Hello and happy 2009 to everybodyI've run out of time in many respects. But it seems that this is becoming urgent too, within hours. So, to Wim's draft:
'All the upward light is contributing to the increase of sky glow. Vertical upward light is meeting few particles, to disturb its path so the probability the light is going down again is rather small. In contrary, horizontal light with a small angle to the horizon, has a much bigger chance of being redirected to the ground again. Several studies by different authors ( Garstang and Cinzano particularly) has tried to calculate the different contributions of different kind of light. It seems there is a 6 till 7 higher chance, light with less than 10 degrees with the horizon is diffused to the ground again in comparison with vertical upward light and so contributing to the increase of sky brightness at great distances from the light sources itself.'
Too complicated... No particles needed, lumps of air molecules do a good part of the job. The blue part especially. Nothing "seems" to scientists or from papers. No chance, no mathematical statistics needed. Everybody's experience is enough.
Most people -- all who are not blind and were outside on a sunny day already -- have noticed that we see even into shadows. It's easy to measure how much light is there and how much the direct sunlight adds outside of shadows. You may do it on a summer noon and just before sunset, for a horizontal open ground (actually, the direct component on a horizontal surface may is almost non-measurable with Sun at h<2, it is better to measure direct illuminance of a sun-facing plane and then to multiply it by sin(h) to get horizontal direct illuminance).
You will find that with sun high in the blue sky, direct light is far stronger than dispersed light. In an extreme case (almost no aerosols present), dispersed fraction is just 8 % at sea level (may be even less with less airmass, as in Tibet). Close to sunset, the situation will be reversed. Almost all light is dispersed, shadows are very shallow.
Light going upwards from lamps and lit objects at night is an obvious analogy to sunlight going down. With no aerosols at all, the dispersion of light in the air is perfectly symmetric, as much light is dispersed in forward directions as in backward directions. If it starts straight up, just 8 % returns down. If it goes almost horizontally, almost all is dispersed. Half of it up, half of it down...
Sooooo simple. You have to multiply by 6 to get 48 % from 8 %. Almost horizontal emissions produce six times more skyglow than near-to-vertical emissions.
Of course, there are some aerosols in the air, sometimes a negligible amount, sometimes they cause more dispersion than the gas (nitrogen and oxygen) itself. In spite of that, the above analogy holds. This is because aerosols scatter near to the original direction of light predominantly, and so the amount of downward returned light stemming from steep-up emissions is little affected by them. The above numbers hold well: for emissions at several degrees above horizontal plane, 6 is the proper ratio. Changing concentrations of non-cloud aerosols affect not this ratio, but the distance which the near-horizontal light emissions travel before being dispersed almost all. With very transparent air, it may be hunreds of kilometres, with turbid air, they pollute within the city itself.
As you see, it's no science, just common sense. Even the most stupid lighting engineers would be able to understand it, provided they wanted to. However, as Al Gore says in his famous Inconvenient Truth, it is difficult to make someone to understand the reality, when his salary depends on not understanding it. So there are professional denialists (criminals) not just at the field of climate change. Artificial lighting has many of them too. (What's the role of Dave in it, that has become a mystery for me in 2003 already.)
For emissions at 10 degrees up, the ratio is sure lower than 6. Look at my graphics scatt_en (google finds it at once). Why is it lower? For the same reason why Sun is not at all faint when at 10 degrees over horizon. One third of its light can often get to you directly, without being dispersed.
S's cavities are a shit, of course, as I remember it from years ago. As all the other "papers" which pretend that the lighting-caused increase of clear-sky luminance at night is a simple function of the total amount of upward-going light.
There was a note that denialists consider light propagation through the air to be "turbulent". Incredible... what's more easy to understand than the (direct) propagation of light, provided we are not blind? Light is really no fluid with vortices.
My programme ies2tab computes an important parameter, introduced by Pierantonio. This is the influence of light sources at vast distances from them. To the air above distant places, just the almost horizontally emitted light comes. The steep upward emissions are irrelevant. You can see the forward-scatter pattern easily: this is the nature of "light domes" above the settlements. It is no dome of turbid air above them. See the foils for my lecture in Bled, http://amper.ped.muni.cz/light/lectures/ds2007/200.pdf for some illustrations (200 could stand for 200 terrestrial miles, or three hundred kilometres). Similarly, you can see skyglow from the distant headlamps, if they are still hidden below the horizon.
For luminaires, 0.1 % emissions above horizontal plane means adding another (completely obsolete) 10 % to the produced skyglow at sites which are tens of kilometres from them. It scales up the same way for shallow bowls, with little light going steep up. So 1% allowance means twice more skyglow at distant sites, compared to the same illuminances produced by FS fixtures.
(FS: our short for 0 cd/klm anywhere horizontally or above. We need not endorse FCO or any other category demanding specific luminous intensity limits at downward directions -- e.g., for illuminating a one-way path from behind, strong emissions at 80 degrees from nadir might be appropriate -- remember headlights).
Jenik, in a hurry.