Sensitivity study of WRF cumulus schemes based on ERA5 remote sensing datas

Under the backdrop of global warming, extreme precipitation events, especially severe rainstorms, are becoming increasingly frequent and severe, making them one of the most significant meteorological disasters worldwide. In this study, we numerically simulated a historical heavy precipitation process in North China during the summer of 2023 using the WRF model. The simulation used NCEP's (National Centers for Environmental Prediction) GFS(the Global Forecast System) datas as the initial field and compared the effectiveness of three different cumulus convection parameterization schemes: the Tiedtke Scheme, Kain-Fritsch Scheme, and Betts-Miller-Janjic Scheme. By contrasting the simulation outcomes with ERA5 (the Fifth Generation of EC-MWF Atmospheric Reanalyses of the Global Climate) reanalysis remote sensing datas, we assessed the influence of these schemes on total precipitation, radar reflectivity, and the wind field at 850 hPa. The Tiedtke scheme, when used within the D01 domain, simulates a uniform and smooth precipitation distribution with concentrated high-intensity areas. Expanding to nested domains (D01, D02, D03) increases the strength and distribution of precipitation centers. This scheme generally predicts a flatter spatial distribution compared to the ERA5 data. In contrast, the Kain-Fritsch scheme shows stronger spatial variability and more localized intense precipitation, particularly in areas of strong convection, with scattered but intense precipitation centers in the D01 domain that become more pronounced in nested domains. The Betts-Miller-Janjic scheme produces precipitation patterns with intensities between those of Tiedtke and Kain-Fritsch, with some concentrated and some scattered precipitation centers. Its performance shows minimal changes across different domain configurations, and it demonstrates significant differences in precipitation in drier areas when compared to ERA5 data.