Which geoid model should be used for GPS heighting on the Toowoomba bypass project?

[Abstract]: Graphical and statistical evaluations of the OSU91A, EGM96, EIGEN2/EGM96, UCPH2/EGM96 and PGM2000A global geoid models and of the bi-cubic and bi-linear interpolations of the SBA Technique, the RBA Technique, AUSGeoid93 and AUSGeoid98 gravimetric geoid models of Australia are made over the Great Dividing Range escarpment at Toowoomba using a set of 116 empirically derived geoid heights. Absolute and relative comparisons made between each geoid model over a 46.2 km Global Positioning System (GPS) and digital levelling traverse, against all 116 control points (6,670 possible baselines) and over Australian Height Datum (AHD) elevations greater than 200 m through to greater than 600 m show that AUSGeoid98 is the superior geoid model for the conversion of GPS-derived ellipsoid heights to AHD elevations on the Toowoomba Bypass project. The results from this study confirm the benefits of including additional topographic data and satellite altimeter-derived gravity data in the production of the SBA Technique, the RBA Technique and AUSGeoid98, which were not used for AUSGeoid93. Conclusions made in this study are qualified by the fact that empirical validation of gravimetric geoid models on land does not provide an unequivocal assessment of the data, theories and techniques used to compute each geoid model due to the errors residing in the empirical geoid heights estimated to be ±0.0371 m evaluated at the 95% confidence level. However, land-based comparisons with GPS and levelling data remain the most practical method of verifying the integrity of gravimetric geoid models