China recently presented a statement [1] on conformity with the Treaty on Principles Governing the Activities of States in the Exploration and Use of Outer Space, including the Moon and Other Celestial Bodies. In their statement, China references two close approaches between Starlink spacecraft and the Chinese Space Station (Tiangong, consisting of the Tianhe core module), which is currently docked with the three-person Shenzhou 13 spacecraft [2]. China judged the Starlink collision risk significant enough to warrant Tiangong avoidance maneuvers.
An excellent technical examination of the actual conjunction events drawing upon open-source data by Jim Shell examined the relative separations and likely maneuvers leading up to those two conjunction events. Rather than duplicate those results, we will examine, in a similarly dispassionate and technical manner, how often the Chinese Space Station comes close to Starlink or another spacecraft.
This question can be answered in several ways: (1) COMSPOC/CSSI’s “Number of Encounters Analysis Tool” (NEAT) , (2) direct application of the CSSI volumetric encounter method [5] to the Tiangong space station and orbit; and (3) CSSI’s technical paper on encounter rates presented at the recent International Astronautical Congress conference in Dubai .
NEAT Tool analysis
The NEAT tool [4] is based on parametric statistical assessments of the frequency of close approach for a spacecraft introduced into the existing space population. Like Jim Shell’s analysis, this tool utilizes TLE data provided by the U.S. government.
Inputs for the NEAT tool were set to a one-year assessment for the Tiangong Space Station, with warning threshold of 10 km (10,000m), maneuver threshold of 5 km, and collision threshold of 10 meters (corresponding to an overall Tiangong length of 20m), a Tiangong orbit altitude of 390 km, orbit inclination of 41.5°, and a space population of 21,995 space objects. This yields estimated warning rates of approximately one unique close approach below 10 km per day, an avoidance maneuver approximately once every four days, and an overall collision probability of one in a million on any given day if an effective collision avoidance maneuver is not taken.
Interestingly, the two reported Starlink conjunctions represent only a small fraction of the total expected close approaches that Tiangong would have experienced during a one-year period - just 2.4% [100 x (2 maneuvers attributed to Starlink per year) / (83.7 total avoidance maneuvers per year)].

Fig. 1. Estimation of Tiangong encounter rates by the NEAT tool
Estimated breakdown of Tiangong encounter rates using volumetric encounter method
The NEAT tool uses parametric evaluations of encounter rates spanning a wide range of orbit altitudes and inclinations, averaged across an ensemble of ascending node values. While useful and insightful, the results do not indicate the country of origin, type, or owner of the objects that Tiangong may conjunct with. In this second assessment method, we directly invoked CSSI's volumetric encounter method [5] to estimate the portion of encounters stemming from tracked debris, Starlink spacecraft, and all other (i.e., non-Starlink) active spacecraft. For this analysis, we adopted the U.S. Government's publicly shared Tiangong orbit elements (which closely match the orbit altitude and inclination specified in the Note Verbale to UN COOPUOS [1] of approximately 390 km in nearly circular orbit inclined at 41.5°).
The results of this second assessment are contained in Table 1. As a sanity check, note that the total number of annual encounters against all publicly tracked debris and active spacecraft totals 101.2 per year, which is about 20% higher than the more approximate value provided by the NEAT tool of 83.7 encounters per year. Starlink only accounts for about 7% of all conjunctions within 5 km, and 17% of all Tiangong encounters with active spacecraft. Debris is identified as the dominant source of close approaches; debris fragments generated by China's own ASAT test in 2007 and the recent Russian ASAT test contribute to this debris risk.
Table 1. Breakdown of Tiangong close approach encounters
Orbit altitude (km) | Tiangong encounters with debris | Tiangong encounters with other (non-Starlink) spacecraft | Tiangong encounters with Starlink spacecraft |
380 to 385 | 6.1 | 1.2 | 1.8 |
385 to 390 | 35.3 | 11.9 | 1.2 |
390 to 395 | 12.9 | 19 | 4.3 |
395 to 400 | 4.6 | 2.9 | 0 |
Total, 380-400 km | 58.9 | 35 | 7.3 |
These findings suggest that Starlink spacecraft do not place undue flight safety burden on the Tiangong Space Station crew or their flight dynamics staff as compared to other active spacecraft that transit Tiangong's orbit altitude band. However, the data highlight the importance of sharing orbit and maneuver information as called for in the UN COPUOS 21 guidelines for Long Term Sustainability (LTS) of space activities in two ways: (1) Starlink, Tiangong, ISS, and other active spacecraft often maneuver (for stationkeeping, orbit raising, orbit lowering/disposal, and collision avoidance), making SSA and Space Traffic Management (STM) challenging, and (2) that Tiangong can be expected to come within 5 km of other active spacecraft every nine days on average.
For their part, Starlink posts their ephemeris, which incorporates planned maneuvers, on the Space-Track.org website in a “public” folder created at the request of Starlink to allow any/all Space-Track.org users to download and use that information for the purposes of flight safety. Unfortunately, the Tiangong operators do not appear to similarly share or post their ephemerides incorporating planned maneuvers.
Evaluation of overall LEO conjunction rates from historical flight safety monitoring systems
As depicted in our recent LEO conjunction rates paper, the rate of conjunctions is increasing dramatically in just the past 4 years, due in part to the increasing amount of debris from both intentional fragmentation events as well as approximately a dozen unintentional fragmentations that occur each year (ESA), but driven primarily by the large constellations introduced in the New Space era. Figure 2 is based on observed (actual) conjunction rates and illustrates this increase, which was corroborated nicely by the data points (blue square symbols) generated by our statistical estimates of encounter rates.

