Autoregressive Modeling of Geomagnetic Data.(Dept.E)

Geomagnetic (GM) data generally appear to have significant organization or structure. We attempted to determine if GM records could be modeled as an autoregressive process with a white noise excitation during the short-time period of one month. The autoregressive (ER) model is then used to synthesize GM signals using AR coefficients and a white noise excitation with the same propability distribution function as those determined from the autoregressive model of an ensemble of monthly records. Both the original and synthesized monthly records are then compared using the root-mean-square (rms) amplitude, the number of zero crossing per month. the number of peaks, and the amplitude distributions of the signals. The results of examining the synthesized GM records indicate that there are no significant differences in the values of these parameters. The use of such synthesized GM records may allow more through testing of forecasting algorithms than is possible with the present limited number of GM records.