Comparison of Different Estimates of the Accuracy of Forecasts of the Earth's Rotation Parameters

Improvement of the prediction accuracy of the Earth's rotation parameters (ERP) is one of the main problems of applied astrometry. In order to solve this problem, various approaches are used and in order to select the best one, comparison of the accuracy of the forecasts obtained by different methods at different centers of analysis are often carried out. In such comparisons, various statistical estimates of the forecast errors are used, based on the analysis of the differences between the predicted and final values of ERP. In this paper, we compare several estimates, such as root mean square error, mean error, median error, and maximum error. It is shown that a direct relationship between the estimates of the forecast accuracy obtained by these methods does not always exist. Therefore, in order to obtain the most informative results of comparison of the accuracy of different forecast methods, it is recommended to use the studies that compare the series of ERP forecasts, especially short-term ones, several estimates together.