Derivation of population distribution for vulnerability assessment in flood-prone German cities using multisensoral remote sensing data

Abstract

Against the background of massive urban development, area-wide and up-to-date spatial information is in demand. However, for many reasons this detailed information on the entire urban area is often not available or just not valid anymore. In the event of a natural hazard – e.g. a river flood – it is a crucial piece of information for relief units to have knowledge about the quantity and the distribution of the affected population. In this paper we demonstrate the abilities of remotely sensed data towards vulnerability assessment or disaster management in case of such an event. By means of very high resolution optical satellite imagery and surface information derived by airborne laser scanning, we generate a precise, three-dimensional representation of the landcover and the urban morphology. An automatic, object-oriented approach detects single buildings and derives morphological information – e.g. building size, height and shape – for a further classification of each building into various building types. Subsequently, a top-down approach is applied to distribute the total population of the city or the district on each individual building. In combination with information of potentially affected areas, the methodology is applied on two German cities to estimate potentially affected population with a high level of accuracy