Why do ~GEO sats seem to cluster around the top portion and bottom portion of the belt, 180 degrees apart?

I made a 3D tle model from https://www.space-track.org/ API. I drew a dashed blue line to show the really "geo" orbits that hardly move with time (ECEF). Then there are a bunch of sats that move up and down and around that are not stuck in the geo plane.

My question is, why do they seem to cluster around the top portion and bottom potion of the belt, 180 degrees apart? Highlighted with the purple arrows. If you go forward a whole day, it almost looks like a spinning disc of satellites that are not strictly on the geo belt. There is also a 24 hour periodicity the up and down clusters follow.

I would assume the sats would be in a sort of random distribution, but there looks like a pattern. I'm curious if this is intentional with their launch, like to follow the sun, or it is from some physics.

I am also not a astro person but I do have an ok understanding of fundamental physics/math.

https://satellitevisualization.pythonanywhere.com/

• See answers to: Why is the ribbon of decommissioned geosynchronous satellites skewed? and links therein. Also see this answer to How stable are satellites in a graveyard orbit and links therein. You don't mention your selection criteria for "~GEO" and how tightly you set your "24 hour" window. Satellites that are really in GEO will have a period of almost exactly one sidereal day (23h 56m 04s) and those moved to "graveyard orbits" at end of life will have a longer period.
– uhoh
Jun 14, 2023 at 23:33
• If you extract the periods of all the objects you show and plot a histogram, you'll get a sharp peak near 1436 minutes, and a noisy bump about 300 km higher which using Kepler's 3rd law will be roughly 15 minutes longer. (Check me, I'm commenting pre-coffee) Once in the graveyard, station-keeping generally ceases and the orbits become inclined due to the gravitational perturbations primarily of the Sun, making them wobbly up and down (what wobbly looks like (but IN GEO))
– uhoh
Jun 14, 2023 at 23:42
• @uhoh Good point, this might help show the criteria... It's the API parameters that created the model space-track.org/basicspacedata/query/class/gp/EPOCH/%3Enow-30/…
– cj91
Jun 14, 2023 at 23:47
• I looked at your first link, and it appears to tackle to question. Thanks All for the links!
– cj91
Jun 14, 2023 at 23:50
• @uhoh quick question, if I had checked this on the equinox, it wouldn't be so obvious then?
– cj91
Jun 15, 2023 at 0:01