FuXi-DA: a generalized deep learning data assimilation framework for assimilating satellite observations
Xiaoze Xu, Xiuyu Sun, Wei Han, Xiaohui Zhong, Lei Chen, Zhiqiu Gao, Hao Li
Abstract
Data assimilation (DA), as an indispensable component within contemporary Numerical Weather Prediction (NWP) systems, plays a crucial role in generating the analysis that significantly impacts forecast performance. Nevertheless, developing an efficient DA system poses significant challenges, particularly in establishing intricate relationships between the background field and the vast amount of multi-source observation data within limited operational time windows. Recently, Deep learning-based (DL-based) weather forecast models have shown promise in matching, even surpassing, the leading operational NWP models worldwide. This success motivates the exploration of establishing DL-based DA frameworks. DL models possess multi-modal modeling capabilities, enabling the fusion of multi-source data in the feature space, which is very similar to the process of assimilating multi-source observational data in DA systems. In this study, we introduce FuXi-DA, a generalized DL-based DA framework for assimilating satellite observations. By assimilating data from Advanced Geosynchronous Radiation Imager aboard Fengyun-4B, FuXi-DA consistently mitigates analysis errors and significantly improves forecast performance. Furthermore, FuXi-DA has been validated against established atmospheric physics, demonstrating its consistency and reliability.