앙상블 칼만필터를 이용한 북서태평양 지역모델 수립

Abstract

Ensemble kalman Filter is implemented on the Northwestern part of the Pacific Ocean including all marginal seas around China, Korea and Japan, configured by Regional ocean modeling system (ROMS). The model resolution is 0.25 degree in horizonal and 20 simga level in vertical, respectively. For the open boundaries the model results from a global ocean model, Estimating the Circulation and Climate of the Ocean (ECCO), are used. The surface forcing is provided from the European Center of Medium Range Weather Forecasting (ECMWF) reanalysis data. The initial ensemble members are generated by adding to all state variables of November 1, 2002 the EOF-based perturbation, calculated over the period of November 1, 1998 - October 31, 2002. The leading five EOFs respectively account for 31.7%, 19.6%, 6.4%, 5.4% and 4.7% of the variance. The fist five modes explain approximately 80% of the total variance of the sample. Satellite SST data at 30 locations are assimilated every 7 day after 14-days spun-up of the initial ensemble members. Localization of background covariance plays an essential role in preventing the collapsing of ensemble spread. Assimilation of SST in the model improves the coherence between satellite observed sea surface height (SSH) and modeled SSH.