This paper introduces a new technique in the automatic detection of storm sudden commencement (SC) using the discrete wavelet transform (DWT). A geomagnetic storm is a global simultaneous phenomenon affecting the whole Earth, which means that all ground magnetometers running online will record this event. An algorithm using different characteristic features of the SC is proposed. The selection of an optimal threshold for feature parameters is critical for the success of SC automatic detection. Therefore, this paper uses particle swarm optimization (PSO) to determine the optimal feature threshold values. The developed algorithm is based on data records from a network of ground magnetometers. This algorithm is implemented via multi-resolution analysis (MRA) of the DWT using the Haar wavelet filter. Four-year data sampled at one sample/s from six ground stations from low to high latitudes were analyzed to develop and test this technique. Data representing 450 days from five stations operating simultaneously are available. The confusion matrix of all possible outcomes shows that the accuracy of the proposed algorithm is 97.33%.
All Science Journal Classification (ASJC) codes