The tropospheric delay has an important influence on the navigation and positioning process. The tropospheric delay calculated from the high -precision meteorological parameters provided by reanalysis data can be used in the positioning process to improve the positioning results. This paper evaluates the accuracy of tropospheric delay calculated from the three types of reanalysis data, and applies them to precise point positioning to analyze their impact on positioning results. First, using the ZTD (Zenith total delay) of 125 IGS (International GNSS Service) stations around the world in 2020 as the true value, the accuracy of the ZTD calculated by three reanalysis data (ERAS, MERRA2, CRA40) were evaluated, and their spatiotemporal distribution characteristics were analyzed. The research results show that the root mean square (RMS) of ERAS-ZTD is the smallest (12. 1 mm), followed by CRA40-ZTD (15. 8 mm) and MERRA2-ZTD (16. 9 mm). The overall accuracy of ERAS-ZTD is the best. The tropical bias is negative, the accuracy of CRA40 is better than that of MERRA2 in the middle and high latitudes, and the opposite is true in low latitudes. The average accuracy of ERAS at each latitude band is the best. When seasonal factors are considered, the ZTD-RMS of all three reanalyses is larger in summer and autumn, and the RMS seasonal variation of ERAS-ZTD is the most stable. After that, the ZTD, Zenith Hydrostatic Delay (ZHD), and Zenith Wet Delay (ZWD) were evaluated with 180 radiosonde stations. It is found that the annual average ZTD of ERAS, MERRA2, and CRA40 were 13. 2, 18. 7, 14. 9 mm respectively. The annual average ZHD were 8. 2, 8. 5, 8. 2 mm respectively, which difference is small, and ZWD were 9. 3, 14. 6, 11. 6 mm respectively. ERAS has again the highest correlation with radiosonde data. Finally, using the ERAS, MERRA2, and CRA40 calculated ZTD to do precise point positioning experiments, the results show that compared with the Saastamoinen model, its accuracy in the N direction is improved by 59%, 56%, 47% respectively, in the E directions 51%, 47%, 41% respectively, and U directions 81%, 78%, 75% respectively. In general, the accuracy of ERAS tropospheric delays is the best in the three reanalysis data.
