For the five ultra-high-degree Earth gravity field models with a degree of 2 190, namely EGM2008, EIGEN-6C4, GECO, SGG-UGM-2, and XGM2019e_2159, which have been published internationally, a study on quasi-geoid refinement was conducted using the Shanxi survey area as an example. First, the accuracy performance of five models in Shanxi area was compared and analyzed using actual GNSS/leveling data. Then, based on the remove-restore technique, the residual height anomalies after removing the results of five ultra-high-degree gravity field models were fitted to the quasi-geoid using the multiquadric function method. A contour line-based method for selecting multiquadric function nodes was proposed, and the fitting accuracy was compared and analyzed. The results show that: 1) The accuracy distribution of five Earth gravity field models in Shanxi area is consistent, with larger errors mainly distributed in the high mountain areas in northern Shanxi. The XGM2019e_2159 model has the highest accuracy with a mean error of 9.1 cm, while the EGM2008 model has the lowest accuracy with a mean error of 11.3 cm. 2) The accuracy of five ultra-high-degree gravity field models does not significantly affect the final quasi-geoid accuracy. Among them, the quasi-geoid derived from considering the EGM2008, EIGEN-6C4, and SGG-UGM-2 models has the highest external consistency accuracy with a mean error of 5.0 cm. The quasi-geoid derived from XGM2019e_2159 model with the highest accuracy has the lowest external consistency accuracy with a mean error of 5.5 cm.

ultra-high-degree Earth gravity field model  /  GNSS/leveling  /  elevation anomaly  /  multiquadric function  /  quasi-geoid