The Development And The Establishment Of Guangxi GRAPES Mesoscale Model System

Numerical Weather Prediction(NWP) is one of the most effective methods for diagnose mesoscale weather analysis and forecasting system, and it’s also the most important tool of meteorological operations and weather research. Study on the mesoscale numerical model can identify evolution details of the small-scale weather processes and its influencing factors. Therefore, the mesoscale model prediction system of Guangxi is built that can be more carefully study the formation mechanism of regional disastrous weather, and improving the ability to refined weather forecasts. Further provide support for the weather forecast, research and decision-making meteorological services in Guangxi Meteorological Administration.This thesis focuses on scale construction of GRAPES(Global/Regional Assimilation and Prediction System) in Guangxi, combined with several research projects, respectively introducing the Dawning TC4000A/TC5000 HPC(floating-point peak value: 1500 gflops/8.192 tflops), and carried out the GRAPES’s transplant and localization in the machines. To complete the GRAPES’s model system and three-dimensional variational(3DVar) assimilation system, the project developed some interface to make GRAPES work well with local data environment, such as background field data, the lateral boundary data, the local sounding data, the observation data, the doppler weather radar’s data, et al, even includes the product’s quality inspection. Further relying on the mass storage device equipped of Dawning HPC, the project built three website: the Guangxi Meteorological Information Sharing System(GMISS), the Guangxi Mesoscale Numerical Model, and the Open Source Code Sharing. From the above, this thesis has been realized the GRAPES’s real localization, which includes download the background field data, process the observation data, prepare the 3DVar, model preprocess, run model, model postprocess, model products production, real-time dynamic quality inspection, and so on. The products of GRAPES numerical model which provide real-time spatial resolution up to 3km, the highest temporal resolution up to 1hr, the longest forecast period up to scale 48 hr, such as GrADS products, regional precipitation/temperature color splash map, reference to the real-time forecast business, weather research, and decision support server.By testing various functional modules of Guangxi GRAPES mesoscale numerical model,this thesis shows that the project’s scheme is established to totally fulfill the requirements of business operations and localization needs. The system provides conversion application designed of NWP’s products in MICAPS, and mesoscale numerical model website is developed to meet the basic needs of the business and researchers, further the meteorological departments at all levels have been widely used and promoted in Guangxi. The GMISS is collaborative development that collected a variety of meteorological data also become an important support for other research projects. Another in this thesis, the source code sharing system is advocated and realized, to improve the reuse of code, effectively reduce the duplication of development, and played a certain reference to other research projects.