Some Refinements to the Gimbal Sim

You won't see very far with a regular camera at night. An infrared camera resolves temperature contrasts, but saying there is no limit to it is directly contradicted by the Gimbal video. It shows clouds features getting blurry and fading out with distance.

Why I'm asking, up to where in the image do people here see these distant clouds, that have their brightness unaffected by distance? Where is this crystal clear cloud-sky line?
 
TheCholla said:
You won't see very far with a regular camera at night. An infrared camera resolves temperature contrasts, but saying there is no limit to it is directly contradicted by the Gimbal video. It shows clouds features getting blurry and fading out with distance.

Why I'm asking, up to where in the image do people here see these distant clouds, that have their brightness unaffected by distance? Where is this crystal clear cloud-sky line?
Click to expand...
The line of sight to the clouds is entirely within the atmosphere, so, of course, atmospheric extinction (mostly water vapor, some aerosols), will affect it with distance. But what distance is it noticable at? Does place some limit on the reconstruction?

Zaine seemed to be saying that inverse square was placing some natural limit that forced them to be closer that in Sitrec. They may well be, but I don't think they HAVE to be.
 
Mick West said:
Does place some limit on the reconstruction?
Click to expand...
Of course it does, the evolution of elevation angle in the reconstructions (a few hundredths of degree) depends on the assumption that we follow a same cloud line, at the exact same elevation angle. Very uncertain without knowing where the clouds stop in the field of view.
 
TheCholla said:
Of course it does, the evolution of elevation angle in the reconstructions (a few hundredths of degree) depends on the assumption that we follow a same cloud line, at the exact same elevation angle. Very uncertain without knowing where the clouds stop in the field of view.
Click to expand...
Well, eventually we'll have better clouds in Sitrec. Hard to get excited about a few hundredths of a degree, with the various other uncertainties.
 
Exactly, a few hundredths of degrees is what you force in Sitrec, from the distance between (what you consider) the top of the clouds and the object. But cloud motion angle suggests significantly more, a few tenths of degree.

Cloud motion is off in Sitrec, look at your katana shape versus the original.

1768461992194.png



1768462002772.png
 
Mick West said:
The line of sight to the clouds is entirely within the atmosphere, so, of course, atmospheric extinction (mostly water vapor, some aerosols), will affect it with distance. But what distance is it noticable at? Does place some limit on the reconstruction?

Zaine seemed to be saying that inverse square was placing some natural limit that forced them to be closer that in Sitrec. They may well be, but I don't think they HAVE to be.
Click to expand...
ATFLIR works with IR wavelenghts of 3 to 5 microns. That's a window of transmission in the atmosphere, meaning that absorption/scatter are minimized.

https://en.wikipedia.org/wiki/Infrared_window
External Quote:

Atmosfaerisk_spredning.png

 
I did some experiments to reproduce the transmittance graph above and then apply it to Gimbal and other UAP cases. You can run the code online with just a couple of clicks.

The most comprehensive atmospheric transmittance models like MODTRAN are not freely available. For now I tried an older lower resolution model, LOWTRAN. The latter is written in Fortran but it has a python wrapper. It would also be worth investigating libradtran. Its latest release is only a year old, and is advertised as being more accurate.

The standard graph above from Wikipedia shows what percentage of the IR gets absorbed/scattered over 1 km at sea level. The transmittance is drastically higher at altitude (25k feet):

1768546900102.png


I tested the case where a jet at 25000 feet is looking at a target at 19k feet, 30 nm away. Preliminary results would indicate that if that is indeed what happened during Gimbal then a significant portion of the IR the target would emit is still transmitted through the atmosphere, so it seems plausible that a distant jet would be visible.

LOWTRAN has a number of atmosphere presets to choose from. 300 nm off the coast of Jacksonville in Jan 2015 the midlatitude winter atmosphere should be the most relevant here, with an atmosphere that is more dry than during the summer according to this. You'd probably need to provide a user defined atmosphere setup for the best results.

1768545607770.png


In Sitrec the clouds in Gimbal are currently at 11740 feet in altitude. The results would indicate that a noticeable portion of the IR from the clouds should still be transmitted even at 100 nm range. Interestingly the graph looks the same for clouds at 40 nm. I don't know yet whether there's a good reason for that or if it's some limitation of the software.

1768546149561.png

The code was unable to reproduce a significant transmittance for the Chilean Navy case. That might be an indication that more refined models are needed.
 
