The combination of terrestrial and satellite aided network observations requires hybrid models of two kinds: a hybrid functional model and a stochastic model being able to process the heterogeneous observations appropriately. An appropriate combination of stochastic information means to adjoin proper weights to the various types of observations on one hand and on the other hand to separate the relevant internal accuracy from that accuracy which is affected by a number of network external influences. The first task can be overcome by the technique of estimating variance components of the individual observation groups, the second one can be handled by applying S-transofrmations to covariance matrices. The paper deals with problems of the hybrid stochastic model. Examples are given.