Abstract:We present an innovative interpretation of Kalman Filter (KF, for short) combining the ideas of Schwarz Domain Decomposition (DD) and Parallel in Time (PinT) approaches. Thereafter we call it DD-KF. In contrast to standard DD approaches which are already incorporated in KF and other state estimation models, implementing a straightforward data parallelism inside the loop over time, DD-KF ab-initio partitions the whole model, including filter equations and dynamic model along both space and time directions/steps. As a consequence, we get local KFs reproducing the original filter at smaller dimensions on local domains. Also, sub problems could be solved in parallel. In order to enforce the matching of local solutions on overlapping regions, and then to achieve the same global solution of KF, local KFs are slightly modified by adding a correction term keeping track of contributions of adjacent subdomains to overlapping regions. Such a correction term balances localization errors along overlapping regions, acting as a regularization constraint on local solutions. Furthermore, such a localization excludes remote observations from each analyzed location improving the conditioning of the error covariance matrices. As dynamic model we consider Shallow Water equations which can be regarded a consistent tool to get a proof of concept of the reliability assessment of DD-KF in monitoring and forecasting of weather systems and ocean currents
What problem does this paper attempt to address?
This paper attempts to address the issues of computational complexity and parallel computing efficiency faced by the Kalman filter (KF) when dealing with large - scale data assimilation problems. Specifically, traditional KF methods require a large amount of computational resources when handling high - dimensional dynamic systems, and due to their sequential time - stepping characteristics, it is difficult to achieve effective parallel computing. Moreover, when applying KF to real - time monitoring and prediction in fields such as meteorology and oceanography, the bottleneck of computational efficiency is particularly prominent.
To solve these problems, the author proposes an innovative method - DD - KF (Domain Decomposition Kalman Filter), which combines the ideas of Schwarz Domain Decomposition (DD) and Parallel in Time (PinT). This method not only decomposes the model in space but also in time, thereby achieving more efficient parallel computing. By introducing correction terms to balance the local solutions in the overlapping regions, it ensures that the final global solution is consistent with the standard KF. At the same time, this method also improves the condition number of the error covariance matrix by excluding distant observations, thereby enhancing numerical stability.
The following are the main contributions of this method:
1. **Space - time decomposition**: Initially divide the entire model along the spatial and temporal directions/steps to generate local KFs with smaller dimensions, and these local KFs can be solved in parallel on different sub - domains.
2. **Introduction of correction terms**: To ensure the consistency of local solutions in the overlapping regions, correction terms are introduced as smoothing regularization constraints to record the contributions of adjacent sub - domains to the overlapping regions.
3. **Improved parallelism**: Through the PinT method, the inherent bottleneck of the time - marching solver is solved, and the smoothness of the analysis field is improved.
4. **Applicability verification**: Using the Shallow Water Equations (SWEs) as a dynamic model, the reliability of DD - KF in weather system and ocean current monitoring and forecasting is verified.
Overall, DD - KF provides a new framework that can effectively reduce the computational complexity of KF, improve parallel computing efficiency, and exhibit good performance in practical applications.