Characterizing precipitation events leading to surface water flood damage over large regions of complex terrain

Abstract

Surface water floods (SWFs) that lead to household losses are mainly localized phenomena. Research on describing the associated precipitation characteristics has previously been based on case studies and on the derivation of local rainfall thresholds, but no approaches have yet been presented on the national scale. Here, we propose a new way to overcome this scaling problem. We linked a gridded precipitation dataset based on both rainfall gauges and radar data with geolocated insurance claims for all of Switzerland. We show that the absolute thresholds vary markedly over complex terrain, and we thus propose basing early warning systems for predicting damage-relevant SWF events on local quantiles of maximum intensity and the total sum of event precipitation. A threshold model based on these two parameters is able to classify rainfall events potentially leading to damage-relevant SWF events over large areas of complex terrain, including high mountains and lowland areas, and a variety of geological conditions. Our approach is an important step towards the development of impact-based early warning systems. Weather warning agencies or insurance companies can build upon these findings to find workarounds for issuing user-targeted warnings at national scale or for nowcasting purposes.