Mendel said:
Your graphs show transmittance reduced by a factor of 2, as expected.
Click to expand...
Sorry, I didn't include the graph here for the clouds being at 40 nm because it really looks identical to the one at 100 nm. You can see both of them in the repo under "Transmittance for the clouds at various distances". I'm not sure what is going on there.
 
logicbear said:
In Sitrec the clouds in Gimbal are currently at 11740 feet in altitude.
Click to expand...

That's high for stratocumulus/low-clouds.

I had acquired radiosonde weather data a while ago, from Jacksonville airport on Jan 25 00UTC (Jan24 7pm local) and Jan 25 12UTC (Jan 25 7am local). From https://www.ncei.noaa.gov/products/weather-balloon/integrated-global-radiosonde-archive

It has different meteorological parameters, including wind speed (interesting for Gimbal), as well as relative humidity (RH) and dew point depression (DPDP), that are used to estimate the presence of clouds along the vertical. I'm not an expert on clouds/radiosonde data but from what I read clouds are considered likely when RH>80%, and DPDP has low values, under 2° (meaning the local air temperature is getting close to dew point temperature, i.e. condensation).

First, at 7pm local (evening of Jan 24), I retrieve this for the wind:
=== WIND NEAREST 25,000 ft (7620 m) ===
Height: 7339 m
Pressure: 400.0 hPa
Wind Direction: 272°
Wind Speed: 107.282 Kts

Which again indicates this was the approximate day/time for the encounter (Jan 20/21 have too low of a wind speed to match). About 110 knots at 25,000 ft, perfectly from the West over Jacksonville.

At 7am (morning of Jan 25), wind speed has decreased a lot already
=== WIND NEAREST 25,000 ft (7620 m) ===
Height: 7666 m
Pressure: 383.0 hPa
Wind Direction: 260°
Wind Speed: 76.436 Kts


Now the potential cloud layers, we can plot RH and DPDP in function of altitude for both times (7pm local on the left, 7am local on the right)
1768586469326.png
1768586500542.png



If the layer of low-clouds that was on Jacksonville is the same as the one we see in the Gimbal video, it wasn't nearly close to 11,000 ft. The stratus/stratocumulus layer was more around 3,000 ft. Of course we'd need more precise data because allegedly the encounter was 300 miles off the coast. But remember GoFast doesn't show clouds. So it doesn't seem like there was a very large low-cloud/stratus layer that was covering the whole region off Jacksonville.

Sharing because I have it, not to prove a specific point. Now about transmission and cloud distance, what I see is that in the video the clouds fade out in the distance so there is some attenuation in how we see them.
 
TheCholla said:
If the layer of low-clouds that was on Jacksonville is the same as the one we see in the Gimbal video, it wasn't nearly close to 11,000 ft. The stratus/stratocumulus layer was more around 3,000 ft. Of course we'd need more precise data because allegedly the encounter was 300 miles off the coast. But remember GoFast doesn't show clouds. So it doesn't seem like there was a very large low-cloud/stratus layer that was covering the whole region off Jacksonville.
Click to expand...
The MODIS satellite (Terra and Aqua) images of the day are available on NASA Worldview. There were little clouds over Jacksonville, but a lot over the ocean some 300 mile to the East.
 
Reanalysis data can help estimating cloud height.
I see the layer of stratocumulus that passes over Jacksonville at 19UTC-Jan24, it's at 900hPa, about 3000ft (the displayed date in my reanalysis file is wrong, don't pay attention to it. This is cloud fraction on Jan 24 19UTC, red is fully covered with clouds)

1768595489602.png



Then the other layers of clouds that passes over the tip of Florida is higher (around 650 hPa or 12,000 ft). That one has a more complex structure, more vertical (has clouds over a few thousands feet), and it is more patchy.

image2.png


Anyway, that would take time to look into more details, but what's clear is that the cloud layer was complex in the region.
 
logicbear said:
I tested the case where a jet at 25000 feet is looking at a target at 19k feet, 30 nm away. Preliminary results would indicate that if that is indeed what happened during Gimbal then a significant portion of the IR the target would emit is still transmitted through the atmosphere, so it seems plausible that a distant jet would be visible.
Click to expand...
Noting Raytheon made claims that its pod can target over 40 miles, a target at 30nm is well within the advertised effective range, and I didnt realise this was contentious.
logicbear said:
The results would indicate that a noticeable portion of the IR from the clouds should still be transmitted even at 100 nm range. Interestingly the graph looks the same for clouds at 40 nm.
Click to expand...
Whilst, when I read that comment, "a noticeable portion", it appears to me that you are indeed agreeing? that there is an effective range for resolving cloud features?
Mendel said:
Your graphs show transmittance reduced by a factor of 2, as expected.
Click to expand...
Is that what you are also saying @Mendel ?

