Comparison of polar motion prediction results supplied by the IERSsub-bureau for rapid service and predictions and results of otherprediction methods

In this paper, four different methods for the prediction of x, y pole coordinates are investigated. We examined the accuracies of the autocovariance (AC), least-squares extrapolation (LS), a least-squares extrapolation and autoregressive combination (LS+AR), and a least-squares extrapolation and neural networks combination (LS+NN) in predicting pole position. The most accurate prediction is the combination of the least-squares extrapolation and the autoregressive prediction of the least-squares extrapolation residuals applied to the complex-valued pole coordinates data. The problem of any prediction method of pole coordinates data in the polar coordinate system is a significant prediction error of the integrated angular velocity.