The amount of clouds on Earth changes from day to day, and as clouds reflect sunlight (with B-V=0.64) more clouds cause B-V of the earthlight to drift towards the solar value, while less clouds shows the Earth bluer. How much do we expect the terrestrial B-V colour to change day to day on the basis of changing cloud masses?
Inspection of GOES West full-disc satellite images of Earth shows that the local noon (i.e. disc is fully illuminated) brightness variations due to changing clouds amount to about 3.2% of the mean brightness. As the Earth below is dark (Pacific Ocean) almost all of the brightness is due to clouds so we are not too far off by saying that the amount of clouds varies by 3.2% around its mean value on a daily basis. With Earth on average being almost 67% cloudy we can re-use the Stam models of expected terrestrial spectra to see what the expected changes in B-V is due to cloud variations.
We find that B-V will vary, with a standard deviation near 0.005 around the observed value of B-V=0.44.
This is smaller than the total observational error we have. We have shown that the theoretical Poison-noise limited uncertainty would be 0.005 in B-V, but we cannot observe that well. Yet.
So – we should not expect to see values for B-V very different from 0.44, which helps explains why we get the same value as Franklin’s mean value using just a single (but precise) observation.
We may be able to see 2 and 3 sigma deviations in the cloud cover, however.
The purpose of our telescope was never to observe daily cloud variations – in the long term we hope to be able to qualify that we can set limits to climate-change induced changes in albedo.
Also, the above is just an investigation into whether we can use COLOUR changes to quantify albedo changes – we still have to quantify how well our direct-photometry measurements can see albedo changes. Paper II.
Yes – I think so: a brightness measurement will tell us more about albedo than a brightness-difference method can. Need to show this with a numerical example.
Excellent work – so this rules out monitoring cloud cover via ES colour, right?
But it does show why we agree with Franklin with just one observation, so that’s good, as it was a source of concern!