Comparison between the three dedicated gravity fields mapping mission, [CHAMP (1996), GRACE (2002) and GOCE (2009)], in geoid modeling (when using the satellite-only model related to these three mission)
Of course there are big differences in terms of spatial resolution (sp. res.) and hence geoid heights (GH) accuracy obtained from each mission.
For example, the maximum sp. res.) that CHAMP provides, is about 286 km (SH d/o 167) based on Weigelt et al., 2013. Let us say this provides GH accuracy of about 10 cm @ 200km.
For GRACE, which has max. sp. res. of about ~ 100 km, provides GH accuracy of about 1 cm @ 200 km which is 10 times better than CHAMP and about 5 - 10 cm@100km (i.e. SH d/o 200). In addition to the detection of temporal varying gravity field signals that GRACE contributes.
By GOCE solutions, which has a sp. res. of about ~66 km (i.e. SH d/o 300), one could reach now GH accuracy of about 1 cm @ 200 km.
That is all, so it is a matter of spatial resolution & mission accuracy.
I forget to say that what I mentioned in my previous reply about the acuracy which each mission provides, plays an important rule to determine the long-medium wavelength of the gravity spectrum required to be subtracted from the terrestrial data while modelling the geoid heights over any region.