Reliability control
Since the global solutions are continuously updated, depending on the new availability of permanent sites, we decided to crosscheck their stability concerning the area under investigation.
Therefore, we performed 3σ and 2σ tests between ITRF97 and ITRF2000. We particularly computed the residuals between corresponding velocity components, the overall RMS (1s) for Vn, Ve, and finally we tested these residuals at two significance level (3σ and 2σ) in order to identify possible outliers, i.e., inconsistencies between the two global solutions.
Moreover, we made the same tests between each ITRF and the solution provided by McClusky et al. (2000) for the common sites, after having estimated and removed the velocity drift between the different reference frames. It is useful to recall that McClusky et al. (2000) solution is mainly based on non-permanent stations and was adopted as geodetic background for the most recent geodynamic interpretations of the Aegean Sea and Anatolian area.
The relevant results of the statistical solution comparisons are the following:
3σ sign. level for all comparisons: all the sites pass the test, except YOZG (ITRF2000)
2σ sign. level ITRF97-ITRF2000 comparison: NSSP, YOZG, ASKI and TELA do not pass the test (mean values σVn=0.9 cm, σVe=1.0 cm)
This means that velocities change more than 2σ and may be interpreted as an improvement in velocity estimate for GPS site (NSSP and TELA, more observations available in 2000) and in low reliability of some SLR mobile sites (YOZG, ASKI)
2σ sign. level ITRF97-McClusky comparison: NSSP, YOZG, MELE, YILG, HELW and TELA do not pass the test (mean values σVn=0.3 cm, σVe=0.6 cm)
2σ sign. level ITRF2000-McClusky: YOZG, TRAB, MELE, YILG, DYON, ASKI, ROUM and HELW do not pass the test (mean values σVn=0.8 cm, σVe=0.7 cm)
It should be noted that the estimated agreement between ITRF solutions and McClusky et al. (2000) solution is remarkably poorer than the claimed precision of the McClusky et al. (2000) solution velocity, which, on the contrary, seems to be too optimistic in spite of the non-permanent data set on which they are based.
The illustrated comparisons show a good coherence between the ITRF solutions at centimetric level, also at the more severe 2σ sign. level, since only 4 sites exhibit high residuals; on the contrary, the agreement with the McClusky et al. (2000) solution seems to be more problematic.
Therefore, in our opinion, the performed assessment of the available geodetic solutions for the Anatolia region and its surroundings suggests: to base geodynamical investigation on the IERS solutions at present; to account the average precision of these solutions, which is at centimetric level.