Strange streak discussion: 2004 Dec 7 APOD

[Ed in Oregon]

Just to add one more fact. Here's the flash curve for an electronic flash. This may explain the bright flash on the rear of the bug without requiring that the bug's rear be reflective. It may also give us the clue that the flash was at the end of the exposure. We may just be seeing the flash curve on a moving bug.

And before the flames start, I'll take back the "one pair of wings" statement. Hymenoptera have two pair, but they are usually deployed together and are seen as one pair.

[Ed in Oregon]

[quote="FastArtCeeToo"]By self-admission I do not know (yet) what caused these phenomena, but I hope someone will come up with the answer.


It seems to me that if you compound these probabilities, you'd be down to vanishingly small.

Go back to page 98 and read Rob's message. He visited the site at dusk. (Yes, we have actually put a man at the scene of the crime!) He says the air was full of bugs. Flying ants to be precise. He thinks it would be very hard NOT to take a picture of a bug.

[Guest]

Just to add one more fact. Here's the flash curve for an electronic flash. This may explain the bright flash on the rear of the bug without requiring that the bug's rear be reflective. It may also give us the clue that the flash was at the end of the exposure. We may just be seeing the flash curve on a moving bug.

And before the flames start, I'll take back the "one pair of wings" statement. Hymenoptera have two pair, but they are usually deployed together and are seen as one pair.

Sorry, Ed in Oregon, but I don't know what to make of this. Can you explain what it is and how it relates to the question please?

[Ed in Oregon]

That's the light output curve of an electronic flash. It ramps up to extremely bright in about 100 microseconds, then slowly decays, in an exponential manner to a low level in about 2 milliseconds. The most intense part of the flash is the beginning, and that lasts only a couple of hundred microseconds. This will expose the bug very brightly for a brief period, then much less brighly over a longer period. The actual "length" of the bug that we see in the image may be a motion blur with the "flash" part of it as just the beginning of the blur where the exposure is the most intense.

[Guest]

Can everyone please put any and all explanations involving meteorites or space debris to rest. We've shown time and time again that the physics doesn't work.

No, you have only shown time and again what meteors usually do.

As others have mentioned, there are shapes and materials which would allow an object to retain the majority of its cosmic speed as it impact the Earth. Granted 10,000 tons would guarantee an object's speed, but other possibilities exist and fresh meteorite falls have very little investigative science behind them as so few have ever been witnessed.

This object may be falling at 200 mph [slow terminal velocity for a round rock but fast for a man - he'd have to be curled into a ball]. If that's a 45 degree trajectory then it has a 200mph horizontal velocity and it's velocity at the water would be 282.84 mph or 20.74 feet during the 1/20 sec. shot. At that speed it would have been 1 1/5 miles away in the previous photo.

But if the smoke/fog is consistent with what would be expected of a very bright flash on these optics, then we might be able to lay the dead bugs to rest.

[Ruidh]

This object may be falling at 200 mph [slow terminal velocity for a round rock but fast for a man - he'd have to be curled into a ball]. If that's a 45 degree trajectory then it has a 200mph horizontal velocity and it's velocity at the water would be 282.84 mph or 20.74 feet during the 1/20 sec. shot. At that speed it would have been 1 1/5 miles away in the previous photo.

You don't appear to understand terminal velocity. If the 200mph were the terminal velocity, then it could not have a velocity of some 280mph at the water. The net effect of the terminal velocity is to rotate the velocity vector toward the vertical. Your 1+ mile away meteor would strike almost vertically long before it got near the scene in Darwin.

But if the smoke/fog is consistent with what would be expected of a very bright flash on these optics, then we might be able to lay the dead bugs to rest.

I don't know why you are so hostile to a bug theory. It is much more likely than an unusual meteorite.

[victorengel]

- It would have to be flying more-or-less parallel to the focal plane of the camera, within a narrow range of distance. (Surely 19 out of 20 bugs would have been flying at some angle to the camera.)

