Abstract: The earth height and normal height of a total of 99 GPS leveling points are used to calculate the elevation anomalies, and DOG spherical wavelet model and polyhedral function are used to fit the elevation anomalies respectively. During the fitting, the points where the residual error of the model is greater than 2 times the median error are eliminated, and the fitting is performed again after the coarse error is eliminated. The accuracy of the model is evaluated by comparing the mean square deviation between the known elevation outliers of the external check points and the spherical wavelet model values and the polyhedral function fitting values. The data results show that the spherical wavelet model has a better fitting precision of 1.65 cm and the multiplanar function of 2.35 cm based on the criterion of minimum mean square error of external check.

Key words: GPS leveling points, elevation anomaly, DOG spherical wavelet, polyhedral function, elevation fitting