[astroCS] cloneni a jas nebe
jhollan at amper.ped.muni.cz
Thu Feb 19 16:00:18 UTC 2004
...a uz posledni vec, totiz prosba o foceni bezmracneho nebe (zenitu, ale
i jinych oblasti) kdyz lezi snih a kdyz zadny neni.
Dominance vlivu primeho svetla z lamp na jas nebe je prakticky hlavni
vec, kterou musime dokazat, a dokazat by mela jit velmi dobre. Fabio
Falchi takove jedno mereni ma, a zda se byt zcela presvedcive (napadlo ho
to uz pred lety).
Na druhe strane, zprava Focusu (ta ma jen 3 MB) rika, ze lidem lampy
samotne prekazeji pri koukani na nebe vic, nez to, ze nebe neni tmave...
takze jejich cloneni, aby z dalky nebyly videt, je hlavni veci uz z tohoto
toz to bylo do tretice a naposled,
preju ciste nebe na tu jasnou kometu pod Algenibem (tou levou
---------- Forwarded message ----------
Date: Fri, 13 Feb 2004 16:15:23 +0100 (CET)
From: Jan Hollan <jhollan at amper.ped.muni.cz>
To: DarkSky-list at yahoogroups.com
Cc: OutdoorLighting-Forum at yahoogroups.com, magnitude6 at yahoogroups.com,
ida at amper.ped.muni.cz
Subject: exact answer on shielding and skyglow
How much would be sky luminance diminished
using fully shielded fixtures:
an exact answer is possible easier than I'd ever think!
How? comparing sky luminance for snow-covered land and a snow-less one.
Yesterday I got an idea, how we can easily assess the contribution of
direct light from the luminaires (together with light from lit vertical
surfaces) to the skyglow. It is applicable in all countries with at
least occassional snow cover. What's needed is to take raw-format
digital images of the sky in two clear nights when all deciduous trees
are without leaves: one night with bare, dark ground, the other one with
If the two nights would have the same transparency of the air
(expressed by the so-called zenith extinction, I'll comment on it
later), the comparison is straightforward: the amount of the light from
the lit terrain rises exactly as the albedo of the terrain does, say from 12
per cent to some 90 per cent (I'll be able to give more accurate albedo
figures within a year), i.e., some eight times, when the ground becomes
completely covered with snow.
The part of the skyglow stemming from the (not completely avoidable)
light dispersed by the ground rises eight times as well. However (as we
know well, of course), the total artificial contribution to skyglow
does not rise so much. Or the skyglow itself, in most urbanised areas,
as the natural skyglow is negligible without aurora. If it would rise
just twice, it would mean that almost half of it is caused by (mostly
avoidable) light from other sources. My guess is that it rises even a
lot less in most areas, esp. in countryside far from the main light
sources, and that the role of the light another than that from the lit
ground will be proven to produce most of the pollution.
Fully shielded fixtures would reduce this part to almost zero, together
with strict limits on billboard and facade lighting luminances. The
minor contribution of lit windows can be assessed by taking images in
the evening and in the late-night hours, when most windows are dark.
Even uncalibrated cameras can be used for this purpose, just giving the
relative sky luminances. Any camera which offers RAW format in its
setup is OK for this purpose. The needed software (compiled for linux only)
is an adapted dcraw programme, and my raw2lum (see
http://amper.ped.muni.cz/light/luminance), but you don't need to
compute it yourselves, just take the images, they can be processed in
Apart from the images of the sky, you should always take an image with
the lens covered, a ``dark frame'', with the same exposure time and ISO
sensitivity, ideally just after taking the sky image. This has to be
subtracted from the sky image, to get reliable luminance/radiance
values (raw2lum does that).
For getting the estimate of air transparency, two methods are
available. One of them is using a luxmeter for the setting (or rising)
Sun. The amount of direct sunlight (you get it by subtracting another
reading made with a small shadow cast onto the luxmeter sensor) is
dependent on the air transparency, and luxmeter errors play almost no
role with the Sun being low in the sky. The accuracy of getting the
``extinction coefficient'' (attenuation of light for stars or Sun being
high in the sky) can be something like 0.02 mag, or two per cent, quite
sufficient for all purposes. Definitely, if you measure several times,
with solar altitudes of some 20, 10 and 5 degrees.
(My program http://amper.ped.muni.cz/light/lun_illum.php can be used to
get the answer, with putting "c0" on the command line for case of Sun
instead of Moon and something more to get the sensor illumination, like
c0 i55:10 t16 ze20
to try zenith extinction of 20 cmag (very clear air) today at 4 p.m.,
in the latitude of +49 degrees.)
The other method assumes you record some stars in the image, like
Capella high in the sky. The star is fainter due to extinction of its
light in our atmosphere, and if the air transparencies were the same,
the attenuation should be the same as well. Of course, to get the sum of
light for the star takes some time, but it is possible e.g. with my
raw2lum. This is of course the standard astronomic way of estimating
extinction. For common digital cameras, the image HAS to be a bit
defocused, so that the star image covers completely four
pixels at least.
In case the transparencies would be different, the sky luminances can
be still compared, just knowing the amounts of ``zenith extinction''.
In the simplest model, the sky luminance is proportional to this
quantity, so having attenuation of light by 0.4 mag instead of just 0.2
mag by ``one airmass'' means doubling the sky luminance. All
measurements can be transformed to standard conditions this way, i.e.
to zenith extinction of 0.3 mag.
So, if you have either a scientific CCD camera, or can borrow a
suitable common digital camera, please take sky (and dark) images on
clear nights. And either log the azimuth and height of the image, or
use a short focus, so that enough stars would be discernible in the
image (imaging zenith only by lying the camera on the ground is an easy
alternative). And of course, log the ground coverage by snow.
We know from theory, that conversion to fully shielded fixtures should
help a lot in all respects, that it's a major cure for light pollution.
But advocates of the current lighting practice won't read any papers on
that. The only way to convince them is to get hard data. We can get lot
of them this winter and so obtain an extremely strong argument this
(If you have no camera, there is still a way of measuring the sky
luminance. It's enough to have a PV panel and a digital voltmeter. For
illumination of tenths of a lux and less, the voltage is simply
proportional to illumination. My panel, one square decimetre with eight
silicon wafers in a series, measures illuminations of one mililux
easily, so even the sky 10 km from a city produces a good signal.)
clear skies for imaging/measuring,
(any questions please off-DSLF and -OLF as well, I don't manage to read
I've just finished a public version of the research report made for the
Czech environment ministry -- for those reading Czech, one of its
addresses should be http://amper.ped.muni.cz/noc perhaps in
As a part of it, we have developed several methods to get values of the
``brightness of the sky'' (sky luminance, or sky radiance).
The purely astronomical way is through standard CCD stellar photometry,
description of the procedure in English is still missing (but it's
rather simple). Some insights into getting reliable estimates of
``extinction'', which are necessary to interpret sky luminance data, have
been gained, with useful implications for astronomical photometry in
general. Any CCD photometry archives should be able to give the
information on the sky luminance, and most photomultiplier archives as
well. If we would get climatic data mentioning the snow cover, we
should be able to use even these old data for getting the answer what's the
contribution of well avoidable light to sky luminance.
More information about the astroCS