Or it could be flying toward or away from the camera but stay within a confined range. The range would depend upon the angle of view of the camera.

- It would have had to be closely aligned with the body of water.

Why? I don't understand this assertion.

- It would have had to have its wings aligned in a nearly straight line at the moment of camera flash.

Again, the flash is not an instant. It is spread over a period of time. It is not a stretch to say the wings lined up during some segment of time.

- If it was a hymenoptera (four-winged) the pairs of wings would have had to be lined up at the moment of the camera flash.

See my previous comment. Furthermore, the fore wings and rear wings typically flap together, so we can eliminate that alignment as a problem. Even if it were an issue, all we'd have to require was that one pair of wings never aligned.

It seems to me that if you compound these probabilities, you'd be down to vanishingly small.

I don't see any of these as being a small probability. Perhaps you could show some calculations.

[Can't use my Bad Buoy]

If the 200mph were the terminal velocity, then it could not have a velocity of some 280mph at the water.

True. So logically 200 mph cannot be the item's terminal velocity. And if the dark streak is not a wispy trail of smoke but instead the object's passing across the landscape, then it is far exceeding mach I [1056 ft/sec].

An item's terminal velocity is highly dependent upon it's shape and mass - the old feather vs. rock proof. Items fall to Earth constantly which do NOT fall vertically. The space shuttle, though a special case, is a case in point. It comes from space with NO propulsion so it is falling. It arrives near the surface in excess of mach I. And at touchdown it's vertical velocity is close to O. It survives with no, a little to the tiles, ablation. Of course here, with this example of the shuttle, I've used an object of a mass your arguement seems to allow.

I suggest there are other shapes which would allow durable, dense matter of much less mass to survive a much higher terminal velocity.

But if we could get some responses on this camera's optics I think it would help us all.

P.S. And if this board's team is approving registrations I'd appreciate being able to use my regular handle of Bad Buoy. I have received no validating email in the past week.

[H0meAl0ne]

Can everyone please put any and all explanations involving meteorites or space debris to rest. We've shown time and time again that the physics doesn't work.

No, you have only shown time and again what meteors usually do.

As others have mentioned, there are shapes and materials which would allow an object to retain the majority of its cosmic speed as it impact the Earth. Granted 10,000 tons would guarantee an object's speed, but other possibilities exist and fresh meteorite falls have very little investigative science behind them as so few have ever been witnessed.

This object may be falling at 200 mph [slow terminal velocity for a round rock but fast for a man - he'd have to be curled into a ball]. If that's a 45 degree trajectory then it has a 200mph horizontal velocity and it's velocity at the water would be 282.84 mph or 20.74 feet during the 1/20 sec. shot. At that speed it would have been 1 1/5 miles away in the previous photo...

Here are the sorts of numbers that make me skeptical of any debris related hypotheses.

Terminal velocity for a meteorite 1m (3') in diameter, made of solid osmium (the densest substance known at 22,500kg per cubic meter vs 8,000 kg per cubic meter for iron) hitting Earth at lowest possible speed (11kms sec) at 45 degrees is approximately 2.4 km/s. Faster than the speed of sound (0.3km/sec in dry air, sea level pressure, 25 degrees C) but not fast enough to melt target or vaporise itself upon impact (need > 12 -15km/sec for that). It would however make a crater about 50m (150') across in dirt/rock).

Boost impact velocity with atmosphere to 72 kms sec (fastest possible impact with Earth for an object orbiting the Sun) and we get an impact fast enough (approx 15.5 km/sec) to allow for a fireball. Should be easy to spot too from the camera position 1km away, as it will appear to be five times larger than the Sun, in the middle of a 100m (350') diameter crater.

