A 10 year intercomparison between collocated Special Sensor Microwave Imager oceanic surface wind speed retrievals and global analyses

To evaluate the scalar ocean surface wind speeds obtained from the Special Sensor Microwave Imager (SSM/I), we compare them over the time period from July 1987 through December 1997 with those from two global analyses: the National Center for Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR) Annual Reanlysis and the European Center for Medium-Range Weather Forecasts (ECMWF)/Tropical Ocean-Global Atmosphere Global Surface Analysis. We perform a statistical analysis for the whole globe and present time series analyses for selected geographical regions in connection with collocated wind speed difference maps. In order to evaluate further geographical biases observed in the SSM/I versus analyses comparisons we use wind speeds from the NASA scatterometer (NSCAT) for the 10 month period from September 1996 through June 1997 as a third data source. The value of the standard deviation for all collocated SSM/I - ECMWF wind speed differences is 2.1 m s(-1) and for all collocated SSM/I - NCEP/NCAR reanalyis wind speed differences is 2.4 m s(-1). When taking monthly or yearly averages in each pixel, which has the effect of cancelling our small timescale wind speed fluctuations, the values are between 0.8 and 1.2 m s(-1) respectively. Global biases range between -0.05 and +0.55 m s(-1) for the various SSM/I satellites. Our analysis allows us to identify regional biases for both the SSM/I and analyses winds. The NCEP/NCAR reanalysis wind speeds appear underestimated in the tropical Pacific and tropical Atlantic. ECMWF wind speeds appear underestimated near the southern Pacific islands NE of Australia. The analyses wind speeds are higher than the SSM/I wind speeds near the Argentinean coast. The SSM/I wind speeds appear high in the extratropical central and eastern Pacific and low in certain coastal regions with eastern boundary currents and in the Arabian Sea. The size of some of these biases are seasonally dependent.