Trend and abrupt changes in long-term geomagnetic indices

Advanced statistical methods are employed to analyze three long-term time series of geomagnetic activity indices (aa, IHV, and IDV) together with sunspot number (Rz) to examine whether or not the aa index can realistically represent long-term variations of geomagnetic activity. We make use of a decomposition method called STL, which is a time domain filtering procedure that decomposes a time series into trend, cyclic, and residual components using nonparametric regression. A Bayesian change point analysis is also applied to the geomagnetic indices, as well as to sunspot number, to detect abrupt changes that may be caused by either instrumental changes, calibration errors, or sudden changes in solar activity. Our analysis shows that all three long-term geomagnetic indices share a similar centennial-scale variation that resembles the long-term trend of sunspot number Rz. The amplitude ratio between the centennial-scale variation and 11-year cycle of aa and IHV are closely comparable. Overall, our analysis suggests that the majority of the changes in the aa index are controlled by solar activity. Instrumental change or site relocation has only a limited effect on the long-term trend of aa. This is in good agreement with those previous studies which have shown aa to be a reliable long-term index