Classification of Volumetric Storm Cell Patterns

Abstract

Meteorological volumetric radar data may be used to detect thunderstorms, storm events responsible for nearly all severe summer weather Discriminating between different types of thunderstorms is a difficult problem due to the high dimensionality of the data, and the imprecision and incompleteness of the data. Several artificial neural network and fuzzy set theoretic classification strategies and preprocessing techniques are tested to determine their usefulness in the discrimination of four types of storm events: tornado, wind, hail and heavy rain.