Mystery satellite in Google imagery

[Easy Muffin]

So it seems an Earth-imaging satellite managed to capture another satellite whilst imaging Earth, I wonder if we can identify them?

https://maps.app.goo.gl/xEKLsAoWnVETiDN47
This is the Google Maps link. We are near the town of Sherman, TX. Pretty much due north of Dallas on the Oklahoma border. I suppose the afterimages are due to the imaging sat taking pictures at different wave lengths in quick succession to combine them into a single colour image, and the other sat moving quickly across the field of view in the short time between each frame.

On the question of what took the image, it is (c) Airbus, that would narrow the candidates down to six sats: Pléiades-1A and Pléiades-1B (of the Pléiades constellation), Pléiades Neo 3 and Pléiades Neo 4 (of the Pléiades Neo constellation), SSTL S1-4 (of the Vision-1 programme), and SPOT 6 (https://www.airbus.com/en/products-services/space/earth-observation/satellite-imagery if you want to know more).

The date given is Nov 30, 2024 - let's assume for now this is correct. As for time for day I went to Suncalc and searched for buildings with straight walls casting a shadow on flat ground. Plenty of those around at the nearby airport. I measure a Sun azimuth of 164°, that's 11:19 local time according to Suncalc.

So... which of these sats might it be? There's one that fits really well, Pléiades Neo 3. It was almost exactly north of Sherman at that time, fitting well with the viewing angles of the imagery, and about to pass that same area. Pléiades-1A made a pass a few minutes earlier and I originally thought this was it, but neither the viewing geometry nor the shadow angles were quite right - so if the date is correct, I would put my money on this being a 3 Neo image.

Now the more interesting question, what sat are we looking at? Unfortunately that's where I haven't made a whole lot of progress so far. Still, might be fun to try and find out somehow. @Mick West, I don't suppose we could rig Sitrec to ingest a TLE so we can put ourselves on top of a sat and watch from its POV like we can do with KMLs and planes?

 

[Mick West]

@Mick West, I don't suppose we could rig Sitrec to ingest a TLE so we can put ourselves on top of a sat and watch from its POV like we can do with KMLs and planes?

It's certainly possible. I've been updating Sitrec to load all the LEO objects history, instead of just Starlink. I got as far as filtering it for Starlink or the ISS, but all satellites are possible. I'd pretty much just need to add a Camera Track node that takes the text name of the satellite.

Fun!

 

[Mick West]

What would be the NORAD designation for those?

 

[Mick West]

When I load the LEO TLE, I'm only getting:

0 PLEIADES 1
1 38012U 11076F   24334.18430395  .00001471  00000-0  32608-3 0  9993
2 38012  98.1898  46.4233 0001528  94.6175 286.9023 14.58566581689641

and

0 PLEIADES 1B
1 39019U 12068A   24334.14559213  .00001462  00000-0  32408-3 0  9998
2 39019  98.1819  46.2548 0001312  93.1397 266.9952 14.58567668638482

I'm possibly doing something wrong. I use this filter

$url = "https://www.space-track.org/basicspacedata/query/class/gp_history/EPOCH/" . $request . "--" . $nextDay . "/MEAN_MOTION/>11.25/ECCENTRICITY/<0.25/OBJECT_TYPE/payload/orderby/NORAD_CAT_ID,EPOCH/format/3le";

 

[Easy Muffin]

What would be the NORAD designation for those?

PLEIADES 1A     38012
PLEIADES 1B     39019
PLEIADES NEO 3  48268
PLEIADES NEO 4  49070
SPOT 6          38755
SSTL S1-4       43618

 

[Mick West]

Hmm, looks like I've got some investigating to do

 

[Mick West]

PLEIADES 1A     38012
PLEIADES 1B     39019
PLEIADES NEO 3  48268
PLEIADES NEO 4  49070
SPOT 6          38755
SSTL S1-4       43618

Common Name	NORAD Catalog Number	TLE Name Field
Pléiades 1A	38012	PLEIADES 1
Pléiades 1B	39019	PLEIADES 1B
Pléiades Neo 3	48268	PNEO3
Pléiades Neo 4	49070	PNEO4
SPOT 6	38755	SPOT 6
SSTL S1-4	43618	SSTL S1-4

 

[Easy Muffin]

Hmm, figured if I looked up the air traffic at that time on ADSBexchange I might just get lucky and find that same plane in the GE imagery.

