For the most part, the photographs are not well explained in the captions or the text. Some of the descriptions are incorrect. Some artifacts in the pictures are identified, but others are accepted as real. The attempts to simulate the observed phenomena in the laboratory are only qualitative, and not very helpful; so I'll try to explain just the solar pictures.
In reading the descriptions that accompany the photographs, one should be aware that the observer's estimates of color were usually strongly affected by adaptation (and, at sunset, by retinal bleaching) effects. The orange-red low Sun is usually described as “yellow” or even “white”, and the much redder colors shown on some photographs are mistakenly attributed to underexposure.
There are so few pictures in the Scientific American article that I'll just note them in passing. If that's all you have access to, here are links to them: the cover, the series on p. 113, the series on p. 114, the red flash from p. 115, and the thin pair of flashes from p. 117.
There's a slightly better-illustrated article in Endeavour, if you can find a library that has it for 1961. Links for it are: Figure 1; Figure 2; and Figure 4. The other pictures there are not in the book.
I'll go page by page through the book, and describe what's in each of the pictures.
The corresponding radiosonde data are on p. 177. The night-time soundings show a considerable thermal inversion, extending up to about the height of the observatory. The afternoon sounding, 3 hours before sunset, shows the normal daytime lapse rate, near adiabatic. The weather map on p. 116 agrees with the local reported wind direction, being driven by the surface pressure gradient from a small high-pressure system in northern Italy; the offshore wind carried the air heated in the afternoon over land out over the cold water. (Note that the date is one week into winter.) This caused the inversions that produced the mock mirages.
The considerable asymmetry of the Sun is due to waves on these inversions; note the diagonal structures that modulate the flashes in the 3rd and 4th panels.
The Sun was overexposed to show the green flashes properly. The overexposure saturated the red-sensitive layer of the film, making most of the Sun appear yellow. This is a common photographic artifact, which we see again and again in pictures of sunsets.
The ship in the bottom panel was photographed 2 hours before sunset. It is plainly considerably closer than the apparent horizon — note that the glitter pattern on the water extends beyond the ship, extending up to almost its deck — so of course it shows no miraging or distortion. (Note that a theorem of atmospheric optics shows that only objects beyond the horizon can be miraged.)
The conditions of no scintillation, and no wind, observed near sunrise, are typical for this time of day. The large literature on atmospheric turbulence is replete with examples of the calm periods for an hour or so around sunrise and sunset, when the ground and air temperatures are similar, and thermal turbulence is usually a minimum.
The “long sharp green spikes” noted in Fig. 2 are due to mock mirages; the Sun is again overexposed, to make the dim green rim visible. Two pictures on p. 53 are supposed to have been taken 2 minutes earlier, but they're full of clouds. Something's wrong, possibly a date.
Plate IV is discussed on p. 45. The following curious passage is readily explained:
Clearly, what's observed here is local patches of looming or towering, which first stretch the green rim enough to be visible, and then cause the bright area inside the rim to have its rising motion magnified as well, when it reaches the zone of great vertical magnification.
… wherever the green shows, the sun's rim seems to be momentarily held back regard [sic] to the rising edge of the sun's disc (middle photograph); then it suddenly shoots upwards and at the same instant the green disappears (upper photograph).
This isn't a mirage, but it's a related phenomenon. The image remains erect, but the vertical magnification varies rapidly with altitude (and, evidently, azimuth as well).
The irregular spatial variations in vertical magnification in the boundary layer are hardly surprising. During the night, different objects on the ground cool by radiation to the cold sky at different rates, depending on their thermal inertias and infrared emissivities. The air in contact with these objects forms stronger local inversions over the colder patches of the surface. In addition, because the line of sight here is slightly inclined, variations in the slope of the surface produce variations in the path length through the surface inversion: where the slope of the ground is similar to the slope of the ray, the path length near the cold surface is longer; but where the ground falls away from the line of sight more steeply, it is less, and the towering is locally less.
Both these plates show the effects of strong overexposure. The actual red color of the low Sun is seen only in the diffuse scattered light at the edge of the apparent horizon.
