摘要: An optimal interpolation assimilation model for satellite altimetry data is developed based on Princeton Ocean Model (POM), which is applied in a quasi-global domain, by the method of isotropic correlation between sea level anomaly (SLA) and sea temperature anomaly. The performance of this assimilation model is validated by the modeled results of SLA and the current patterns. Comparisons between modeling and satellite data show that both the magnitudes and distribution patterns of the simulated SLA are improved by assimilation. The most significant improvement is that meso-scale systems, e.g., eddies, are well reconstructed. The evolution of an eddy located in the northwest Pacific Ocean is traced by using the assimilation model. Model results show that during three months the eddy migrated southwestward for about 6 degrees before merging into the Kuroshio. The three dimensional structure of this eddy on 12 August 2001 is further analyzed. The strength of this warm, cyclonic eddy decreases with the increase of depth. The eddy shows different horizontal patterns at different layers, and the SLA and temperature fields agree with each other well. This study suggests that this kind of data assimilation is economic and reliable for eddy reconstruction, and can be used as a promising technique in further studies of ocean eddies as well as other fine circulation structures.