Bias correction of ENSEMBLES precipitation data with focus on the effect of the length of the calibration period

Abstract

Bias correction (BC) has become a standard procedure in climate change impact studies, since climate model output often shows a bias when compared to observed data. Especially for daily precipitation, we expect the performance of the BC to depend on the length of the period used for the BC calibration. In this study we analyzed how the length of the calibration period affects the BC performance of quantile mapping (QM). We subsequently reduced the length of the calibration period, starting with a calibration period length of 30 years, and analyzed the effect on the BC performance based on three skill scores.The results show that already a small reduction in the length of the calibration period can result in a significant decrease of the BC performance. However, the critical calibration period length at which this decrease occurs, varies strongly. Nevertheless, it is larger than ten years in all experiments for all skill scores. Furthermore, the critical calibration period length is found to depend on the choice of the control period and especially on the choice of the QM method. But it has to be noted that these results are slightly different for the three skill scores. Overall, the results indicate that QM methods with many degrees of freedom, especially the empirical QM, are more vulnerable to a reduction of the calibration period length. Based on our results, we recommend to use a calibration period as long as possible and to apply QM methods with few degrees of freedom, when using QM for the BC of data that was not used in the calibration.

Keywords

• bias correction
• bias adjustment
• quantile mapping
• ENSEMBLES
• precipitation
• regional climate model
• RCM