But I would be remiss if I didnt mention the limitations of the encounter setups for ranging the clouds.
For clarity.
1. We know the first LOS direction
2. We know the last LOS direction
3. We have been able to extract the horizontal spread of the FOVs, roughly 11 Fovs for a total of 3.7 degrees
4. We can use triangulation to range the start and end of the clouds.
5. Where last LOS direction intercepts, First ray (initial los angle plus 3.7 degrees) we have the end range for the clouds.
6. Repeat for where first LOS direction intercepts last ray (last los angle minus 3.7 degrees) we have the end range for the clouds.

To get even close to the reported 100nm for the clouds, the F-18 needs to have an initial heading of 057 degrees. But that would then mean the target is going with the wind, more of a north-east heading before reversing directions, so it can be ruled out.

To model the encounter more accurately, noting the 305 degree initial heading from the Gimbal paper, presented at AIAA, the clouds distance reduce to only 35nm.


Source: https://x.com/LathanielS5437/status/2012291992693432353


So, we all appreciate the lengths @logicbear is going to with these calculations and commentary about software, but, just like the "pilots comfort rotation", can we first sustain the reason it is being bought up in the first place? IE before saying IR transmission from 100NM results in "a noticeable portion" of the IR from clouds being transmitted, can we place the clouds at the range required first?
 
Zaine M. said:
For clarity.
1. We know the first LOS direction
2. We know the last LOS direction
3. We have been able to extract the horizontal spread of the FOVs, roughly 11 Fovs for a total of 3.7 degrees
4. We can use triangulation to range the start and end of the clouds.
5. Where last LOS direction intercepts, First ray (initial los angle plus 3.7 degrees) we have the end range for the clouds.
6. Repeat for where first LOS direction intercepts last ray (last los angle minus 3.7 degrees) we have the end range for the clouds.
Click to expand...
I don't follow this. Can you draw a diagram?
 
This is illustrative, NOT to scale, and I hope this assists. Where the triangulation figures come from


Source: https://x.com/LathanielS5437/status/2012301713722450425


1. Black semi circle is F-18 path
2. Solid blue line is initial LOS direction (from the video time=0s, Az -56 degrees)
3. Solid orange line is last LOS direction. (from the video time=33s, Azimuth +7 degrees)
4. The purple dotted line is a ray = (initial LOS direction plus the FOV spread. Derived from the panorama and known fov settings)
5. The red dotted line is another ray = (last LOS direction minus the FOV spread. Derived from the panorama and known fov settings)

Where purple dotted line intersects solid orange line, gives the range to the end of the clouds from video

Where red dotted line intersects solid blue line, gives the range to the start of the clouds from the video.

Is that easier to understand?
 
Last edited:
TheCholla said:
Reanalysis data can help estimating cloud height.
Click to expand...
I was also looking into this kind of data.

Reanalysis data can be downloaded from Copernicus Data Store (https://cds.climate.copernicus.eu/datasets
"ERA5 hourly data on pressure levels from 1940 to present") and the files can be opened an plotted easily with GRADS (http://opengrads.org/) although I think there are also python modules to open NetCDF files to access the data.

Similar data is also available at NOAA, (https://psl.noaa.gov/data/gridded/reanalysis/),

I was looking at JAN,21st, but now it seems like JAN,25nd (UTC) is more likely, so I downloaded data for the region, between 0 UTC to 15 UTC on JAN, 25.
I have checked the wind speed at 25.000 feet (400 HPa). I have taken as reference the point at coordinates 31N - 76W, (around the point where @TheCholla puts the event). Winds above 100 knots in that region occur from 0 UTC to 2 UTC (24JAN, 19 - 21h local time), and also at 8 UTC (25JAN, 3 am local time) there is an increase of wind speed near that point (actually, at 32ºN)

25JAN25_00.png
25JAN25_08Z.png



I recall gimbal happened at night, just before sunrise, but I may be wrong. But I guess the exact date and time (and location) has never been confirmed.

Anyway, I the looked at the cloud cover at 31N-76W as a function of altitude (pressure, actually). The cloud cover is a variable going from 0 (totally clear) to 1 (totally covered)

.
Cloud_Cover_0Z_b.png
Cloud_Cover_8Z_b.png


So the cloud layer was at about 850 HPa, roughly 5.000 feet.

Finally, these are the cloud cover maps at 850 HPa at 0 UTC and 8 UTC (red being a cloud cover near 1, and purple a cloud cover of 0)
Cloud_Cover_0Z_map.png
Cloud_Cover_8Z_map.png



Now, I am not expert in meteorology, nor the interpretation of these data, so I don't know how reliable the data is, or all the nuances one has to take into account to have a rigurous assessment.
 
