Strange streak discussion: 2004 Dec 7 APOD

[Mike K]

This is only a guess after studying the interesting photograph.

I suspect that the photographer captured the precise moment when the lamp bulb in the light pole exploded causing a bright flash and projecting the shadow from the light pole into the darkening sky back toward the photographer.

Other observations:

The shadow in the sky seems to be narrower near the pole and broader (and increasingly diffuse) at height which would be expected if the dark streak is the shadow of the light pole was projected back toward the photographer.

Near the light pole, there appears to be an irregular yellowish cloud below the pole. It could be that the lamp pole contained a high wattage industrial lamp bulb (Metal Halide, Mercury Vapor or High Pressure Sodium) and the small cloud below the pole is the vapor residue from the exploding bulb.

Industrial outdoor lighting lamps can require several hundred watts of electrical power and can generate up to 50K lumens. During the ignitor stage (when the lamp is first turned on) the ballast generates much higher short term voltage to get the lamp started. The atmospheric conditions indicate that the time of the photograph is near sundown - the entire scene seems to be backlit by high clouds illuminated by the near horizontal rays from a setting sun. Most outdoor lighting is turned on automatically at dusk. As an earlier contributor commented, the lamp may have failed and flashed brightly just at the moment it was started.

A question to ask the photographer is whether the lamp housing at the top of the pole extended outward and away from the photographers view point thus allowing the pole shadow to be projected back toward the photographer.

[victorengel]

OK. Now that I know how to use the poll software, and since I unintentionally omitted the contrail theory, I've created a new poll. Please don't use the other poll anymore. If you already voted in the previous poll, please vote in this poll as well. Only one vote per computer will be counted this time.

http://www.opinionpower.com/Surveys/377021291.html

[Dave]

Thanks victorengel. I have now been able to see the streak, so I can understand better what is being discussed.

[victorengel]

OK. Now that I know how to use the poll software, and since I unintentionally omitted the contrail theory, I've created a new poll. Please don't use the other poll anymore. If you already voted in the previous poll, please vote in this poll as well. Only one vote per computer will be counted this time.

http://www.opinionpower.com/Surveys/377021291.html

If you don't want to vote, but you're curious about the results, you can use this results link to view them.
http://www.opinionpower.com/results.cgi?id=377021291

[SC Guest]

Has the photographer confirmed that the photographs were made using the G3's flash? It seems very odd (and pointless) that he would choose to take flash photographs of clouds. Yet the bug hypothesis seems to rest on the use of the flash and its synchronization.

I apologize if this question has been answered elsewhere on this forum and I missed it.

[PassingBy.za]

Also, there seems to be a curve up top left side...

[Guest]

The shadow of a contrail falls on hazy air.
Often contrails are straight lines.
When you are inside the shadow looking along it you can see the shadow. Look at right angles and you cannot see the shadow.
When the contrail drifts due to wind, you are no longer inside the shadow and cannot see it.
This is why it is visible only briefly.

Paul Doherty
Senior Staff Scientist
The Exploratorium

I couldn't have said it better. tks. As for the event on the wharf, who cares, it has nothing to do with the shadow.

My $.02, as with the water in the forground the lightpost/bulb/hood/glass is reflecting the bright light from the growing thunderhead in the background and in the picture it appears as a flash (allow for wind movement). As for the "smoke", maybe someone just started up a diesel engine (maybe one with a leaky headgasket).

A little bit of information can be dangerous.

John (head in the clouds) vB
Williamstown, Ont.

[victorengel]

Also, there seems to be a curve up top left side...

Can you give some context for this image? I can't identify what it corresponds to.

[wattie42]

this looks to me,a ground to sky lighting trail

[Guest]

As far as the flash is concerned, is the only evidence for it firing the imbedded file info? Perhaps there was a flash in the boot that was not turned on? I would think the photographer could easily verify this, and it would put a lot of people at ease since the bug theory depends entirely on having the flash fire.

