Performance of the WRF model in simulating convective rainfall events in the humid subtropical monsoon climate region-Poyang Lake basin

The Weather Research and Forecasting (WRF) model can provide high-resolution rainfall and is important for flood management and prevention. In order to evaluate the performance of WRF and select the best combination of physical parameterizations, as well as exploring how cumulus parameterizations influence the rainfall simulation at 3-km grid spacing over the Poyang Lake basin, the WRF model was configured with horizontal grid spacing of 27, 9, and 3 km using the three microphysical parameterizations (MPs), three cumulus parameterizations (CUs), and two planetary boundary layer parameterizations (PBLs) for three representative 24-h rainfall events. Results show that (1) the selection of physical parameterizations has a greater influence than horizontal grid spacing on rainfall simulation. The maximum relative error (RE)difference between three domains is 2.51%, 1.11%, and 2.52% for three rainfall events, while that for R1 to R3 is 29.13%, 26.68%, and 39.89%, respectively. (2) Among 18 schemes, S11, containing parameterizations of New Thompson and YSU, shows overall better performance for spatial and temporal rainfall simulation. S5, containing parameterizations of WSM6 and YSU, and S15, containing parameterizations of Morrison and YSU, are also good choices for rainfall simulation. (3) In most cases, the use of cumulus schemes at 3-km resolution does not show obvious advantages over these not use cumulus schemes in rainfall forecasts. (4) The numerical simulations are more sensitive to the selection of PBLs than MPs in the study area. Among individual parameterizations, YSU is the best choice for PBLs in simulating rainfall. Overall, the results demonstrated that WRF can provide useful rainfall simulation, and the study provides a reference in rainfall forecast using WRF over the Poyang Lake basin.