基于非线性集合四维变分同化方法NLS-4DVar的土壤温、湿度协调同化研究

In the project, we will start from the land data assimilation system (referred as IAP-LDAS) continually developed at IAP. This system is built based on the community land model and the ensemble POD-4DVar method. This system already has the capabilities to assimilate both in-situ and satellite observations. The POD-4DVar aimed at exploiting the strengths of the two forms (i.e., 4DVar and EnKF) of data assimilation while simultaneously off-setting their respective weaknesses. A novel nonlinear least-squares enhanced POD (Proper Orthogonal Decomposition)-based 4DVar algorithm (referred as NLS-4DVar) is further proposed for the nonlinear ensemble-based 4DVar. In the algorithm, the Gauss-Newton iterative method is employed to handle the non-quadratic nonlinearity of the 4DVar cost function while the overall structure of the algorithm still resembles the original POD-4DVar algorithm. It is proved that the original POD-4DVar algorithm is a special case of the proposed NLS-4DVar algorithm under the assumption of the linear relationship between the model perturbations (MPs) and the simulated observation perturbations (OPs). Particularly, the NLS-4DVar approach can produce satisfactory results only using a very small iteration number and thus doesn’t bring any additional computational cost compared with the original POD-4DVar method. The NLS-4DVar is expected to address the high non-linearity issues arisen from the land surface processes. Further, we will incorporate the satellite retrieved land surface temperature data into IAP-LDAS and thus achieve joint assimilation of soil moisture and temperature. Also, the surface turbulent heat fluxes will be simultaneously optimized through the joint data assimilation.

以项目申请人自主研发、基于通用陆面过程模式可同时同化多源土壤湿度观测资料（台站与微波亮温）的IAP陆面数据同化系统为基础，将其核心的同化算法POD-4DVar进一步发展，通过严格的理论推导将其代价函数转换为非线性最小二乘最优化问题，实现非线性最小二乘最优化问题的理论体系与集合数据同化方法的融合，形成基于非线性最小二乘最优迭代的集合四维变分同化方法NLS-4DVar，摈弃一般集合同化方法所惯用的线性化假设而采用非线性最优迭代，可克服由于陆面过程模式高度的非线性所造成的虚假噪声，节约计算量并使得同化结果准确可信；进一步纳入卫星反演的地表温度观测资料，在四维变分同化的框架内进行土壤温、湿度之协调同化，由此实现土壤温、湿度与陆表水、热通量以及陆面过程模式整体模拟性能的改善。鉴于微波亮温数据及地表温度反演资料的高时空分辨率及其在全球的极易获得性，该系统的建立使得在全球获取高精度的陆表数据集成为可能。

以项目负责人自主研发、基于通用陆面过程模式可同时同化多源土壤湿度观测资料的IAP陆面数据同化系统为基础，将其核心的同化算法POD-4DVar进一步发展，实现非线性最小二乘最优化问题的理论体系与集合数据同化方法的融合，发展先进的NLS-4DVar、将其与IAP陆面数据同化系统相耦合，在四维变分同化的框架内进行土壤温、湿度协调同化，由此发展新一代IAP陆面同化系统。在该项目的支持下，我们1）进一步完善发展了先进的集合四维变分同化方法NLS-4DVar, 该方法实现了数据同化领域En4DVar、4DEnVar以及EnKF统一的公式表达；将大数据的概念引入NLS-4DVar,由此形成了大数据驱动的NLS-4DVar,极大地促进了集合四维变分同化的发展；2）推出了最新版本的IAP陆面数据同化系统，该同化系统采用先进的NLS-4DVar，在四维变分同化的框架之内实现了对进行土壤温、湿度的协调同化；3）构建了多重网格NLS-4DVar陆气耦合数据同化系统SNAP。目前NLS-4DVar已经该方法已被成功应用于陆面数据同化、目标观测、雷达资料同化、常规资料同化、卫星资料同化、核辐射数据同化以及碳循环数据同化之中，都取得了相当不错的同化效果，显示了极大的潜力；而最新版本的IAP陆面数据同化系统正在与我们所构建的多重网格数据同化系统SNAP耦合，待成熟后将在海南省气象局进行业务应用。这些同化算法与同化系统的发展对于促进资料同化算法、数值天气预报以及气候模拟水平的提高具有巨大的推动作用。