资源环境科技发展态势分析平台

The need for good quality information on climate and its future changes has become increasingly important for society. Of particular interest are analysis and predictions of rainfall at pertinent spatial scales. An option is the use of regional climate models (RCMs) as a physics-based downscaling tool to retrieve higher spatial resolution information from coarser present and future climate datasets. In order to verify if simulations provide added values that result in an appropriate representation of the state of climate at finer spatial resolutions, the adoption of RCMs requires prior optimization. The performance of the Weather Research and Forecasting Model RCM in simulating precipitation over Indonesia is examined by a series of sensitivity experiments using different parameterized convective physics. Among four tested schemes, the best performance is provided by the Betts-Miller-Janjic (BMJ) parameterization. RCM multiannual seasonal rainfall bias outperforms or matches the reanalysis. Modeled results provide added value in simulating rainfall-related climate indices but show low skill in recreating the annual rainfall pattern at monthly resolution. RCM precipitation exhibits complex spatial response to different ENSO phases, with El Nino conditions resulting in a general loss of model skill during the southern hemisphere spring. A series of regional climate simulations using the BMJ convective scheme forced by a future climate projection dataset show changes in rainfall aligned with previous studies.