The foliage in the foreground does not appear to have speculars from a flash. In particular, the bushes in the lower left appear to be entirely backlit, which I would not expect with a flash and a wide-angle lens.

Has anyone addressed this?

[Guest]

for the fourier-analysis crowd, mapping this image into fourier space (a.k.a. frequency space or k-space) will give you **spatial** frequency information, because this is a spatial image. I fail to see how this spacial frequency can be accurately interpreted as temporal frequency (a.k.a. time frequency, what the laymen normally think of when discussing frequency), and using the shutter speed as a time metric would not be very accurate or justifiable.

What metric or mapping was used to convert spatial frequency to temporal frequency? It seems to me that any fourier analysis could only be interpreted if this mapping was perfectly linear (i.e. everything in the picture was moving at a constant velocity for the duration of the shutter speed)... which could be an accurate assumption for slow-moving objects.

[Steve Bright]

I apologise for my earlier post "Insufficient Evidence", which was obviously incorrect.
I can now supply a full explanation:
Spectral analysis shows the streak is composed of Unobtainium. A minor flaw in the invisibility field of the Mother Ship (estimated to be >>50km in length) momentarily revealed part of its structure.
The extreme energy in the flash means it must be an X Ray laser being used to analyse construction of lamposts on Earth by our curious interstellar visitors.

[Cloudbait]

for the fourier-analysis crowd, mapping this image into fourier space (a.k.a. frequency space or k-space) will give you **spatial** frequency information, because this is a spatial image. I fail to see how this spacial frequency can be accurately interpreted as temporal frequency (a.k.a. time frequency, what the laymen normally think of when discussing frequency), and using the shutter speed as a time metric would not be very accurate or justifiable.

What metric or mapping was used to convert spatial frequency to temporal frequency? It seems to me that any fourier analysis could only be interpreted if this mapping was perfectly linear (i.e. everything in the picture was moving at a constant velocity for the duration of the shutter speed)... which could be an accurate assumption for slow-moving objects.

The goal was to attempt to isolate the wingbeat pattern of an insect. This would manifest as a regular intensity variation along the flight path.

Nobody was looking for a temporal signal (except to the extent that the temporal pattern of the insect wingbeat was translated to a spatial pattern). The entire image was not examined in frequency space, only the portion of the image along the shadow path.

The absence of such a spatial frequency component showing up doesn't really argue against the insect theory, since the transparent wings would be expected to produce a very weak signal in silhouette, and the intrinsic noise level of the image is high.

[Luis]

I do not like the idea of the bug theory. So I did a bit of geometrical optics to find if it is possible to capture a bug as blurry as that with the known settings of the camera.

My assumptions are:
f/5.6
f=9.1mm (and this equal to the distance from the lens to the CCD chip)
pixel size 3.1microns (From G3 canon spec)
Assume an ideal thin lens

With these values I can find the aperture of the iris through the relation ship
f/number=f/a=f/2r; a the diameter and r the radius of the aperture

with this, a couple of diagrams and the lens equation

1/l'=1/l+1/f; f focal length; l' image distance and l object distance.

I wrote the following function in matlab

function [lp,bl,bl_px,m]= thin_lens_blur(l,l0,f,fn);

r=f/(2*fn); %Diameter of the iris aperture


f0=1/(1/f-1/l0); %Focal value needed to bring the object at distance l0 (light pole) on the CCD detector

lp=1/(1/l+1/f0); %Distance from the lens to the perfectly focused image of the bug with the focus value calculated above

bl=(lp-f)*r/lp; %Size of the blurred image of a sinlge point source at the sames distance as the bug

bl_px=bl/3.1e-3 % Above value in pixels

m=lp/l/3.1e-3; %Magnification pixel/mm. That is multiply the size in mm of your bug and find out size in pixels on the image. e.g. if m=-6, the size of a 10mm bug will be 10mm*6pixel/mm= 60pixels

and looped it up

k=0;
for n=100:10:1000
k=k+1;
[lp(k),bl(k),bl_px(k),m(k)]=thin_lens_blur(n,1000000,-9.1,5.6);
lbug(k)=n
end

Then did a few plots

plot(lbug,bl_px)

If someone tells me how to upload a picture I'll put it on.

