Impacts of climate change on extreme floods in Finland – studies using bias corrected Regional Climate Model data

Abstract. Bias correction of precipitation and temperature of five Regional Climate Models (RCMs) was carried out using Distribution Based Scaling (DBS) method with two versions for precipitation adjustment: single gamma and double gamma. This data were then used as input for a hydrological model to simulate changes in floods by the end of this century, and the results were compared to corresponding changes simulated using delta change approach. The results show that while the DBS adjustment significantly improves the RCM precipitations and temperatures compared to observations, especially the double gamma distribution does not always preserve trends of the uncorrected RCM data. The simulation of floods in the control period is improved by the DBS adjustment with no significant differences between single and double gamma. However, some scenarios are still unable to match the observed hydrology adequately due to remaining biases especially in near zero winter temperatures. These scenarios may produce an unrealistic climate change signal and should therefore be discarded from further use. A simple criterion for evaluating the adequate performance of the RCMs and hydrological models compared to observed floods is presented. The results of climate change simulations show that extreme summer precipitations increase more than average values in Finland. The changes in floods by 2070–2099 vary in different regions depending on season and the main flood producing mechanism (snowmelt or heavy rain). The changes in floods simulated with the DBS adjusted RCM data are mostly similar as with delta change approach, but the DBS method produces larger range of changes.