Gridded datasets of precipitation are of great importance to evaluate recent climate models and are frequently applied to select a subset of available models. As climate models are still prone to biases on the regional scale, gridded datasets are also essential to correct or adjust these biases. Various studies revealed considerable differences, that is, observational uncertainty, in the available gridded datasets of precipitation, especially over complex terrain. This study focuses on the impacts of observational uncertainty on the evaluation, selection, and bias correction of 15 regional climate model (RCM) simulations provided through the EURO-CORDEX initiative over the alpine Adige catchment located in northern Italy. Nine reference datasets originating from observations, reanalysis, and remote sensing are applied to evaluate the performance of RCMs and select a subset based on validity. These reference datasets are then applied to bias correct the RCM ensemble using a standard quantile mapping method, and the resulting changes in the projections are assessed. The presented results show a selection of similar RCMs, indicating that observational uncertainty is lower than model uncertainty. The influence of the choice of the reference dataset on bias correction is negligible for the climate change signals. Small differences in projected change signals can be attributed to model selection. As expected, the choice of the reference dataset strongly influences future projections of precipitation even more pronounced for the extremes. The findings of this study highlight the need to account for observational uncertainty for bias correction of RCM simulations for impact modeling studies.
Ludwig Maximilians Univ Munchen, Dept Geog, Munich, Germany
Recommended Citation:
Gampe, David,Schmid, Josef,Ludwig, Ralf. Impact of Reference Dataset Selection on RCM Evaluation, Bias Correction, and Resulting Climate Change Signals of Precipitation[J]. JOURNAL OF HYDROMETEOROLOGY,2019-01-01,20(9):1813-1828