P-band and UHF foliage-penetration emitter detection from orbit
Low-frequency emitters in dense forest are effectively invisible to ground direction-finding networks, yet their signals propagate upward with little canopy loss. Spaceborne RF receivers at LEO can intercept these emissions and geolocate sources using time-difference-of-arrival across clustered satellites.
Sensors
- HawkEye 360 cluster constellation: Three-satellite clusters in 550–600 km LEO, separated by tens of kilometres, collect RF signals from roughly 100 MHz to 18 GHz. TDOA/FDOA geolocation across the cluster yields position estimates with published circular error probable in the range of 1–10 km depending on emitter duty cycle, signal duration, and cluster geometry at the moment of overflight. Revisit over a fixed point is typically several times per day globally, improving with latitude.
- Spire STRATOS RF payload: Broadband RF monitoring payload hosted on Spire's 3U cubesat constellation. Covers VHF and UHF bands including the 200–500 MHz range of interest. Designed for signal detection and characterisation rather than precision geolocation; useful for spectrum occupancy mapping and confirming emission presence before tasking a cluster geometry optimised for TDOA.
- ESA BIOMASS P-band SAR (published design): Designed to operate at 435 MHz (P-band) with a synthetic aperture radar in fully polarimetric stripmap and tomographic modes. Although its primary mission is forest biomass measurement, its published design documentation establishes the foliage-penetration phenomenology at this frequency: canopy two-way attenuation is typically 3–8 dB for dense tropical forest at P-band, compared with 20–30 dB at C-band. This calibrates the link-budget assumptions for passive intercept at similar frequencies.
- Future dedicated SIGINT constellations (generic LEO class): Several government-sponsored and commercial programmes are developing or operating LEO SIGINT satellites covering VHF/UHF. Orbital parameters (400–600 km altitude) give a ground footprint of roughly 4,000–5,000 km diameter per satellite, meaning a single pass can intercept emissions across a large forested region. Specific system sensitivities are classified; publicly available link-budget analysis sets a theoretical noise floor for a 1 W emitter detectable at 600 km with a modest aperture antenna.
Why low frequency defeats the canopy
At frequencies above 1 GHz, dense tropical forest attenuates radio signals by tens of decibels. Ground-based direction-finding networks rely on this: an emitter under a closed canopy is simply too weak to detect beyond a few kilometres. P-band and UHF (broadly 200–500 MHz) behave differently. Wavelengths at 300 MHz are roughly one metre, comparable to branch spacing rather than leaf spacing, so scattering is far weaker. ESA's published BIOMASS design documentation, grounded in decades of airborne P-band SAR campaigns, records typical one-way canopy attenuation of 1.5–4 dB in dense tropical forest at 435 MHz. That is not zero, but it is far from the 15 dB or more that kills a C-band signal.
The consequence for military concealment is well understood by operators of tactical VHF/UHF radios, HF-to-UHF relay nodes, and certain radar systems. A unit transmitting on 300–400 MHz from beneath a forest canopy can remain invisible to ground sensors while remaining detectable from above. The canopy loss that protects the emitter from horizontal interception is largely irrelevant to a receiver looking down from orbit at a geometry that is close to vertical.
How orbit intercepts what the canopy hides
A LEO satellite at 550 km altitude has a geometric view of a ground footprint roughly 4,500 km across. Any emission that clears the canopy and propagates upward into that cone is potentially interceptable. The critical variable is emitter effective isotropic radiated power (EIRP). A tactical radio running 10 W into a whip antenna under canopy might exit the forest with 5–8 W equivalent, still detectable against the thermal noise floor of a well-designed spaceborne receiver at that altitude, provided the signal lasts long enough for the receiver to integrate.
