摘要
The performance of the fifth generation of the Canadian Regional Climate Model (CRCM5) in reproducing the main climatic characteristics over India during the southwest (SW)-, post- and pre-monsoon seasons are presented in this article. To assess the performance of CRCM5, European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) and Interim re-analysis (ERA-Interim) driven CRCM5 simulation is compared against independent observations and reanalysis data for the 1971-000 period. Projected changes for two future periods, 2041-070 and 2071-100, with respect to the 1971-000 current period are assessed based on two transient climate change simulations of CRCM5 spanning the 1950-100 period. These two simulations are driven by the Canadian Earth System Model version 2 (CanESM2) and the Max Planck Institute for Meteorology’s Earth System Low Resolution Model (MPI-ESM-LR), respectively. The boundary forcing errors associated with errors in the driving global climate models are also studied by comparing the 1971-000 period of the CanESM2 and MPI-ESM-LR driven simulations with that of the CRCM5 simulation driven by ERA-40/ERA-Interim. Results show that CRCM5 driven by ERA-40/ERA-Interim is in general able to capture well the temporal and spatial patterns of 2?m-temperature, precipitation, wind, sea level pressure, total runoff and soil moisture over India in comparison with available reanalysis and observations. However, some noticeable differences between the model and observational data were found during the SW-monsoon season within the domain of integration. CRCM5 driven by ERA-40/ERA-Interim is 1-?°C colder than CRU observations and generates more precipitation over the Western Ghats and central regions of India, and not enough in the northern and north-eastern parts of India and along the Konkan west coast in comparison with the observed precipitation. The monsoon onset seems to be relatively well captured over the southwestern coast of India, while the monsoon withdrawal occurs too late in comparison with observations. Boundary forcing errors are generally of the same magnitude or larger for the SW-monsoon season and smaller for the other seasons when compared to performance errors. For the two future periods analyzed, both CanESM2 and MPI-ESM-LR driven CRCM5 simulations imply a general warming over India in the twentyfirst century, with maximum increases projected for the pre-monsoon season. However, conflicting climate change signal is noted for precipitation for both future periods with respect to the 1971-000 current period, with CanESM2 driven simulation suggesting an increase in precipitation in future climate, which is consistent with the projected increase in evapotranspiration, while MPI-ESM-LR driven simulation suggests a decrease in precipitation and evapotranspiration. The conflicting climate change signal in precipitation is also reflected in those variables that are tightly linked with precipitation, such as total runoff, highlighting the need for multi-model ensembles to better represent the uncertainties related to future projections.