Learning from Colors

	This is a color picture; you can tell the sky is blue, the vegetation is green and so on. If you look closely, however, you'll see the clouds have colored fringes. That's because I made this picture the hard way: taking black-and-white images through colored filters, scanning the results and combining them in software. The clouds moved slightly while I was in the process, so the blue, green and red frames don't quite register. So eventually, with much trouble, I was able to do what your smartphone does in 1/100 second. But this was more than a simple demonstration.
	I also took pictures through an infrared filter (this was done on Kodak infrared film, which is no longer made, alas). Now I mapped the green image to the blue channel, red to green, infrared to red; so it's what you'd see if your eyes were shifted down a color in sensitivity. The sky gets darker, it's easier to see into the distance, and living vegetation turns bright red (I think it's cholorphyll that reflects so well in the near infrared). Kodak used to make a color film that did just this (also long gone, alas). Satellites in orbit often use their electronic detectors to track vegetation by making `false-color' maps in this way.
	I also took a frame in the ultraviolet. This shows what you'd see if your eyes were shifted up a color in sensitivity: UV mapped to blue, blue to green, green to red. It's not as spectacular as the previous picture, which probably accounts for the fact you don't see this done, and Kodak (as far as I know) never made a film to do this. Since UV gets absorbed in the air easily, you don't see as far; the mountains and the far range of hills are less distinct. (If you're interested in haze, though, this might be just what you need.) Note the difference in color between the semi-desert plants in the foreground and the crops in the valley.
	Here is what you get by running a mathematical operation with the red and infrared: (IR - R)/(IR + R). If these were properly calibrated images, this would be the Normalized Difference Vegetation Index (NDVI). It's a good measure of the quantity of growing plants, and is a common product of satellite imaging. Here there are some artifacts from moving clouds; also, since some of the vegetation in this image is seen through far more atmosphere than other parts, NDVI is not as good as a quantitative measure in this situation.
	Finally, I put all five colors together. UV is mapped to blue; blue to a middle blue-green; green to green; red to a middle red-green; IR to red. This is something like what you'd see if your eyes were sensitive to the invisible colors on either side of the visible spectrum. In principle, there's much more information in this picture than in the first one. I invite you to examine this and see what you can find. (I think I know why the sky is turquoise instead of blue. . .)