Overbounding residual zenith tropospheric delays to enhance GNSS integrity monitoring

Tropospheric delay is one of the main error sources that should be considered in global navigation satellite system (GNSS) positioning and integrity monitoring. Usually, it is first corrected by an empirical tropospheric zenith total delay (ZTD) model and an elevation-dependent mapping function during the preprocessing procedure, and then, the residual ZTDs are further compensated in the mathematical model, either functional or stochastic. Therefore, a tight and conservative stochastic model of the residual ZTDs is of great benefit to GNSS integrity monitoring. Since the residual ZTDs usually show significant geographical and seasonal variations and are not Gaussian distributed, using a spatiotemporal-invariant root mean square or standard deviation (STD) value to describe their stochastic characteristics would be either overly optimistic or overly conservative. We present a global and spatiotemporal-varying overbounding method to quantitatively assess the residual ZTDs, with a view to enhancing GNSS integrity monitoring. The proposed method combines the hierarchical clustering and Gaussian overbounding techniques to tightly envelop the residual ZTDs, by a constant bias and a periodic-varying STD in each latitude band. Modeling results of three conventionally used ZTD models (GPT2w, UNB3, and Saastamoinen) are presented. Also, we demonstrate how the proposed overbounding model can enhance the availability of GNSS integrity monitoring.