Visible Light Techniques

Clouds are visible due to reflected light. I remember a night in Mexico at about 8,000 feet elevation when I was out doing a star party for some folks in a remote village with absolutely no light pollution from the surface. I noticed that there were missing stars, whole constellation asterisms, in the sky. Very disturbing until I realized there were clouds blocking the view. But the clouds were not visible so the idea of detecting clouds from the St. Louis area will actually depend upon light pollution!

Look at these images I got from Dan Crowson's all sky camera in New Mexico. While those skies are undoubtedly much darker than those around St. Louis, even there there is enough light from the surface of the ground to detect clouds. I have applied labels to these images according to my personal tastes, ie, my feelings about whether I would jook forward to observing or just staying in bed. Your mileage may vary and we may seek some sort of consensus going forward.

I am looking for something about these images that would help me decide whether the labels are merited. In my judgment, dark is good, bright is bad. How much dark? How much bright?

These images are, of course, visualizations of arrays of numbers. Smaller numbers are visualized as dark, larger numbers are visualized as bright. Numbers are things we can work with a computer. We can easily count how many numbers of a given value (from 0 to 255) are in each image file. If we plot them out, it's called a histogram with which many imagers are quite familiar. I've visualized the histograms for each of the figures and I'm happy to report they are quite different and might be another means to recognize good, fair or poor. But they still require the human eyeball/brain combination.

On the other hand, if we (easily) compute what is called the cumulative distribution by adding up the total number of pixels by number value we can find out how many of the pixels are at or below a giving number. That can be hard to keep up with so we divide the sums by the total number of pixels to express it in percentages. We can visualize these cumulative distributions as I have done below.

Once again, we have some distinctive indications that the human eye/brain system can use to judge between and among the choices. But what the curves are showing is that, eg, the brightness levels where the curves cross, say, the 50% (or whatever) level. The computer can easily identify these points. Would you like to know that 90% of the sky is darker than some level you have judged (from experience) to be okay for a trip outside?