Turns out, there's this Delta A319. Exporting the KML and having a look around in Google Earth and there's this...

33.675135, -96.797430

Left side is a general view, right side is zoomed in. It sort of lines up but it also sort of doesn't - would parallax explain the offset? The contrails seem to suggest a slightly different course too but winds aloft might be playing a part? Hmmm. On the one hand, things somehow don't seem to fit together quite right but on the other I was rather surprised to even stumble upon that plane in the first place!
I'm not too great on ID'ing planes either but an A319 looks like a possible match? The time in the screenshot above is Nov 30, 2024 at 17:19:32 UTC (11:19:32 local).

 

[Mick West]

Coming soon!

This is showing the view from the ISS. Needs more testing, and work to user-define which satellite to attach the camera to (or use as a target)

This should be able to replicate ISS videos.

Unfortunately, domestic duty calls. Maybe later.

 

[Mick West]

Okay, I've deployed the satellite imporvments
https://www.metabunk.org/sitrec/?cu...mazonaws.com/1/Easy Muffin/20250420_194355.js

You can set any satellite as the camera. Here I've also put in target coordinates.

It does not seem like any satellites pass between PNEO3 and the target, or even really anywhere close, during this pass.

Adding in the plane might help

 

[Easy Muffin]

Wow, that's great! This could be incredibly useful.

I added the KML and while the contrails and the plane's orientation still don't seem quite right, the actual position lines up really well. It's pretty much an exact match, I'd say.

Not sure what to make of this. It'd be quite the coincidence for a random plane to simply show up there but as you point out there's no obvious satellite that could have passed in front of the camera.

I then loaded the 1A pass from a few minutes earlier and noticed this sat crossing the FOV.

It's an Iranian imaging satellite. Again the position is a little bit off, I wonder if small errors in the TLE would become noticeable at these resolutions? We're talking milliseconds here.
On the other hand, this time there's no plane to be seen anywhere. Curiouser and curiouser!

 

[Ravi]

I am not convinced it was/is a satellite.

 

[Trailblazer]

while the contrails and the plane's orientation still don't seem quite right,

Remember that the orientation of the plane does not necessarily match the actual heading. If there is a crosswind then the plane would be pointing at an angle to the actual direction of travel (which is that is shown on the flight tracker). And of course the crosswind would also be pushing the contrail at an angle.

Here's a tentative ID for the satellite:


Source: https://x.com/planet4589/status/1910900719252627895

 

[Easy Muffin]

I am not convinced it was/is a satellite.

Question then is, what else could it be? The spacing of the individual images strongly suggests an object travelling at orbital velocities. See for example the plane from the screenshot above - that's one of the fastest objects we have travelling around down here and it was only changing its position enough for a faint red trace to appear.

Remember that the orientation of the plane does not necessarily match the actual heading. If there is a crosswind then the plane would be pointing at an angle to the actual direction of travel (which is that is shown on the flight tracker). And of course the crosswind would also be pushing the contrail at an angle.

Here's a tentative ID for the satellite:


Source: https://x.com/planet4589/status/1910900719252627895

I don't think this is a solid case tbh. We can see the north-facing walls of buildings in the Google imagery which means the picture was taken from a point north. So for the Starlink sat to cross the line-of-sight it would have had to cross somewhere between those two points, i.e. to the north of Sherman - but it moved across Dallas, to the south of it.

It would have had to be over Oklahoma, instead.

 

[Ravi]

@Easy Muffin
Thanks for your post, it is now more convincing to me.

 

[Mick West]

Turns out, there's this Delta A319. Exporting the KML and having a look around in Google Earth and there's this...

33.675135, -96.797430

Left side is a general view, right side is zoomed in. It sort of lines up but it also sort of doesn't - would parallax explain the offset? The contrails seem to suggest a slightly different course too but winds aloft might be playing a part? Hmmm. On the one hand, things somehow don't seem to fit together quite right but on the other I was rather surprised to even stumble upon that plane in the first place!
I'm not too great on ID'ing planes either but an A319 looks like a possible match? The time in the screenshot above is Nov 30, 2024 at 17:19:32 UTC (11:19:32 local).

