Scoping study for precipitation downscaling and bias-correction

Various methods exist for correcting biases in climate model precipitation data. This study has investigated four of these bias-correction methods, here called linear, non-linear, gamma and empirical, and extensively tested their performance and suitability for biascorrecting daily precipitation outputs from a Regional Climate Model (RCM) for use as inputs to hydrological models over six test regions spanning the Great Britain. The RCM daily precipitation data were taken from the unperturbed variant of the Met Office Hadley Centre Regional Model Perturbed Physics Ensemble (HadRM3-PPE-UK), and observed daily precipitation data were taken from the Continuous Estimation of River Flows gridded precipitation dataset. Spatial downscaling (re-gridding) and correction of the fraction of rain-days were undertaken as pre-processing steps before the bias-correction procedure, which translated the RCM data from a 0.22° grid sca le to the 1 km grid scale of the observed dataset. Re-sampling tests were used to assess the performance of the bias-correction methods in terms of the first four statistical moments, and cumulative distribution functions (cdfs) were produced to compare the distribution of the bias-corrected precipitation with respect to the observed and pre-processed RCM precipitation. We found that whilst the first and second moments of the precipitation frequency distribution can be corrected robustly, correction of the third and fourth moments of the distribution is much more sensitive to the choice of biascorrection procedure and to the selection of a particular calibration period. Overall, our results demonstrate that, if both precipitation datasets can be approximated by a gamma distribution, the gamma-based quantile-mapping technique offers the best combination of accuracy and robustness. In circumstances where precipitation datasets cannot adequately be approximated using a gamma distribution, the non-linear method is more effective at reducing the bias but the linear method is least sensitive to the choice of calibration period. The empirical quantile mapping method can be highly accurate, but results were very sensitive to the choice of calibration time period. Examination of the seasonal variation of the non-linear bias-correction factors showed that the bias-correction applied to the HadRM3 daily precipitation varied with season, location, topography and precipitation intensity, suggesting that the method is capable of reproducing many features of the complex spatial and temporal patterns of UK daily precipitation. Taking the known limitations into account this study concluded that the gamma-based quantile-mapping technique is the most suitable for bias-correcting daily HadRM3 precipitation for use in hydrological modelling in the UK