Comparative analysis of BDGIM, NeQuick-G, and Klobuchar ionospheric broadcast models

Elimination of ionospheric delay using a dual-frequency ionospheric-free combination is impossible for single-frequency global navigation satellite system (GNSS) users. Therefore, effective empirical ionospheric models are essential. Three ionospheric models used in GNSS systems, namely, BDGIM, NeQuick-G, and Klobuchar, are evaluated in this study. We selected five stations for 28 days of data in 2020 to examine the performance of models in predicting total electron content (TEC) and improvement in accuracy of single point positioning (SPP) with different ionospheric models. Compared with that using the Klobuchar model, root mean square error (RMSE) values of global ionospheric map (GIM) predicted using the NeQuick-G and BDGIM models reduce by 26.9% and 48.2% on quiet days and 24.7% and 46.6% on disturbed days, respectively. Both the Klobuchar and BDGIM models present a significant deviation in TEC values close to 0. However, overestimation of the NeQuick-G model is particularly serious and underestimation is observed in the BDGIM model when the TEC value is large. Slant TEC (STEC) estimated using GPS observations is used as the reference in this work. The accuracy of STEC from the NeQuick-G model at high latitude stations is high, while RMSE values of STEC from the BDGIM model are small at mid- and low-latitude stations. We examined the results of SPP with different ionospheric models. Results showed that all three models improve the accuracy of SPP, especially in the U direction, and the NeQuick-G and BDGIM models achieve satisfactory performance.