The impact of wind corrections and ocean-current influence on wind stress forcing on the modeling of Pacific North Equatorial Countercurrent

Highlights

• The weak Pacific NECC remains in ocean models even if JRA55-do forcing data is used in OMIP-2.

• The wind correction with QuikSCAT is the main factor affecting the Pacific NECC simulations.

• The surface ocean current is counted-twice incorrectly when calculating the wind stress in ocean models.

Abstract

Bias correction of reanalysis-based wind stress using scatterometer derived equivalent neutral wind has been a common practice in producing the forcing datasets used in recent global ocean model intercomparisons (OMIPs). Here we systematically evaluate the effect of this wind correction procedure on the simulation of the Pacific North Equatorial Countercurrent (NECC) with multiple sets of model experiments. The weak NECC evident in earlier OMIPs employing the Coordinated Ocean-ice Reference Experiments (COREs) forcing dataset persists with the new JRA55-do (Japanese 55-year Reanalysis) forcing dataset. Two factors appear to significantly affect the Pacific NECC in forced ocean simulations: i) the bias correction procedure using QuikSCAT derived winds and ii) whether or not the ocean current is considered in the bulk formula. In the forced ocean simulations, the QuikSCAT correction weakens the averaged NECC transports by about 60%. Taking the ocean currents into account in the bulk formula may weaken the averaged NECC transports by about 26%–30%. Under the current OMIP protocol the above two procedures are used together to force the ocean model resulting in a double-counting of ocean surface current feedback on wind stress because the QuikSCAT estimates the equivalent 10-m neutral winds relative to surface current. We further systematically verify and investigate the impacts of this double-counting of the ocean surface currents on the modeled Pacific NECC using offline linear Sverdrup transport analysis, in which the observational data of vector wind and surface current are used to calculate the surface wind stress. It shows that including the ocean current in the bulk formula may reduce the zonal Sverdrup transport (ZST) by about 6.6 Sv (33%) and the further double-counting of the ocean current leads to an additional reduction of 6.4 Sv (48%). Next, using “perfect model” experiments with output from a coupled ocean–atmosphere model we further identify that the double-counting of current feedback in the bulk formula results in approximately 21% weakened volume transport. The built-in nonlinear processes in the model, such as the advection and friction terms, may partly damp the reduction due to the double-counting of the ocean current. However, the double-counting bias can only explain 26%–30% of the Pacific NECC simulation bias and the other part of the bias, around 30%, caused by the correction with QuikSCAT has not been explained. We speculate that this part may be explained by the retrieval biases in QuikSCAT wind data and the use of annual mean climatological wind adjustment factors.