In understanding this this post, we now compare the results to those from the edge fitting method. We remind thereader that the Force method gives us the difference in B-V [magnitudes] between the BS and the DS, whil ethe dge-fitting method gives us albedos fro Earth in e.g. B and V bands. We look at the same range of phases in the two methods and repeat the relevant ploits here. First the edge-fitting method results:
Upper panel shows us B-band albedo minus V-band albedo against lunar phase (upper panel) and against the average alfa (B-alfa and V-alfa averaged).
We repeat the plot from the Force method here:
Here are the pdfs:
In the upper panels we see a slight dependence on lunar phase - B albedo minus V albedo as well as B-V [mags] do tend to rise towards Full Moon. Note that this implies opposite colours! The rise in B albedo relative to V albedo means that the Earth appears bluer as we approach Full Moon in the edge-fitting method, while larger B-V [mags] in the Force method implies a redder Earth as we approach Full Moon.
In the lower panels we see that the alfas used in the edge fitting method and the derived albedo differences have little relationship with each other, while the 'incremental alfa' needed in the Forcing method bears a strong relationship with B-V, as discussed before.
This seems to rule out the worst of the possibilities given in the Force posting - namely that if B and V albedos depended as strongly on alfa as B-V does then we would have a serious problem. In the present situation we are not in that position - but where are we then?
For both methods we seem to have outliers for phases less than -120 degrees. Let us take a closer look at these points. They are from night JD 2456073. The upper plot shows that the B and V albedos we could derive from edge fitting' differs by .1 between the upper group and the lower group. From the Force method we see that B-V [mags] differs by 0.5 between these two groups.
The images involved have these names :
There are more data points for the edge-fitting method, in the plots above, than there are for the Forcing method. This is because more B and V combinations were picked tested for the former. The images picked were observed withing half an hour of each other. These images should be visually inspected.
[later:] Tried that - it gets messy: differences in the level-differences of the DSs in B and V can be spotted but do seem to depend on knowing the exposure times, and this is one of the things we do not think worked - better to trust the edge-fitting method, since it is a 'common mode rejecting method'.