资源环境科技发展态势分析平台

Future projections using regional climate models (RCMs) are driven with boundary conditions (BCs) typically derived from global climate models. Understanding the impact of the various BCs on regional climate projections is critical for characterizing their robustness and uncertainties. In this study, the International Center for Theoretical Physics Regional Climate Model Version 4 (RegCM4) is used to investigate the impact of different aspects of boundary conditions, including lateral BCs and sea surface temperature (SST), on projected future changes of regional climate in West Africa, and BCs from the coupled European Community-Hamburg Atmospheric Model 5/Max Planck Institute Ocean Model are used as an example. Historical, future, and several sensitivity experiments are conducted with various combinations of BCs and CO2 concentration, and differences among the experiments are compared to identify the most important drivers for RCMs. When driven by changes in all factors, the RegCM4-produced future climate changes include significantly drier conditions in Sahel and wetter conditions along the Guinean coast. Changes in CO2 concentration within the RCM domain alone or changes in wind vectors at the domain boundaries alone have minor impact on projected future climate changes. Changes in the atmospheric humidity alone at the domain boundaries lead to a wetter Sahel due to the northward migration of rain belts during summer. This impact, although significant, is offset and dominated by changes of other BC factors (primarily temperature) that cause a drying signal. Future changes of atmospheric temperature at the domain boundaries combined with SST changes over oceans are sufficient to cause a future climate that closely resembles the projection that accounts for all factors combined. Therefore, climate variability and changes simulated by RCMs depend primarily on the variability and change of temperature aspects of the RCM BCs. Moreover, it is found that the response of the RCM climate to different climate change factors is roughly linear in that the projected changes driven by combined factors are close to the sum of projected changes due to each individual factor alone at least for long-term averages. Findings from this study are important for understanding the source(s) of uncertainties in regional climate projections and for designing innovative approaches to climate downscaling and impact assessment.