EXPERIMENTS ON ASSIMILATION OF INITIAL VALUES IN NUMERICAL PREDICTION OF A WARM-SECTOR PRECIPITATION IN SOUTH CHINA

In order to understand the impact of initial conditions upon prediction accuracy of short-term forecast and nowcast of precipitation in South China, four experiments i.e. a control, an assimilation of conventional sounding and surface data, testing with nudging rainwater data and the assimilation of radar-derived radial wind, are respectively conducted to simulate a case of warm-sector heavy rainfall that occurred over South China, by using the GRAPES_MESO model. The results show that (1) assimilating conventional surface and sounding observations helps improve the 24-h rainfall forecast in both the area and order of magnitude; (2) nudging rainwater contributes to a significant improvement of nowcast, and (3) the assimilation of radar-derived radial winds distinctly improves the 24-h rainfall forecast in both the area and order of magnitude. These results serve as significant technical reference for the study on short-term forecast and nowcast of precipitation over South China in the future.