Evaluation of relative performance of QuikSCAT and NCEP re-analysis winds through simulations by an OGCM

The response of an ocean general circulation model (OGCM) to two different wind products, viz., NCEP/NCAR reanalysis and QuikSCAT scatterometer, was examined. OGCM-simulated thermodynamic variables from the two simulations, hereafter referred to as NCEP-R (NCEP/NCAR wind forced) and QS-R (QuikSCAT wind forced) were intercompared and also were compared against observations for a period of 3 years (2000-2002). In the tropical Indian Ocean (10), the sea-level anomaly (SLA) simulated by QS-R has less root mean square error (RMSE) and higher correlation with respect to TOPEX/Poseidon SLA observations than SLA simulated by NCEP-R. Intraseasonal variability of currents observed by TRITON buoy in the 10 was closely captured by QS-R, although the magnitudes are somewhat underestimated. Surface currents simulated by QS-R have less RMSE than those simulated by NCEP-R in the Pacific. However, the sub-surface currents are much weaker in magnitude in both the solutions, possibly because of deficiencies in the diffusion and viscosity parameterization. Sea-surface temperature (SST) simulated by QS-R has a cooler bias. The RMSE of SST simulated by NCEP-R. is less than the RMSE of SST simulated by QS-R, with the latter capturing the variabilities more realistically. The large differences between SST simulated by QS-R and observations could be partly due to physical inconsistency between the momentum and heat fluxes. Scatterometer-forced model simulations of 20 degrees C thermocline depths (D20) are in better agreement with in situ-derived D20 than the D20 simulated by NCEP-R. Variations in the mixed layer depth at the TRITON buoy are better captured by QS-R than by NCEP-R. Speed of Kelvin and Rossby waves and the strength of upwelling/downwelling features in the 10 are closer to observations in QS-R than in NCEP-R simulations. (c) 2007 Elsevier Ltd. All rights reserved.