In this blog-entry we discussed a method to force halos to cancel by convolving the sharper of two images in a B/V pair, until their halos matched and could be cancelled by subtraction.
From the many possible B and V image pairs that we have, a number has been selected, and the method applied. The criterion used was that the linear slope on the DS disc should be within 1 sigma of 0. The result, for negative lunar phases, is this:
Here is a sharper pdf:
In the upper frame above, we plot the B-V difference between the BS and the DS. On average, B-V seems to be 0.4 bluer on the DS than on the BS. Given that the BS has B-V near 0.92 [Allen, 3rd ed.] we can convert the above to B-V for the DS by subtracting it from 0.92. That would give us B-V values for the DS near 0.5. Fred Franklin found less blue values since his B-V for DS was 0.17 below the BS value. That is at the lower limit of what we find.
We plot +/-1 error bars - they seem to get bigger as phase approaches Full Moon. We note that the scatter is larger than the errors - so something other than the errors we can estimate are causing the scatter.
Is there a trend in the above? By eye, it seems like there is - as phase approaches FM we have more positive ordinates - implying that the DS gets bluer with lunar phase.
The points fall in 'columns' corresponding to single nights - so something is causing a colour-change as the night progresses. All data are of course extinction-corrected so we doubt that the nightly colour-change is due to extinction. Is it a true color-change in earthshine?
We can compare the photometric B-V derived in the above way with B-V differences in albedo found by fitting the DS disc edge.
Now for the lower panel:
In the lower frame we plot B-V against the PSF-parameter alfa needed to bring the sharper image into the same state of blurriness as the less sharp image - so that halo cancellation could occur. There is a very clear relationship between alfa and the B-V achieved. I guess this is bad news? Nature (i.e. the B-V colour of Earth) should not depend on a PSF parameter used in the above convoluted way. Oh dear.
I still think it is worth comparing the present results with the ones based on fitting the DS edge. If we get a contradiction it may mean anything - but a strong agreement would imply that the 'edge-fitting method' has some gremlins hiding in it - i.e. that results are very dependent on those alfas.
It is also possibly revealing that the nightly points fall in columns [probably sequences once telescoped out to single nights - will check]. It is telling us something about things that change on an hourly basis - this includes airmass (i.e. extinction) and also the Earth-Moon-Observer aspect - so, still interesting!