We present the second realization of the International Celestial Reference Frame (ICRF2) at radio wavelengths using nearly 30 years of Very Long Baseline Interferometry observations. ICRF2 contains precise positions of 3414 compact radio astronomical objects and has a positional noise floor of ∼40 μas and a directional stability of the frame axes of ∼10 μas. A set of 295 new "defining" sources was selected on the basis of positional stability and the lack of extensive intrinsic source structure. The positional stability of these 295 defining sources and their more uniform sky distribution eliminates the two greatest weaknesses of the first realization of the International Celestial Reference Frame (ICRF1). Alignment of ICRF2 with the International Celestial Reference System was made using 138 positionally stable sources common to both ICRF2 and ICRF1. The resulting ICRF2 was adopted by the International Astronomical Union as the new fundamental celestial reference frame, replacing ICRF1 as of 2010 January 1.

No abstract

No abstract

No abstract

Time-series of zenith wet and total troposphere delays as well as north and east gradients are compared, and zenith total delays (ZT D) are combined on the level of parameter estimates. Input data sets are provided by ten Analysis Centers (ACs) of the International VLBI Service for Geodesy and Astrometry (IVS) for the CONT08 campaign (2008 August 12-26). The inconsistent usage of meteorological data and models, such as mapping functions, causes systematics among the ACs, and differing parameterizations and constraints add noise to the troposphere parameter estimates. The empirical standard deviation of ZT D among the ACs with regard to an unweighted mean is 4.6 mm. The ratio of the analysis noise to the observation noise assessed by the operator/software impact (OSI) model is about 2.5. These and other effects have to be accounted for in order to improve the intratechnique combination of VLBI-derived troposphere parameters. While the largest systematics caused by inconsistent usage of meteorological data can be avoided and the application of different mapping functions can be considered by applying empirical corrections, the noise has to be modeled in the stochastic model of intra-technique combination. The application of different stochastic models shows no significant effects on the combined parameters but results in different mean formal errors: the mean formal errors of the combined ZT D are 2.3 mm (unweighted), 4.4 mm (diagonal), 8.6 mm (variance component estimation), and 8.6 mm (operator/software impact, OSI). On the one hand, the OSI model, i.e. the inclusion of off-diagonal elements in the cofactor-matrix, considers the reapplication of observations yielding a factor of about two for mean formal errors compared to the diagonal approach. On the other hand, the combination based on variance component estimation shows large differences among the variance components and exhibits a comparable scaling of formal errors. Thus for the combination of troposphere parameters a combination of the two extensions of the stochastic model is recommended.