Multi-variable error correction of regional climate models

Climate change impact research needs regional climate scenarios of multiple meteorological variables. Those variables are available from regional climate models (RCMs), but affected by considerable biases. We evaluate the application of an empirical-statistical error correction method, quantile mapping (QM), for a small ensemble of RCMs and six meteorological variables. Annual and monthly biases are reduced to close to zero by QM for all variables in most cases. Exceptions are found, if non-stationarity of the model's error characteristics occur. Even in the worst cases of non-stationarity, QM clearly improves the biases of raw RCMs. In addition, QM successfully adjusts the distributions of the analysed variables. To approach the question whether time series and inter-variable relationships are still plausible after correction, we evaluate the root-mean-square error (RMSE), autocorrelation and inter-variable correlation. We found improvement or no clear effect in RMSE and autocorrelation, and no clear effect on the correlation between meteorological variables. These results demonstrate that QM retains the quality of the temporal structure in time series and the inter-variable dependencies of RCMs. It has to be emphasised that this cannot be interpreted as an improvement and that deficiencies of the RCMs in those features are retained as well. Our results give some indication for the performance of QM applied to future scenarios, since our evaluation relies on independent calibration and evaluation periods, which are affected by climate variability and change. The effect of non-stationarity, however, can be expected to be larger in far future. We demonstrate the retainment of the RCM's temporal structure and inter-variable dependencies, and large improvements in biases. This qualifies QM as a valuable, though not perfect, method in the interface between climate models and climate change impact research. Nonetheless, in case of no correlation between re-analysis driven RCM and observation, one should consider that QM does not correct this correlation.