Applying prior correlations for ensemble-based spatial localization

Abstract. Localization is an essential technique for ensemble-based data assimilations (DAs) to reduce sampling errors due limited ensembles. Unlike traditional distance-dependent localization, the correlation cutoff method (Yoshida and Kalnay, 2018; Yoshida, 2019) tends localize observation impacts based on their background error correlations. This was initially proposed as a variable localization strategy coupled systems, but it can also be utilized extensively spatial localization. study introduced examined feasibility of alternative with local ensemble transform Kalman filter (LETKF) preliminary Lorenz (1996) model. We compared accuracy correlation-dependent localizations explored potential hybrid strategies. Our results suggest that deliver comparable analysis more efficiently faster DA spin-up. These benefits would become even pronounced under complicated model, especially when sizes are reduced.