The feasibility of using satellites for providing surface winds or wind stress data was explored. Three popular methods were compared using nearly colocated data to assess the accuracies of each and the coverage that each could provide. The three methods tested were 1) the use of the sun glitter reflection seen on visible images of the ocean surface; 2) the use of active microwave sensors (flown on SEASAT) which reflect microwaves off the ocean surface; and 3) the use of cloud motions as indicators of the surface winds.
Close agreement in wind speed estimates was found among the three methods. The biases were <0.6 m s−1 for comparisons between comparable methods of estimating surface winds (1 and 2). Cloud motion comparisons to the other methods exhibited biases of <3.0 m s−1. Individual point-by-point comparisons between wind measurements had an average scatter of 2.0 m s−1 (rms) or less after the mean biases were removed. Atmospheric variability caused as many of the differences as the instrumental errors indicating that meaningful wind information could be obtained from all three methods.
Very detailed spacial coverage was obtained with the sun-glitter method for wind speeds. However, the coverage was restricted to a narrow band 5° of latitude wide in the tropics. SEASAT also provided good coverage for two swaths (4° longitude wide) on each side of the satellite's orbit. Gaps between the swaths and orbits (polar non-synchronous orbits) were left unsampled. Both methods required external data on the wind directions which were obtained from cloud motions. The cloud motions provided coverage over larger areas than the other two methods because of the abundance of low-level cumuli.