Dear authors,

the referee was essentially happy with the modified manuscript. There is one suggestion for a technical correction.

However, I do have a request. One question was already raised in the quick review of the initial submission, and this point wasn't addressed properly. In my view, Rayleigh scattering cannot be the cause for the wavelength-dependence of the "PDF" averaged over the globe (line 331 of the manuscript). Rayleigh scattering causes a smoothing of the image, in the sense that some radiation from outside the shadow is scattered into the shadow region. But i don't see why this should reduce the loss. Certainly, the shadow itself is brighter, but outside the shadow it's darker and in first approximation the integrated reflectance would be the same. Also, if Rayleigh scattering was the cause, I would have expected a wavelength dependence at the lower wavelengths, not just at the 4 higher ones. Between 317 and 551nm the Rayleigh scattering cross section varies by a factor of 10 (4th power of the wavelength) - yet all reduction values are the same. between 680 and 780nm, Rayleigh varies much less, yet the reflectance changes from 10 to 13%. As suggested before, this has probably nothing to do with Rayleigh scattering but with the vegetation edge. The stronger attenuation starts exactly at the wavelengths were the vegetation reflectance starts increasing, and since the shadow is over land, the longer wavelengths are attenuated disproportionally since the shaded area is much brighter than the ocean and hence a shadow causes larger reduction compared to lower wavelengths where the land is equally dark as the ocean. Could you please consider that, or convince me that I am wrong?

Sincerely,
Bernhard Mayer.

Thanks for considering my final comments!

Best wishes,
Bernhard.