Intercomparisons of SMS Wind Sets: A Study Using Rapid-Scan Imagery

Abstract

In this study, the effect of variations in the time and space resolutions of satellite images on satellite-derived drift winds is examined. Rapid-scan satellite data are used as a basis for computing cloud-tracked wind fields over the south central United States on three synoptically different days in 1978. Winds from 30, 15, 6 and 3 min intervals are produced, discussed and compared. The latter includes examination of flow patterns, comparisons of mean and standard deviation values for speed and direction, and a discussion of operator observations. Low-level clouds are tracked on three separate days, the cirrus level is tracked on two separate days. Nearly 10 times as many low cloud winds were computed on days of rapid cloud growth and dissipation using a 3 min interval rather than the conventional 30 min interval; spatial coverage of vectors was also significantly greater. The difference in the number of cloud winds was less pronounced on days with longer cloud lifetimes. Cirrus clouds, with lifetimes much greater than those of low-level cumuli, could be tracked more easily with a 30 min interval which also produced a smoother wind field. Whether tracking cumulus or cirrus, viewing a 30 min interval sequence, as well as examining the synoptics to get a “feel” for the situation, always proved valuable before any tracking was performed. Short-interval sequences caught the mesoscale features which the longer intervals missed and consequently the resulting wind fields often differed. Much more operator time was needed for tracking on the rapid scan images, however. It is recommended that the conventional 30 min interval with single-pixel tracking methods be used on cirrus clouds, while a shorter interval (6–10 min) be used on low-level clouds.