Detection alone is not geolocation. To fix a position, you need either multiple simultaneous receivers or a single receiver making multiple measurements as it moves. HawkEye 360's published cluster architecture uses three satellites separated by tens of kilometres in the same orbital plane, collecting the same signal at slightly different times. The time difference of arrival (TDOA) between pairs of satellites, combined with frequency difference of arrival (FDOA) from the Doppler shift introduced by each satellite's velocity, constrains the emitter to a hyperbolic locus on the ground. Two independent TDOA measurements from three receivers reduce that to a point, with accuracy depending on the signal-to-noise ratio of the TDOA measurement and the cluster geometry at the moment of collection.
Sensitivity limits the honest buyer should know
TDOA accuracy degrades sharply for low duty-cycle emitters. A burst-mode tactical radio transmitting for 50 milliseconds every few minutes may not produce a signal long enough for all three satellites in a cluster to record a clean cross-correlation. Published geolocation research on TDOA systems indicates that position error grows roughly inversely with the square root of signal duration and signal-to-noise ratio. A 1 km CEP result quoted for a continuous emitter may become 5–10 km for a short burst at the same power.
Canopy attenuation, though lower at P-band than at higher frequencies, is not negligible. Very dense wet tropical forest at the high end of the 200–500 MHz range can impose 4–8 dB two-way loss; for a passive intercept receiver looking downward, only the upward path matters, so roughly half that figure applies. This still reduces detectable range and SNR. Emitters below roughly 1 W EIRP after canopy loss become marginal at 550 km altitude with a small aperture receiver. Revisit is another honest constraint: a cluster passing over a specific 100 km² forest block may do so only a handful of times per day, and an emitter that is silent during those windows will not be detected regardless of its peak power.
Geolocation geometry and what drives the error budget
The TDOA hyperbola from one satellite pair is a wide arc on the ground. The intersection of two such arcs from three satellites narrows it to an ellipse whose minor axis is roughly aligned with the baseline between the outer satellites. Cluster geometry at the moment of collection therefore matters enormously. A cluster passing directly overhead produces a near-circular error ellipse; a cluster at low elevation angle on the horizon produces a highly elongated one, sometimes stretching the 95th-percentile error to 20 km or more along the radial direction.
FDOA adds independent information. Because each satellite moves at roughly 7.5 km/s, the Doppler shift of the received signal differs between satellites by an amount that depends on the emitter's position relative to the velocity vectors. Combining TDOA and FDOA tightens the solution, particularly for emitters near the edge of the cluster's ground track. In practice, commercial RF geolocation vendors report CEP figures for representative scenarios rather than worst cases; a buyer should ask specifically about performance at low elevation angles and for emitters in the target frequency range.
What a monitoring programme actually looks like
A practical foliage-penetration RF monitoring programme combines persistent detection with periodic geolocation tasking. Spectrum occupancy data from a broad-coverage payload such as Spire STRATOS flags when and where energy appears in the 200–500 MHz band over a forest region of interest. That detection cues targeted collection by a cluster geometry optimised for TDOA, scheduled for the next pass with a favourable elevation angle over the area.
Satellize structures analytics pipelines of this kind for government clients, integrating open and commercial RF data streams with geometric quality indicators to flag only those passes where the error budget meets the client's stated accuracy requirement. The approach is similar in principle to the geolocation workflow underlying our Tonga crop-estimation programme, where pass-selection logic filters for the collection geometry that maximises analytic value rather than simply ingesting everything. For RF geolocation, that geometry filter is the difference between a 2 km fix and a 15 km smear.
Output formats typically include geolocated emitter reports with error ellipses, time-stamped activity logs, and, where multiple detections exist, pattern-of-life summaries showing emission schedules. These feed directly into GIS environments or structured intelligence reports depending on the client's workflow.
Regulatory and operational context
Passive RF intercept from orbit occupies a legally complex space. Receiving radio emissions is generally not regulated by the ITU Radio Regulations in the same way as transmitting; however, the use of intercepted signals and the nationality of the operator introduce national legal constraints that vary significantly. Buyers should obtain their own legal assessment before commissioning a monitoring programme targeting specific emitters.
