Impacts of WRF Model Domain Size on Meiyu Rainfall Forecasts over Zhejiang, China

摘要

We explore impacts of model domain size on rainfall forecasts over Zhejiang Province during the Meiyu season in this paper. The existing modeling system is based on the Weather Research and Forecasting (WRF), which is referred to as the Zhejiang WRF with Advanced Data Assimilation System (ADAS) Real-time Modeling System (ZJWARMS). First, we identify the regions with large uncertainties of model forecasts using correlation analysis of different reanalysis datasets. Five key regions affecting precipitation in Zhejiang are identified by calculating the correlation-coefficients between precipitation in Zhejiang and large-scale atmospheric factors. A new WRF model domain, called ZJdomain, is then chosen to cover the key regions and to avoid the regions with large uncertainties. To evaluate the differences of precipitation forecasts in each precipitation category between using the new domain, ZJdomain, and using the current operational model domain, ZJWARMS, two sets of experiments are conducted, including two case studies and a batch test based on daily 72-h forecasts. The results suggest that the ZJdomain can better simulate the key weather systems and their associated precipitation in Zhejiang. Quantitative tests of different-category precipitation also show that the equivalent threat score (ETS) increases significantly, and the frequency of missed (FOM) and the false alarm ratio (FAR) decrease with varying degrees. In the batch test, the ETSs of the torrential rain in forecasts with the lead times of 24-48 h and 48-72 h increase by 100.3% and 62.1%, respectively. Compared with the Shanghai Meteorological Service WRF ADAS Real-time Modeling System version 2.0 (SMSWARMS), it is found that the unsuitable WRF model domain is one of the main reasons that the precipitation forecast error of ZJWARMS is always larger than that of SMSWARMS. This study demonstrates that the selection of a suitable model domain is a key to improve the forecast of regional operational numerical weather prediction systems.