I'm Andrew T. Young, an astronomer by training (which explains my connection to a Department of Astronomy). I got into the green-flash business in spite of a determination not to.
Several years ago, George Kattawar at Texas A&M University got me involved in a Navy project, in which we are studying the marine boundary layer by means of sunset images. I had always thought there was useful information in the shape (often highly distorted) of the setting Sun, and indeed there is.
I already had some awareness of boundary-layer meteorology, as I'd done some work back in the late 1960s on stellar scintillation and what astronomers call “seeing.” I had a long-standing interest in atmospheric optics, particularly atmospheric extinction (see my third chapter in Carleton's Methods of Experimental Physics, vol. 12, Astrophysics; Part A: Optical and Infrared) and calculations of airmass and extinction. Extinction involves both absorption and scattering, so I was fairly well prepared to handle sunset phenomena (I thought, anyway). After all, wavelength-dependent extinction causes the reddening and dimming of the setting Sun.
As I live not far from the Pacific Ocean, I became the photographer for this sunset project. So I began photographing sunsets with a telephoto lens.
I had expected to see some green flashes in the course of this work, and indeed I did. But I also realized that the Navy wasn't very interested in the phenomenon, so at first the green flashes were just a pleasant fringe benefit of the sunset work. I had read O'Connell's book when it first appeared, and figured “the green flash” was well understood. So I resolved to ignore green flashes.
But the more green flashes I saw, the more I became convinced that the story in the textbooks wasn't entirely correct. First, there was the variability: some sunsets show big flashes and others show little ones. Why? The variations in green flashes are much larger than the variations in refraction at the horizon. But the standard story is that the dispersion, and hence the length of the flash, is proportional to the horizontal refraction. Something is wrong here.
Then there was the discrepancy between the classical “last glimpse” flash (which I still supposed was just due to the horizon cutting off the green rim) and the flashes that appeared when much of the Sun was still above the horizon. How could those be explained? I took another look at O'Connell's book, and found no explanation of them at all (but several pictures).
There were theoretical discrepancies as well. One of the first papers I read in my literature search on sunset distortions was Alfred Wegener's 1918 paper on mirages, which is where the term “blind strip” used by O'Connell came from. I quickly realized that what O'Connell called a “blind strip” was quite different from what Wegener had intended. After adapting Wegener's simple but elegant model for the effects of inversion layers to computer simulation, I found that what we were seeing was produced by an inversion layer below eye level, not (as in Wegener's original model) above.
So we published a paper explaining this curious phenomenon that Wegener should have discovered, but by chance overlooked; we call it a “mock mirage” because its optics are quite different from ordinary mirages. Actually, it's closer to a phenomenon Wegener did discover, which he called a Nachspiegelung or “late mirage”.
While I was doing the computer simulations for the mock-mirage paper, I figured it might be useful to do them for two wavelengths, red and green, and display the results. This immediately showed me that the green flashes caused by strips that separate from the upper limb while the Sun is still above the horizon are indeed mock-mirage flashes. Then it was easy to try GF simulations for the ordinary inferior mirage; and they showed me that this mirage greatly enhances the “last glimpse” green flash.
But there is a striking difference between the two kinds. The inferior mirage produces a zone of sky in which the image is stretched vertically, and the flash appears near the height of maximum stretching. But the mock mirage produces image compression, not stretching; yet it too is associated with a green flash of greatly exaggerated magnitude. Clearly two very different effects are at work here.
In November, 1996, I was trying some additional simulations that correspond more nearly to Wegener's model (i.e., with an inversion layer overhead), and discovered yet another kind of green flash. This corresponds to the extension of Wegener's Nachspiegelung into the space below an optical duct. Optically, it has some characteristics of the inferior-mirage flash (image magnification) and some of the mock-mirage flash (refraction through inversion layers). I began to wonder just how many more kinds of green flash Nature might have up her sleeve.
Considering that neither the mock-mirage flash nor my flash below a duct had been recognized in the atmospheric-optics literature as distinct phenomena, I began to think about finding a source of money to work on these problems.
Meanwhile, I'd noticed that most of the green flashes I saw turned out yellow on color film. At first I thought this was simply because I was too fast on the shutter release, and was jumping the gun on the developing flash. But, even after learning to wait a second or so for the green to develop fully, I found I was still getting yellow flashes. Could I really be seeing the wrong color?
I wrote to Pekka Parviainen, our Finnish co-author who is far more experienced in sunset photography than I am, and suggested that maybe visual adaptation was playing a part in what I saw. “Of course,” he told me. “Compare what the eye that stares through the viewfinder sees with what the other eye sees.” And of course he was right; I'd been keeping my other eye closed, and it saw the Sun much redder than the eye that stared at the Sun. Then it was only a matter of digging up the appropriate references from the literature of vision to understand what is happening.
The major effect here is bleaching of the photopigment in the long-wavelength cones by the very high brightness of the setting Sun. No other everyday light source is bright enough to produce significant bleaching; and adaptation effects in the unbleached retina are much milder than those in an eye lacking erythrolabe. So the effects are both spectacular, and unfamiliar to most people. In fact, the result has often been described as temporary color-blindness (protanopia).
You can easily duplicate the effect if you have a flashlight with a red lens, as most astronomers do. Put the flashlight up to your eye and stare into it, looking at the bright red image of the filament in the bulb. After a minute or two, you'll notice that the light no longer looks deep red, but a less-saturated orange. Now look at a low-pressure sodium-vapor street light: instead of its usual yellowish-orange color, it will appear yellowish green.
The atmosphere has a transmission window just to the short-wavelength side of the sodium D lines, so the yellow flash that is seen in this wavelength window when the longer wavelengths have been cut off appears “emerald green” to the red-bleached eye. So most “green” flashes seen at sunset are really yellow ones, as the color photographs show. This fact is hinted at on the Bishop Museum Web page, where this appearance is called the “fool's flash,” though it's mistakenly called an afterimage there. (Actually, the solar afterimage is much dimmer and bluer, and is so inconspicuous that I know of only 2 or 3 published observations of it.)
Yet the Party Line in the green-flash literature is that because the flash is not (as has often been suggested) an afterimage, there are no visual effects involved. Well, that's bad logic. Ruling out afterimages does not rule out other kinds of visual effects. And there are even experiments of textbook quality in the GF literature (conveniently ignored by everyone from O'Connell onward) that demonstrate this. So some corrections are needed to the standard accounts.
Once I started digging into the GF literature in earnest, I naturally found many other errors in need of correction. Everybody makes mistakes; but not, in my experience, with the frequency they have been committed to paper in this field. I think some of this is due to people regarding the GF as a sort of fringe phenomenon, not worthy of serious study or careful scholarship. And part, surely, has been due to the crude, qualitative, natural-history approach that was the only thing available before computers became powerful enough to churn out a whole sunset simulation in a few minutes.
Don't think this is all the work of one person. I've had help from a great many people whose cooperation has made a big contribution to what you see on these pages. They are too numerous to list here, but I have a separate page to acknowledge their part in this effort.
You can reach me by email here.
© 1999, 2003, 2005 – 2007, 2010 Andrew T. Young