The Construction Of A Simplified2-D Cloud Resolving Model And Its Superparameterization Application Experiment In GRAPES

Atraditionaldefinitionofsuperparameterization(SP)isexplicitlyrepresentingsub-gridprocesses in a large-scale modeling system by embedding a Cloud Resolving Model (CRM)in each column of the large-scale model. This thesis expands the utility area of SP to ameso-scale regional weather model GRAPES. The thesis consists of two parts:In the first part, a lightweight but powerful simplified two-dimensional (2D) CRM isdesigned and developed, aiming to realize the application of SP in GRAPES. The perfor-mance of the CRM is evaluated with classical benchmark tests such as Density Current,Convective Thermal Bubble, Internal Gravity Waves and also an ideal thunderstorm test.The ideal simulations show that with RK3-5discretization scheme in a staggered Arakawa-C and Charney-Phillips grid, the CRM performs well for typical atmospheric phenomenon,especially convective and vortex phenomenon. The dynamic framework is robust and qual-ified for high resolution simulation, and the microphysics paramterization is reasonable forcloud-scale simulation. Since the design of the CRM takes the utility of SP as its ultimategoal, and it takes the characteristics of GRAPES under consideration, the CRM is handy andsuitable for GRAPES to apply SP.In the second part, the simplified2D CRM is coupled with GRAPES, so as to form aSP version of GRAPES—SP-GRAPES. The case of Beijing “721” heavy rainfall is chosedto evaluate the precipitation forcasting performance of SP-GRAPES, and the control experi-ment(CTL)andsuperparameterizationexperiment(SP)areconducted. Becauseoflimitaionof computing resource, SP is only adopted in the region near Beijing, and the setting of themodel is the same as the domain out of Beijing. The simulation of the case shows that SPcan successfully make an adjustment about the location of the rainfall and make up for thedeficiency of the maximun amount of the precipitation. Meanwhile, the simulation of cloud fraction is obviously improved. However, the SP scheme produces excessive precipitatioinamount in general. And also, since the SP is adopted only in a certain area of the simula-tion domain, the improvement of the prediction of rainfall location and precipitation amounttakes the bias of the lower reach regions of Beijing as cost.By now, the framework of SP-GRAPES has been initially built up. However, this isonly a preliminary attempt of SP in a NWP model, many improvements can be made in thetechnical specifications. For instance, more specific and delicated physical processes couldbe constructed inner the CRM so as to get closer to the final target of physics coupler. Andthe method of coupling these two models in a harmonious way is always a subject worth ofdeep thinking and research.