Handling temporal correlated noise in large-scale global GNSS processing

Abstract

Global Navigation Satellite System (GNSS) products are an integral part of a wide range of scientific and commercial applications. The creation of such products requires processing software capable of solving a combined station position and GNSS satellite orbit estimation by least squares adjustment, also known as global GNSS processing. Such processing is routinely performed by the International GNSS Service (IGS) and its Analysis Centers. For the IGS Reprocessing Campaign 3 (repro3), Graz University of Technology (TUG) participated as an AC using the raw observation approach, which uses all measurements as observed by the receivers. However, a common feature of almost all global multi-GNSS processing strategies is the use of diagonal covariance matrices as stochastic models for simplicity. This implies that any spatial or temporal correlations are ignored. However, numerous studies have shown that GNSS processing is indeed affected by spatial and temporal correlations. For global GNSS processing, research on stochastic modeling and its challenges is rather scarce. In this work, a detailed insight into the problems of stochastic modeling in global GNSS processing using the raw observation approach is given along with a detailed overview of the intended TUG approach. An analysis of the impact of temporal correlation modeling on the resulting GNSS products and GNSS frame estimation is also given.