The overexposure is probably responsible for another effect: the correlation of the green parts in Fig. 2 with the higher features of the irregular horizon, and the apparent widening of the solar image (and consequent drowning out of the green rim by the locally overexposed disk) above the lower parts of the horizon. In the description on p. 45, O'Connell says the rim “seems pinched in” above the humps on the horizon; but it's really the other way around: the overexposed image spreads out more where the the horizon is lower, and overwhelms the green rim, “which was extremely faint on this occasion”. It's a sort of photographic “irradiation” effect.
The caption says these were taken Aug. 8, 1954 — the same date as Fig. 2 on p. 43 — but if this one was taken 11 minutes before sunset and that one was taken 9 minutes before sunset, the upper limb wouldn't have had enough time to get out of the could in those 2 minutes; yet the other picture shows a fine green rim and no trace of cloud. Maybe one of the dates is wrong?
Fig. 1 is supposed to have been taken 1.5° above the horizon, which would make the red flash features due to waves on an inversion just above the observer.
Fig. 3 here is Fig. 2 in the Endeavour article. The explanation of a ray appearance to the naked eye by a stack of mock-mirage flashes seems to be adapted from Mulder's book. Don't believe it.
The cause of the spikes must be much further off and at a great altitude, as Wegener also supposes.
is complete nonsense: the cause has to be close to the observer, where the line of sight is most nearly horizontal; and, in addition, Wegener never supposed anything of the kind — see my discussion of his principle. Additional pictures from this sunset are on p. 158, Figs. 1 and 3.
The cross-references to other pictures from the second sunset are also very useful: pp. 95, 162, and 163. These confirm the presence of an inversion with waves on it.
The extreme compression of the inferior mirage is due to the great height (450 m) from which these pictures were taken. As O'Connell remarks several times, the sea horizon is about 80 km away. As the layer of air in which the inferior mirage is produced is usually only a few meters thick at most, we can roughly estimate the angular extent of the mirage as the angular subtense of a few meters at 80 km. If we adopt 8 m as an upper limit to the thickness, the angle is 10-4 radians or 20 seconds of arc as the expected thickness.
The width of the inferior mirage (measured from the apparent horizon to the fold line, as indicated by the left-most extremity of the island image) is about 1 mm on the printed page. The Sun's horizontal diameter, which is unaffected by refraction, is 121 mm, which corresponds to about 32′ or 1920 seconds of arc; 1/121 of this is about 16 arc seconds — as good an agreement as one could ask for, considering the roughness of the thickness estimate.
Of course the connection with the upper atmosphere that is suggested on p. 62 is complete nonsense, as I have pointed out elsewhere, both in these pages and in print.
Because of the considerable altitude above the astronomical horizon, the irregularities here must be attributed more to “seeing” (the effect of turbulence on image distortions) rather than layered structure in the atmosphere. O'Connell's use of the word “scintillation” to describe these effects is somewhat misleading.
Really, a textbook example of a inferior-mirage flash observed from an unusually great height. (Earlier stages of this sunset are shown on p. 167 and Fig. 1 of p. 169, where one sees some mock-mirage features as well.)
The tip-off is the phrase “rays … which seem to start radially from a point below the horizon” — typical for crepuscular rays, which always appear to radiate from the Sun (whose center is below the horizon here).
Curiously, although O'Connell refers here to the demonstration of overexposure effects on p. 173, he fails to realize that this is the cause of the yellow color in the overexposed part of the Sun's image.
The radiosonde data on p. 81 do not extend to sea level, but do show a boundary-layer lapse rate near the adiabatic value of 10°/km. However, at this time of year (September and November) the water is still warm from the summer's heat, and the generally northerly winds were probably bringing in air cooler than the water. The weather maps on pp. 82–83 show cold fronts to the north and west.
What O'Connell is calling “layers of discontinuity” here is just a series of mock mirages, due to a corresponding series of thermal inversions.
The asymmetries visible in this and the previous picture are due to waves on the inversions. These waves have produced the upper-left to lower-right tilt of many fine details, due to the oblique viewing of the wave fronts. This accounts for the “upturned corners” noted by O'Connell. The lack of wind, and the “distant layers of haze” also are symptoms of thermal inversions.
The doubling of the inferior mirage shown in the top panel is puzzling.
The third image shows that the fold line is about 3 mm above the sea horizon on the printed page. But if you measure up from the extremities of the miraged image at the horizon to the lower limb directly above, you find that the zone of sky that's reflected is an order of magnitude larger than this. The compression of the mirage is about a factor of 10. (The note on p. 93 points out that the apparent asymmetry of the line is due to truncation of the right end of this miraged image by the film holder.)
