Effects Of Different Cloud Microphysical And Cloud Parameterization Schemes On Simulation Of The Torrential Rainfall Event Occurring In Beijing On 21 July 2012

In this study, seventeen cloud microphysical parameterization schemes in the weather research and forecasting modeling system (WRFV3.5.1 version) are used to simulate the torrential rain event that occurred over Beijing area during July 21-22 2012. It aims to evaluate the ability of the cloud microphysical parameterization schemes in simulating the rainstorm event and discuss the reason for the difference of the simulation. Based on the best microphysical parameterization scheme, the study is carried out with four cumulus parameterization schemes, Kain-Fritsch(KF), Betts-Miller-Janjic(BMJ), Grell-Devenyi(GD) and Simplified Arakawa-Schubert (SAS), respectively. It focuses on temporal and spatial feature of convection provocation and the relation between rain forecast and the status of convection provocation. Combined with the diagnosis of the basic physical quantities and convection/stratiform precipitation, the rainfall process could be better understood.Results show that simulated precipitation is quite sensitive to the different microphysical parameterization schemes used in the WRF model. Relatively, as the NSSL 1-momlfo scheme is used in the WRF model, the model well simulates the different classified rainfall patterns such as moderate rainfall and torrential rainfall, in particular, their location and intensity agree well with observations. When the Eta (Ferrier) scheme is used in the WRF model, the heavy rain is well simulated. Overall, for this torrential rain case, "good schemes" such as NSSL 1-mom, NSSL 1-momlfo and Milbrandt 2-mom schemes are used in the WRF model, the model performs well, while the model performs worst as the WDM6 scheme is used. In addition, when the aforementioned "good schemes" are used, the simulated variations of cloud water, cloud liquid water, cloud ice, and graupel are consistent with that of the observed precipitation. For these "good schemes", the WRF model also simulates much more ice particles and super-cooled water with larger range, which promote the development of ice-phase process, finally resulting in more precipitation.At last,the ensemble average is performed and it produces more stable and credible result,reducing the uncertainty of the scheme members.Cumulus parameterization schemes have an important impact in simulating heavy rainfall centers. By comparison, the simulated rainfall location and intensity with the KF scheme are closer to observed ones. BMJ scheme exaggerates the scope and intensity of the heavy rainfall. However, GD and SAS schemes have poor simulation results with almost no simulated heavy rain center, scattered distribution of the heavy rain and lower TS(Threat Score) score. Initial provocation status and time change of convection in four schemes are quite different and relate closely to their 24-hour precipitation simulations. Generally speaking, KF scheme predicts the convection better and it simulates a strong upward motion and abundant moisture over heavy rain area during the heavy rain period. However, the rainfall rate in deep convection area is relatively high with KF and BMJ schemes. The precipitation with SAS scheme is dominated by the stratiform type and the SAS scheme is unable to simulate the heavy rainfall center.