That seems a bit large, so we can reduce the diameter of the meteorite back to 0.1m (approx 100kg of osmium) but as soon as we do air resistance takes its toll and the impact velocity drops sharply back to sub-sonic speeds (0.09km/s). If you assume it is on a hyperbolic orbit and keep boosting its initial velocity then the terminal speed never increases while more and more energy is expended in ploughing through the upper layers of the atmosphere so that eventually (at an initial velocity of around 5,000 km/s by my calcs) it will breakup in the atmosphere several tens of kms up.

So, unless we do something to keep the mass while reducing the air resistance we will have problems getting something small enough to hit the Earth fast enough to have noticable but negligible effects.

If it's going fast it will be incandescent (persistent smoke trails that would reach to the edge of the image , sonic booms, bright streaks on the image - not dark ones, radar reflections from the ionised air, large waves/craters). If you postulate a small object then air resistence will slow it down till it is sub-sonic.

It can't be big enough to show in the image and yet small/slow enough to not have a big impact unless you are willing to hypothesise something like a solid osmium crowbar with fins. Even then you'll have to show me the math before I will believe it is possible. And even then I'll be skeptical at the probability of such a thing existing.

For those that are interested / destructive there are some impact simulators on the web (eg http://www.lpl.arizona.edu/impacteffects/ ) that let you toy around with input parameters, but unfortunately they do not seem to include the option to modify the shape (all assume spherical as far as I can see).

[victorengel]

Velocity is a vector. That means it has a direction. In the case of terminal velocity, that direction is normally straight down. Not always. Take a strip of paper and drop it. It falls at a terminal velocity that has a tragectory similar to the angle of our featured object. Mighty coincidental, don't you think?

P.S. And if this board's team is approving registrations I'd appreciate being able to use my regular handle of Bad Buoy. I have received no validating email in the past week.

I don't think the moderators will see your P.S. I don't remember where it is ... probably on the main forum page, there is a link to contact a moderator. Several options, like AOL, ICQ, etc. are given. When I had trouble registering, I selected one of those and connected with a moderator right away. He resolved my registration problem while I waited.

[cdsmith]

I've read several posts that tried to discount the bug theory due to the fact that the "bug" is aligned with the light post. They say that this is unlikely to happen.

This is a completely false assumption. If it is in fact an object near the camera, then whatever is in the background has no effect on where the "bug" might be when the camera flash goes off. In fact, any pixel in the image is just as likely as any other to be the point where the "bug" is when the flash goes off.

If it is improbable that the bug would be aligned with the light post, then I ask, what should it have been more likely aligned with?

[Cloudbait]

Items fall to Earth constantly which do NOT fall vertically. The space shuttle, though a special case, is a case in point. It comes from space with NO propulsion so it is falling. It arrives near the surface in excess of mach I. And at touchdown it's vertical velocity is close to O.

The space shuttle is aerodynamically unstable. If it didn't have a computer continually adjusting its control surfaces, it would arrive at the surface with a vertical speed a good deal faster than zero, and at a considerably more vertical angle then it does when controlled.

Objects do not fall to the Earth regularly at angles significantly far from vertical. I know of no cases at all where low mass objects have done so. The only conditions under which this could be an object falling from space with supersonic or hypersonic velocity would be if it massed several tons, was powered, or was actively controlled. Any of these are unlikely in the extreme.

[FastArtCeeToo]

Several posters have stated that there are lots of insects of suitable size, with suitable speed, where the photo was taken. Somebody opined that it would be difficult to get a photo without catching a few insects.

I don't know if the original photographer is looking in on this discussion, or if anyone has his email address, but I wonder if he could set up in the same way at the same time of day and take a photo every, say, 10 seconds for, say, 200 photos (in batches, if necessary).

Surely in that sample we should be able to see a few streaks and flashes if the cause was an insect.

In fact, as there are many identical cameras around the world, and many buggy places, couldn't we have ten volunteers each take 1000 photos with a similar setup?

Anybody second this motion?

