Investigation of Different Interpolation Models Used in Network-RTK for the Virtual Reference Station Technique

Abstract

This paper thoroughly investigates several approaches to implementing the GNSS network-based real-time positioning technique, which requires the estimation of atmospheric corrections on an epoch-by-epoch basis for RTK. In this study, a network of Continuously Operating Reference Stations in New South Wales, known as CORSnet-NSW, was utilised to: 1) obtain atmospheric residuals from each reference station, and 2) determine network correction for a rover operating in the area covered by the network using several interpolation methods. Applying the atmospheric corrections obtained by the interpolation methods, “synthetic” measurements at a virtual reference station are generated and then used for rover positioning. Field tests with various masterrover baseline lengths ranging from 21 to 62km indicate that a range of 1.9 to 6.5cm of horizontal positioning accuracy is achieved. In this study, the performance of geostatistical (Oridinary Kriging Method and Least Squares Collocation Method) and deterministic (Linear Combination Method, Linear Interpolation Method, Low-order Surface Method and Multiquadric Surface Fitting Method) interpolation methods used in GNSS network-based RTK positioning were also analysed in order to identify the optimal method for mitigating atmospheric effects for real-time kinematic applications under different network geometries.