Like many stargazers, I refer frequently to the World Atlas of the Artificial Night Sky Brightness, more conveniently called the Light-Pollution Atlas. I use it to seek out dark sites for deep-sky observing and also to compare my own experience with that of other people who live too far away for me to visit.
By the admission of its own homepage, the Light Pollution Atlas has many failings. That's nothing to be ashamed of considering what it achieved. In one fell swoop, we went from a situation where light pollution had been measured at a handful of scattered locations in a few countries to having a map covering essentially every inhabited spot in the world.
The assumption has always been that whatever the failings, observing conditions in (say) the red zone — the second brightest — were roughly the same everywhere. But several people had cast doubt on this assumption for various different reasons. Subjective descriptions of red-zone conditions in the U.S. Southwest seemed systematically worse than descriptions of red-zone conditions in the Northeast by equally experienced observers. A year ago, I noted in a blog that the light-pollution blobs for some northern cities seemed implausibly big. How could Montreal possibly have the same amount of light pollution as Los Angeles, a metropolitan area nearly five times as big?
In a recent post in Cloudy Nights' Light Pollution Forum, David Lorenz published data that very strongly suggest that the original Light Pollution Atlas was systematically biased by the fact that snow was on the ground when the underlying satellite measurements were taken. Lorenz recalculated the light pollution for the U.S. and southern Canada based on snow-free satellite observations, and the whole northern part of the area came out roughly one full zone darker. That means that the original atlas overestimate the skyglow in this area by a factor of three.