Geostationary satellites execute station keeping burns to avoid "wandering", from the POV of an observer on the ground.
Something like a satellite in a Lagrange point might also execute tiny station keeping burns.
A LEO satellite such as starlink has thrusters, yes, and it executes burns, yes, but its burns are executed prograde in the direction of travel to increase orbital altitude. Likely in several small burns. A single prograde burn made when on one side of the planet has the effect of increasing the height of apogee on the opposite side of the planet, so after you do your first burn and are halfway around a full orbit, you need to do at least a second burn to re-circularize your orbit.
Nobody that's involved in LEO satellite operations would call it a station keeping burn because it's literally not keeping station anywhere.
I also doubt the validity of the data in this article because without internal information from spacex on things like how often they burn, when they burn, duration of burn and expected delta-v gain per burn (which spacex is OBVIOUSLY not going to share with anyone because making an extremely fuel efficient small satellite is a competitive advantage and a trade secret), you can't know how much the satellites are actually thrusting in any given day, week, month or year to account for varying increased atmospheric density.
The one thing that the article did get right is that the superdove satellites, which are 3U cubesats, have no thrusters at all, so watching the degradation of their orbit over time (and the rate of change in any given week or month) might be a more valid source of data. There's no need to "guess" at what a satellite operations center at SpaceX might be commanding them to do in terms of fuel expenditure because there is no fuel to expend or thrusters to turn on to burn prograde.
This article is a very bad summary of a research report [1]. To summarize, SpaceX publishes very detailed ephemerides of the Starlink satellites three times a day. You can use that to calculate the total energy of each satellite, and every orbit-raising maneuver is quite apparent in that data as a sharp spike in orbital energy. The remaining slow dissipation in energy is correlated with the thermospheric density. There's also a press release from Kyoto University with a high-level overview [2].
> To summarize, SpaceX publishes very detailed ephemerides of the Starlink satellites three times a day.
And for very good reasons indeed. With the number of satellites they have and the number of other things in LEO, there's a legit need for automated planning/alarming of future collision avoidance maneuvers. Both by spacex and by third party operators. I guess what I meant is that despite them publishing the ephemerides of the satellites one thing that shall remain opaque to an outsider is details like how long a prograde burn lasts, how much fuel it consumes, how much calculated delta-v each burn imparts to a satellite (you can't know this unless you know the precise weight of the satellite and its estimated remaining onboard propellant).
I think calculating an accurate expected degradation in orbit from the point of view of a spacex outsider would also be hard unless you were in possession of the 3D modeling datasets that document exactly how much drag the satellite has in different orientations. Reasonably decent estimates can be made, but I'm sure SpaceX has much better data. One of the most interesting satellites in recent memory is one that was specifically designed to be long and somewhat streamlined to prolong its lifespan at very low orbital altitude: https://en.wikipedia.org/wiki/GOCE
Yeah, I agree that SpaceX has access to much more data, but from reading the paper it appears that you can get quite far if your goal is to map atmospheric drag. I would suggest skimming through it if you're interested in the practical challenges.
> how long a prograde burn lasts, how much fuel it consumes, how much calculated delta-v each burn imparts to a satellite
Yes, to the first two, but surprisingly not to the third. Since the maneuvers are short and much higher thrust than the average atmospheric drag, you can accurately pinpoint where those maneuvers are occurring. The energy gain can only come from the thrusters, and you can also use your atmospheric model to subtract out the (very small, comparatively) energy lost to drag. So you can actually calculate the delta-V of each maneuver quite accurately even if you don't know its exact duration or propellant consumption.
> One of the most interesting satellites in recent memory is one that was specifically designed to be long and somewhat streamlined to prolong its lifespan at very low orbital altitude
GOCE was an astounding mission! They were trying to accomplish something much harder though, mapping tiny distortions of the Earth's gravitational field, which required an ion thruster to precisely cancel out the fluctuations due to atmospheric drag and maintain an inertial trajectory. You need several orders of magnitude less precision to map the atmospheric drag itself.
I think the coolest aspect of this work is that without any involvement from SpaceX, and taking advantage of the large numbers of Starlink satellites, the researchers were able to use some nice modelling and statistics to get some very high-quality data.
It's fascinating to see that the GOCE satellite had only 40kg of fuel onboard and managed to last significantly longer than expected, ultimate mission duration of 4 years 7 months.
Such obnoxious Claudish in this article. I don't know enough verify the analysis and I genuinely don't know if the author does either.
It also doesn't "know" basic orbital mechanics.
From the article:
> and station-keeping burns corrupt them
Geostationary satellites execute station keeping burns to avoid "wandering", from the POV of an observer on the ground.
