A global assessment of ray-traced and blind tropospheric models in the retrieval of tropospheric parameters from ground-based GPS observations

Abstract

The number of ground-based GNSS reference stations has increased significantly in recent years, which provides very favorable conditions for ZTD/ZWD retrieval. The methods of estimating ZTD/ZWD based on three a-priori troposphere values are discussed using GNSS reference stations without meteorological equipments. GPS observation data for two weeks in different seasons from 22 global IGS stations were processed using online GAPS software with static and kinematic PPP mode based on ECMWF grid product and GPT2.1w/UNB3m blind tropospheric models. Vienna Mapping Function 1 (VMF1) is used for all the a-priori models. The numerical results show that: compared with the IGS ZTD products, mm-level and about 2 cm-level ZTD products can be obtained based on the three a-priori troposphere values using static PPP and kinematic PPP respectively. About 1 cm-level and 2 cm-level ZWD products can be obtained based on ECMWF grid product/GPT2.1w model using static PPP and kinematic PPP respectively. The discrepancy between the ZTD estimates based on the three a-priori troposphere values is no more than 3 mm, and the differences of the averaged ZTD Mbias, RMSE and Std of the 22 IGS stations based on the three a-priori troposphere values are no more than 0.3 mm. While there is obvious differences in ZWD estimates based on the three a-priori troposphere values. The differences of the averaged ZWD Mbias/RMSE between ECMWF grid product and GPT2.1w model is about 0.2 cm, and the differences of the averaged ZWD Mbias/RMSE between ECMWF grid product and UNB3m model are about 0.5–1 cm. There is a systematic bias of about ±1 cm between the ZWD estimates based on GPT2.1w and UNB3m. This study shows that GNSS PPP technology based on ECMWF grid product and blind tropospheric model has great potential for ZTD/ZWD retrieval using ground-based GNSS reference stations.