Zaine M. said:
This is illustrative, NOT to scale, and I hope this assists. Where the triangulation figures come from


Source: https://x.com/LathanielS5437/status/2012301713722450425


1. Black semi circle is F-18 path
2. Solid blue line is initial LOS direction (from the video time=0s, Az -56 degrees)
3. Solid orange line is last LOS direction. (from the video time=33s, Azimuth +7 degrees)
4. The purple dotted line is a ray = (initial LOS direction plus the FOV spread. Derived from the panorama and known fov settings)
5. The red dotted line is another ray = (last LOS direction minus the FOV spread. Derived from the panorama and known fov settings)

Where purple dotted line intersects solid orange line, gives the range to the end of the clouds from video

Where red dotted line intersects solid blue line, gives the range to the start of the clouds from the video.

Is that easier to understand?
Click to expand...

Not really. For one you seem to be neglecting the platform heading, and treating Az as absolute? Also I don't see how you would get an accurate distance from that. It seems like a misapplication of the FOV traversal number, conflating static with moving measurement points.

If I put those lines into Sitrec, they do not intersect. I think I did it as you describe. Blue is the first LOS, organge is the last. Magenta (purple) is the 3.7° to the left of the first los (so ostensibably pointing at the end of the panorama. Red is 3.7° to the right of the last LOS (so pointing, in your idea, to the start)

2026-01-16_16-45-14.jpg

The intersections you describe just don't exist in front of the camera, as the lines diverge.
 
jplaza said:
ERA5 hourly data on pressure levels from 1940 to present
Click to expand...
That's the one I was looking at too.
jplaza said:
around the point where @TheCholla puts the event
Click to expand...
This is only a vague location, based on R. Graves saying they were "about 300 miles off the coast of Jacksonville". Inside the 100+ kt wind in that general area is I think a good estimate. Note that the clouds could be 50-100 miles from there, depending on their distance in the footage.
Zaine M. said:
There is metadata in the footage. If memory serves, corrected time, is just after 9pm on the 24th (local time).
Click to expand...
Yes and given the fact that R. Graves said it was a night flight, plus the 100+ -kt wind between 7-9pm (confirmed by @jplaza below) sometimes in the evening (6-9pm) seems very plausible.
jplaza said:
Winds above 100 knots in that region occur from 0 UTC to 2 UTC (24JAN, 19 - 21h local time)
Click to expand...

From our 2023 paper (ERA5 wind data)
Screenshot_20260116-180203-858.png
 
Last edited:
Zaine M. said:
Are you able to demonstrate this with the "bank and speed" setting, as opposed to "match clouds" and let me know the results?
Click to expand...
2026-01-16_22-31-55.jpg


75 miles. But I don't think this method is really valid. It's an oversimplification, ignores several things, and minute changes in angles can result in the distance being 10s of miles different. It's a waste of time.
 
I truly appreciate the effort Mick.

We can agree to disagree as to it being a waste of time, IMHO it provides a necessary, repeatable, falsifiable justification for, at least my, reconstruction work.

So again, thank you for taking the time
 
Zaine M. said:
IMHO it provides a necessary, repeatable, falsifiable justification for, at least my, reconstruction work.
Click to expand...
I don't see any indication that it could ever give useful results. Try it yourself, I deployed the code:
2026-01-16_22-53-46.jpg

The white line is between the intersections (first and last LOS intersecting with the lines angles away by "Angular Traverse")
 
@Zaine M. is it close to what you find with that particular F-18 heading?

@Mick West I repeat myself but I do not see why your "Match clouds" config would deliberately not match the clouds. You're missing about 25% of the angular distance between the first and last LOS.
 
TheCholla said:
@Mick West I repeat myself but don't see why your "Match clouds" config would deliberately not match the clouds. You're missing about 25% of the angular distance between the first and last LOS.
Click to expand...
It's not deliberate. "Match clouds" uses a spline editor to try to match the clouds. I eyeballed it, but felt at the time that the data were of insufficient fidelity to do fine-tuning. As seen above, the tiny angles involved create huge instabilities in solutions. I just lost interest.

I think ultimately a more accurate reconstruction could be useful, and I've done some work (motion tracking, clouds) that might be useful there. But it's complicated, requires rigor, and I have limited time.
 
Didn't think I would be making this post but here we are.

1. With Grok now not able to code for me, my attempt to challenge @mickwest with 3d recreation software has been halted, I'm anticipating all AI platforms will soon be implementing this for safety and security reasons. I'm going to take a break from this, if not bow out completely.