PS. And Mr Bike Wheel, maybe you could get a pet bee to fly on a tether while being gently guided around a circular path at speed. You might be able to contribute a few useful photos as well! javascriptemoticon('')
Wink

[FastArtCeeToo]

If there were many bugs flying about, something of them should have shown up in the photo. If they were just a bit further from the camera we should see a shorter streak and a weaker flash. If they were still further away, we might not see any streak, but we should still see a weak flash. Past some distance, of course, we would see neither streak nor flash even if there were dozens flying about.

But between streaks and/or flashes for some kinds of bugs, and maybe just short streaks (all the way down to just black dots) for larger bugs that were flying either directly toward, or directly away from camera.

I've scanned the photo by eye, and looked at the results of subtracting images, and I can spot only one candidate flash, at location (x=888, y=1346), but it's at exactly the same location in all three photos. Could it be a camera artefact? The colour seems somewhat similar to the main flash. Any close examination is impossible because of jpeg artefacts.

If this is a reflection of some sort, I guess we can't say anything about its true location except to say that it's somewhere along a line from the camera flash to the water surface.

I don't see any obvious dark streak or dark dot candidates, either in water areas of the photos or in the sky areas.

Has anyone else tried to pursue this angle?

[Can't use my Bad Buoy]

As others have mentioned, there are shapes and materials which would allow an object to retain the majority of its cosmic speed as it impacts the Earth.

It can't be big enough to show in the image and yet small/slow enough to not have a big impact unless you are willing to hypothesise something like a solid osmium crowbar with fins.

...there are some impact simulators on the web, but unfortunately they do not seem to include the option to modify the shape (all assume spherical as far as I can see).

Yes, too bad shape is not a variable as it has a major impact on atmospheric drag. And just a rod, sliver, shaft of material with a slight flare in one end would be enough to act as an arrow.

What a fun impact tool. Now everyone can test the effects of their neighborhood nuclear superiority. 8)

And I note that the impact tool states "The most probable angle of impact is 45 degrees".

But, whatever the item is [most likely something from a jetliner taking off or landing at Darwin's International Airport close by], it is entering the water at subsonic speed and has kicked up a tall splash/spray. Either the object or the splash has picked up the last of the setting sun [6:56pm at Darwin on Nov 22, 04]. The color spectrum of the item/splash has been noted by others as very similar to the light on the water's surface or from the cloud.

The event may have unfolded high over the viewer's left shoulder and is making an almost verticle plunge into the water a short distance in front of the viewer [giving an apparent 45 degree track]. And in that case the dark streak is the object during it's last 20th second and represents less than 880 feet. Anything further would be supersonic.

And as for anyone else noticing this event, the odds are small. The event took place during one frame [at best two] of the eye. The impact would have seemed less than of seeing a whale blow, which events themselves are fleeting enough. It is very dubious that, even if aimed in the same direction, any security camera would have recorded anything, though their exposure rate is usually greater than every 15 seconds.

[STONEY]

I ONLY STUMBLED ACCROSS THIS TODAY (12/16/04) CALIFORNIA TIME.

AT FIRST I WANTED TO KID ABOUT IT BEING A "WORM HOLE" (THEY MIGHT REALLY EXIST).

THEN I THOUGHT OF A LASER BEAM FROM A MILITARY SATELITE, BUT I READ OTHER REPLIES THAT MENTIONED THIS AS WELL A METERIORITES.

BUT MY BACKGROUND AS AN ELECTRONICS ENGINEER FOR 35 YEARS
MADE ME MORE SUSPIOUS ABOUT THE IMAGE SENSOR OF THE CAMERA.

SOMEONE ASKED FOR IDEAS TO ASK THE CANNON ENGINEERS SO HERE
ARE SOME OF MY THOUGHTS.

WHAT IF A COSMIC RADIATION PARTICLE HIT THE IMAGE SENSOR.
(I AM TALKING ABOUT A REAL THING. "RADIATION PARTICLES" ARE THOUGHT TO BE ONE OF THE WAYS THAT GENES IN REPRODUCTIVE EGGS CARRIED BY FEMALE ANIMALS CAN CHANGE THE GENETIC STRUCTURE WHEN IMPACTED BY SUCH RADIATION AND CAUSE ERADIC MUTATUIONS.)

