Establishment of a global zenith troposphericdelay model using multi-source data

Abstract

Radio space-based geodesy techniques suffer from atmosphere propagation delays, of which the ionospheric delay can be largely eliminated by iono-free carrier phase combination techniques, and then the tropospheric delay becomes the main error source. In general, we project the slant delay to zenith direction with mapping function in GNSS navigation and positioning, so modeling the zenith tropospheric delay (ZTD) is a common method to reduce the tropospheric influence on signal travelling. Currently, the commonly used tropospheric delay models are all based on single data source. In this paper we initially used multi-source ZTD data to construct a tropospheric delay model. The ZTD data used for modeling are gridded data from GGOS Atmosphere and IGS troposphere products respectively. The Global Geodetic Observing System (GGOS) Atmosphere is a project that aims to establish atmospheric models. It provides gridded data of global zenith delays with temporal resolution of 6 h (0:00, 6:00, 12:00, 18:00UTC) and spatial resolution of 2.5°×2° (Lon× Lat), which are derived from the reanalysis data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF). IGS troposphere products are derived from position solutions of IGS stations with a high temporal resolution of 5 min and a high accuracy. Firstly, we used IGS ZTD data to analyze the characteristics of short periodic variations (diurnal cycle and semidiurnal cycle). Then, on basis of this analysis and previous researches on ZTD periodic variations conducted by other researchers, we combined GGOS ZTD data and IGS ZTD data to establish ZTD model in accordance with the idea of direct ZTD modeling. We obtained the temporal parameters via cycle fitting for ZTD data and saved the parameters in the form of global Delaunay triangles. For the proposed ZTD model, the ZTD values can be estimated by the methods of height reduction and linear triangle interpolation. The cycle fitting residuals of GGOS ZTD data and IGS ZTD data show that the internal accuracy are 3.62 and 3.60 cm respectively, indicating that our function model with diurnal cycle and semidiurnal cycle can well describe the temporal variations of ZTD. We validated the multi-source data model with respect to ZTD grid data from 94 globally distributed IGS stations during 2013, which were not involved in modeling. The obtained results show that the global average bias and RMS are - 0.31 cm and 4.16 cm respectively. Compared to other tropospheric delay models, multi-source data model is significantly superior to GZTD model (bias: - 0.48 cm; RMS: 4.46 cm) which ignores the short cycles of ZTD and is better than the GGrid model without IGS ZTD involved in modeling. With an accuracy reduce of only 0.2 mm, the multi-source data model is comparable to the most accurate and complicated GPT2w model (bias: - 0.04 cm; RMS: 4.14 cm). At the same time, we validated the stability of multi-source model in space and time, the testing results show the good correction performance and reliability. In addition to the globally applicable convenience, the multi-source model also show a characteristic of regional augmentation, which can greatly improve the tropospheric correction performances in areas around IGS sites. With the gradual increase of regional Continuously Operating Reference Stations (CORS) and global IGS stations and the development of meteorological data sources, how to assimilate diverse ZTD data sources to establish a more comprehensive global multi-source tropospheric delay needs more detailed and further study.