Implementation of 3D-VAR and assessment of forecast skill for Regional

Abstract

We applied 3-dimentional variational data assimilation technique (3D-Var) to the MOM3-based regional oceanmodel in the East/Japan Sea (EJS) which is an East Asian marginal sea. Satellite-borne sea surface temperatureand sea surface height, and temperature profiles have been assimilated into the model. The numerical performanceof the regional ocean model with 3D-Var system appears to be efficient enough to be used in an operational oceanforecast system.Model-data comparison shows that the reanalysis produced by 3D-Var system fairly well represents not onlythe mean circulation but also meso-scale eddies in the EJS. The overall root-mean-square error between 100mtemperature fields of reanalysis and observation is 2.1 oC, and the spatially averaged grid-to-grid correlation ishigh with a mean value of 0.79 for the inter-comparison period.For the assessment of forecast skill of the regional ocean model, we compared the measured temperature by thePressure-equipped Inverted Echo Sounder and hindcasted one initialized by means of the 3D-Var system in thesouth-western basin of the EJS. The forecast skill of our regional ocean model will be displayed.