The answer to this conundrum may lie in some of the terms that you have used.
Pesumably the term "scheduled V1 " means v1 figures derived from tables/graphs.
But in using the term "published Vmcg" do you mean a single value for Vmcg that is stated in some part of the manual? If this is the case then your Vmcg value is probably set against ISA msl or some similar fixed standard.
The actual value of Vmcg and Vmca will decrease with increasing altitude so a fixed figure cannot be valid for all altitudes.
If you are taking values of V1 Vr and V2 that have been corrected for altitude and comparing them with a value for Vmcg that has not been corrected, then are likely to find the kind of problems that you describe.
Is it perhaps the case that for the aircraft in question the "minimum V1, Vr and V2 curves depicted on the chart" are actually based on the altitude-corrected vaulues of Vmcg and Vmca?
Your statement that "......In all cases, the v-speeds derived are at least equal to the "minimum" V1, Vr and V2 curves depicted on the chart." suggests that this is the case.