Projecting Regional Climate Change through Statistical and Dynamical Downscaling Techniques

Abstract

Climate change has comprehensive and profound influences on every aspect of the natural and human systems such as the hydrological systems, ecosystems, food systems, infrastructure, human health and livelihoods. Not only are the impacts diverse in expressions, they are also geographically heterogeneous. In order to manage and reduce the risk of climate change impacts with the consideration of its heterogeneity, region-specific mitigation and adaptation strategies needs to be taken. The foundation of regional climate impact studies and the subsequent design of regional mitigation and adaption strategies is the projection of future regional or local climate. To this end, this study focuses on exploring how the regional or local climate is affected in the context of global warming, using both statistical and dynamical downscaling techniques. Specifically, (1) the stepwise cluster analysis (SCA) is used for downscaling of the local climate of the City of Toronto, and (2) the RegCM is used for the dynamical downscaling of the regional climate over China. From the statistical downscaling study of the temperature of the City of Toronto, the capability of the SCA for capturing the relationship between the global atmospheric variables and the local surface variables is demonstrated. Geophysical Fluid Dynamics Laboratory (GFDL) data of the historical period, representative concentration pathways (RCP4.5 and RCP8.5) is used for the projection of the future local climate. The results show that the future daily maximum, mean and minimum temperature (Tmax, Tmean, and Tmin) of the City of Toronto is likely to increase, with the speed of increase higher under RCP8.5 than under RCP4.5. The Tmin is projected to have a slightly larger increase than the Tmax. Significant increasing trend can be found under both scenarios for the entire 80- year future periods for Tmax, Tmean, and Tmin. In terms of the seasonal variations, large temperature increase happens in summer under RCP4.5, while it happens in both summer and winter under RCP8.5. In terms of the extreme climate, the occurrence of extreme cold weather will decrease and extreme warm weather increase, whether the index is percentile-based or value-based. The diurnal temperature range will decrease, which is consistent with the aforementioned conclusions. Potentially beneficial for the agriculture, the growing season length is projected to increase for the City of Toronto. For the dynamical downscaling study of the regional climate over China, the results indicate that RegCM is capable of reproducing the spatial distributions of temperature and precipitation over China. Particularly, the high temperature centers in the Tarim Basin and the Sichuan Basin, which GFDL fails to capture, are reasonably represented by RegCM. RegCM also demonstrates good performance in eliminating the unrealistic high-precipitation center between the Tibetan Plateau and the Sichuan Basin produced by GFDL. Future projections from RegCM suggest that an increase of 2 °C in mean temperature is expected in China by the end of the twenty-first century under RCP4.5 while an increase of 4 °C would be seen under RCP8.5. The Tibetan Plateau is likely to expect the largest increase in temperature in China. The magnitude of increase in minimum temperature is apparently higher than that of mean and maximum temperature. In comparison, the annual total precipitation over China is projected to increase by 7% by the end of the twenty-first century under RCP4.5 and by 9% under RCP8.5. The projected changes in precipitation show apparent spatial variability due to the influences of local topography and land cover/use.