Sensitivity Study of Cumulus Convective Parameterization at a Gray Zone during a Heavy Precipitation Process over the Northeastern Tibetan Plateau

Model simulations of a heavy convective process that occurred in the northeastern Tibetan Plateau (TP) on August 17 and 18, 2016, were performed using the 3-km high-resolution Weather Research and Forecasting V4.0 model. Based on the observed hourly precipitation products, FY2E satellite, observation sounding, and reanalysis datasets, a set of four sensitivity tests were performed to compare and analyze the differences and influence mechanisms of convective precipitation using a fully explicit convection scheme (EXP experiment), traditional parameterized convection scheme (KF experiment), and scale-aware parameterization scheme (multiscale KF; MSKF experiment) in a model convective gray zone. The major conclusions were as follows. Considerable variations in heavy precipitation simulation of intensity and location in grayscale over the northeastern TP were observed with various treatments of moist physical processes. Compared with those simulated by the EXP and MSKF experiments, the 18-h total precipitation simulated by the KF experiment was closest to the observed precipitation in terms of precipitation amount and location, which is associated with the relative contribution of Kain–Fritsch Eta generated precipitation to total precipitation in the central rain area. Furthermore, the 300–400 hPa positive potential vorticity (PV) anomalies generated by the latent heat release of precipitation in the EXP and MSKF experiments tended to tilt more downstream with altitude because of their strong vertical wind shears, leading to more advanced positive PV anomalies at a lower level of 500 hPa than that in the KF experiment. This further resulted in the alteration of the low-level local wind fields, moisture flux transport fields, and atmospheric stability, ultimately generating heavy precipitation northeastward (downstream) versus the KF experiment and observation. The above investigation results suggested that it may be necessary to employ the convective parameterization scheme at the gray zone over the TP region, treating the subgrid-scale convective process implicitly to compensate for the simulation deviation from the explicit scheme. However, the MSKF scheme has some uncertainties in the expression of subgrid-scale convection over this region.