The reason for the great reduction in height of the reflected image is that the effective reflecting surface nearly follows the curve of the Earth. You can think of the inferior mirage as a minified reflection in a convex surface. Because of the height of the observer (450 m), an appreciable part of the curved Earth is seen, foreshortened, at the horizon.
Notice how the line at the horizon (which is the inverted and compressed, miraged image of the zone of sky a few minutes of arc wide just above it) grows wider as more of the lower limb sinks into the miraged zone.
The fold line is also marked by the raised extremities of the miraged upper limb in the first two panels. These extremities are about 2 mm above the apparent horizon seen at the center of the image; but note that only 10 or 12 mm of the upper limb is still visible here. So the lowest point on the solar image is not really the apparent horizon, but is just the inverted (miraged) image of the upper limb.
This series of 5 appears on p. 114 of the Scientific American article.
The highly structured inferior-mirage flash shown in the right column appears to be modulated by bad “seeing” (turbulence) between the observatory and the horizon. This sunset also appears at the bottom of p. 61 and on pages 162 and 163.
The online AMS Glossary contains the term “surface of discontinuity”, which it seems to equate to a front. But a front is hardly a discontinuity; and in fact the density and refractive index of air cannot change discontinuously.
Note that the sequence of images on this and the succeeding pages is down each column, rather than across the rows.
What O'Connell calls a “blind strip” (his translation of Wegener's term “blinder Streifen”, for which I have preferred to use the phrase “blank strip”) is, in fact not one. As Wegener's drawing (adapted for use on p. 101) shows, the true blank strip remains constant in width as the Sun sinks behind it; it does not close up, as the gap between the main image and the red flash in the second picture on the left side of this page does. (For a good example of the difference between a mock mirage and a blank strip, see Fig. 8 of our mock-mirage paper.)
There does, however, appear to be a persistent discontinuity in extinction and refraction at the inversion, which might be a duct seen from above. However, the simulation of a mock-mirage sunset without a duct seems adequate to reproduce the main features (but without the inferior mirage at the horizon).
O'Connell's “luminous sea horizon” is just the inferior mirage.
These pictures form Figure 1 in the Endeavour article.
Once again, this is just a mock mirage, not Wegener's obscuring strip.
The high (cirrus) clouds shown in these pictures have, of course, nothing to do with the solar distortions.
The last column of pictures shows the development of the characteristic “hourglass” shape that often precedes a mock-mirage flash; see the simulations. The mock-mirage flash is O'Connell's “high detached strip which vanished with an intense emerald green.”
This is the first of a series of 5 plates showing the development of a single sunset in considerable detail. (See p. 176 for the radiosonde data.)
In the lower image, the inferior mirage has joined the main image, forming an Omega, as O'Connell notes.
As I have pointed out in my 2004 A. J. paper, the sharpness of the ship and the distortions of the Sun are explained by the ship's location between the principal planes of the atmospheric lens. It's also worth pointing out that the zone of sky against which the ship's superstructure appears is not very distorted in any case.
The picture is made more difficult to understand by the distant clouds, which are also miraged. These are the black features; the sky is gray, and the flash is white in these reproductions.
As usual, the “layers of discontinuity” are just mock mirages due to a stack of thermal inversions. This kind of complexly-layered inversion structure is quite typical after a few days of inversion conditions, when the sea-breeze circulation has had enough time to chop up the inversion structure thoroughly. The boundary layer becomes something like a piece of puff pastry, with many thin layers piled up in a stack.
The red flash appears on p. 115 of the Scientific American article; the green flash is on the cover of the Jan., 1960, issue.
There are, however, a few more sunset images to consider:
The deep red color reported for the first 3 images is typically associated with sunsets having abnormally large refraction, and hence abnormally large reddening. The large extinction is quite plain at the bottom, where the Sun fades out.
Fig. 4 shows features that are probably due to waves on the inversion responsible for this mock mirage. See p. 93 for later stages of this sunset with similar structure.
The pair of mock-mirage flashes in Fig. 5 is very pretty. Such pairs are surprisingly common. I believe they may be due to the splitting of a thick inversion layer by the onset of turbulent mixing in the middle of it, which produces a thin region with a nearly adiabatic lapse rate. (This black-and-white image appears on p. 117 of the Scientific American article. See Fig. 3 on p. 169 for a color version.)
