Destin from "Smarter Ever Day" posted a video where he noticed two satellites photobomb his solar eclipse. This GH is an attempt to reverse engineer which satellites photobombed the images.
Based on the video from Smarter Every Day:
- The location is Jackson, Missouri (Drury Hotel: 37.42925803643183, -89.64323382671625)
- The data is the solar eclipse, 4/8/2024
- Duration of Totality*: 4m 9.6s
- Totality Start*: 13:58:11 (CDT (GMT-5))
- Mid-eclipse*: 14:00:16 (CDT (GMT-5))
- A satellite slightly away from Baily's beads
- A satellite right after the previous one but on the solar disk
For a satellite to cause a flare during a solar eclipse, the satellite must be out of the shadow of the eclipse. The satellite could be reflecting light from the earth back down and since the sky is dark, the satellite is now noticeable. Otherwise, the angle from the sun to the satellite and then to earth would have to be rather extreme. I believe the case of the satellite reflecting light from the Earth would be the more likely case. In this scenerio, the satellite would be close to the observer but relatively in line with the eclipse.
Here is a cool SpaceX video showing the eclipse from LEO:
https://www.youtube.com/watch?v=ust2eR3nhJc
This python script uses a TLE filed pulled from Celestrak and then:
- Given a lat/lon and date/time range computes the Azimuth and Elevation to the sun
- It then computes Azimuth and Elevation range for active satellites from Celestrak for the location and date/time window
- It then compares to find satellites within a field of view of "specified" degrees. The script started with a rather wide field of view for the observation but we can downselect candidates after (like a sieve)
- The video's observations are rather fast so the transit's duration is a bit ambigious but it seems like seconds after the beads appear. To confirm visuals, we'll plot a few minutes of time in our python ground plots.
The python script ran for 3 minutes after the solar eclipse's end (starts just before the end), which is more than enough time to catch a satellite near Baily's beads and one right after in the solar disk. How the script works, it essentially tried to find if a "line" is in or interesects a circular field of view.
Active satellites were loaded into the script but no results seemed plausible. A OneWeb satellite seemed close but didn't fit the time window.
This Orbitron plot shows active satellites. There are lot.
I then proceeded to try to get debris TLEs. Once, in Sanibel, I saw what I thought was a satellite. It turned out to be rocket debris. In fact, most "satellites" seen in space tend to be debris. Celestrak (Thank you Dr. Kelso for your service over they years) has deprecated various TLE sets because of user abuses. A catalog including debris was downloaded from Space-Track.org.
Visualised in Orbitron, the added density of objects when debris is included is seen:
This is where the script comes in handy. Running the script with a wide net of 10 degrees of field of view and outputting the relevent TLEs gives the result of:
Further refining to 3 degrees of field of view gives us 4 satellites, of which 2 seem to fit the requirement of back to back passes, one at baily's beads and the other on the solar disk:
TLEs for this set can be found:
Python Ground Plots for Cosmos 1859:
Python Ground Plots for Thorad Agenda D Deb
Python Ground Plots for Fregat Deb
The solution seems to be centered around:
- Cosmos 1859, launched June 1987 and no longer in active use (NY20 data on Cosmos 1859
- (A) Thorad Agena Debris from 1970 (Wiki page on Throad Agena (SATCAT information
- (B) Fregat Debris from 2011 (45649U) (N2YO page
Cosmos 1859 is travelling at 7.163 km/s. This kind of jives with the suspected velocity of the object. Thorad debris is in LEO and travelling at 7.34 km/s, so it jives. Fregat is 7.78 km/s @ LEO as well. The fact one object is inactive and the other 2 potentials are debris means they won't show up in the active satellite list.
Using older TLEs (closer to the accuracy of April 8th), Fregat appears out of play. Cosmos 1859 remains a likely candidate. Thorad debris as well.
Lots of cool stuff, much debris in space!
I still have to clean up the code, push it and update the GH but I wanted to get the results out for peer review from anyone interested. The April 6th to 7th Space-Track set I have is filled with a fair amount of objects that have no label. I'll rerun this with an Apil 8th and try to clean it up. Here is the space track query: