Interference into radio astronomy protected bands from LEO
ITU-protected radio astronomy bands are being measurably degraded by LEO constellation emissions and terrestrial leakage. This reference covers how interference is characterised, quantified and attributed using ground observatory monitors and space-based RF sensors.
Sensors
- Green Bank Observatory RFI monitoring system: Continuous wideband spectral monitoring across protected and adjacent bands, including 1400–1427 MHz (hydrogen line) and 10.6–10.7 GHz. Published interference logs document both impulsive LEO overpass events and persistent terrestrial leakage at levels well below ITU-R RA.769 thresholds.
- ITU-R RA.769 threshold framework: Not a sensor but the regulatory reference against which all measurements are normalised. RA.769 defines detrimental interference thresholds for each protected band; the 1400–1427 MHz threshold sits around -251 dB(W/m²/Hz) for a 2000-second integration, making it among the most sensitive limits in the radio spectrum.
- ESA Sentinel-1 SAR (C-band, 5.405 GHz): Sentinel-1's own published RFI analyses document how strong terrestrial emitters in the 5 GHz region corrupt synthetic aperture radar imagery. The same detection logic, identifying anomalous spectral energy in nominally passive or restricted bands, transfers directly to astronomy-band monitoring from orbit.
- Space-based RF spectrum monitors (e.g. HawkEye 360, Spire Global payloads): Commercial RF monitoring satellites can record wideband spectrograms during LEO passes. Frequency coverage and sensitivity vary by payload; Spire's STRATOS product covers L-band and adjacent regions. Spatial geolocation accuracy for emitters is typically 1–10 km depending on TDOA geometry and signal duration.
- VLBI and phased-array observatory flagging pipelines (e.g. MeerKAT, LOFAR): Modern observatory correlators flag RFI in near-real time. Published flagging statistics from MeerKAT and LOFAR quantify what fraction of L-band data must be excised during Starlink and OneWeb overpass windows, providing an independent measure of operational impact.
What RA.769 actually protects, and why it is almost impossible to honour
The ITU-R RA.769 recommendation lists threshold power flux densities below which interference is considered detrimental to radio astronomy observations. For the 1400–1427 MHz band, the threshold is approximately -251 dB(W/m²/Hz) integrated over 2000 seconds. That figure is not a bureaucratic rounding: it reflects the physical reality that hydrogen-line emission from distant galaxies arrives at Earth with power levels close to the thermal noise floor of cryogenically cooled receivers. The band is passive by allocation, meaning no transmissions are permitted within it. In practice, out-of-band emissions from adjacent services, and direct in-band leakage from LEO payloads, routinely exceed the threshold.
The 10.6–10.7 GHz band faces a structurally different problem. It sits between active radar and radiometry allocations, and thermal emission from Earth's atmosphere at these frequencies is scientifically valuable for retrieving water vapour and cloud liquid water. Interference here is harder to flag because the spectral environment is busier and the science signal is spectrally broad rather than a narrow line.
How LEO constellations produce interference even when they are not transmitting into the band
Starlink and OneWeb downlink signals are allocated to Ku-band (10.7–12.75 GHz) and Ka-band (17.8–20.2 GHz). Neither allocation overlaps the 10.6–10.7 GHz radio astronomy band. The problem is spectral leakage: transmitter harmonics, intermodulation products and imperfect filtering can deposit energy just below the allocated band edge. Published analyses by Hainaut and Williams (2020), though focused primarily on optical brightness, prompted the IAU to commission broader electromagnetic impact studies. Subsequent work by observatory teams confirmed measurable L-band leakage from Starlink satellites during overpass, attributable partly to on-board oscillator harmonics near 1.4 GHz.
The geometry compounds the problem. A single Starlink shell at 550 km altitude has an orbital period of roughly 96 minutes. At Green Bank's latitude, multiple satellites are above the horizon simultaneously for most of each hour. Each overpass lasts two to eight minutes depending on elevation angle. The duty cycle of contamination is therefore not a rare event: it is a near-continuous background modulated by constellation density. As shell populations grow toward the licensed figures of 12,000 or more satellites, the fraction of observing time recoverable by flagging and excision shrinks.
Ground observatory monitoring: what the Green Bank data shows
The Green Bank Observatory operates within the National Radio Quiet Zone, a 13,000 square mile area in West Virginia where terrestrial transmissions above certain power levels are restricted by federal regulation. Despite this, GBO's published RFI monitoring logs record interference events in the 1400–1427 MHz band from both satellite overpasses and distant terrestrial leakage. The observatory uses a wideband spectrometer capable of recording the full band at millisecond time resolution, which allows individual satellite passes to be identified and their spectral signatures characterised.
The key finding from published GBO monitoring is that interference events in the protected band cluster strongly in time with known LEO overpass windows, and that peak power levels during those windows can exceed the RA.769 threshold by 20 dB or more for short intervals. Excision of affected time samples is possible but not lossless: for transient phenomena such as fast radio bursts or pulsar single pulses, a contaminated two-minute window is simply lost. No post-processing recovers a signal buried under 20 dB of excess noise.
Space-based RF monitors as a complement to ground truth
Ground observatories measure what arrives at the dish. They cannot easily distinguish a satellite emitting slightly out of band from a terrestrial source whose signal scatters off the ionosphere. Space-based RF monitors, flying in orbits that allow them to observe LEO satellites from below or alongside, can record the satellite's own emission spectrum directly. This geometry removes the propagation ambiguity.
