenous forest cover. The results of the investigation can be summerized as follows. The comparison of the precisely rectified KFA-image with orthophotos and topographical maps showed that geometrical accuracy of the satellite data meets the requirements of the Topographic Model With the exeption of shadowed regions the classification into the different forest types showed sufficient quality, too. Because of the very heterogenous spatial distribution of the forest classes the classification result had to be pre-processed before vectorisation.