Ideal Case Study of Adaptive Localization in Storm-scale Ensemble Kalman Filter Assimilation

This study explores the use of the hierarchical ensemble filter to determine the localized influence of observations in the Weather Research and Forecasting ensemble square root filtering (WRF-EnSRF) assimilation system. With error correlations between observations and background field state variables considered, the adaptive localization approach is applied to conduct a series of ideal storm-scale data assimilation experiments using simulated Doppler radar data. Comparisons between adaptive and empirical localization methods are made, and the feasibility of adaptive localization for storm-scale ensemble Kalman filter assimilation is demonstrated. Unlike empirical localization, which relies on prior knowledge of distance between observations and background field, the hierarchical ensemble filter provides continuously updating localization influence weights adaptively. The adaptive scheme improves assimilation quality during rapid storm development and enhances assimilation of reflectivity observations. The characteristics of both the observation type and the storm development stage should be considered when identifying the most appropriate localization method. Ultimately, combining empirical and adaptive methods can optimize assimilation quality.