The honest limit of current commercial RF monitoring satellites is sensitivity. Detecting a weak harmonic from a Starlink satellite at a level relevant to RA.769 thresholds requires a receiver noise floor well below what most broadband monitoring payloads achieve. Published Spire and HawkEye 360 capabilities are optimised for detecting strong emitters, not sub-threshold spectral leakage. Purpose-built science payloads, such as those proposed under ESA's PECAS study, would be needed to close that gap. Until such payloads fly, the most reliable characterisation comes from combining ground observatory flagging statistics with orbital mechanics to infer per-satellite emission budgets.
Attribution, mitigation and the limits of coordination
Attributing a specific interference event to a specific satellite requires time-correlated orbital data (from public TLE catalogues) matched against the observatory's flagging timestamps. This is tractable for a constellation of known size with published ephemerides. It becomes harder as constellation operators update orbital planes, as TLE accuracy degrades over hours, and as multiple satellites are simultaneously above the horizon.
Mitigation options fall into two categories. Operational mitigation means scheduling observations to avoid known overpass windows, or flagging and excising contaminated samples. Both reduce available observing time. Technical mitigation means requiring constellation operators to suppress out-of-band emissions through better filtering or transmitter design. The ITU coordination process for this is slow relative to the pace of constellation deployment. SpaceX has engaged with NRAO and other observatories on a voluntary basis, and some firmware changes have reduced L-band leakage from later Starlink batches, but the published record does not yet show compliance with RA.769 thresholds across all operational modes. Satellize monitors the published literature and observatory flagging reports to track how that picture evolves.
The broader tension is not resolvable by engineering alone. Radio astronomy's sensitivity requirements are set by physics. The universe does not emit more loudly to compensate for a noisier electromagnetic environment. Spectrum sharing between passive science allocations and active commercial services is a coordination problem with no technically clean solution, only a series of trade-offs between scientific access and commercial capacity.
What characterisation data can and cannot tell a government buyer
A government operating a national radio astronomy facility, or considering building one, needs to know three things: how much observing time is currently lost to LEO interference, how much will be lost as constellations grow, and what regulatory or technical levers exist to limit that loss. Characterisation data from ground monitors and space-based RF sensors can answer the first question with reasonable precision. The second requires modelling constellation growth against published licence filings, which is uncertain but bounded. The third is a policy question that the data informs but does not answer.
The honest limit of any monitoring programme is that it is retrospective. It tells you what happened during the observation window. Predictive interference budgeting, projecting forward from current satellite counts and emission measurements to future observatory impact, requires combining measured per-satellite emission levels with orbital mechanics and receiver sensitivity models. That analysis is technically feasible and is the natural next step for any observatory or spectrum regulator that wants to move from complaint to evidence-based negotiation.
Typical figures
| Primary protected bands covered | 1400–1427 MHz (hydrogen line, OH line vicinity); 10.6–10.7 GHz (passive microwave) |
| ITU-R RA.769 detrimental threshold (1400–1427 MHz) | Approximately -251 dB(W/m²/Hz) for 2000-second integration time |
| Typical LEO overpass duration at observatory | 2–8 minutes per satellite at 550 km altitude, depending on elevation angle |
| Ground monitor time resolution | Millisecond-scale for wideband spectrometers (e.g. Green Bank VEGAS backend) |
| Space-based RF monitor geolocation accuracy | 1–10 km (TDOA-dependent; varies by signal duration and receiver geometry) |
| TLE orbital prediction accuracy (for attribution) | Typically 1–2 km in-track within hours of epoch; degrades over 24–48 hours |
| Observed excess interference above RA.769 threshold (published GBO data) | Up to ~20 dB above threshold during LEO overpass events in 1400–1427 MHz band |
| Archive depth (Green Bank RFI monitoring logs) | Continuous monitoring records extend back over a decade; publicly accessible subsets available |
| Starlink operational shell altitude (primary) | 540–570 km; orbital period approximately 95–96 minutes |
Analytics Satellize can run
| Per-satellite interference contribution estimate | Correlation of observatory flagging timestamps with TLE-derived overpass windows; statistical attribution across multiple events to isolate per-satellite emission budget | Tabular report ranking satellites or orbital planes by estimated L-band leakage contribution, updated as new flagging data is ingested |
| Observing-time loss projection | Orbital mechanics modelling of constellation growth (from published ITU filings) combined with measured per-satellite contamination durations; Monte Carlo over elevation-angle distributions | Time-series forecast of recoverable observing hours per month at a specified observatory latitude, under low/mid/high constellation growth scenarios |
| Regulatory evidence dossier | Aggregation of ground monitor spectrograms, flagging statistics and RA.769 threshold exceedance events into a structured evidence base aligned with ITU-R documentation formats | PDF and structured dataset suitable for submission to national spectrum regulators or ITU working party meetings |
| Interference source discrimination (satellite vs terrestrial) | Time-correlated matching of flagging events against TLE overpass catalogue; residual events not matching overpass windows classified as candidate terrestrial sources for follow-up direction-finding | Event log with source classification (satellite ID or terrestrial candidate), timestamp, frequency centroid and estimated power above threshold |
| Mitigation scheduling advisory | Forward propagation of TLE catalogue to generate overpass-free observation windows at specified coordinates; optimised against science programme scheduling constraints | Machine-readable observation schedule (VOSI or plain CSV) flagging high-risk windows for the following 7 or 30 days |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.