Disentangling drivers of historical river flood losses

Recent studies of past changes in precipitation patterns suggest regionally varying but clearly detectable trends of global warming on physical flood indicators such as river discharge. Whether these trends are also visible in economic flood losses, has not yet been clearly answered, as changes in trends of damage records may be induced by either climatic or socio-economic drivers. In general, the socioeconomic impact of an extreme weather event is composed of three components: The hazard, the exposure of socioeconomic values to the event, and the vulnerabilities of the values, i.e., their propensity or predisposition to be adversely affected. In this work, we separate the historically observed trends in economic losses from river floods into the three contributions. We then quantify the effect of each driver on the overall change in economic losses from river floods between 1980 - 2010 for different world regions. In particular, this allows us to determine in which regions anthropogenic warming has already contributed to the observed trends in damages. We use flood depth as biophysical hazard indicator calculated by combining discharge simulations from 12 global hydrological models of the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) model ensemble with the river-routing and flood inundation model CaMa-Flood. The hydrological models are driven by observed weather data and our simulations account for present-day protection standards from the FLOPROS database. Asset losses are estimated by combining gridded asset data with state-of-the-art flood damage functions translating flood depth to the fraction of affected assets employing the open-source socioeconomic impact modelling framework CLIMADA. Trends in modeled historical flood damages are then compared to observational damages by Munich Re’s NATCATService database in order to explain residual differences in trends by the three types of drivers. 

We first show that the method permits to reproduce the year-to-year variability observed damages on the regional level. We identify changes in exposure as the main driver of rising damage trends, but also observe significant - rising as well as declining - trends in flood hazards in several regions. Thus, effects of anthropogenic climate change that have already shown to unfold in discharge patterns, partly manifest already in economic damages, too. Residual trends in observed losses, that cannot be explained by changes in the hazard and the exposure alone, are caused by changes in vulnerability that can be well explained with trends in GDP per capita.  Mostly, rising regional income results in declining vulnerability to river floods, in particular in less developed world regions. However, we also find indications of maladaptation, i.e., in some regions, vulnerability increases with GDP per capita.