Something like a satellite in a Lagrange point might also execute tiny station keeping burns.
A LEO satellite such as starlink has thrusters, yes, and it executes burns, yes, but its burns are executed prograde in the direction of travel to increase orbital altitude. Likely in several small burns. A single prograde burn made when on one side of the planet has the effect of increasing the height of apogee on the opposite side of the planet, so after you do your first burn and are halfway around a full orbit, you need to do at least a second burn to re-circularize your orbit.
Nobody that's involved in LEO satellite operations would call it a station keeping burn because it's literally not keeping station anywhere.
I also doubt the validity of the data in this article because without internal information from spacex on things like how often they burn, when they burn, duration of burn and expected delta-v gain per burn (which spacex is OBVIOUSLY not going to share with anyone because making an extremely fuel efficient small satellite is a competitive advantage and a trade secret), you can't know how much the satellites are actually thrusting in any given day, week, month or year to account for varying increased atmospheric density.
The one thing that the article did get right is that the superdove satellites, which are 3U cubesats, have no thrusters at all, so watching the degradation of their orbit over time (and the rate of change in any given week or month) might be a more valid source of data. There's no need to "guess" at what a satellite operations center at SpaceX might be commanding them to do in terms of fuel expenditure because there is no fuel to expend or thrusters to turn on to burn prograde.
https://www.google.com/search?client=firefox-b-d&q=superdove...
This article is a very bad summary of a research report [1]. To summarize, SpaceX publishes very detailed ephemerides of the Starlink satellites three times a day. You can use that to calculate the total energy of each satellite, and every orbit-raising maneuver is quite apparent in that data as a sharp spike in orbital energy. The remaining slow dissipation in energy is correlated with the thermospheric density. There's also a press release from Kyoto University with a high-level overview [2].
[1] Yamamoto, M. Tomography of thermospheric density from Starlink Ephemeris: initial report. Earth Planets Space 78, 175 (2026). http://dx.doi.org/10.1186/s40623-026-02509-5
[2] https://www.sciencedaily.com/releases/2026/08/260812015212.h...
> To summarize, SpaceX publishes very detailed ephemerides of the Starlink satellites three times a day.
And for very good reasons indeed. With the number of satellites they have and the number of other things in LEO, there's a legit need for automated planning/alarming of future collision avoidance maneuvers. Both by spacex and by third party operators. I guess what I meant is that despite them publishing the ephemerides of the satellites one thing that shall remain opaque to an outsider is details like how long a prograde burn lasts, how much fuel it consumes, how much calculated delta-v each burn imparts to a satellite (you can't know this unless you know the precise weight of the satellite and its estimated remaining onboard propellant).
I think calculating an accurate expected degradation in orbit from the point of view of a spacex outsider would also be hard unless you were in possession of the 3D modeling datasets that document exactly how much drag the satellite has in different orientations. Reasonably decent estimates can be made, but I'm sure SpaceX has much better data. One of the most interesting satellites in recent memory is one that was specifically designed to be long and somewhat streamlined to prolong its lifespan at very low orbital altitude: https://en.wikipedia.org/wiki/GOCE
Yeah, I agree that SpaceX has access to much more data, but from reading the paper it appears that you can get quite far if your goal is to map atmospheric drag. I would suggest skimming through it if you're interested in the practical challenges.
> how long a prograde burn lasts, how much fuel it consumes, how much calculated delta-v each burn imparts to a satellite
Yes, to the first two, but surprisingly not to the third. Since the maneuvers are short and much higher thrust than the average atmospheric drag, you can accurately pinpoint where those maneuvers are occurring. The energy gain can only come from the thrusters, and you can also use your atmospheric model to subtract out the (very small, comparatively) energy lost to drag. So you can actually calculate the delta-V of each maneuver quite accurately even if you don't know its exact duration or propellant consumption.
> One of the most interesting satellites in recent memory is one that was specifically designed to be long and somewhat streamlined to prolong its lifespan at very low orbital altitude
GOCE was an astounding mission! They were trying to accomplish something much harder though, mapping tiny distortions of the Earth's gravitational field, which required an ion thruster to precisely cancel out the fluctuations due to atmospheric drag and maintain an inertial trajectory. You need several orders of magnitude less precision to map the atmospheric drag itself.
I think the coolest aspect of this work is that without any involvement from SpaceX, and taking advantage of the large numbers of Starlink satellites, the researchers were able to use some nice modelling and statistics to get some very high-quality data.
It's fascinating to see that the GOCE satellite had only 40kg of fuel onboard and managed to last significantly longer than expected, ultimate mission duration of 4 years 7 months.