Space Debris Cleanup Hack: Using the Gravity of the Moon

The impact point text in the final image of this blog has been updated. As correctly stated in the blog, the correct impact site is 4.58 N, 129.06 W.

Just look up! On March 4, 2022, around noon in Greenwich, the third stage of a spent Long March 3C is expected to crash into the Moon. AGI reconstructed the circumstances surrounding this impending event using STK and ODTK. We’d like to thank Bill Gray (presented on the ProjectPluto website) and countless other independent astronomers around the world freely sharing optical tracking data, as well as Tom Johnson at ExaResearch for his assistance developing an orbit determination solution. 


On October 23, 2014, the China National Space Administration (CNSA) launched the Chang’e 5-T1 spacecraft aboard a Long March 3C rocket from Xichang Satellite Launch Center on a free-return trajectory around the Moon. The mission ultimately served as a testbed for the Chang’e 5 mission, a lunar sample return mission in 2020. The test mission was declared a success and the service module is still operational today.

Collision Course

With the launch complete, the third stage was all but forgotten. Its final orbit was high enough that there was no collision risk with Earth-orbiting satellites, and it was assumed that the empty 2800 kg booster would continue in its orbit forever. However, astronomers searching for asteroids found it shortly afterward, initially supposing that it was an asteroid. Once the object was recognized as human-made, it was incorrectly identified as the second stage of a Falcon 9, from the launch of NOAA’s DSCOVR spacecraft in January 2015. A small group of astronomers kept tabs on it just in case, observing it intermittently over the coming years. Then, on February 12, 2022, these scientists realized that the orbit did not coincide with DSCOVR’s trajectory. Jon Giorgini, at NASA’s Jet Propulsion Laboratory (JPL), pointed out to Bill Gray that DSCOVR never went past the Moon on its way to the Sun-Earth L1 libration point. Instead, the object’s cislunar trajectory was matched to the Chang’e 5-T1 mission’s third stage. The corrected identification was soo published on the ProjectPluto website.

The following plot shows the evolution of the booster’s altitude above Earth along with points representing the times that tracking measurements were taken. In response to a call for data by Bill Gray — to confirm an earlier estimate indicating potential lunar impact — the rate of observations increased beginning in 2022.

Plot diagram

Each blue dot represents a set of telescope observations of the booster.

Nearly seven years after launch and several harmless flybys of the Moon, the orbit became altered. A series of three flybys set up the booster for its eventual demise. These flybys occurred on September 18, 2021, January 5, 2022, and February 5, 2022. Our orbit determination solution reveals the effect of the flybys on the booster’s semimajor axis and inclination, as depicted in the graph below.

Plot diagram 2


This plot of the booster’s semimajor axis and inclination clearly illustrates how its orbit is affected by each flyby.

The flyby in January was the most important, significantly altering the semimajor axis, eccentricity, and inclination of the booster’s orbit. At its closest approach, the booster passed less than 12,000 km from the Moon’s surface. We calculated that the January flyby imparted an equivalent Delta-V of nearly 640 m/s on the booster! After the third gravity assist on February 5, the booster’s next and final lunar encounter will result in impact with the surface of the Moon.

Initial Orbit Solution

Equipped with measurements collected and published by astronomers around the world, we used Orbit Determination Tool Kit (ODTK) and its optimal sequential filter to process the right ascension and declination measurements and verify Bill Gray’s conclusion that the booster will impact the Moon. This orbit solution included measurements from the beginning of September 2021 through January 20, 2022, and covered two lunar flybys. Plots of the right ascension and declination measurement residuals are shown in the plots below. The residuals represent the difference between where the orbit solution predicted the booster to be and where it was measured to be in optical observations. Good agreement between orbit solution prediction and optical observation is required for confidence in the prediction.

Plot diagram 3

Plot diagram 4

Each point represents an individual measurement and each color represents an observatory. Note that the majority of the residuals are under an arcsecond in size.

The position uncertainty of our orbit solution, including all tracking measurements up to January 20 , are presented in the plot below. While uncertainty was initially high, we gained more confidence in our solution as more observations were collected in 2022 after the January flyby.

Plot diagram 5

Position uncertainty of our orbit solution using ODTK. Note that uncertainty increases significantly around each flyby date, but it diminishes soon after new observations are recorded and processed.

With this data, we propagated our orbit determination solution to visualize the booster’s covariance ellipsoid in Systems Tool Kit (STK). The tumbling of the booster results in a time-varying force caused by solar radiation pressure impinging on the booster’s sunlit side and introduces new complexities for an accurate model of the forces on its body.

