After some deliberations we now return to the 'edge fitting method' and its results. There are 19 nights on which more than 17 B and V pairs of closely spaced images exist. We consider those B and V image pairs with negative lunar phases, and look at how the B minus V (albedos, not colours!) values are distributed against lunar phase. We see this:
The error bars are estimated on the basis of error propagation in the B and V albedos from the edge-fitting uncertainty procedure. We have added small offsets in phase to get separation of points. The above actually contains data from 19 nights. To better see things we calculate the nightly median value Of B and V albedo, and now plot these:
Here, error bars are given by the standard deviation of the nightly values. Each point is labelled with the JD number. We see three night-after-night sequences: 2456073-76, 2456016-17, and 2456045-47. Given the error bars, these sequnces follow the same pattern - a rise in B minus V (albedo) as you go from new moon (left) to half moon (-90 degrees). Since the three sequences painmt the same pattern we are confindet in saying that 'Something Is Going On Here!'.
At the moment we do not know if this is geophysics in the shape of Earth albedo changing with phase - or some aspect of the halo (which increases from left to right above) being harder to model as phase grows.
Next it may be appropriate to study what the Earth actually looked like in the above three sequences. The negative phases selected for the above plots all correspond to the Sun illuminating Earth from Western Pacific/Australia/East ASia and westwards. As time passes on a given day the Moon sinks further in the West (as seen from Hawaii) and more of Asia contributes to the earthshine. On days in a sequence, at the same local hour, less of Earth is illuminated as seen from the Moon so the contribution to earthshine drops but is more and more 'sicle-like'. SUmmarizing:
For the negative phases selcted we expect:
a) during a single day - contribution to earthshine by continental areas increases, ocea contributions decrease,
b) same time of day but consequitive days - earthshine contribution comes from areas closer and closer to the edge of Earths disc.
I think this means that - if clouds are evenly distributed - we should see reddening of the albedos during a sequence taken on one day, and a blueing in a series of days.
We see b)! Have we seen a)? Below is a plot of the B albedo minus V albedo values for each of the days in question plotted against JD day fraction.
sharper pdf here:
We see little evidence of any up- or down-turns in this. Perhaps 2456104 shows an upturn - i.e. the opposite of what we thought we'd see. The remaining narrow sequences above seem to show a slight downward trend as the day passes - i.e. reddening. Potentially, clouds dominated these days so that little of the surface was visible?