Due to the steady progress in global positioning system (GPS) technology an
d methods of data evaluation, it is possible to obtain highly precise relat
ive point positions also for extensive geodetic networks. However, some lim
iting influences such as temporal correlations of observational data are ne
glected in most of the GPS processing programs. Therefore, it is necessary
to consider the impact of these neglected correlations on the coordinates a
nd their accuracy measures. In this paper the sensitivity and reliability o
f GPS positioning over long baselines when temporal correlations are modell
ed with the aid of a correlation function are discussed. The implementation
in the variance-covariance matrix and the subsequent evaluation process re
quire a considerable amount of computing time and memory. Therefore it is n
ecessary to use appropriate numerical methods such as approximated matrix i
nversion in order to reduce the numerical requirements. After the descripti
on of the methodical and numerical handling of the temporal correlations. e
xemplary results are given to illustrate the effects on position and accura
cy of GPS stations.