MPAS-GSI Assimilation Prediction Framework And Preliminary Application Based On Conservation Remapping

Data Assimilation provides a vital approach in enhancing accuracy of numerical forecasts.In recent years,with the development of NWP models and high-performance computers,global atmosphere models with unstructured meshes are becoming popular.Compared with traditional structured grid,unstructured mesh features flexible geometric adaptability,and has better parallel computing performance.However,data assimilation techniques with regard to unstructured mesh is not well-developed yet,which provides an obstacle to further application of models with unstructured meshes.In this article,with highlight on MPAS-A,a new global non-hydrostatic model with unstructured SCVT mesh,in addition with operational GSI global data assimilation system,a MPAS-GSI data assimilation and forecast framework is constructed.Cycled data assimilation and forecast run is achieved through conservative remapping technique,which transforms meteorology variables between unstructured meshes of MPAS-A and structured grids of GSI.Principles of transformation and the process of construction is explained in detail in this paper,followed by validation and application of the framework.The main results are as follows:(1)Conservative remapping is capable of transforming meteorological variables between unstructured grid of MPAS-A and structured grid of GSI.Second-order conservative remapping yield smaller transformation error than the first-order approach,and the incremental transforming strategy could further improve the accuracy of transformation.(2)Results from grid transformation tests and pseudo-observation experiments show that MPAS-GSI framework achieved model variable transformation reasonably.Analysis results from GSI could be correctly projected to MPAS-A,regardless of employing quasi-uniform and variable-resolution meshes.(3)Full-cycled data assimilation and forecasting experiments revealed the impacts of assimilating conventional and satellite radiances to model analysis and forecasts.It is shown that with the aid of GSI data assimilation,forecast results of dynamical and thermodynamical fields,both on quasi-uniform and variable-resolution meshes of MPAS-A,are improved in general.Compared with conventional observations,assimilating satellite radiances could further reduce forecast error in the north and south hemisphere,implying observations of multi sources are assimilated effectively.Case study of Meiyu precipitation process showed that different meshes features their unique characters of simulating weather systems.The study of typhoon case showed that assimilating sea surface winds reduced forecast error of wind field surrounding the typhoon center.In general,the enhancements of model forecast variables could show positive effects to forecasting precipitation typhoon tracks.