Viewed from PNE03, that plane goes within a few feet of that exact spot at 2024-11-30 17:19:10.376 UTC

https://www.metabunk.org/sitrec/?cu...mazonaws.com/1/Easy Muffin/20250422_062032.js

The angle is consistent with a crosswind.

 

[Mick West]

You can get them both on the screen at the same time

This appears to be all from an image or images from the same pass. There's no discontinuity, and the angle of the shadows of the utility posts seems the same.

 

[Easy Muffin]

https://earth.nullschool.net/#2024/...ographic=-98.88,33.55,3290/loc=-96.800,33.676
This is the 250-mbar level around the same time. Works out to about 34,000 ft in altitude.

There was a pretty gnarly crosswind from the west. That would indeed push the contrails east quite fast and it would also result in some pronounced crabbing by the plane itself. I still find it surprisingly difficult to picture the whole scene in my mind to any precise level.

You can get them both on the screen at the same time
[...]
This appears to be all from an image or images from the same pass. There's no discontinuity, and the angle of the shadows of the utility posts seems the same.

Yup. This is the full image I think, having traced around the edges in Google Earth.

It's a rectangle roughly 13.5x38.5 km / 8.5x24 mi (520 km^2 / 200 mi^2).

 

[Mick West]

Nothing is showing up with LEO objects, so I've been trying some increasing inclusive queries. First, removing the orbital restrictions to include eccentric orbits
https://www.space-track.org/basicsp...payload/orderby/NORAD_CAT_ID,EPOCH/format/3le

Then removing the OBJECT_TYPE/payload, to maybe get spy sats or other things classed differently.
https://www.space-track.org/basicsp...4-11-29/orderby/NORAD_CAT_ID,EPOCH/format/3le

Neither of these had any satellites fly anywhere near the lines of sight.

I've flirted with:

I am not convinced it was/is a satellite.

That's a lot of motion blur, and it appears transparent on the left side. How exactly would a satellite create this image? One comment:

Source: https://x.com/4tis/status/1911123663572656486


Explained here:

Article:

Pan sharpening uses a higher-resolution panchromatic image (or raster band) to fuse with a lower-resolution multiband raster dataset. The result produces a multiband raster dataset with the resolution of the panchromatic raster, where the two rasters fully overlap.

Pan sharpening is a radiometric transformation available through a raster function or from a geoprocessing tool. Several image companies provide low-resolution, multiband images and higher-resolution, panchromatic images of the same scenes. This process is used to increase the spatial resolution and provide a better visualization of a multiband image using the high-resolution, single-band image.

Source: https://pro.arcgis.com/en/pro-app/latest/help/analysis/raster-functions/fundamentals-of-pan-sharpening-pro.htm

So the white image isn't a combination of the red, green, and blue, it's a high-resolution panchromatic image (i.e. a black and white image that used the full visible light spectrum). The thre red, green, and blue images are lower resolution, and are combined with the panchromatic image to create the final image.

So it has to be something there in each exposure, so not a sensor artifact. A satellite seems the only thing that can make sense.

So why isn't it showing up in Sitrec. The most likely explanation is some error on my part. Possibly to do with the oblateness of the Earth.

 

[Easy Muffin]

So it has to be something there in each exposure, so not a sensor artifact. A satellite seems the only thing that can make sense.

So why isn't it showing up in Sitrec. The most likely explanation is some error on my part. Possibly to do with the oblateness of the Earth.

Unlikely as it may be but I'm beginning to suspect this could be a satellite that's not yet been added to the TLE database, i.e. one launched only recently.

There were several launches at the end of November 2024. 4 of them in just the 24 hours preceding the PNEO image.

https://en.wikipedia.org/wiki/List_of_spaceflight_launches_in_July–December_2024

One possibility is Starlink Group 6-65. It launched 12 hours before and deployed 24 Starlink v2's. They would have been starting to form a train and happily orbiting Earth half a day later, but NORAD only started tracking on Dec 6.

https://in-the-sky.org/spacecraft_e...tyear=2024&endday=10&endmonth=12&endyear=2024

This is one of the 24 from that group.

