The role of disaggregation of asset values in flood loss estimation: a comparison of different modeling approaches at the Mulde River, Germany.

In loss estimation there is a spatial mismatch of hazard data that are commonly modeled on an explicit raster level and exposure data that are often available only for aggregated administrative units. Usually disaggregation methods that use ancillary information to distribute lumped exposure data in a finer spatial resolution help to bridge this gap. However, the actual influence of different mapping techniques and ancillary data on the final loss estimation has not been analyzed yet. In this paper three methods are applied to disaggregate residential building assets using two kinds of land use/land cover (LULC) data. The resulting disaggregated assets are validated and compared using census data of the residential building number on the community and constituency level. In addition, the disaggregated assets are taken to estimate residential building losses due to the flood in August 2002 in 21 municipalities on the River Mulde in Saxony, Germany. Losses are calculated with the help of four loss models. In general, disaggregation helps to decrease the error variance within the loss estimation. It must, however, be stated that the application of sophisticated disaggregation methods does not lead to significant improvements compared to the straightforward binary method. Therefore more effort should instead be put into the provision of high-resolution LULC data. Finally, the remaining uncertainties in loss estimation are high and demand further improvements in all modeling aspects.