Tropospheric correction for long-range high precision GNSS positioning with regional numerical weather prediction models

Abstract

Static positioning, Precise Point Positioning (PPP) and Real-Time-Kinematic (RTK) positioning are three main high-precision positioning techniques used for Global Navigation Satellite Systems (GNSS). As current models for tropospheric delays are not accurate, sufficient long observation time is required by static positioning and PPP to estimate tropospheric delay parameters directly from GNSS observation, and network-based RTK is normally limited to tens of km for the interpolation of tropospheric delays from reference stations. In this research, we try to use regional Numerical Weather Prediction (NWP) models, the meteorological (MET) data from ground, and GNSS real-time tropospheric delays estimated at reference station to generate tropospheric correct model for precise positioning for PPP and network RTK with long baseline separation (i.e. 200-300 km). In this way, we can significantly improve GNSS positioning efficiency and the cost of network RTK for wide area coverage. Firstly, the accuracy of tropospheric delays derived from a regional NWP model is evaluated using long-term GNSS and Radiosonde data in East China region. A new Fast and Accurate Iterated Ray Tracing (FAIRT) algorithm is developed to estimate the tropospheric delays along with the GPS signal path from the NWP model. As a by-product, the precision of the tropospheric delay mapping functions are also evaluated. In general, the NWP derived tropospheric delays agree with GPS and MET data derived delays, with the accuracy of 20 mm. The accuracy of the tropospheric delays is better in dry seasons than that in wet seasons. Using real-time wet pressure observations on the ground and GNSS derived tropospheric delays at reference stations, the tropospheric delay estimation can be further improved with a new data assimilation method developed in this study. The tropospheric delay models developed in this study is also evaluated by applying them to GNSS positioning. With the model, GNSS positioning accuracy of centimeter and decimeter level can be achieved in horizontal and vertical components respectively. To further improve positioning accuracy, the NWP derived tropospheric model is used as initial value and a new tropospheric search algorithm is proposed to improve GNSS positioning convergence time.