Data Assimilation by Coupling Uncertain Support Vector Machine with Ensemble Kalman Filter

Data assimilation is widely applied to improve prediction accuracy. In common assimilation routine the prediction and assimilation are performed alternatively. However, prediction directly using the original data requires high computation costs and low accuracy. In this paper, a method of data assimilation by coupling variable precision rough set, ensemble Kalman filter and SVM is proposed. The rough set is adopted to reduce the redundant inputs. Prediction is performed by SVM with the reduced inputs. Then, ensemble Kalman filter is adopted to assimilate prediction results from SVM. The experimental results demonstrate that the proposed method reduces the training time and improves data assimilation accuracy.