Digital Camera as Sky Meter -- Ongoing Report

Here's another report in my ongoing quest to use an
inexpensive digital camera to measure sky brightness.
My primary purpose in posting this is to urge other
people to try the same thing. Cheap digital cameras
are more than sensitive enough to measure normal
suburban light pollution, although they're only
marginally capable of measuring dark skies.

All you need is a digital camera capable of taking
timed, multi-second exposures and an image-processing
program cable of resampling and of measuring the
intensity (or saturation) or any given pixel or
area in an image.

Take the camera, set it to maximum time and widest
f-stop, set it face up on the ground, and press the
button. When you get home, download the image,
resample roughly 10:1 to eliminate pixel-level noise
and faint stars, and measure the saturation of the
center of the image. On my cameras (Canon A60 and A80),
the response is nearly linear with sky brightness up
to 25% saturation and begins to climb more slowly
after that. But the linear or not, the readings
are quantitative, repeatable, and objective, which
is more than you can say more most measures of sky
brightness.

For instance, Rich Anderson recently posted that
the sky darkness seen from the top of a tall building
was half that seen from its base. I'm sure that it
did indeed decrease significantly, and I'm nearly
as sure that the decrease was nowhere near 50%.
Two readings with a digital camera would have
settled the question with just minutes of effort.

Is the sky near my city home brighter or darker than
an otherwise dark site at full Moon? Brighter, but not
by much. How do I know? My camera told me.

The sky is nearly 50% brighter once the leaves fall from
the trees, and another 50% when there's snow on the ground.
The city center is only 50% darker than where I live,
but I can decrease my sky brightness by a factor of 2
with a 5-mile bicyle ride and by 4 with a 12-mile drive.
After that, things fall off much more slowly. Facing the
best direction, my sky is 25% brighter 45 degrees above
the horizon than it is at the zenith, but 50% brighter
at 30 degrees. Even in the context of urban light
pollution, the sky is significantly brighter within
a mile of a lighted ballfield. The light dome of New
York City 120 north of Times Square is quite bright
a few degrees off the horizon and almost undectable 30
degrees up. I'm not guessing or biased -- I've
measured all of this.

One advantage of a digital camera, which may or
may not be significant, is that it can measure sky
brightness quickly over quite a large swath of sky --
about 40x50 degrees in the case of the Canon A60 and
A80. I have recently perfected the technique of
stitching 20 such shots into a brightness map of
the entire sky dome; it now takes me about 20 minutes
to set up and take the shots and another 40 minutes
to stitch them together on a computer with pretty
good accuracy.

In practice, a single zenith reading characterizes
a site reasonably well, but to get a really good idea
of directionality, you need 8 additional readings,
ideally about 30 degrees above the horizon. 45
degrees is easier to arrange, but different headings
blend together a lot at that height. Also, under
nearly-dark skies, light domes are a lot easier
to detect at 30 degrees than at 45. Once you've
characterized how your camera vignettes (a lot!),
you can get 30-degree measurements with the
camera angled 45 degrees up.

Two disadvantages of using a digital camera as a sky
meter are that the equipment is fairly expensive and
that the reading can only be obtained after downloading
the image to a computer, which is usually not possible
in the field. The expense isn't an issue if you already
own a suitable camera, of course.

I still haven't solved the thorniest problem of any
sky-brightness meter, namely callibration. I have
a very approximate estimate of how my camera readings
correlate to magnitudes per square arcsecond, but
that could easily be off by a factor of two.
Likewise, crude correlation between different models
of camera should be possible by measuring things
like the zenith at full Moon or the Cygnus star
cloud, but those could be off by at least 50%.
But that's still much better than comparing
zenith limiting magnitude among observers,
who frequently vary by a factor of 3 or more.

- Tony Flanders