Today it is common practice to derive terrestrial reference frames (TRF) by the combined analysis of the following space-geodetic techniques: VLBI, GPS, SLR, DORIS. Various validations and cross-check calculations show a high consistency in the range of a few millimetres for recent solutions of the International Terrestrial Reference Frame (ITRF). However, several factors limit the quality of intra-technique and inter-technique combinations such as the use of different parameterizations, different correction models for the same physical effects, individual constraints for stabilization or regularization, or low-quality local ties. In addition, not all applied analysis procedures are rigorous from a mathematical point of view. This is mainly due to the fact that intra-technique solutions which are based on identical observation data are considered as independent. As the normal equations or the variance-covariance matrices of the intra-technique combined solutions are used for the weighting of the inter-technique combinations (and thus for the final product 'reference frame') a dedicated study is urgently required. The scope of this paper is twofold. On the one hand there are the above-mentioned factors which are relevant for the proper assessment of the quality of intra-technique and inter-technique combinations. They are discussed and modelled from a general point of view. This includes effects of different software packages and operators. On the other hand the impact of identical observation data on the quality measures of intra-technique combinations is modelled and quantified. For this purpose some typical observations of space-geodetic techniques (i.e. VLBI) are analysed. Finally, recommendations are given for improved intra-technique and inter-technique combinations.
