Analysis and Evaluation of Various Tropospheric Modeling Approaches for High-Precision GPS Kinematic Positioning over Medium Ranges and at High Altitude: Case Study

• Published in: Journal of surveying engineering 2009-05, Vol.135 (2), p.66-71
• Main Authors: Zhao, S.-H, Lu, X.-S, Wang, Q.-F, Hu, G.-R, Ogaja, C
• Format: Article
• Language: English
• Subjects: TECHNICAL PAPERS
• ISSN: 0733-9453; 1943-5428
• DOI: 10.1061/(ASCE)0733-9453(2009)135:2(66)

In global positioning system (GPS) positioning, the tropospheric delay is a systematic error. Mismodeling of the tropospheric delay results in a degradation of the estimated height component, and thus constitutes a limitation to high-accuracy GPS applications. As such, it is obvious that the tropospheric delay should be modeled as accurately as possible. Modeling the tropospheric delay for some applications, such as precise airborne kinematic differential positioning, is further complicated due to the large altitude difference between the ground-based receiver and the airborne roving receiver. This study tests and analyzes three methods for modeling the tropospheric delay, in an attempt to improve the accuracy of the height component for airborne GPS kinematic positioning. As there are several other error sources in GPS, the test and evaluation have to be carefully designed, as any improvements in accuracy due to the use of any tropospheric modeling approaches could be masked by other effects, such as residual orbit errors and ionospheric delays. The analysis is performed for real airborne GPS data and data from a multiple-base station network, and model performance evaluation is based on an independently well-determined aircraft trajectory. Test results show that the estimation of residual tropospheric zenith delay simultaneously with the position parameters after applying the tropospheric delay corrections derived from a network of GPS receivers is seen as the best option with an additional 62% improvement for the height component in terms of standard deviation when compared with results from using a UNB3 tropospheric model only.