The result shows that for a a bug at about 40 to 60cm from the lens the blur size would be between 5 to 10 pixels... which is roughly right.

if you then plot

plot(lbug,m*10)

you'll see that a bug of 10mm accross would be imaged over approx 60 pixels if the thing is 50cm away from the lens.

This together with the other plot shows that the blur of a point source at 50cm from the lens would be about 5 pixels accross.

I hate to admit it, but it all makes sense, if it was a bug, it was about 1cm in length and was about 50cm from the lens.

Of course all this provided my maths are right and that there are no bugs in my code

[anthony]

The solution must take into account all the observations that can be discerned in the picture. The shadow lines up exactly with the light; this is most likely not a chance observation. There is a flash that seems to emminate from the lamp. There is also something that looks like "smoke" emminating from a point near the base of the lamp post. If this is smoke from the lamp, it strongly suggests that there should be some mark on the pole. That should be checked. If there is no mark, we can reasonably disassociate the smoke from the other observations. All acceptable solutions must then take into account the shadowy line and the flash.

[Luis]

The solution must take into account all the observations that can be discerned in the picture. ... All acceptable solutions must then take into account the shadowy line and the flash.

And the only one that seems to stand robustly is the buggy one... Although I don't like it. It seems to explain all the bits in the picture. Nasty bug!

[Guest]

This is a very long thread! I have not read all messages, but I recently created this image in Photoshop and thought I would share it



This image can be seen by using the unsharp mask in Photoshop with settings around the following:

Amount: 393
Radius: 28
Threshold: 148

(Lowering the Radius value significantly gives some very interesting, striking images.) It is very clear this image is NOT of a "bug", nor of a flare in the water, nor of a great number of other things. The image is, at the VERY least, that of an exploding lamp.

In my opinion, this image is EXACTLY what it pruports to be - the image of a meteroite striking (or NEARLY striking) the light pole.

JC

[sfarmer]

Find out who the guy is in beside the car in the woods. What ever it was, was enough to scare him into ducking in the 18:52:52 and 18:53:07 images. There is also another guy sitting on the rocks near the shoreline. He jumps up in the 18:53:07 image. I don't know about the flash on the light pole or the "smoke trail". I did notice many lights are on in the area and others are not. Maybe its getting dark enough for the lights to start coming on and this one blows as it is turning on. It has a spotlight effect and makes the "smoke trail" appearance in the haze. As far as the smoke around the light pole, it could be someone welding. This looks like some kind of boat docking station. The welder does his weld and the smoke arises. As he "sparks" the electrobe to continue his weld, the flowing current flashes and momentarily lights up the smoke in the air. An arc welder would give off the bluish colored smoke. Maybe he wasnt welding in the other images. Maybe the welder is welding near the light pole and the power surge from the welder causes the light to blow. Maybe the light has a transformer inside it as many do. This would make a great flash as I have seen these blow before.

Maybe.

sfarmer

[sfarmer]

Find out who the guy is in beside the car in the woods. What ever it was, was enough to scare him into ducking in the 18:52:52 and 18:53:07 images. There is also another guy sitting on the rocks near the shoreline. He jumps up in the 18:53:07 image. I don't know about the flash on the light pole or the "smoke trail". I did notice many lights are on in the area and others are not. Maybe its getting dark enough for the lights to start coming on and this one blows as it is turning on. It has a spotlight effect and makes the "smoke trail" appearance in the haze. As far as the smoke around the light pole, it could be someone welding. This looks like some kind of boat docking station. The welder does his weld and the smoke arises. As he "sparks" the electrobe to continue his weld, the flowing current flashes and momentarily lights up the smoke in the air. An arc welder would give off the bluish colored smoke. Maybe he wasnt welding in the other images. Maybe the welder is welding near the light pole and the power surge from the welder causes the light to blow. Maybe the light has a transformer inside it as many do. This would make a great flash as I have seen these blow before.

