Comparison of tropospheric delay models using ground based GPS ZTD values in the atmosphere of Iran

There are several sources of error that must be considered for accurate GPS positioning. One of these sources of error is the tropospheric delay of the signal, whose accurate estimation leads to an increase in the accuracy of positioning in navigation, as well as the accurate calculation of precipitable water vapor for meteorological and climatological applications. One of the accurate methods in determining ZTD values is to estimate it along with the coordinate components of ground stations using GPS observation processing. However, it is not possible to access permanent GPS receivers in all places and it is expensive. In addition to permanent GPS station data processing, the use of atmospheric profiles obtained from the radiosonde launch at each station is one of the other conventional methods for calculating the tropospheric zenith delay. The low temporal resolution of radiosonde observations (usually twice a day) and the high cost are the main limitations of this method.

Moreover, the use of global or regional empirical models or models based on surface meteorological data are among the methods of calculating Zenith Tropospheric Delay (ZTD). It is necessary to evaluate the accuracy and precision of these models in each region before using them in the intended application. Iran has diverse topography and climatic conditions, so different tropospheric delay models may have different statistical quality in Iran compared to other regions. On the other hand, until today, no comprehensive research has been done in the region of Iran to evaluate the different tropospheric delay models presented in recent years.

Empirical ZTD models presented in recent years are a function of position, place and time and some models, such as Hopfield and Sastamoinen ZTD models are known as famous models based on surface meteorological data. Also, according to research conducted in other parts of the world, HGPT2 and GTrop models are among the successful global emperical models in ZTD estimation that have been proposed in recent years and operate independently of surface meteorological parameters.

In this study, with the help of one year of ZTD estimates obtained from the processing of GPS observations in 28 stations located in the region of Iran, the statistical quality of Hopfield, Sastamoinen, HGPT2 and GTrop models was investigated. Based on the results, the average RMSE values of one year of ZTD calculated with the help of Hopfield, HGPT2, Gtrop and Sastamoinen models were estimated to be 75, 38.8, 31.7 and 26.1 mm, respectively. Also, the average bias of ZTD values obtained from Hopfield, HGPT2, GTrop and Sastamoinen models in the whole region was -8.69, 18.9, 5.4 and -10.8 mm, respectively. The model of Hopfield and Sastämöinen was the most consistent with the ZTD values obtained from the processing of ground GPS observations. The results of this study showed that, in general, the Sastamoinen model is statistically more efficient than the other three models, but in order to achieve proper accuracy, it is necessary to develop a model suitable for the region of Iran.