Radar emitter geolocation from low Earth orbit
Passive RF-sensing satellites can fix the position of ground-based and shipborne radar transmitters without illuminating them, using time- and frequency-difference-of-arrival geometry across multi-satellite formations. Accuracy, revisit and ambiguity limits depend heavily on constellation geometry and emitter behaviour.
Sensors
- HawkEye 360 RF constellation: Three-satellite formation clusters flying in close trail, separations of tens to low hundreds of kilometres, enabling TDOA/FDOA geolocation of emitters from roughly 2 MHz to 18 GHz. Published geolocation accuracy is on the order of hundreds of metres to low single-digit kilometres depending on geometry and signal duration. Revisit varies by latitude; equatorial sites may see passes every few hours, polar sites more frequently.
- Kleos Space RF constellation: Four-satellite clusters in 37-degree and sun-synchronous orbits providing TDOA-based geolocation. Kleos targets maritime and land emitters across HF through microwave bands. Cluster geometry determines the dilution of precision; the company targets sub-kilometre accuracy for strong, stable emitters under good geometry.
- Spire Global RF payload (LEMUR-2): Spire's LEMUR-2 buses carry software-defined radio payloads capable of recording signal presence and basic spectral characterisation across a wide frequency range. The large constellation (100-plus satellites) provides high revisit for signal detection, though single-satellite passes cannot resolve emitter position without cross-cueing to a multi-satellite formation or ground-based reference.
- PICOSAR / experimental SAR-RF hybrid: Some SAR missions detect strong surface radars as interference artefacts in imagery, providing a coarse independent cross-check on emitter location. This is a secondary effect rather than a designed geolocation capability, but it has been documented in published literature for high-power coastal surveillance radars.
Why a radar cannot hide from a formation of listeners
Every active radar is, by definition, a radio transmitter. It broadcasts its presence at high power, often on predictable frequencies, in patterns that encode its mode, waveform and intent. A passive RF-sensing satellite does not need to transmit anything; it simply records what arrives at its antenna.
The geolocation trick rests on two physical differences between signals received at satellites flying in formation. Time-difference-of-arrival (TDOA) measures the tiny lag between the same pulse reaching satellite A versus satellite B. Frequency-difference-of-arrival (FDOA) measures the difference in Doppler shift experienced by each satellite as it moves at roughly 7.5 km/s relative to the emitter. Neither measurement alone pins a location. Together, with a third satellite providing a second baseline, they resolve a unique ground position. HawkEye 360 has published methodology describing exactly this three-satellite geometry. The maths is classical interferometry; the engineering challenge is synchronising satellite clocks and ephemerides to the nanosecond precision that TDOA demands.
Pulse-Doppler versus continuous-wave: the emitter makes a difference
Not all radars are equally cooperative targets for passive geolocation. A continuous-wave emitter, or one with a high duty cycle, gives the receiving satellites a long, stable signal to correlate. TDOA estimation improves with signal duration; a pulse lasting milliseconds is harder to time precisely than one lasting seconds.
Pulse-Doppler radars, which dominate military air-search and fire-control applications, transmit brief bursts at high peak power. The short pulse duration compresses the available correlation window, degrading TDOA accuracy unless the satellite receivers capture many pulses across the formation pass. Agile-frequency radars, which hop across bands between pulses, fragment the signal further. Geolocation is still possible but requires waveform-specific processing to reassemble the pulse train across frequency hops before cross-correlation.
Continuous-wave maritime navigation radars, by contrast, are among the easiest targets. They transmit almost continuously on well-known X-band or S-band frequencies, at predictable sweep rates, with high effective radiated power. A three-satellite formation passing overhead will typically accumulate enough signal energy for a reliable position fix within a single overpass.
What accuracy is actually achievable, and when it falls apart
Published figures from HawkEye 360 indicate geolocation accuracy in the range of several hundred metres to a few kilometres for well-behaved emitters under favourable geometry. That range is not evasion; it reflects real physics. The dominant error sources are satellite clock synchronisation residuals, orbital position uncertainty, ionospheric delay variation across the formation baseline, and the geometric dilution of precision that worsens when the formation baseline is short relative to slant range.
At 500 km orbital altitude with a formation baseline of, say, 100 km, the TDOA measurement must resolve time differences of microseconds or less. A one-nanosecond timing error translates to roughly 30 cm of range error on a single baseline, but errors compound across the geometry. Emitters near the sub-satellite track suffer worse dilution than those off to the side. Signals that are weak, intermittent or heavily modulated degrade correlation quality and push accuracy toward the worse end of the published range.
The honest limit: passive geolocation from LEO is not a precision targeting tool. It is a cueing and monitoring tool. It tells you a radar is active, approximately where it is, and something about its waveform. Confirmation to finer accuracy requires additional passes, cross-cueing to other sensors, or ground-based direction-finding.
Revisit, latency and the gaps in coverage
A single three-satellite cluster in a roughly 500 km orbit completes about 15 passes per day globally, but any given point on Earth sees perhaps two to four passes per day from that cluster, depending on latitude. High-latitude sites benefit from orbital convergence; equatorial sites see fewer passes per day from a single cluster.
