Urban structuring using multisensoral remote sensing data: By the example of the German cities Cologne and Dresden

Abstract

The urban landscape is a highly complex and small-structured, heterogeneous area as a result of permanent human settlement. Urban structure is scale-dependant and can be expressed on various levels of detail by satellite imagery. Very high resolution satellite (VHR) sensors are capable of mapping and monitoring cities - on house/block level - with their high degree of landcover diversity. However, detection of morphological features such as shape and elevation of single objects is performed much better when a digital surface model (DSM) - e.g. derived by airborne laserscanning - is incorporated. An object-oriented methodology for the joint analysis of optical satellite data and a digital surface model is presented for the classification of the urban morphology in terms of urban structural types. These are spatial units - mostly on block level - with aggregated information on the classified single features like landcover/landuse and urban fabric. Hence, a hierarchical, modular segmentation and classification workflow is implemented to extract the required information. The methodology is applied on two study areas in the cities of Cologne and Dresden, Germany, and a validation of the capability of the potential for transferability of the rulebase is shown.