An Improved Data Assimilation Localization Method Based on Fuzzy Logic Control

Abstract

To determine the best estimates of the ocean or atmospheric state, Data Assimilation (DA) is a methodology to combine all available information from numerical models and observation information. With small ensemble numbers in the Ensemble Kalman filter, the more effective utilization the observation data are, the higher the assimilation performance become. In this study, we proposed a new fuzzy control-based data assimilation system, named FA (fuzzy analysis) and CF (covariance fuzzy), which coupled with fuzzy logic control algorithms to improve assimilation performance. A number of numerical experiments are designed using a classical nonlinear model (the Lorenz-96 model) to explore the effects of the new algorithms on the ensemble transform matrices. The experiments show that the FA algorithm can be selected when the system is in weak assimilation, whereas both algorithms can be implemented in medium assimilation situations. If the system is strongly assimilated, the CF algorithm has demonstrated more robust performance.