El Niño Impact on Polar Motion Prediction Errors

The polar motion prediction is computed as a least-squares extrapolation of the polar motion data. The least-squares model consists of a Chandler circle with constant or variable amplitude, annual and semiannual ellipses, and a bias. The model with constant amplitude of the Chandler oscillation is fit to the last three years of polar motion data and the model with variable amplitude of the Chandler oscillation is fit to the whole time series ranging from 1973.0 to 2001.1. The variable amplitude of the Chandler oscillation is modeled from the envelope of the Chandler oscillation filtered by the Fourier transform band pass filter from the long-term IERS EOPC01 polar motion series. The accuracy of the polar motion prediction depends mostly on the phase variation of the annual oscillation, which is treated as a constant in the least-squares adjustment. There were two significant changes of the annual oscillation phase of the order of 30° before the two El Niño events in 1982/83 and 1997/98.