基于BGM法结合Hybrid同化开展暴雨短期集合预报方法研究

Although rainstorm disasters are often directly caused by mesoscale convective system，the basic theory and principles on how to make mesoscale rainstorm short-range ensemble forecast have not been mature up to now. In this situation，this project begins with the research of SREF (short-range ensemble forecast) based on BGM (Breeding of Growing Modes) method in the limited area model WRF model，followed by the study on the application scheme and design principles of BGM method mesoscale rainstorm EPS(ensemble prediction system) for the sake of better rainstorm SREF skills . Several basic and important problems (containing how to constrain perturbations conditions, how to generate perturbations structure, how to determine ensemble size and so on) were studied respectively in this SREF EPS based on BGM method. Through these studies，it will tell the forecaster how to determine the schemes for EPS of BGM method when facing with hazardous mesoscale heavy rainfall system.. Besides these above studies and considering the progress of data assimilation theory，the latest data assimilation development namely Hybrid data assimilation scheme was introduced into WRF model to improve SREF skills in a new manner. This new type SREF EPS was established with BGM method EPS inherently combining with Hybrid data assimilation system in WRF model to study how to further improve mesoscale rainstorm SREF skills. This new manner EPS is able to carry out inherent positive feed-back mutual promotion between EPS and Hybrid data assimilation system. More importantly,this Hybrid data assimilation system based on the existing WRF 3DVAR (three-dimensional variational data assimilation)system，unlike many other operational 3DVAR systems and conventional ensemble-based data assimilation systems，is able to incorporate flow-dependent ensemble covariance information into existing variational systems. Thus the better analysis field must be obtained by assimilating real observations in this new system than conventional 3DVAR system. Then the better analysis fields lead to better mesoscale rainstorm SREF skills compared with SREF EPS only based on BGM method alone. Therefore, a suitable scheme on how to carry out hazardous weather SREF for region of China will be presented in this study.. In the end, it is easy to find that this new manner of SREF EPS is superior to the current mesoscale rainstorm SREF EPS in theory and method. Especially, the application and study of current project of SREF EPS based on the combination of Breeding method and Hybrid data assimilation is original and practical, and it must fill in the gaps in the field of mesoscale rainstorm SREF.

中尺度系统是直接造成暴雨灾害的天气系统。基于目前中尺度暴雨集合预报理论还不完善，本项目在WRF模式中引入BGM法开展短期集合预报研究，并确立BGM法中尺度强降水短期集合预报的应用方案和设计准则（包括扰动的约束条件、结构、幅度等关键问题）。同时结合资料同化的理论进展，引入目前最新的同化思想-Hybrid理论，建立一个BGM法结合Hybrid同化的内在呈"正反馈"相互促进的新型中尺度强降水短期集合预报系统，开展短期集合预报研究，并将其效果与BGM法短期集合预报效果进行对比分析，以此提出适合我国区域特点的短期集合预报方案。这种新思路的主要思想是把BGM法集合预报结果中反映中尺度不确定性的"流相关"的集合方差引入到Hybrid同化系统中,同化观测资料并形成最优分析场，从而改进集合平均预报，提高中尺度强降水短期集合预报水平，较目前的理论和方法有明显的创新性，可填补我国暴雨集合预报研究在该领域的空白。

中尺度系统是直接造成暴雨灾害的天气系统，然目前中尺度暴雨集合预报理论仍在不断的尝试和探索中，其中主要的问题有初值不确定性、物理过程不确定性和模式动力框架的不确定性，国内外对这些问题在短期集合预报领域的研究进展都很重视，本项目主要围绕短期集合预报的核心问题进行深入研究，即产生更优的初始扰动集合。首先，项目完成了WRF模式BGM集合预报系统方案与程序设计，研究了BGM法扰动培育饱和时间判定及其影响因素试验研究，以及开展了初始随机模态与初始模幅度对扰动发展的影响研究，并分析了自由繁殖与动态调整两种方案的扰动演变特征，获得了一系列有意义的结论。.利用BGM法开展短期集合预报研究的研究成果，基于“集合预报与资料同化一体化”的思路，引入目前最新的同化思想—Hybrid理论，建立了一个BGM法结合Hybrid同化的内在呈“正反馈”相互促进的新型中尺度强降水短期集合预报系统，并在该系统框架下，讨论了水平风场增量、位温扰动和位势扰动增量与3DVAR系统的不同，研究了不同静态背景误差协方差对同化效果的影响，局地化尺度对同化的影响，并将该系统集合预报效果与BGM法短期集合预报效果进行了初步分析，新型的集合预报系统其平均预报结果较单独的BGM法集合预报结果系统性误差小，这种新思路的主要思想是把BGM法集合预报结果中反映中尺度不确定性的“流相关”的集合方差引入到Hybrid同化系统中,同化观测资料并形成最优分析场，从而减少集合平均预报的系统性偏差。