In this paper two different methods of how to handle topography in geoid de
termination is investigated. First method employs the Residual Terrain Mode
l (RTM) remove-restore technique and yields the quasigeoid, whereas the sec
ond method is the classical Helmert condensation method, yielding the geoid
. Both methods were used with the Earth Gravity Model (1996) (EGM96) geopot
ential model as reference, and results are compared to precise Global Posit
ioning System (GPS) levelling networks in Scandinavia, especially an accura
te GPS data set from the very rugged Sognefjord region, where the topograph
y was represented by either a detailed (100 m) or a coarse (1000 m) digital
terrain model. The inclusion of bathymetry in the terrain model was also i
nvestigated.
Even if two different methods were used, they produced almost identical res
ults at the 5 cm level in the mountains, but small systematic differences e
xist. Results show the importance of comparing the right types of geoid (cl
assical geoid or quasigeoid), since differences in residuals are significan
t. (C) 1999 Elsevier Science Ltd. AII rights reserved.