Fig. 2. Volumetric non-fratricide
encounter rates added to the SOCRATES and SDA depiction
Fig. 3 presents our overall assessment of LEO encounter rates of spacecraft with debris (blue dots) and active spacecraft (magenta diamonds). The increase in encounter rates corresponding to the altitudes of large constellations is readily apparent. This is not the result of any specific constellation, but rather of the sheer quantity of spacecraft in the constellation, the altitude range(s) that the constellation occupies, and to what extent other spacecraft occupy (or transit) that space. If realized, the Chinese Hongyan large constellation consisting of 12,992 spacecraft [7] would experience similarly close approaches with other active spacecraft.
One important consideration is that we may be seeing a “clash of best practices” occurring with large constellations. The Starlink constellation is initially launched into a relatively low altitude of 350 km to allow the spacecraft operator to carefully check out and verify the proper operation of each spacecraft prior to doing an orbit raise for insertion into the operational altitude. This operational procedure is considered a best practice from an orbital debris mitigation standpoint. Similarly, if a spacecraft is experiencing an anomaly or degraded functionality, it is considered best practice to lower its orbit altitude to either cause it to re-enter or to minimize its post-mission orbit lifetime; again, an orbital debris mitigation best practice. These lower initial and disposal altitudes cause the spacecraft to share orbits with human-habitable space stations such as the ISS and Tiangong.
Jim Shell captured it well with his
statement that perhaps such close approaches between human-inhabited and
robotic spacecraft may be just the forcing function the global space community
needs to prioritize (1) space cooperation; (2) two-way exchange of information;
and (3) space traffic coordination and management.

Fig. 3. Comparison of debris and
active satellite encounter rates as a function of altitude.
This series of events also highlights the cooperation that the U.S. Government's Departments of Commerce and Defense could leverage to facilitate and conduct comprehensive commercial industry assessments and monitoring of such situations. The commercial community routinely performs assessments and detailed analyses in support of space safety.
While the “normal” distance and probability conjunction assessment metrics provide legitimate approaches to assess collision risk, in the case of Starlink (and any New Space large constellation spacecraft, for that matter) that use low thrust to frequently maneuver, any STC or STM system that does not reflect planned maneuvers for all satellites can improperly assess miss distance and collision risk. To address this SSA gap, serious efforts to collaborate, share space data and employ commercial space safety services will be required.
It was also quite interesting that, as Jim Shell pointed out on one of the conjunctions, an automated Starlink collision avoidance maneuver (which, while freely shared on the Space-Track website in ephemeris form, is apparently not being incorporated into China’s Tiangong flight safety assessments) may have led China’s analysts to assume that Starlink was doing uncertain maneuvers and therefore should be avoided. But as we have learned in such situations on previous occasions, performing a collision avoidance maneuver “in the blind” is ill-advised because it can potentially steer your spacecraft toward the other space object, rather than away from it.
“Evaluation of LEO conjunction rates using historical flight safety systems and analytical algorithms,” IAC-21-A6.7x65213, Dan Oltrogge, Salvatore Alfano, Jim
Wilson, Pascal Wauthier, October 2021, Dubai, UAE.