Using QuikSCAT-derived surface winds and a GCM to improve predicted wind speed variability and ocean surface fluxes

Surface momentum, heat, and gas fluxes depend non-linearly on wind speed so that the wind speed distribution tails disproportionately influence these fluxes. Scatterometer measurements offer the opportunity to characterize wind speed probability density functions (PDFs) across climatically significant spatial and temporal scales. Studying the air-sea interaction driven by these PDFs in general circulation models (GCMs) allows us to assess the magnitude of sub-gridscale air-sea interactions, and their non-linear feedbacks, on global scales. We characterize and intercompare ocean surface wind PDFs derived from QuikSCAT and TAO/Triton measurements with GCM (NCAR CCSM) predictions for the period 2000-2006. Our results show how sub-gridscale wind PDFs contribute small but significant fluxes which feedback onto the mean climate state.

We find extensive and significant positive mean wind speed biases of 1.0 m/s in the northern hemisphere (NH) trade and southern hemisphere storm track regions. These biases are largely due to non-linear contributions to surface drag by gusty winds that are seen in all observations but neglected (as sub-gridscale) in the GCM. Seasonally-persistent negative shape and mean wind speed biases were found along the ITCZ. Weibull shape parameter biases as large as 4.0 were reported within the NH and southern Indian Ocean trades.

The observed spatial and temporal wind speed PDFs arise from sources of variability including mechanical shear, buoyancy-induced turbulence, and convective downdrafts. We have parameterized these processes into a new GCM representation of sub-gridscale wind variability. Sensitivity tests show that representing the sub-gridscale wind PDFs reduce ocean surface wind biases (by about 1.0 m/s) and corresponding biases in air-sea exchange.