A spatiotemporal analysis of Germany's largest urban agglomeration

摘要

This paper deals with the Ruhr area, Germany's largest urban agglomeration. Since the late 1960s, this area is characterised by urban shrinkage including an increase of brownfields and a decrease of population. The aim of this study is the creation of a bi-temporal land cover database which is planned to be used for the investigation of the urban landscape's transformation. For this purpose, Landsat 5 (TM) images from 1987 and 2009 are used. In addition, thresholding of band derivatives as well as decision tree prediction (C4.5 algorithm) on segmentation level are applied. Using image-to-image comparison, the derivatives are also applied for the creation of a change/no-change mask which helps to specialise the C4.5 algorithm on the change areas. Furthermore, the mask is used to spatially restrict a post-classification map comparison and, thus, solves the problem of error propagation, which is inherent to the post-classification approach. Finally, we present an example application of the database by integrating the Ruhr areas' population counts referring to its 15 districts. These districts show different characteristics and dynamics, whereas the overall tendency underlines the shrinking processes accompanied by an increasing per-person consumption of built-up land.