I guess SpaceX would have manoeuvred them around quite a bit in the early deployment stage so extrapolating an existing TLE back a couple days wouldn't tell us all that much I'm afraid.

 

[jarlrmai]

Seems to the fit the profile of a Starlink Block 2 mini

https://space.skyrocket.de/doc_sdat/starlink-v2-mini.htm

 

[Easy Muffin]

I guess SpaceX would have manoeuvred them around quite a bit in the early deployment stage so extrapolating an existing TLE back a couple days wouldn't tell us all that much I'm afraid.

Quoting myself but figured we might get lucky turning to the satellite observer community. Somebody tracked the sat I randomly picked above from a place in New Jersey some six hours after the GE image.

***  2024 Nov  30  Sat evening  *** Times are PM UT-5 ***  1725  6 7

  H  M  S Tim Azi El C Dir  Mag Dys F  Hgt Shd  Rng  EW Phs  R A Dec
71862 STARLINK-32497                            6.0
17 49  7     160 70 C 270  4.1   0 4  185  56  196 4.1  75 23 1 21.2

https://www.satobs.org/seesat/Dec-2024/0003.html

That might be a start.

 

[jplaza]

That's a lot of motion blur, and it appears transparent on the left side. How exactly would a satellite create this image? One comment:

Source: https://x.com/4tis/status/1911123663572656486


Explained here:

Article:

Pan sharpening uses a higher-resolution panchromatic image (or raster band) to fuse with a lower-resolution multiband raster dataset. The result produces a multiband raster dataset with the resolution of the panchromatic raster, where the two rasters fully overlap.

Pan sharpening is a radiometric transformation available through a raster function or from a geoprocessing tool. Several image companies provide low-resolution, multiband images and higher-resolution, panchromatic images of the same scenes. This process is used to increase the spatial resolution and provide a better visualization of a multiband image using the high-resolution, single-band image.

Source: https://pro.arcgis.com/en/pro-app/latest/help/analysis/raster-functions/fundamentals-of-pan-sharpening-pro.htm

So the white image isn't a combination of the red, green, and blue, it's a high-resolution panchromatic image (i.e. a black and white image that used the full visible light spectrum). The thre red, green, and blue images are lower resolution, and are combined with the panchromatic image to create the final image.

So it has to be something there in each exposure, so not a sensor artifact. A satellite seems the only thing that can make sense.

So why isn't it showing up in Sitrec. The most likely explanation is some error on my part. Possibly to do with the oblateness of the Earth.

Just noticed there is a 5th image of the satellite, a dark one :

The spanish version of wikipedia states that there are in fact 5 sensors: https://es.wikipedia.org/wiki/Satélites_Pléiades
Blue / Green / Red / Near Infrarred / and the Panchromatic-wide band


The black "image" would be the shadow in the near infrarred of the satellite. (the sat blocking the near IR light from the ground)

 

[Mick West]

I guess SpaceX would have manoeuvred them around quite a bit in the early deployment stage so extrapolating an existing TLE back a couple days wouldn't tell us all that much I'm afraid.

Even with established orbits, extrapolating backwards does not work very well

Quoting myself but figured we might get lucky turning to the satellite observer community. Somebody tracked the sat I randomly picked above from a place in New Jersey some six hours after the GE image.

*** 2024 Nov 30 Sat evening *** Times are PM UT-5 *** 1725 6 7 H M S Tim Azi El C Dir Mag Dys F Hgt Shd Rng EW Phs R A Dec 71862 STARLINK-32497 6.0 17 49 7 160 70 C 270 4.1 0 4 185 56 196 4.1 75 23 1 21.2https://www.satobs.org/seesat/Dec-2024/0003.html

That might be a start.

This data sounds like it would have worked.

Source: https://x.com/TSKelso/status/1862907170758832228


But Celestrack does not seem to archive.

 

[Mick West]

https://earth.nullschool.net/#2024/...ographic=-98.88,33.55,3290/loc=-96.800,33.676
This is the 250-mbar level around the same time. Works out to about 34,000 ft in altitude.

There was a pretty gnarly crosswind from the west. That would indeed push the contrails east quite fast and it would also result in some pronounced crabbing by the plane itself. I still find it surprisingly difficult to picture the whole scene in my mind to any precise level.

