Severe Hail Identification Model Based On Saliency Characteristics

There are always high false alarm ratios when warning against the severe hailwith the Severe Hail Index (SHI) which is supplied by digital weather radar system.To solve this problem, several novel features, such as “Overhang”, extractionalgorithm are designed and realized, and such ability is attempted to give thesefeatures that they can describe the severe hail conceptual model from different aspectsseparately. Then we take short-time heavy rainfall cells which are easy to be confusedwith severe hail cells as counter examples to do statistic analysis for these featuresand the SHI. Test results show that they have more significant difference between twokinds of samples and hence every of them can reflect one aspect’s characteristic ofsevere hail cells. Then a severe hail recognition model that is the Support VectorMachine with radial primary kernel function is learned. Lastly, the normalizeddistance between the sample to be recognized and the optimal separating hyper-planeis defined as a new SHI to warning against the severe hail. Experiment results showthat the method proposed in this paper makes severe hail hit ratio higher than the SHIof being used and the false alarm ratio is reduced substantially.