For 11 years I monitored the zenith sky brightness at the 2800-m level of Mauna Kea. These are observations in the V-band (550 nm) and B-band (440 nm), using a single-channel photometer that employs an uncooled RCA 931A photomultiplier tube. The telescope is a 15-cm f/5.82 Newtonian reflector. The area of the sky measured is 6.522 square arc minutes. Some details of the observations can be found in this paper:

Krisciunas, K., 1990, "Further measurements of extinction and sky brightness on the island of Hawaii," Publications of the Astronomical Society of the Pacific , 102 , 1052-1063.

I have just submitted another article to the Publications of the Astronomical Society of the Pacific which further discusses these issues.

Given below are the V-band data (top) and B-band data (bottom) averaged by year.

I now have covered an entire solar cycle. If you compare the graph of the 10.7 cm solar flux below (from the Algonquin Radio Observatory, Ottawa, and the Dominion Radio Astrophysical Observatory, Pentiction, British Columbia) with the sky brightness graph, you will see clearly that when the sunspot activity is at a minimum, the sky is darker. This is because the solar wind activates chemical reactions in the Earth's upper atmosphere. These chemical reactions give rise to the permanent aurora, or nightglow.

Now consider a plot of the zenith V-band sky brightness vs. the 10.7-cm solar flux, where the solar flux values are those data obtained on the day prior to the evening on which the night sky brightness was measured:

The open triangle is for the year 1985. The open circle is for the year 1993. The latter point may have been affected by ash from the Philippine volcano Pinatubo, which affected Hawaii's skies for two years after its explosion in June of 1991. The least-squares line is as follows:

B_zen = (47.0 +/- 2.78) + (0.2019 +/- 0.0222) X (10.7 solar flux),

where B_zen is the V-band sky brightness at the zenith in nanoLamberts. (To convert to S_10(V), the number of 10th magnitude stars per square degree, divide nL by 0.263.) This means that without the airglow we would still have light from "blank" sky. I find that this amounts to 13.0 nL from V = 13 to 19 stars, 33.9 nL due to zodiacal light, and 0.14 nL due to extragalactic background light. The airglow contribution ranges from 11.0 to 49.5 nL, a factor of 4.5 over the solar cycle.

The largest source contributing the internal scatter of night sky measurements appears to be the variations on time scales of tens of minutes like those observed by the Whole Earth Telescope project. An example is shown below, where the sky counts have been corrected to the zenith.

If you are trying to image the faintest galaxies, you will want to observe at solar minimum when there is also no evidence of volcanic effluvia in the atmosphere. Otherwise you must take longer integrations, since imaging the faintest objects is a "contrast effect".