To Abstract List
Most current realistic applications of ocean data assimilation focus on approximations to the error covariance statistics in order to achieve computational viability. Monte Carlo methods provide one possible way to evolve the forecast error statistics in a realistic manner while limiting the computational burden. Evensen (MWR, 1996) showed the viability of this approach for assimilating GEOSAT data into a two-layer QG model of small ocean region in the Agulhas Current. This paper explores the efficacy of this approach for assimilating surface altimeter data from TOPEX/Poseidon into a mesoscale QG model of the North Pacific Ocean circulation. The assimilation of the larger spatial scales observed by TOPEX into the mesoscale model is effected through a multi-resolution analysis, further enhancing the computational viability of this approach. The size of the ensemble is contained by singular-value decomposition. Identical twin experiments show that the assimilation of the large scale data can reduce the uncertainty in the surface variability by about 50% in the ocean interior and by at least 20% in the high eddying regions of the western boundary currents. The basin-wide rms error at 1000 m is reduced by about 33% after 5 assimilation cycles.
To Pacific Workshop
To WOCE Home Page