Evaluation of the ICRF stability from position time series analysis
The celestial reference frame is realized by absolute positions of extragalactic sources that are assumed to be fixed in the space. The fixing of the axes is one of the crucial points for the International Celestial Reference System (ICRS) concept. However, due to various effects such as its intrinsic activity, the apparent position of the extragalactic sources may vary with time, resulting in a time-dependent deviation of the frame axes that are defined by the positions of these sources. We aim to evaluate the axis stability of the third realization of the International Celestial Reference Frame (ICRF3). We first derive the extragalactic source position time series from observations of very long baseline interferometry (VLBI) at the dual $S/X$-band (2.3/8.4~GHz) between August 1979 and December 2020. We measured the stability of the ICRF3 axes in terms of the drift and scatter around the mean: (i) we estimate the global spin of the ICRF3 axes based on the apparent proper motion (slope of the position time series) of the ICRF3 defining sources; (ii) we also construct the yearly representations of the ICRF3 through annually averaged positions of the ICRF3 defining sources and estimate the dispersion in the axis orientation of these yearly frames. The global spin is no higher than $\mathrm{0.8\,μas\,yr^{-1}}$ for each ICRF3 axis with an uncertainty of $\mathrm{0.3\,μas\,yr^{-1}}$, corresponding to an accumulated deformation smaller than $\mathrm{30\,μas}$ for the celestial frame axes during 1979.6--2021.0. The axis orientation of the yearly celestial frame becomes more stable as time elapses, with a standard deviation of 10--20$\mathrm{\,μas}$ for each axis. The axes of the ICRF3 are stable at approximately 10--20~$\mathrm{μas}$ from 1979.6--2021.0 and the axis stability does not degrade after the adoption of the ICRF3.