Abstract:
Repeated Very Long Baseline Interferometry, VLBI, measurements exhibitautocorrelation before the baseline rates are estimated and removed from the data.Autocorrelations reduce markedly once the baseline rates are estimated. This raises thepossibility that they may adversely influence the estimated baseline rates if the measurements arein fact autocorrelated but not modeled. We modeled and incorporated autoregressivedisturbances into the estimation of baseline rates using an iterative procedure to investigate theirpresence as well as to assess their impact on the estimated baseline rates and on thecorresponding statistics. Our findings indicate that, overall, the estimated baseline rates, theircorresponding standard deviations (at one sigma level, based on the a posteriori varianceof unit weight for each baseline model) are not markedly influenced by the introduction ofautocorrelated disturbances. On the other hand, the estimated autocorrelation functions of theresiduals show significantly higher order correlations that are due to stochastic and deterministiccyclic effects in the baseline measurements. Corellograms also have signatures of a slippageeffect on all baselines that involve HRAS station.