IF A RADIATION PARTICLE HIT THE IMAGE SENSOR IT COULD CAUSE A BRIGHT SPOT THAT WOULD BE SEVERAL PIXELS IN DIAMETER.

THE PERFECTLY STRAIGHT LINE (VAPOR TRAIL AS SOME DESCRIBE IT) COULD BE AN ATRIFACT OF THE PROCESS OF DOWNLOADING THE IMAGE FROM THE ("CHARGED COUPLED DEVICE" TECHNOLOGY) INAGE SENSOR. THIS PROCESS SHIFTS THE ANALOG CHARGED COUPLED PIXEL INFORMATION FROM THE SENSOR SURFACE TO DETECTORS AT THE EDGE OF THE DEVICE. THIS IS PERFORMED BY ROWS AND COLUMNS BEING SHIFTED BY A FORMAT DEFINED BY THE DESIGN OF THE IMAGE SENSOR.

PUT THESE 2 THINGS TOGETHER AND YOU COULD HAVE A REALY RARE COMBINATION OF REAL THINGS.

A RADIATION PARTICLE THAT HITS THE IMAGE SENSOR AT THE RIGHT TIME TO CAUSE A BRIGHT SPOT AND CAUSE A NOISE DISTURBANCE THAT EFFECTS SOME KIND OF STRAIGHT LINE SHIFTING OF THE DOWNLOAD PROCESS OF THE COLUMN AND ROW ARANGEMENT OF THE PHOTO SENSITIVE PIXELS MADE UP OF ANALOG (NOT DIGITAL) CHARGE COUPLED ELEMENTS.

I KNOW THIS SOUNDS LIKE FICTION BUT THIS IS AN APPROXIMATION OF HOW THE BLOOD AND GUTS OF DIGITAL CAMERAS WORK.

FEEL FREE TO PRESENT THIS TO THE "CANNON CAMERA ENGINEERS" AND SEE WHAT THEY THINK.

MY THOUGHTS ARE THAT THIS IS A MUCH MORE REALISTIC POSSIBILITY AND SHOULD QUIET MANY FEARS OF SUPER NATURAL SPACE OBJECTS.

ALL OF THIS COULD HAVE TAKEN PLACE WITHIN THE CAMERA.

DENNIS STONE
ELECTRONIC ENGINEER

[Can't use my Bad Buoy]

ALL OF THIS COULD HAVE TAKEN PLACE WITHIN THE CAMERA.

DENNIS STONE
ELECTRONIC ENGINEER

Then the smoke/fog could not be an artifact of the lens since there was no flash out in the real world.

I understand your theory, Dennis. Anything consistent with your theory which would explain the final element to this puzzle?

[victorengel]

It would be nice if you posted in mixed case instead of all uppper case, which is harder to read.

SOMEONE ASKED FOR IDEAS TO ASK THE CANNON ENGINEERS SO HERE
ARE SOME OF MY THOUGHTS.

It's Canon, not Cannon.

IF A RADIATION PARTICLE HIT THE IMAGE SENSOR IT COULD CAUSE A BRIGHT SPOT THAT WOULD BE SEVERAL PIXELS IN DIAMETER.

OK. But that doesn't explain the dark trail.

THE PERFECTLY STRAIGHT LINE

It's NOT perfectly straight.

(VAPOR TRAIL AS SOME DESCRIBE IT) COULD BE AN ATRIFACT OF THE PROCESS OF DOWNLOADING THE IMAGE FROM THE ("CHARGED COUPLED DEVICE" TECHNOLOGY) INAGE SENSOR.

Please explain how this could be the case.

ALL OF THIS COULD HAVE TAKEN PLACE WITHIN THE CAMERA.

