Long polar motion series: Facts and insights

Abstract

Polar motion (PM) series from various observation methods and sources are analyzed using one and the same simple analysis tool. The maximum time resolution of the Earth Rotation Parameter (ERP) series considered here is one day, which leaves out ERP series with sub-daily time resolution. A longer spacing between subsequent epochs is permissible, a variable spacing between epochs like in the VLBI (Very Long Baseline Interferometry) series as well. The shortest length of the PM time series analyzed, based on GNSS (Global Navigation Satellite Systems), is 26+ years, the longest one, the IERS C01 series, 173+ years.Following the tradition, PM is decomposed into a mean motion, i.e., the motion of the center of the PM curve on the surface of the Earth, a prograde Chandler motion with a nominal period of 432.25 days, and a prograde annual motion with a nominal period of 365.25 days. In our analysis, Secular PM is not assumed as linear, but as piecewise linear and continuous. The defining parameters of the periodic parts of polar motion are not assumed as constant either, but as piecewise linear and continuous functions of time as well. All parameters of the PM model based on one particular ERP series are determined in one and the same linear least squares estimation. Parameters of the same type may be purely deterministic or filtered (smoothed) by constraining the second time differences of the parameters (the second time derivative of the parameter functions) of the same kind.The key results achieved by applying our method to a wide variety of ERP series are visualized and discussed. They include: (a) the description of a sharp bend of mean PM around 1996, (b) strong evidence for a multi-component model for Chandler motion with periods separated by few days, similar to that proposed by Chandler (1901), (c) a reconstruction of mean PM from 1846 to present taking into account the composite nature of the series, and (d) a prediction of Chandler motion for the next decades.A sharp bend of mean PM was found around 1995/96, along the meridian at about 70° West to the meridian at about 12–15° West. All PM curves analyzed show a mean motion in this new direction after 1995/95. Smaller excursions w.r.t.d this mean exist, with periods <10 years. A periodic variation with a period of about 180 years was observed in the Chandler signal of PM with an amplitude of more than 200 mas. The variation may be explained approximately by a composite nature of Chandler motion, similar to the model already proposed by Chandler (1901). This simple model was generalized using a Fourier series separated in period by few days. This alternative model of Chandler motion is used to predict the Chandler part of PM, with an amplitude close to zero currently, to reach values of the order of 100 mas around 2028–2030.