Detection of strong convective weather based on manifold learning

With the rapid development of radar technology, the radar data people can access is growing exponentially. For a mass of high-dimensional data, it is necessary to reduce the data dimension while maintaining the data information in order to minimize the impact of the dimension disasters. The detection method of the strong convective weather (hailstone and rainstorm) is based on manifold learning algorithm in this paper. Firstly the dimension of 22-dimensional features of the strong convective weather is reduced by manifold learning algorithm-Local Tangent Space Alignment, then in low-dimensional (8-dimensional) data space the useful and hidden rules for the detection of strong convective weather is dig out, finally the effective rules are obtained to detect the strong convective weather. Compared with the non-dimensionality reduction method, the proposed method improves the detection accuracy and reduces the time complexity through experimental test.