Severe Storm Cell Classification Using Support Vector Machines and Radial Basis Function Approaches

Meteorological volumetric data are used to detect thunderstorms that are the cause of most of the summer severe weathers. There are systems that may convert the volumetric data into a set of derived products. Based on these derived features, this work compares three classifiers to determine which approach will best classify a storm cell data set coming from Environment Canada. The criterion for comparison is the accuracy in the classification over a testing set. The three approaches compared are the support vector machine (SVM) classifier, with radial basis function (RBF) kernel; the classic RBF classifier, with the centres found using the orthogonal least squares approach; and the hybrid RBF, with the centres corresponding to the support vectors found using the SVM approach. The results show that the SVM approach is the best of these approaches, in terms of accuracy, for the storm cell classification.