Downscaling regional climate model estimates of daily precipitation, temperature and solar radiation data

The smallest spatial scale of representation by regional climate models (RCMs)-i.e. 50 x 50 km-is greater than that at which site-specific studies on climate change impacts, mitigation and adaptation studies are conducted. An approach is therefore needed to evaluate the quality of data from RCMs used for these purposes, to identify systematic errors and adjust future projected estimates accordingly. The present study uses a simple downscaling approach for recalibrating RCM estimates of precipitation, maximum and minimum air temperature (Tmax and Train), and solar radiation. We compared the Hadley Centre HadRM3 RCM-hindcast estimates for 1960 to 1990 with observed data from 15 meteorological stations in the UK. Downscaling factors (DFs) were applied to improve the match between hindcast and observed data. The DFs were then applied to the RCM data for the A2 2070 to 2100 scenario, assuming that the systematic deviations present in the hindcast estimates will persist. The hindcast RCM data included a considerable excess of small (< 0.3 mm) precipitation events, whilst significantly overestimating the mean annual total at some sites and underestimating it at others. Estimates of Tmax were closer than for Tmin, which the model tended to overestimate by an average of VC. Estimates of lower Tmax and upper Train values were generally good, but the model was less effective in representing extreme warm and cold conditions. The model systematically overestimated solar radiation. Because the use of DF substantially improves the fit of hindcast estimates with observed data, their use with RCM projections should considerably increase confidence in model outputs for studies on impacts, mitigation and adaptation.