Wind Sets from SMS Images: An Assessment of Quality for GATE

Abstract

In this study we explore the accuracy, representativeness and reproducibility of tracer winds in the area of the 1974 GARP Atlantic Tropical Experiment (GATE). These winds were generated by tracking clouds in Synchronous Meteorological Satellite (SMS) images displayed on the University of Wisconsin's Man-Computer Interactive Data Access System (McIDAS). Two questions are addressed: 1) How accurately can the cloud displacements be measured? and 2) To what extent do the cloud displacements represent the wind field? Accuracy is evaluated in terms of data characteristics, McIDAS precision and consistency. We find that for full-resolution visible data neither navigation nor resolution errors significantly affect the tracking of clouds. An examination of consistency, defined as similarity of wind sets independently produced by several scientists tracking clouds from the same set of images, yields an rms reproducibility of 2 m s?1 for cirrus level and 1.3 m s?1 for cumulus level winds. This is smaller than the ?random? error generally attributed to cloud winds. In addition, the vorticity and divergence fields are qualitatively reproducible. The discussion of representativeness centers about cloud height determination, and relating cloud motion to winds. Representativeness is examined through 1) the internal consistency of consecutive sets; 2) the consistency of the cloud wind field, including divergence and vorticity with such features as clusters, vortices, and clear areas; and 3) the difference between proximate satellite and ship winds. These differences were all under 3 m s?1, which is close to the noise level of ship winds and better than radiosonde-radiosonde comparison. We conclude that the representativeness of cloud tracers to cumulus and cirrus level flow is good to within the accuracy of currently available ground truth data.