In this entry we started exploring the Stam 2008 set of model Earth spectra. That entry was based on the 'land only' set of Stam models. Ocean-only models are also provided, so we can mix these with the land models to get more realistic Earth spectra. In Stam ocean models the albedo is assumed to be 0 apart from a specular contribution, so all colours are dominated by the Rayleigh scattering and a bit of sunlight reflected at the glint point. In land-only models the vegetation red edge has a large contribution, so we expect some changes when a land-ocean mix is introduced. We assume that Earth is 26 % land surface in what follows.
As before we calculate the spectrum of a range of cloudyness percentages. We find that:
B-V for Stam cloud free forest model is: 0.19
B-V for Stam cloudy forest model is : 0.54
B-V for Stam cloud-free ocean model is : -0.23
B-V for Stam cloudy ocean model is : 0.53
Note how similar the 100% cloud cases are for land-only and ocean-only, as expected. There is a small effect of the surface below the clouds (which is one of the features of the Stam set of models - others have ignored this effect).
Note also how blue a cloud-free ocean only Earth would be - this is, I expect, the B-V of our own blue skies on a cloud-free day, as we have discussed elsewhere. I think we should try again to measure blue-sky B-V with our DSLR calibrations.
As before we estimate slopes of the B-V vs cloud % relationship.
Change in % cloud cover per mmag change in observed B-V: 0.31 %pts/mmag
Change in % cloud cover for total observed error in B-V: +/- 6.3 %pts
Change in % cloud cover for internal observed error in B-V: +/- 1.6 %pts
The total B-V errors we have are +/- 0.02, mainly due to uncertainties in the Solar B-V as pointed out by Chris. If we leave all measurements relative to the fixed (but somewhat unknown solar B-V) we have internal errors of just +/- 0.005 mags and that implies we could determine Earth's cloud cover to within something like +/- 1.6%-points on the basis of observed colour alone. Not bad!
With a technique for observing B-V on all nights we could thus complement direct albedo-determinations (found with edge-fitting for instance). The colour-method has drawbacks and benefits relative to the direct albedo determination in that a difference is used instead of an absolute fit.
This is a point towards why we need to understand how the halos are formed and just why they cancelled on that night!
It is also fuel for fixing the SKE because with an SKE more images could be obtained where the remnant halos in B and V almost cancelled on the DS.
With the CCD working again, we have some hopes of understanding the SKE problems, as well as performing some very careful lab observations with halo due to optics only.