Kleos and HawkEye 360 operate multiple clusters in different orbital planes, which improves revisit. Even so, the gap between passes at a mid-latitude site can be four to six hours. For a radar that transmits continuously, that gap is merely an observation gap. For a radar operating in brief emission windows, the probability of detection per day is substantially lower than the pass count suggests.
Data latency from detection to analyst delivery has been quoted by commercial operators in the range of one to a few hours after downlink, depending on ground station access and processing pipeline load. Near-real-time alerting for persistent emitters is operationally plausible; detecting a single brief emission and geolocating it within minutes is not.
Reading the waveform: what signal intelligence adds beyond position
Position is only the first product. A recorded signal also carries waveform characteristics: pulse repetition interval, pulse width, carrier frequency, scan rate and modulation type. These parameters form a radar's electronic fingerprint. Matching that fingerprint against a library of known emitters is the domain of electronic order-of-battle analysis, a discipline with a long history in airborne and shipborne SIGINT that is now being extended to spaceborne platforms.
From orbit, the recording fidelity is limited by receiver bandwidth, dynamic range and the signal-to-noise ratio achievable at several hundred kilometres of slant range. Published commercial systems do not claim to match the sensitivity of purpose-built SIGINT aircraft. What they offer is persistent, global, politically unconstrained access. A satellite overflying a denied maritime zone faces no diplomatic barrier that a reconnaissance aircraft would encounter.
Satellize ingests geolocation and waveform metadata from licensed commercial RF operators and fuses it with vessel tracking, SAR imagery and optical passes to build activity patterns over time. The Overhead column has covered how this fusion approach was tested in the context of the Kingdom of Tonga crop-estimation programme's data pipeline, where multi-source fusion methods were validated before being applied to security use cases.
Sovereignty, legality and the spectrum question
Passive reception of radio emissions from orbit is not regulated by the ITU in the same way as active transmissions. A satellite that only listens does not require a frequency assignment for its sensing function. The legal landscape is more nuanced for recording and distributing signal content, particularly for encrypted military waveforms, but the geolocation of an emitter based on signal timing and frequency is generally treated as a remote-sensing activity rather than an interception.
For government clients building a sovereign RF monitoring capability, the relevant question is not just legality but attribution confidence. A geolocation fix with a two-kilometre uncertainty ellipse, corroborated by a vessel AIS position or a SAR ship detection, is actionable intelligence. The same fix in open ocean with no corroborating data is a lead worth following on the next pass. Designing a monitoring programme means accepting that confidence accumulates across passes and sensors, not from a single observation.
Typical figures
| Frequency coverage | Approximately 2 MHz to 18 GHz across commercial RF-sensing payloads; specific band coverage varies by operator and payload generation |
| Geolocation accuracy (TDOA/FDOA, three-satellite) | Hundreds of metres to low single-digit kilometres depending on emitter stability, signal duration, formation geometry and clock synchronisation quality |
| Orbital altitude (typical) | 450 to 575 km LEO for HawkEye 360 and Kleos clusters |
| Revisit at mid-latitudes | Two to four passes per day per cluster; multiple clusters across operators can reduce gap to under two hours at some latitudes |
| Minimum detectable emitter power | Not publicly specified with precision; maritime navigation radars (typically 25 W to 50 kW peak) are reliably detected; low-power tactical radars present a harder target |
| Data latency (detection to delivery) | Approximately one to a few hours post-downlink under normal operations; near-real-time alerting for persistent emitters is feasible |
| Archive depth | HawkEye 360 has been operational since 2018; Kleos from 2020. Multi-year historical signal databases exist for licensed customers |
| Emitter types addressed | Maritime navigation radar (X-band, S-band), air-search radar (L-band, S-band), coastal surveillance radar, fire-control radar (where power permits detection) |
| Waveform characterisation outputs | Pulse repetition interval, pulse width, carrier frequency, scan period, modulation class (where SNR permits) |
Analytics Satellize can run
| Emitter geolocation fix with uncertainty ellipse | TDOA/FDOA cross-correlation across three-satellite formation baselines, following published HawkEye 360 methodology | GeoJSON point feature with semi-major/semi-minor axes of uncertainty ellipse, timestamp, frequency band and pass metadata |
| Radar activity timeline for a defined maritime zone | Aggregation of per-pass detection events across multiple clusters and operators; temporal binning to identify emission patterns and operating schedules | Time-series report and interactive chart showing emission frequency, gaps and apparent operational tempo over a client-specified period |
| Vessel-to-emitter association | Spatial and temporal correlation of RF geolocation fix with AIS position reports and SAR ship detections; dark vessel cross-check where AIS is absent | Linked vessel record with confidence score, supporting evidence log and map layer |
| Waveform fingerprint classification | Extraction of pulse descriptor words (PDW) from recorded signal; comparison against open-source radar parameter databases and client-supplied electronic order-of-battle libraries | Emitter classification report with candidate radar type, confidence level and parameter table |
| Persistent emitter monitoring alert | Automated detection of a known emitter reappearing within a defined area of interest across successive passes; threshold set by client on frequency, location and waveform match | Push alert with fix coordinates, time, frequency and link to full pass data; delivered via API or secure email |
| Multi-sensor activity fusion layer | Integration of RF geolocation, SAR imagery, optical tasking and AIS into a common operational picture; change detection against baseline period | Georeferenced activity layer in GIS-compatible format, updated per new pass, with analyst summary note |
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.