When traversing the earth's neutral atmosphere, GPS radio signals are affected significantly by the variability of its refractive index, which causes primarily a delay, usually referred to in the literature as the tropospheric delay. An inaccurate modeling of this delay results in degradation of position estimates, affecting mainly the height component. The tropospheric delay is commonly divided into two components, "hydrostatic" and "wet", each one consisting of the product of the delay at the zenith and a mapping function that projects the zenith delay onto the desired line-of-sight. A few high-accuracy mapping functions, parameterized by either specific meteorological parameters or other site-dependent parameters, have been developed in recent years. As regards the prediction of the zenith delay, the problem is far more complicated essentially due to the high spatial and temporal variability of the wet component. We have determined mean bias and r.m.s. scatter for a great number of zenith delay prediction models developed in the last few decades, including the models generally used in airborne navigation, by comparing the models against ray-tracing using a one-year data set of radiosonde profiles from 50 stations distributed worldwide. We have concluded that the hydrostatic zenith delay can be predicted with submillimetre accuracy, provided accurate measurements of station pressure are available. As regards the wet zenith delay, the models differ significantly in accuracy but show very similar r.m.s. scatter. Our analyses show that the wet zenith delay can typically be predicted with a precision of ～3 cm (at the one-sigma level), using meteorological data. The prediction of the total delay by models typically used in airborne navigation indicates a much poorer accuracy, leading to prediction bias ranging from -6 cm up to more than 20 cm. In general, all the models tested perform significantly better at mid-latitudes than at low latitudes.
展开▼