Taking all the forces of Earth, the Moon, and the Sun into account, our initial model indicates that the booster will impact the Moon in the vicinity of the large Hertzsprung Crater on March 4, 2022, at about 12:27:46 UTC with a speed of 2.60 km/sec. The Hertzsprung Crater is located on the far side of the Moon, which means that we will not be able to observe the collision from here on Earth. Using STK’s Astrogator capability to propagate our orbital solution and employing a “zero-altitude above topography” stopping condition, we initially predicted that the impact site would be approximately 4.41° N, 131.19° W4.41° N, 131.19° W. But, the story doesn’t end there.

Refining Our Solution

Using the results of the initial solution, we can increase our confidence in the impact location and timing whenever new tracking data is obtained. From February 5-9, 2022, astronomers observed the booster once again to collect more data. We used this data to restart our filter and produce a refined orbit solution with small uncertainty in the impact conditions. The measurement summary and refined filter residuals and position uncertainty plots are displayed below.

Plot diagram 6Plot diagram 7Plot diagram 8

With this newly processed data, we found that the booster will impact the Moon at 4.58° N, 129.06° W on March 4, 2022, 12:26:58 UTC. This changes our impact time by about one minute and moves the impact site prediction by just over 250 km, closer to the center of the Hertzsprung Crater. The prediction from our original orbit solution is identical to our new orbit solution up until February 5, when new tracking data is obtained. At that time, the new solution’s covariance is reduced due to the infusion of measurement information, but the updated position uncertainty ellipsoid is contained in the position uncertainty from the prior solution. This statistical consistency between the initial and refined solutions is necessary for having a realistic covariance for the initial solution and gives us a sense of confidence in the realism of the updated solution and associated uncertainty. Because of the booster’s lighting conditions, optical telescopes will be unable to image the booster again before impact. This means that this refined solution will provide our best estimate for where the booster will ultimately land.

What Now?

While we won’t see the impact in real time, satellites orbiting the Moon may see the aftermath. In particular, NASA’s Lunar Reconnaissance Orbiter (LRO) and ISRO’s Chandrayaan-2 may be able to pick out the new crater created by the booster’s impact. Both satellites have onboard cameras capable of imaging the predicted crash site, and if we’re lucky we might find it! To predict when these satellites will have their first opportunities to image the site, we loaded ephemeris data for them into STK and determined when each satellite will next pass over Hertzsprung while it is lit by the Sun. We used a known ephemeris published by JPL for the LRO, but we had to simulate Chandrayaan-2’s future orbit with Astrogator because limited predictions are available. With a predicted state vector from February 23, 2022, we used Astrogator to propagate the ISRO’s satellite forward, including expected station keeping.


Each sharp increase in perigee radius shows a station keeping burn required to keep Chandrayaan-2 in its nominal orbit, about 100 km above the surface of the Moon.

With these orbits, we expect that the LRO’s first post-impact pass over the sunlit Hertzsprung Crater will occur on March 28, 2022, and Chandrayaan-2’s first pass will come shortly after on April 4, 2022. Pending instrument availability and a little bit of luck, we might be able to catch a glimpse of the booster’s remnants! Additionally, the impact crater is likely to expose some fresh lunar crust for imaging by these satellites, and the resulting data may have scientific value for geologists researching the Moon. Using STK’s Electro-Optical Infrared (EOIR) capability, we can model the LRO’s wide angle camera with publicly available technical specifications and create a rendering of what it might see on its pass at the end of March: 

EOIR-simulated WAC view

Background image shows previous composite observations taken by the LRO’s NACs, and inset shows EOIR-simulated WAC view of the impact site as the LRO passes overhead.

The impact poses no threat to missions operating on the lunar surface, including historical Apollo landing sites and other lunar probe locations. However, as humanity looks to make greater use of cislunar space in its bid to return to the Moon, the global community needs to be more cognizant of tracking debris and properly disposing of it when possible. Without the hard work of amateur astronomers keeping an eye on this booster, it is likely that its collision with the Moon would have gone completely unnoticed. Cislunar space introduces a host of complex multibody physics, and we can take advantage of it to develop novel trajectories and provide precise OD solutions. AGI is a pioneer in providing simulation tools that can be leveraged to fit this need — and so much more. Check out some more resources about cislunar space here:



Design high-fidelity spacecraft trajectories for mission planning and operations.

Orbit Determination Tool Kit (ODTK)

Process tracking data and generate orbit ephemeris with realistic covariance.

Systems Tool Kit (STK)

Modeling and simulation software for digital mission engineering and systems analysis.