Has anybody done a scientific study of the sources of light
pollution in a typical American city? It sounds like something
that would be fairly easy to derive from satellite photos or
reconaissance aircraft.

One tends to think of street lighting as the primary culprit
because it is ubiquitous, but I have my doubts. For one thing,
even the worst street lighting is fairly well shielded, as these
things go. I bet a conventional "bad" cobra light doesn't send
more than 20% of its light upward. The new fashionable acorn
lights are much worse, but even they are pretty well shielded
on top.

Recently, attempting to measure sky brightness at various spots
around town, it has dawned on me that ball fields are a truly
major source of light pollution. In Boston, Fenway Park is an
obvious culprit, but the Boston University field is every bit
as bright as Fenway, and is illuminated much more of the time.
It has a large, measurable effect on sky brightness within a
radius of one or two miles. Then there are Harvard's fields
and MIT's fields and the various municipal fields, which remain
illuminated increasingly many nights and increasingly long into
the night. I have measured nearly a 50% drop in sky brightness
between 9PM and 5AM, and I bet that a good chunk of that is due
to ball fields.

Other sources I can think of:

3. Automobile headlights
4. Parking-lot lights (municipal, industrial, commercial)
5. Industrial/commercial security lights (hard to separate from #4)
6. Residential security lights
7. Car dealers
8. Prisons

Anything else? Note that of these, ball fields, car dealers, and
prisons are particularly threatening, since they aim (for perfectly
valid reasons) to make outdoor areas as bright as day. By and large,
they use fairly well shielded lights to achieve that aim, ball fields
being probably the worst exceptions, but even a small patch of day
emits a mighty lot of light towards the sky.

- Tony Flanders