Evaluation of Empirical Tropospheric Models Using Satellite-Tracking Tropospheric Wet Delays with Water Vapor Radiometer at Tongji, China.

Bofeng Li,Miaomiao Wang

doi:10.3390/s16020186

Abstract

An empirical tropospheric delay model, together with a mapping function, is commonly used to correct the tropospheric errors in global navigation satellite system (GNSS) processing. As is well-known, the accuracy of tropospheric delay models relies mainly on the correction efficiency for tropospheric wet delays. In this paper, we evaluate the accuracy of three tropospheric delay models, together with five mapping functions in wet delays calculation. The evaluations are conducted by comparing their slant wet delays with those measured by water vapor radiometer based on its satellite-tracking function (collected data with large liquid water path is removed). For all 15 combinations of three tropospheric models and five mapping functions, their accuracies as a function of elevation are statistically analyzed by using nine-day data in two scenarios, with and without meteorological data. The results show that (1) no matter with or without meteorological data, there is no practical difference between mapping functions, i.e., Chao, Ifadis, Vienna Mapping Function 1 (VMF1), Niell Mapping Function (NMF), and MTT Mapping Function (MTT); (2) without meteorological data, the UNB3 is much better than Saastamoinen and Hopfield models, while the Saastamoinen model performed slightly better than the Hopfield model; (3) with meteorological data, the accuracies of all three tropospheric delay models are improved to be comparable, especially for lower elevations. In addition, the kinematic precise point positioning where no parameter is set up for tropospheric delay modification is conducted to further evaluate the performance of tropospheric delay models in positioning accuracy. It is shown that the UNB3 model is best and can achieve about 10 cm accuracy for the N and E coordinate component while 20 cm accuracy for the U coordinate component no matter the meteorological data is available or not. This accuracy can be obtained by the Saastamoinen model only when meteorological data is available, and degraded to 46 cm for the U component if the meteorological data is not available.

Highlights

When traveling through the atmosphere, global navigation satellite system (GNSS) signals are typically affected by two kinds of error sources; namely, ionosphere and neutral atmosphere
Function 1 (VMF1), Niell Mapping Function (NMF), and MTT Mapping Function (MTT); (2) without meteorological data, the UNB3 is much better than Saastamoinen and Hopfield models, while the Saastamoinen model performed slightly better than the Hopfield model; (3) with meteorological data, the accuracies of all three tropospheric delay models are improved to be comparable, especially for lower elevations
The total slant tropospheric delay (STD) consists of a slant hydrostatic delay (SHD) and a slant wet delay (SWD), and both of them can be expressed by a relevant zenith tropospheric delay (ZTD)

Summary

Introduction

When traveling through the atmosphere, global navigation satellite system (GNSS) signals are typically affected by two kinds of error sources; namely, ionosphere and neutral atmosphere. The empirical wet delay correction performance depends on the accuracy of the zenith wet delay model and the wet mapping function. Precise GNSS applications need to correct the tropospheric delays as accurately as possible, for which one needs to identify an accurate empirical tropospheric model together with a mapping function. We will evaluate different empirical tropospheric models together with mapping functions by comparing the model-calculated wet delays with those precise ones measured by water vapor radiometer. The ground-based water vapor radiometer measures sky emission at two or more well-separated frequencies with high temporal resolution It can scan and track all visible satellites individually and precisely measure the slant wet delays at their azimuths and elevations.

Tropospheric Delay Calculation

Data Collection

Comparison Experiments

Standard Atmosphere

Meteorological

Histograms

Satellite-Specific Wet Delay Errors

Kinematic PPP with Different ZTD Models

11. The positioning errors of kinematic

22 November

22–30 November

13. Statistics