O'Connell's comment that “individual waves can be distinguished in front of the sun's disc” calls for a rejoinder. The diameter of the Sun would be about 20 cm on the printed page, which corresponds to about 0.01 radian. The “waves” are several tenths of a millimeter high on the page; half a millimeter would correspond to 1/400 of a radian. At the horizon distance of 80 km, that would make the waves 2 meters high.
This seems unreasonable for fair weather in a body of water as restricted as the Tyrrhenian Sea. Besides, the “waves” are just as big in the upturned corners of the image, which are merely the upper limb of the Sun reflected by the inferior mirage. Finally, the irregularities in the apparent horizon — which are probably due as much to thermal fluctuations as to height fluctuations of the sea surface — are also responsible for the broken-up structure of the inferior-mirage flash (cf. p. 75), which O'Connell calls “disturbed” on both pages.
I have pointed out elsewhere in these pages the peculiar appearance of the sea horizon when there is an inferior mirage; this is yet another example.
The picture at the bottom of the page shows the complex structure of a mock-mirage flash that seems to be perturbed by multiple sets of waves on inversions.
Figs. 2 and 3 show in color the paired mock-mirage flashes that were shown in black and white on Fig. 5 of page 161.
Fig. 4 is a color version of the inferior-mirage flash shown in Fig. 3 on page 175.
Fig. 5 nicely shows the green tips on a thin mock mirage. However, O'Connell's comment about a “red lower border” is off the mark: the detached strip, as well as the main solar image, is yellow because of overexposure. The red border here is correctly exposed, and shows the true photographic color of the low Sun.
The flash “was brighter than the portion of the sun's rim just below” because of the strong gradient in extinction near the horizon; see the simulations of this effect. (The bottom row of pictures shows it best.) The extinction gradient is more obvious in the upper of O'Connell's pictures.
Something has to be said about O'Connell's remark that “Perhaps the mixture of blue and yellow may have appeared as green …”. Adding blue light to yellow light makes white; this notion that blue + yellow = green is a common misunderstanding. See the introduction to color for examples and details. In any case, this little flash near the horizon is much smaller than the rays reported by green-ray observers.
Here's an attempt to match things up. I've decided to put the various dates together by season, regardless of year. That's because low-Sun phenomena are seasonal, so that what happens in September is likely to be similar from one year to the next.
|Date||Image pages||Weather pages||ZD data|
|Mar. 22, 1954||61|
|Mar. 25, 1955||143|
|Mar. 27, 1956||133, 135||131||99|
|Apr. 10, 1956||171|
|May 10, 1955||63|
|May 20, 1956||59|
|May 31, 1955||169, no.5|
|June 13, 1954||73|
|July 6, 1954||143|
|July 7, 1956||51|
|July 12, 1955||41|
|July 15, 1954||43|
|July 17, 1955||49|
|July 17, 1956||47|
|Aug. 6, 1956||156|
|Aug. 7, 1956||100 (note)|
|Aug. 8, 1954||43, 53|
|Aug. 11, 1954||53|
|Aug. 12, 1954||75, 167, 169|
|Aug. 13, 1954||119 – 129||176|
|Aug. 30, 1956||105||102||99|
|Aug. 31, 1954||161|
|Sep. 13, 1956||107|
|Sep. 15, 1956||109||102, 178||99|
|Sep. 16, 1956||111||102, 178||99|
|Sep. 17, 1956||113||102, 179||99|
|Sep. 18, 1956||115||102, 179||99|
|Sep. 16, 1955||79||81, 82||99|
|Sep. 17, 1954||161, 169|
|Sep. 19, 1955||167||99|
|Oct. 7, 1955||95||97|
|Nov. 13, 1954||169, 175|
|Nov. 22, 1955||77||81, 83||99|
|Nov. 26, 1955||61, 95, 162, 163||97|
|Nov. 27, 1955||85 – 93, 161||98||99|
|Dec. 8, 1955||137, 139, 141|
|Dec. 28, 1955||3||116, 177||99|
I've kept the series of days in Sept., 1956, together. This has slightly displaced nearby dates in other years.
Copyright © 2006, 2012 Andrew T. Young
GF pictures page
main mirage page
GF home page
or the website overview page