Adjusting Internal Model Errors through Ocean State Estimation

Abstract Oceanic state estimation is a powerful tool to estimate internal model parameters simultaneously with the model’s initial conditions and surface forcing field that jointly would bring a model into consistency with time-varying large-scale ocean observations. Here an attempt to estimate geographically varying fields of horizontal and vertical viscosity and diffusivity within a 9-yr-long estimation procedure is presented. The estimated coefficients are highly efficient in preserving watermass characteristics and frontal structures by reducing the model temperature and salinity drift, especially around the Southern Ocean. The estimated mean circulation results in stronger transports of western boundary currents and of the Antarctic Circumpolar Current. Moreover, an increase of about 10% in the strength of the meridional overturning circulation and in the poleward heat transport can be found. Estimated changes in the horizontal mixing coefficients seem to agree with the notion that diapycnal mixing is superfically high with Laplacian mixing formulations, especially close to frontal structures in the ocean. In comparison with adjustments in tracer diffusivities (vertically and horizontally), adjustments of viscosity coefficients are fairly minor outside lateral boundary regions, suggesting that state estimation attempts might be most successful in providing enhanced insight into tracer mixing.