The Benefits of Using Short-Interval Satellite Images to Derive Winds for Tropical Cyclones

Abstract

During the 1975, 1976 and 1977 North Atlantic hurricane seasons, NOAA's National Environmental Satellite Service (NESS) and NASA's Goddard Space Flight Center (GSFC) conducted a cooperative program to determine the best resolution and frequency now available from satellite images for deriving winds to study and forecast tropical cyclones. Rapid-scan images were obtained in 1975 at 7.5 min interval from SMS 2 for Hurricane Eloise on 22 September and of tropical cyclone Caroline on 28, 29 and 30 August; in 1976 at 3 min intervals from GOES 1 for tropical storms Belle on 5 August and Holly on 25 October; and in 1977 at 3 min intervals Corn GOES 1 for tropical cyclone Anita on 30 and 31 August and 1 September. Cloud motions were derived from these images using the Atmospheric and Oceanographic Information Processing System (AOIPS) at GSFC. Winds that were derived from the movement of upper (∼200 mb) and lower tropospheric (∼900 mb) level clouds using rapid scan data were compared with the 15 and 30 min interval data. This was done using visible images having 1, 2, 4 and 8 km resolution for the areas within 650 km of the storm center for the 1975 and 1976 tropical cyclones. Greater than 10 (5) times as many clouds could be tracked to obtain winds at both levels using 3 and 7.5 min rapid-scan images as when using 30 min (15 min) interval images. In addition, by using the frequent images, it was possible to track a few bright areas within the central dense overcast which appeared to be moving with the winds at low levels. For Hurricanes Eloise and Caroline the winds that were derived by tracking these bright areas within the central dense overcast had speeds differing in the mean by only 2.5 m s −1 from the wind speed measured by aircraft flying at ∼0.5 km above the surface in the same quadrant 4 h later. Full-resolution visible images (1 km) were needed to track slow moving low-level cloud elements, since on a degraded resolution image, subpixel movement would introduce additive inaccuracies to the wind measurements. Rapid-scan full-resolution GOES 1 data for tropical cyclone Anita (1977) provided representative wind fields only outside the central dense overcast at the lower tropospheric level. For this area aircraft-measured wind speeds differed in the mean again by only 2.5 m s−1.