基于水物质控制变量的卫星云观测资料同化方法研究

The effective utilization of the cloud-affected satellite data will be an important approach to improve numerical weather forecasting. Dealing with hydrometeor variables scientifically under cloud areas by data assimilation system is a key work to improve the performance of cloud-affected satellite data assimilation system, and it is also a difficult and hot issue on data assimilation. This project will carry out researches on cloud-affected satellite data assimilation methods using the hydrometeor variables as control variables. The hydrometeor variables transform method in the data assimilation system will be discussed, to investigate the method of getting hydrometeor control variables that satisfy independence and unbiased Gaussian distribution as much as possible. And the ensemble-variational hybrid assimilation method with hydrometeor control variables will be studied, to explore the character of background error covariance in cloud areas under different weather system. Moreover, single observations assimilation tests, cycling assimilation and prediction experiments will be conducted, to analyze the impact of ensemble members, weight coefficient and parameterization schemes on the assimilation and prediction. This project aims to obtain proper hydrometeor control variables, and to get flow-dependent and multivariable correlation hydrometeor variables background error covariance for data assimilation system, then to develop an assimilation system that can analyze hydrometeors in cloudy regions more properly and directly, and to improve the assimilation performance of the system in cloud areas. This project also will lay a more solid foundation to directly assimilate cloud-affected satellite radiance observations.

云区卫星资料的有效利用将是提高数值天气预报水平的重要途径。同化系统对云中水物质变量的科学处理是提高云区卫星资料同化效果的关键工作之一，也是目前资料同化研究的难点和热点。围绕水物质变量的处理这一核心问题，项目拟开展以水物质为控制变量的云区卫星资料同化方法研究。探讨同化系统中水物质控制变量转换方法，探索获得尽可能独立和满足无偏高斯分布的水物质控制变量的途径；研究以水物质为控制变量的集合-变分混合同化方法，探究不同天气背景下云区及云区水物质变量的背景场误差协方差相关特征；开展单点观测理想试验和循环同化及预报试验，讨论集合成员、权重系数和物理参数化方案对同化和预报的影响。旨在为同化系统引入合理的水物质控制变量，并获得流依赖及多变量相关的云区及云区水物质背景场误差协方差，建立能够对水物质进行更合理直接分析的同化系统，改善系统在云区的同化性能。本研究也将为云区卫星辐射率资料的直接同化奠定更好基础。

有云环境下卫星资料同化还存在很多科学问题有待解决，而同化系统对云区水成物变量的科学处理是提高云区卫星观测资料同化系统性能的关键问题之一。本项目围绕资料同化系统对水物质变量的合理处理这一核心问题，开展了以水成物变量为控制变量的卫星云观测资料同化方法研究。研究了同化系统中水物质变量的变量转换方法，提出了一种softmax函数的水物质控制变量转换方法，获得了更为高斯分布的水物质控制变量；建立了包含水物质控制变量的集合-变分混合同化模块，引入了流依赖和多变量相关的水物质变量背景场误差协方差并讨论了其具体特征；开展了一系列单点观测理想试验和循环同化与预报试验，讨论了水物质控制变量的集合-变分混合同化模块在实际同化和预报中的效果。通过本项目研究，建立了以水物质为控制变量的集合-变分混合同化模块，使得同化系统能够更好地对水物质变量进行直接分析和处理，提高了同化系统的性能和实际预报效果，为云区卫星资料的直接同化奠定了更好的基础。