The automation of permanent networks monitoring: remarks and case studies

Abstract

GNSS permanent networks materialize and distribute the reference frame. The continuous monitoring of the stations involves the daily adjustment of the network in order to obtain the daily estimates of the coordinates. Moreover, the resulting time series should be analyzed to estimate the time models of the coordinates. Many tasks precede and follow the raw data processing in order to prepare the data and to analyze the results: several tasks are purely technical, but some of them involve methodological choices and numerical analyses. A complete automation of all the processing flow is clearly useful in the network management. During the last 7 years, we have been involved in the monitoring of different permanent networks that are heterogeneous in the spatial scale, the number of stations, and the scope. The raw data processing is performed by the Bernese GPS Software 5.0. To automate all the pre-processing tasks and the post-processing analyses, a software package has been implemented: it is called RegNet, is based on C and Matlab source code, and will be freely distributed at the time of the paper publication. At first, the paper describes the RegNet structure and its output reports. One example of RegNet application to a permanent network of about 200 stations, based on more than 1 year of data, is presented. The adjustment of a large network requires its splitting into small subnetworks, the separate adjustment of all the subnetworks, and finally, the recombination of the solutions by the so-called Normal Equation Stacking. In the second part of the paper, different network splitting strategies are compared from a methodological point of view. All these strategies have been implemented in RegNet. A quite realistic case study, based on about 100 stations of the European Permanent Network, 1 month of data, is presented, and the results are compared.