The Kp index nowcast method based on neural network

The planetary three-hour magnetic condition index Kp, which widely used in space physics research and space weather services, is a versatile geomagnetic index that reflects the global geomagnetic disturbance level. But the release of the official Kp index has been delayed for two weeks, making it impossible to use it directly for space weather services. Therefore, the high-precision nowcast of the Kp index is an urgent problem to be solved.The Kp index is constructed from the three-hour magnetic condition index K. This paper first proposes a K index nowcast method to solve the problem of insufficient accuracy, which can meet the real-time requirement under the condition of ensuring K index high-precision. On this basis, the 20 parameters related to the Kp index are determined because of the analysis about the factors which influence the Kp index, and these parameters are used as inputs to construct a neural network to nowcast the Kp index. Finally, using the eleven-year data to verify the Kp index nowcast method proposed in this paper, the verification results show that the proposed Kp index nowcast method can meet the real-time and accuracy requirements about the Kp index.