A multivariate intercomparison between three oceanic GCMs using observed current and thermocline depth anomalies in the tropical Pacific during 1985–1992

The thermocline depth anomalies during 1985–1992 and the 15-m current anomalies during 1988–1992 simulated by three oceanic general circulation models (OGCMs) of the tropical Pacific forced by different surface wind stress fields are compared to observations using a multivariate model testing procedure. The Hadley Centre model, OPA from LODYC, Paris, and HOPE from MPI, Hamburg, were forced by wind stress fields derived from the pseudo-wind stress produced from surface marine reports at the Florida State University (FSU) or from the response of atmospheric general circulation models (AGCMs) (ARPEGE, ECHAM3 and the Hadley Centre model) to the observed sea surface temperature (SST). It is found that the three oceanic models simulate the thermocline depth anomalies rather well, although large discrepancies with the observations remain, in particular, in the eastern part of the Pacific, and that they simulate them better than the surface current anomalies, without clear correspondence in model skill between the two variables. The simulated thermocline depth anomalies are also less sensitive to the choice of the wind stress. In general, the FSU wind stress leads to reliable simulations for both variables, while a given atmospheric model field may lead to a good simulation with one ocean model and to a bad one with another. Overall, the analysis suggests that the model parameters, which optimize a variable using a particular forcing, are not necessarily appropriate for another variable or another forcing, which suggests that some of the physics are not correctly represented in the oceanic models.