The impact of bias correcting regional climate model results on hydrological indicators for Bavarian catchments

• Published in: Journal of hydrology. Regional studies 2018-10, Vol.19, p.25-41
• Main Authors: Willkofer, Florian, Schmid, Franz-Josef, Komischke, Holger, Korck, Jane, Braun, Marco, Ludwig, Ralf
• Format: Article
• Language: English
• Subjects: Bavaria; Bias correction; Climate change signal; Hydrological modeling; Regional climate model; Runoff indicators
• ISSN: 2214-5818; 2214-5818
• DOI: 10.1016/j.ejrh.2018.06.010

•Ensemble of three different regional climate models applied to drive a process based, deterministic hydrological model (WaSiM).•Application of three bias correction methods (linear and local intensity scaling, quantile mapping – monthly and yearly correction factors).•Correction of essential meteorological parameters required for the hydrological model setup.•Bias correction has considerable impact on the climate change signals of specific runoff indicators (e.g. high flows). The Mindel river catchment, gauge Offingen, Bavaria, Germany. The study investigates the potential interference of climate change signals (CCS) in hydrological indicators due to the application of bias correction (BC) of regional climate models (RCM). A validated setup of the hydrological model WaSiM was used for runoff modeling. The CCS, gained by the application of three RCMs (CCLM, REMO-UBA, RACMO2) for a reference period (1971–2000) and a scenario period (2021–2050), are evaluated according to eight hydrological indicators derived from modeled runoff. Three different BC techniques (linear scaling, quantile mapping, local intensity scaling) are applied. New hydrological insights for the region: Runoff indicators are calculated for the investigated catchment using bias corrected RCM data. The quantile mapping approach proves superior to linear scaling and local intensity scaling and is recommended as the bias correction method of choice when assessing climate change impacts on catchment hydrology. Extreme flow indicators (high flows), however, are poorly represented by any bias corrected model results, as current approaches fail to properly capture extreme value statistics. The CCS of mean hydrological indicator values (e.g. mean flow) is well preserved by almost every BC technique. For extreme indicator values (e.g. high flows), the CCS shows distinct differences between the original RCM and BC data.