Maybe.

sfarmer

[redxeth]

I think we've come down to the conclusion that it's either a bug in the image or a lamp exploding.

We all need to contact the author of the photograph in order to get more info regarding the state of the lamp, which as implied on the APOD page "The light pole near the flash has been inspected and does not show any damage, although the light inside was not working."

We need more photographs of the state of the lamp before we can proceed further.

My first analysis was that it was a bug -- however I am biased toward this conclusion because I assumed an undamaged light meant that the light was not the cause.

If, however, it can be shown that a light burning out CAN indeed cause this image-- I may change my verdict.

We also need to know from the author is if there were indeed bugs in the area near the time of the photograph.

[victorengel]

As far as the flash is concerned, is the only evidence for it firing the imbedded file info? Perhaps there was a flash in the boot that was not turned on? I would think the photographer could easily verify this, and it would put a lot of people at ease since the bug theory depends entirely on having the flash fire.

The foliage in the foreground does not appear to have speculars from a flash. In particular, the bushes in the lower left appear to be entirely backlit, which I would not expect with a flash and a wide-angle lens.

Has anyone addressed this?

I mentioned it, but I have no answer. This, to me, is the biggest problem with the bug theory.

[victorengel]

for the fourier-analysis crowd, mapping this image into fourier space (a.k.a. frequency space or k-space) will give you **spatial** frequency information, because this is a spatial image. I fail to see how this spacial frequency can be accurately interpreted as temporal frequency (a.k.a. time frequency, what the laymen normally think of when discussing frequency), and using the shutter speed as a time metric would not be very accurate or justifiable.

What metric or mapping was used to convert spatial frequency to temporal frequency? It seems to me that any fourier analysis could only be interpreted if this mapping was perfectly linear (i.e. everything in the picture was moving at a constant velocity for the duration of the shutter speed)... which could be an accurate assumption for slow-moving objects.

The point is that the photo was taken during a finite amount of time, which, if the bug theory is correct, maps directly to a line in the picture. One end of the line corresponds to the start of exposure. The other end to the end of the exposure. So there is a direct correlation between time and space. The insect image is what makes the connection.

[victorengel]

The shadow lines up exactly with the light

Not really. It is a bit high. The bottom edge of the shadow comes close to hitting the top of the lightpole. But consider this. At the angle of the shadow line, and given the width of the shadow line, the odds of some part of the shadow line intersecting the top of a light pole by my eye seem to be close to 50:50. That's pretty good odds.

this is most likely not a chance observation.

If my odds just mentioned are anywhere near correct, it very well could be, especially since the alignment is a bit off.

There is a flash that seems to emminate from the lamp. There is also something that looks like "smoke" emminating from a point near the base of the lamp post. If this is smoke from the lamp, it strongly suggests that there should be some mark on the pole. That should be checked.

I think others have stated that this was checked and nothing was found.

If there is no mark, we can reasonably disassociate the smoke from the other observations.

Or conclude that what looks like smoke is really something else.

[Ed in Oregon]

Victor,
I can't find it now (in this horribly long thread. Is there a way to search the thread?) but I believe I saw a confirmation from the photographer that the camera flash was intentionally on to force the 1/20 second exposure to get the clouds right.

I wish that all the other images he took that night were available, so we could search for other bugs in those images. They've got to be there. I've looked at the three we have and am starting to see things. (As you have noticed.)
Ed

[victorengel]

Find out who the guy is in beside the car in the woods.

Facts not in evidence. Seriously, there is a lot of jpeg artifact noise around the perimiter of the car. The size of a person standing next to the car would be about the same size as the jpeg artifacts. I seriously doubt whether you can even tell if someone is beside the car.

What ever it was, was enough to scare him into ducking in the 18:52:52 and 18:53:07 images.

Those other images are different images, hence the jpeg artifacts are slightly different. I challenge you to demonstrate that there is someone actually there.

There is also another guy sitting on the rocks near the shoreline.

Again, where? I think you're just seeing jpeg artifacts.