Medium-short-term prediction of polar motion combining the differencing between series with the differencing within series

The accuracy of polar motion forecasting has been the focus of attention in the fields of satellite navigation and deep space exploration. However, the traditional or differential methods for forecasting X or Y series based on LS and AR models are straightforward and monolithic, and cannot reduce the range of forecast errors. Therefore, this study proposes a new method (called the between-within, B-W method) that combines the X, Y and Y-X series forecasts of the traditional and differential methods in pairs according to the mathematical relationship of Y-X. This approach is one way to obtain the minimum range of forecast errors by making full use of the advantages of each method in the combination. A total of 262-hindcast experiments were conducted during 2010-2020 with strictly simulated time delays. For forecasts of 1-180 d at the x-pole, the average improvement is 10.7 per cent over Bulletin-A. For the y-pole at 1-90 d an average improvement of 11.7 per cent over Bulletin-A is achieved. In addition, further incorporation of the last 1 d IGS (International Global Navigation Satellite System Service) Ultra-rapid (IGU) data can effectively improve the MAE at 1-10 d. The 2016-2018 performance of the B-W method at the x-pole may be related to the amplitude and phase of the Chandler wobble, and the 2013-2016 performance at the y-pole may be related to El Nino climate change events. In terms of overall stability, the B-W method is superior to the IERS Bulletin-A in the medium-short-term and has potential practical application.