UKNanowaves Yahoo Group in a posting by Barry Chambers.
The free software is available at
http://www.philiplaven.com/mieplot.htm and allows us to work out the intensity of a scattered optical signal as a function of scattering angle, droplet size and wavelength. Scattering is best when the angle of incidence is at grazing incidence and the droplet size is small. If I've understood the results correctly, aiming at the underside of a cloud at 45 degrees would result in a scattered signal some 50dB weaker than if at grazing incidence. So, depending on how far apart the 2 stations are and the angle at which the optical signal hits the underside of a cloud then signals can be quite strong or extremely weak. This is why weak signal modes like QRSS60 may be needed to work a given path by cloudbounce.