I guess I missed something. Please describe the steps that occur within the camera. Items that need explanation:

* The trail does not go all the way to the edge of the picture.
* The trail is angled at 33.6 degrees, approximately. This does NOT conform to a natural geometry of the sensor elements.
* The trail is not perfectly straight. It undulates gradually first one way and then the other, grossly. On a finer scale, it seems to have a frequency of about 220 Hz.
* What process or set of processes would produce both the lighter "flash" and the darker trail?
* Why are the before and after pictures not affected?
* Explain how the effects observed do not result in saturated pixels.
* Explain what was unique inside the camera for just the middle picture. Why wasn't the effect visible in the before and after pictures?

[DC]

The first time I did a reasonable Diff file, I had to compress the color space to make it work in my image processing program, making it lose a little color resolution. Probably someone has done this correctly earlier. I just finished another one with a mathematical ((Event + 128) - (Before + After)). I also created an (Event - (Before + After) with negative differences clipped to black, called "PixelsGettingLighter".

Using the calculated diff, I recreated the compressed trail picture like the one someone else did earlier, hoping for a little better color resolution. This time I didn't get the orange color I got previously. The questions are whether the trail is straight, wobbly, or bends downward at the end.



This is a portion of the PixelsGettingLighter image, first shown unmodified, then with auto levels and a threshold mask, then with just auto levels. One slender black line below the flash seems to suggest the flash and the smoke/reflection/whatever is behind the pole, not in front of it. Some of the filter images below suggest the opposite, so the truth may be obscured by JPG artifacts.



I was also able to cobble together a homespun bandpass filter. In the following pictures, it shows a bandpass for the colors around the object, then about the same area in color with the band subtracted, then the whole picture for that band subtracted.



I notice two things in these pictures. First the angular part of what I call the event reflection is generally at the same angle as the color changes in the water. This might increase the probability that is actually a reflection of some sort. In the larger view, however, notice that the black band of sky reflections off the water (where the light of one specific color used to be) curves around the back to join the other side, while the event "reflection" is very straight. I followed up on this and superimposed a mask from my PixelsGettingLighter image over the filtered image with the black band. That shows that if my "reflection" really is a reflection, it is not following the pattern of the reflections from the sky, because the mask goes past the black band.



The other thing i was looking for in these filter images was some color component in the foreground water suggesting a reflection of a flash. In some images I see a suggestion of that, and in others any lightness in one spot seems to be a continuation of the same light band in the background water.

As to what this thing really is, I'm still clueless. (But I don't think it's a bug)




This is the picture where the extra foreground water reflections seem to be a continuation of background water reflections.

[Can't use my]

* The trail is not perfectly straight. It undulates gradually first one way and then the other, grossly.

Not that a JPEG picture isn't acceptable for this analysis as the position of the compression scheme's decision to change contrast may vary over the length of the trace.

But if this is falling debris, then as it twists and turns the trace could appear darker to either side of the path as rotation and possible tumble change during descent.

[Can't use my Bad Buoy]

* The tail is not perfectly straight. It undulates gradually first one way and then the other, grossly.

If this is falling debris, then as it twists and turns the trace could appear darker to either side of the path as rotation and possible tumble change during descent.

[Ernst Lippe]

I've found someone at Canon Support who might be able to answer questions about the PowerShot G3 and the EXIF data. By policy, Canon Support can't participate directly in the discussion group. If anyone has specific questions on the G3 and EXIF, please post them - and please be concise - I'll compile and send them on to this guy (Anthony), and we'll see what we get back. I'll give it 24 hours to collect questions.

Thanks,

JJ

That would be very interesting.

The most important point is the duration of the flash and if it was first or
last curtain sync. If the bug was moving at a uniform speed, the flash must
have been shorter than 1/1000 s. It would be very nice if we could have a
graph of the amount of light versus the time.