Source: https://x.com/grok/status/2013437831209742512



Source: https://x.com/grok/status/2013438694040961242



Source: https://x.com/grok/status/2013439941808718141
 
Here is an attempt to see how the cloud cover looked like on the evening of Jan 24, 7pm local (ET), in the area.

This is using the ERA5 reanalysis mentioned above, we retrieve the general cloud distribution seen on the satellite image (post #453)
https://www.metabunk.org/threads/some-refinements-to-the-gimbal-sim.12590/post-361401

Because plotting a 3-D view of the clouds gets messy, the plot shows where low clouds (below 5,000 ft), mid-level clouds (5,000<cc<10,000ft) and higher clouds (>10,000ft) were located in the area.
It also marks Jacksonville with a red circle, and a 300-mile radius circle around it.
The dashed grey contour marks where wind speed at 400hPa (~25,000ft, where the F-18 is) is greater than 110 knots.
Cloud distribution is for at least 30% of cloud cover in the altitude range (under 30% is masked).

1769030769648.png


If we put the F-18 where the circle and 110+kt wind intersect, we see that we could be looking at low-level clouds (under 5,000ft), or at a region with a more vertical distribution of clouds, depending on where we are looking.

The problem with the low-level cloud layer, infinite stratocumulus, solution, is that they need to be below 5,000ft and this is really at the limit of what can be seen in the FOV inside the -1.5/-2.5° elevation angle range in Sitrec. It also sends the cloud-sky line very far (~150Nm). Other potential clouds have a more vertical structure.

Will need to look more into it to see if we can infer more from this.
 
Update with one category that was missing and masked, the grid points that have low and high clouds (no mid-level clouds).
And with another way to look at it on the right, the max altitude (or top) of the clouds.
(still for cloud cover >30%)

1769128495932.png
1769127348404.png


Mostly shared for @Zaine M. , to have this referenced somewhere and help us with our investigation (and in case of possible interest for Sitrec in the future).

I have a hard time to see how the F-18 could look at the low clouds (blue shade in the right plot), given they are mostly below 4,000 ft, and they would need an elevation angle lower than -2.5° to be in the FOV (easily verifiable in Sitrec). Looks more plausible to me that we are looking at the clouds in red, between 10,000 and 15,000 ft (altocumulus?).
 
TheCholla said:
I have a hard time to see how the F-18 could look at the low clouds (blue shade in the right plot), given they are mostly below 4,000 ft, and they would need an elevation angle lower than -2.5° to be in the FOV (easily verifiable in Sitrec). Looks more plausible to me that we are looking at the clouds in red, between 10,000 and 15,000 ft (altocumulus?).
Click to expand...
Unless the effective elevation angle is incorrect. Maybe that dark line IS the horizon.

However I don't think we are 100% on the exact date/time., are we?
 
Mick West said:
Unless the effective elevation angle is incorrect.
Click to expand...
You mean it's less than -2.5°?

Mick West said:
However I don't think we are 100% on the exact date/time., are we?
Click to expand...
We can't be 100% exact of course, but there aren't many options with a 120-kt westerly wind at 25,000 ft, in late January 2015. It's even the only window really. I'll try to plot it out. And it's a perfect match with video extraction (metadata), like they would do after a flight.
 
Mick West said:
View attachment 88063
https://local.metabunk.org/sitrec/?sitch=video&video=Gimbal

Seems unlikely. But I'm not sure if it was ever ruled out.
Click to expand...
From this post from @Mendel: https://www.metabunk.org/threads/could-the-gimbal-video-show-an-atlas-v-launch.12078/post-259386
The horizon should be at -2.6⁰ if the refraction is correct.
Click to expand...

One may wonder if 2º refers to the range [1.5 - 2.4] (what I guess most of us assume), or to the range [2.0 - 2.9]
 
You must log in or register to reply here.

Similar threads

Zaine M.
GImbal Glare, Rotation, Clouds, and Angles
2 3
Replies
86
Views
3K
Mick West
Mick West
Mick West
3D Analysis of the Yemen Orb
2
Replies
75
Views
11K
Mendel
Mendel
A
Mick vs Marik (rotation glare gimbal)
Replies
19
Views
4K
jackfrostvc
jackfrostvc
J
Hank Green vs. Luis Elizondo, absent summaries, and the AI rabbit hole
Replies
15
Views
3K
Minus0
Minus0
TheCholla
Gimbal derotated video, using clouds as the horizon
2
Replies
57
Views
13K
logicbear
L

Latest posts

Trending content

Back
Top