[G01-2] Celestial reference frame and VLBI

[G01-2-03] Correlated atmosphere noise in VLBI analysis

^○Hana Krasna¹, John Gipson² (1.TU Wien, Vienna, Austria, 2.NVI, Inc., Greenbelt, USA)

The standard assumption in the routine VLBI analysis is that the observations are station and time independent which manifests itself in a diagonal observation covariance matrix. But this simplification causes a mis-characterisation of the measured group delays leading to incorrect estimation of parameters and too optimistic formal errors. In the first step we compare the estimated baseline length scatter from CONTinuous VLBI campaigns using two ways of reweighting obsevations, i.e. adding baseline dependent and elevation dependent noise. In the second step we introduce correlations into the observation covariance matrix focusing on mis-modeling of the atmosphere and taking into acount correlations between observations at a common time. We demonstate that this reduces the baseline length scatter, indicating that the results are more consistent day-to-day. We also show that introducing the correlations improves the agreement between VLBI and GPS measured polar motion.