Book Description
A spiral analysis technique is developed which quantifies the spiral patterns of cloud bands observed in IR satellite images of tropical cyclones. The technique utilizes the Navy's SPADS (Satellite Data Processing and Display System) minicomputer for processing the digital IR data. The technique consists of best fitting spherical-logarithmic spirals to tropical cyclone spiral cloud bands and performing multiple Fourier analyses of the radiance field along spirals orthogonal to the bands. Linear regression techniques are used to establish a relationship between spiral parameters derived from the Fourier analyses and tropical cyclone intensity. Algorithms for current intensity, and 12 and 24 h forecasts are developed. Tests on an independent data set show significant skill in estimating current intensity and in making 12 h forecasts. The most important predictors selected are presistence and parameters related to the dominant spiral signal and maximum IR count of the analyzed radiance field. Biases in the forecast algorithms suggest that other parameters are necessary to more accurately predict tropical cyclone intensity. The results, however, demonstrate the usefulness of the technique as an aid to tropical cyclone forecasters.