On the technical side, the 400–406 MHz range includes internationally protected frequencies for meteorological and distress applications. A monitoring programme should apply frequency masks to avoid conflating tactical emitters with protected-band signals, and should maintain audit logs of collection parameters. This is not a hypothetical concern: the 406 MHz distress beacon band sits squarely within the upper edge of the target range, and false attribution of a distress signal as a tactical emitter would be both operationally and legally serious.
Typical figures
| Target frequency range | 200–500 MHz (P-band and upper UHF); HawkEye 360 cluster covers approximately 100 MHz–18 GHz, with UHF included |
| Orbital altitude (typical LEO RF collectors) | 400–600 km; HawkEye 360 clusters operate at approximately 550–600 km |
| Ground footprint per pass | Approximately 4,000–5,000 km diameter visibility circle at 550 km altitude |
| Geolocation accuracy (TDOA/FDOA, continuous emitter) | Published CEP figures for HawkEye 360 range from approximately 1 km (favourable geometry) to 10 km (low elevation, short signal) |
| Geolocation accuracy (burst/low duty-cycle emitter) | Degrades significantly; 5–20 km CEP is a realistic range for sub-100 ms bursts at typical tactical power levels |
| Canopy attenuation at P-band (one-way, dense tropical forest) | Approximately 1.5–4 dB at 435 MHz per published BIOMASS campaign data; higher toward 500 MHz |
| Revisit over a fixed point | Several passes per day globally for HawkEye 360; passes with geolocation-quality geometry (favourable elevation) are a subset of total passes |
| Minimum detectable EIRP (indicative, not guaranteed) | Approximately 1–5 W EIRP after canopy loss at 550 km altitude for a continuous signal; lower duty-cycle emitters require higher peak power |
| Latency from collection to geolocated report | Typically hours for commercial RF geolocation services; near-real-time pipelines are technically feasible with direct downlink infrastructure |
| Archive depth | HawkEye 360 commercial archive available from approximately 2019; Spire STRATOS RF data from approximately 2020 |
Analytics Satellize can run
| Emitter detection alert | Energy detection in target frequency band using broadband RF payload data; threshold set against background noise floor with frequency masks for protected bands | Timestamped alert with approximate bearing cone, frequency, and estimated signal duration; delivered as structured feed or email notification |
| Geolocated emitter fix with error ellipse | TDOA/FDOA across three-satellite cluster; cross-correlation of received waveforms to derive time differences; hyperbola intersection with geometric dilution of precision (GDOP) quality flag | GeoJSON point with associated error ellipse polygon, GDOP score, and collection geometry metadata; suitable for GIS ingestion |
| Pass geometry quality forecast | Orbital propagation of cluster satellites over area of interest; computation of elevation angles and baseline geometry for each upcoming pass; ranking by expected GDOP | Collection opportunity schedule (24–72 hour horizon) showing predicted CEP for each pass; used to prioritise tasking requests |
| Emission pattern-of-life summary | Aggregation of multiple detection and geolocation events over days to weeks; clustering by frequency, estimated location, and time-of-day; statistical characterisation of duty cycle and schedule | PDF or structured report showing emission timeline, location cluster map, and inferred activity schedule with confidence intervals |
| Frequency occupancy map for forest region | Spectral energy aggregation from broadband RF payload across multiple passes; gridded into 0.1–0.5 degree cells; normalised for receiver gain and pass geometry | Raster GIS layer (GeoTIFF) showing mean signal energy by frequency sub-band and grid cell; updated on configurable cadence |
| Change detection: new emitter onset | Comparison of current-period occupancy map against historical baseline; anomaly flagging where energy exceeds baseline by a defined sigma threshold in a cell with no prior emission history | Alerting layer highlighting grid cells with statistically anomalous new RF activity; timestamped and frequency-tagged |
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.