Statistical characterization of zonal and meridional ocean wind stress

Four years of ocean vector wind data are used to evaluate statistics of wind stress over the ocean. Raw swath wind stresses derived from the Quick Scatterometer (QuikSCAT) are compared with five different global gridded wind products, including products based on scatterometer observations, meteorological analysis winds from the European Centre for Medium-Range Weather Forecasts, and reanalysis winds from the National Centers for Environmental Prediction. Buoy winds from a limited number of sites in the Pacific Ocean are also considered. Probability density functions (PDFs) computed for latitudinal bands show that mean wind stresses for the six global products are largely in agreement, while variances differ substantially, by a factor of 2 or more, with swath wind stresses indicating highest variances for meridional winds and for zonal winds outside the Tropics. Higher moments of the PDFs also differ. Kurtoses are large for all wind products, implying that PDFs are not Gaussian. None of the available gridded products fully captures the range of extreme wind events seen in the raw swath data. Frequency spectra for the five gridded products agree with frequency spectra from swath data at low frequencies, but spectral slopes differ at higher frequencies, particularly for frequencies greater than 100 cycles per year (cpy), which are poorly resolved by a single scatterometer. In the frequency range between 10 and 90 cpy that is resolved by the scatterometer, spectra derived from swath data are flatter than spectra from gridded products and are judged to be flatter than omega(-2/3) at all latitudes.