The Sky Quality Meter (SQM) and its close cousin the lensed Sky Quality Meter (SQM-L) have revolutionized the study of artificial sky glow. For the first time in history, amateurs can purchase a relatively inexpensive device that can measure sky brightness well down below the level produced by natural sources such as airglow, starlight, and the zodiacal light.
On a purely individual basis, this makes it possible to calibrate one's observations against an objective measure of light pollution. Like many others, I take SQM and/or SQM-L readings for every serious observing session. These help me confirm my own gut feeling about which nights are better than others at my three or four customary observing sites, about how those sites compare with each other, and about how they stack up against places that I only visit once or a few times during the course of my travels.
At the social level, the SQM (and SQM-L) make it possible to calibrate other people's observations against one's own experience. If I hear somebody say "I can see NGC 3628 from my suburban backyard," I have very little idea how bright or dark that backyard really is. A suburb of a small city in the western U.S., surrounded on three sides by desert, is a very different thing from a suburb of New York City. And even within a single metropolitan area, suburbs have a huge range of brightness. But if someone tells me "I can see NGC 3628 from my backyard, SQM reading 19.0," then I have a very good idea how bright that backyard really is.
However, there is a tendency to overinterpret SQM readings, because SQMs have numerous problems that severely limit their accuracy.
First of all, SQMs are based on a sensor whose sensitivity varies depending on the temperature. The SQM then adjusts those raw sensor readings according to its own internal thermometer, so that in theory it always gives the same result regardless of the temperature. Lab tests indicate that it does this pretty well. But the real world isn't a lab. It is a known fact that SQMs give wildly variable readings until they equilibrate thermally, just as telescopes give poor images until they equilibrate. For this reason, I and many other take a series of readings, and only start recording the results once the readings have stabilized, which in my experience usually happens within a few minutes.
Once the SQM has equilibrated, I then take five consecutive readings, and record my results with a standard deviation, e.g. 21.12+-.03. My deviations are usually down to +-.02 or +-.03, but occasionally as big as +-.06 or worse. That's a pretty big range!
Second, SQMs have a truly gigantic field of view. They are most sensitive to the region within 45 degrees of the optical axis, but they do get some light from as far as 75 degrees from the optical axis. That means that any directly visible light more than 15 degrees above the horizon can render the results all but meaningless. It is, for instance, almost impossible to get meaningful SQM readings when the Moon is up. In those cases, of course, the SQM reads too bright. More subtly, visual obstructions such as trees -- including trees behind one's back that one's not even aware of -- can make the SQM read too dark. In my part of the world (the eastern U.S.), it's a pretty rare observing site that doesn't have some obstruction high enough to affect SQM readings.
That is why the SQM-L was invented. It still has a vast field of view by telescopic or binocular standards, but it does have near-zero sensitivity to light 45 degrees off axis, making it usable (with care!) in typical suburban settings, or when the Moon is up but not too high.
A less obvious fact is that the optical axis of both SQM and SQM-L units is not necessarily well aligned with the physical axis. When you think they're pointed to the zenith, they may in fact be reading 10 or 15 degrees from the zenith.
Finally, although any individual SQM or SQM-L gives pretty stable readings over moderate time periods, there is quite a bit of variation among different units. I am now proud owner of two SQMs that read roughly 0.15 magnitude different from each other, and two SQM-Ls that read roughly 0.19 magnitude different from each other. In a recent thread (sorry I can't find it!) someone else reported inter-unit variations as high as 0.4 magnitude.
To make matters even worse, individual units can drift over time. I began to suspect this of the first SQM that I ever purchased, because its difference from my SQM-L had clearly changed over a matter of years. I shipped the SQM back to Unihedron, who confirmed that it was no longer calibrated correctly, and supplied a new unit for free.
Now for some purposes a variation of +-.2 around the norm is not a big deal. A typical darkish urban site or very bright suburban site reads around 17.8 to 18.2; there's not a huge difference in the appearance of deep-sky objects within that range. Both are quite different from skies in the 18.8 to 19.2 range, which is characteristic of medium-dark suburbs.
But at the dark end of the rage, where most of skyglow is natural and artificial light pollution is a trace contaminant, a difference of +-.2 is very significant indeed. I can vouch that there is a vast difference in the appearance of the Milky Way -- or of galaxies through a telescope -- between skies that read 21.3 on my original SQM and skies that read 21.7 on the same unit.
I have to conclude that attempting to draw conclusions about two different dark sites based on readings of two different SQMs are futile. If one site reads 21.8 and the other 22.0, that's as likely due to variations between the two SQM units as it is due to actual differences between the sites.
Likewise, drawing conclusions based on comparison between an SQM unit and an SQM-L unit at the same site are meaningless. Some time ago, Don Pensack noted that his SQM-L always read darker than his SQM -- which one would expect based on theoretical considerations. But my own SQM-L usually read brighter than my SQM, which baffled me considerably. Now I know that was due to the fact that my SQM had drifted dark. But Don's observations on this subject are equally suspect; there's no way to determine how much of the difference between his SQM and SQM-L is due to physical reality and how much is due to inaccuracies of the units themselves.
And if (like me) you want to measure how skyglow is changing at any given site over a matter of years or decades, you are well advised to own at least two different SQM units, and preferably more, to compensate for any possible instrumental drift.