Nonlinearity in Kp and its implication to Kp forecasting

Kp index has been used as a proxy for global geomagnetic activity level. Cumulant based analysis of Kp from 1932 to 2003 reveals linear and nonlinear correlations of Kp(t) and Kp(t - tau). The linear correlation peaks at tau = 0 hour and drops off rapidly after a few hours, suggesting that the persistence in Kp is in the order of a few hours. While the linear response is roughly the same for solar minimum and solar maximum, there is sharp difference in the nonlinear response. Near solar maximum, the nonlinear correlation follows closely the behavior of the linear correlation. In contrast, near solar min, the nonlinear correlation has a peak at tau = 0 hour and a secondary peak around tau = 40 hours. In fact, the nonlinear response appears to roughly anti-correlate with the sun spot number in every solar cycle since 1932. Several solar wind driven Kp forecast models were evaluated with data that spans over two solar cycles, 1975 - 2001. The evaluations show that the models predict Kp more accurately during solar maximum than solar minimum. All together, the results suggest that geospace is more dominated by internal dynamics during solar minimum than solar maximum, when it is more directly driven by external inputs, namely solar wind and IMF.