So that's 117 knots from 280. Adding that into the sitch.
https://www.metabunk.org/sitrec/?cu...naws.com/1/Plane with wind/20250423_202840.js

It gives us the expected wind correction angle, which almost perfectly matches the image.

The green line here is the actual track. The contrails are always in line with the engines.

I need to add contrails to Sitrec.

 

[Mick West]

Yup. This is the full image I think, having traced around the edges in Google Earth.

It shows up on the ApolloMapping site:

Just visible in the preview:

The plane and contrail are less obvious, but they are there.

 

[Easy Muffin]

It shows up on the ApolloMapping site:

Couple seconds difference between acquisition time and Sitrec's best match. Maybe the sat needs a few secs to spin up its gears, so to speak? I've been wondering if the hardware is capable of taking a picture instantly like you would with your phone, or if the process is a tad more involved and time consuming. I'm guessing it's the latter.

Anyway, a quick thought that popped into my mind. With some spherical trig it should be possible to calculate the heading of an object in orbit given its orbital inclination and respective latitude, e.g. as described here: https://space.stackexchange.com/questions/17679/

Measuring the trail of the sat trace in GE gives a heading of ~34° at 33.7 N but that's assuming a top-down view and a location on the ground - in reality PNEO took the image from an oblique angle and the other sat would have passed at a higher lat, let's say 35 N.

For the 5-65 Starlink launch (inc 43°) that should result in a heading of arcsin(cos(43°)/cos(35°)) = 63.2° - That seems too far off what one might expect?
The NROL-126 launch (inc 70°) seems a better fit: arcsin(cos(70°)/cos(35°)) = 24.7° - this launch too deployed a number of v2 mini sats for Starlink and Starshield.

 

[Ashley Pomeroy]

Couple seconds difference between acquisition time and Sitrec's best match. Maybe the sat needs a few secs to spin up its gears, so to speak?

I'm going off on a tangent here, but I'm puzzled by the number of sensors on the imaging satellite. Back in the 1990s there were monochrome digital medium format cameras, but they could create full-colour images by taking three images in quick succession through a filter wheel that had red, green, and blue filters. The technique was only useful for static subjects but the results were sharper than a single colour exposure taken with a Bayer filter. I assume that early imaging satellites had a similar system.

Presumably this satellite doesn't actually have separate red, green, and blue sensors - which is why the images have a little bit of separation, because the same sensor takes three images very quickly with a short pause for the colour filter to slide into place - but that raises the question of why they don't use the high-resolution panchromatic camera instead.

The only thing that springs to mind is the possibility that it doesn't have a big enough buffer to take three images in quick succession. The samples on Airbus' website are around eight megapixels, but I know that space-rated equipment tends to lag behind the state-of-the-art on Earth, and of course they also need to store and transmit that imagery as well, so presumably there is a good technical reason.

 

[Mick West]

Presumably this satellite doesn't actually have separate red, green, and blue sensors - which is why the images have a little bit of separation, because the same sensor takes three images very quickly with a short pause for the colour filter to slide into place - but that raises the question of why they don't use the high-resolution panchromatic camera instead.

They do have separate sensors. There are no moving parts. It's a "push-broom" camera where there's 6-pixel strips of:

Article:

Single pushbroom linear array, TDI instrument
Camera – 12 detectors (6 panchromatic, PAN; 6 multispectral, MS)
Detectors: PAN – 8,016 pixels, 12 µm detector size; MS – 2,004 pixels x 6, 48 µm detector size
Image size: PAN – 48,096 pixels (8,016 x 6); MS – 72,144 pixels (2,004 x 6 x 6)

Source: https://apollomapping.com/pleiades-neo-satellite

These are staggered in the focal plane in the direction of travel. A given 6-pixel long strip on the ground will move across that focal plane at a fixed rate from the motion of the satellite, so they have a time offset when combining the image. That offset is based on the non-moving ground, so goes wrong for high or moving objects.

 

[Mendel]

Compare:

https://maps.app.goo.gl/ESxmf7v7kp82pNf28

Happened across this one too in Google Earth. I don't see this often. This one has green, blue, and red layers.

36°58'45"N 76°30'31"W