Real time Kp predictions from solar wind data using neural networks

Multilayer feed-forward neural network models are developed to make three-h our predictions of the planetary magnetospheric Kp index. The input paramet ers for the networks are the B-z-component of the interplanetary magnetic f ield, the solar wind density n, and the solar wind velocity V, given as thr ee-hour averages. The networks are trained with the error back-propagation algorithm on data sequences extracted from the 21(st) solar cycle. The resu lt is a hybrid model consisting of two expert networks providing Kp predict ions with an RMS error of 0.96 and a correlation of 0.76 in reference to th e measured Kp values. This result can be compared with the linear correlati on between V(t) and Kp(t + 3 hours) which is 0.47. The hybrid model is test ed on geomagnetic storm events extracted from the 22(nd) solar cycle. The h ybrid model is implemented and real time predictions of the planetary magne tospheric Kp index are available at http://www.astro.lu.se/similar to fredr ikb. (C) 2000 Elsevier Science Ltd. All rights reserved.