Semantic analysis of space imagery for mapping purposes

Abstract

From the area around Zonguldak several space images with different ground sampling distance (GSD) from 30m down to 62cm are available. The semantic information of Landsat images is too poor for the generation of topographic maps, while QuickBird images do allow a mapping in the scale 1:5000. Not only the GSD is determining the possible map scale, also the spectral and radiometric resolution is important. The imaging conditions, especially the sun elevation with the corresponding length of shadows, do lead to quite different results especially in cities with narrow streets. The possibilities and limitations of mapping with the different space images, partially with the same image type taken in different month, has been analysed.