Can we improve the representation of modeled ocean mixed layer by assimilating surface-only satellite-derived data? A case study for the tropical Pacific during the 1997-1998 El Niño

Abstract

In the tropical oceans the density mixed layer does influence the ocean-atmosphere interactions, a feature that is fundamental to the development of energetic climatic events such as the El Niño-Southern Oscillation (ENSO) phenomenon. The aim of this work is to take advantage of existing sea surface temperature (SST) and velocity together with future sea surface salinity (SSS) satellite-derived data by assimilating these data in a primitive equation model in order to improve the modeled mixed layer. As satellite SSS data are not yet available, and to better analyze how assimilation works, we performed twin experiments in a context that renders assimilation conclusive with regard to real experiment. Two simulations using different forcing were used for the experiments. They have errors that are comparable to the ones between any simulation and real observations. The assimilation scheme is an adaptive version of the SEEK filter [Pham et al., 1998] readily usable for assimilating real data. An assimilation experiment was conducted covering the 1997-1998 ENSO event to analyze the main characteristics of the actual mixed layer. Looking in particular at the relevance of SSS, SST, mixed layer depth, and barrier layer thickness, satellite-derived data prove to be useful to better simulate the oceanic mixed layer. Velocity data are specially needed to control the zonal equatorial current. Interestingly, assimilation of surface-only data still worked well below the mixed layer, and some improvements were detectable in terms of barrier layer thickness, even though the limit of the assimilation scheme was reached.