Another important point is the response curve of the camera under these
conditions. It seems very likely that it is non-linear. Along the dark streak
the intensity of the light will be reduced by approximately 1% (this is a
rough approximation because we don't know the size of the insect) and we would
like to know if this is consistent with the pixel values in the image.

Also if we had a full response we would be able to compute better difference
images. At the moment the difference images are based on subtraction of the
pixel values, but the correct model for the streak is based on multiplication
of the light intensity. In the subtraction images there is still some minor
dependency on the brightness of the background, but with a response curve we
might be able to eliminate this.

Of course, all other pieces of information about the image are also welcome.

[Luis]

By self-admission I do not know (yet) what caused these phenomena, but I hope someone will come up with the answer.

- It would have to be flying more-or-less parallel to the focal plane of the camera, within a narrow range of distance. (Surely 19 out of 20 bugs would have been flying at some angle to the camera.)

No it would not!
If you see the graph below, you will see that the size of the insect would vary about 2 pixels if it was moving from 40 to 60cm away from the camera. If the bug trajectory moved from 40 to 50cm the bug size would vary only 1 pixel. And as you get away from the camera the change in size is even smaller.



Anyway I think that the apparent constant size of the streak, only tells us that the bug trajectory was "parallel" to the lens with an error of about 50%. That is if the bug was at about 40cm, the length of the streak was of an insect moving 20cm from the beginning of the streak to the flash. In this space of time the bug moved away from the camera no more than 10cm...

So that one you can rest assured is not a vanishingly small probability

[/quote]

[Guest]

Does this look familiar?

http://antwrp.gsfc.nasa.gov/apod/ap010219.html

[victorengel]

The first time I did a reasonable Diff file, I had to compress the color space to make it work in my image processing program, making it lose a little color resolution. Probably someone has done this correctly earlier. I just finished another one with a mathematical ((Event + 128) - (Before + After)). I also created an (Event - (Before + After) with negative differences clipped to black, called "PixelsGettingLighter".

These equations don't make sense. Maybe you mean (Before + After)/2 whenever you say (Before + After). I will assume so.

Using the calculated diff, I recreated the compressed trail picture like the one someone else did earlier, hoping for a little better color resolution. This time I didn't get the orange color I got previously. The questions are whether the trail is straight, wobbly, or bends downward at the end.

To my eye, it looks like from left to right it goes down, up, straight, down.

This is a portion of the PixelsGettingLighter image, first shown unmodified, then with auto levels and a threshold mask, then with just auto levels. One slender black line below the flash seems to suggest the flash and the smoke/reflection/whatever is behind the pole, not in front of it. Some of the filter images below suggest the opposite, so the truth may be obscured by JPG artifacts.

Yes, Jpeg artifacts get in the way here. Just compare the other lamp posts. For a good diff picture of ideal pictures, you shouldn't see any of the lamp posts. You can make them out, though, because the jpeg artifacts differ between frames.

I notice two things in these pictures. First the angular part of what I call the event reflection is generally at the same angle as the color changes in the water.

Did you take any measurements? I respectfully disagree with this assertion. First, let me establish the a positive angle means leaning to the right at the top (like 1 o'clock on a clock face) and a negative angle means leaning to the left (like 11 o'clock). The top "wing" has a negative angle relative to the bottom "wing". The general trend of the reflections, however, goes the other way. The "insect" feature, furthermore is angled about 25 degrees right of vertical. The general trend of the overal water reflections, on the other hand, is angled about 67 degrees right of vertical. This doesn't seem particularly close to me.

This might increase the probability that is actually a reflection of some sort. In the larger view, however, notice that the black band of sky reflections off the water (where the light of one specific color used to be) curves around the back to join the other side, while the event "reflection" is very straight.

If anything, the event "reflection" curves the opposite way.

I followed up on this and superimposed a mask from my PixelsGettingLighter image over the filtered image with the black band. That shows that if my "reflection" really is a reflection, it is not following the pattern of the reflections from the sky, because the mask goes past the black band.

I don't follow.