Ground-based space surveillance radar emission monitoring from orbit

High-power ground-based space surveillance radars emit signals detectable by RF receivers in LEO. Characterising their waveforms, frequencies, and duty cycles from orbit reveals operational tempo and mode changes, with significant caveats around classification and ambiguity.

Sensors

HawkEye 360 cluster satellites: Three-satellite formation flying at roughly 575 km altitude, using time-difference and frequency-difference of arrival (TDOA/FDOA) across the cluster to geolocate emitters. Geolocation accuracy is publicly cited at 1–5 km CEP depending on geometry and signal duration. Frequency coverage spans HF through Ku-band. Revisit over a fixed point is irregular, driven by orbital mechanics, typically several passes per day at mid-latitudes.
Spire LEMUR-2: A constellation of over 100 LEO satellites carrying GPS-RO and AIS payloads, with some units carrying general-purpose RF monitoring receivers. Spire does not publish detailed sensitivity figures for its RF-monitoring capability, but the constellation's density gives frequent revisit, potentially multiple passes per hour at high latitudes. Useful for persistent spectral occupancy monitoring rather than precision geolocation.
US Space Fence (Lockheed Martin / USAF, Kwajalein Atoll): Not an orbiting sensor but the primary publicly documented subject emitter. Operates at S-band (approximately 1.3–2.0 GHz range, specific frequency classified). Transmit power is in the megawatt class. Its emissions are the reference signal for this use case: a LEO receiver overflying Kwajalein will receive a strong S-band pulse train whose PRF, pulse width, and scan pattern are in principle measurable.
GRAVES (France, ONERA/DGA, Dijon): French space surveillance radar operating in the VHF band at 143.050 MHz, one of the few space surveillance radars whose operating frequency is openly published in ONERA literature. Transmit power approximately 40 kW. Continuous-wave frequency-modulated waveform. Its relatively low frequency and CW nature make it distinctly identifiable in wideband RF captures from orbit, and it has been used in amateur radio and scientific bistatic experiments.
Russian Voronezh-series early-warning radars: UHF and VHF phased-array systems deployed at multiple sites across Russia (Lekhtusi, Armavir, Orsk, Barnaul and others). Operating frequencies are not officially published but open-source analysis places them in the VHF/UHF range. Transmit power is in the megawatt class. Their large aperture and high duty cycle make them plausible targets for orbital RF detection, though mode discrimination from orbit is speculative without classified reference data.

Why a megawatt radar is hard to hide from a receiver in orbit

Space surveillance radars are, by design, among the most powerful ground-based emitters on Earth. The US Space Fence on Kwajalein operates at S-band with transmit power in the megawatt class, illuminating a curtain of space to detect objects as small as a baseball in low Earth orbit. That power level does not stay neatly pointed upward. A LEO satellite passing within several hundred kilometres of the beam's sidelobe structure, or crossing the main beam at a high elevation angle, will receive a signal many orders of magnitude above the noise floor of a modest RF receiver.

The physics is straightforward. Free-space path loss at S-band over 600 km is roughly 148 dB. A 1 MW transmitter produces 60 dBW of radiated power. Even accounting for sidelobe suppression of 30–40 dB relative to the main beam, a receiver with a modest antenna gain and a noise figure of a few decibels can detect the pulse train. GRAVES, operating at 143.050 MHz with a CW waveform, is even more accessible: its published frequency makes it trivially identifiable in a wideband spectrogram, and its continuous-wave nature means there is always signal to integrate.

What the waveform actually tells you, and what it does not

A captured pulse train carries pulse repetition frequency, pulse width, and, if the receiver has sufficient bandwidth and dynamic range, intra-pulse modulation such as linear frequency chirp. From these parameters an analyst can infer duty cycle, approximate range resolution (inversely proportional to bandwidth), and, over multiple passes, whether the radar is operating in a search mode, a tracking mode, or is simply on standby. Mode changes are operationally significant: a radar shifting from routine fence surveillance to a higher duty-cycle tracking mode suggests it has acquired a target of interest.

The honest limits here are severe. First, the classification status of Space Fence waveform parameters means that any inference about mode from an unclassified orbital capture is necessarily ambiguous without a validated reference library. Second, distinguishing between a deliberate mode change and a routine maintenance cycle requires persistent monitoring over weeks or months, not a single pass. Third, at S-band the Space Fence beam is narrow enough that a LEO satellite may only be in a detectable sidelobe geometry for seconds per pass, limiting the data volume per opportunity. GRAVES is the exception: its published frequency and CW waveform make it the closest thing to a calibration source for this type of monitoring.

Geolocation from orbit: the cluster geometry problem

HawkEye 360's published approach uses TDOA and FDOA across a three-satellite cluster to fix an emitter's position. For a fixed, known installation like Space Fence or GRAVES, geolocation is not the analytical goal. The emitter's coordinates are already public. What matters instead is signal characterisation: frequency stability, spectral purity, pulse timing, and any anomalies that deviate from baseline. This shifts the problem from 'where is it' to 'what is it doing and has anything changed'.

For the Voronezh systems, whose precise operating frequencies are not officially published, geolocation from a cluster satellite does add value by confirming which site is active and at what frequency, cross-referenced against the known site coordinates from open-source imagery. The 1–5 km CEP accuracy of commercial cluster systems is more than adequate to discriminate between Voronezh sites separated by hundreds of kilometres. The challenge is that a single cluster pass may not capture enough signal duration to produce a reliable FDOA solution if the radar is scanning rather than fixed-beam.

Operational inferences a government client can draw

Persistent orbital RF monitoring of known space surveillance radar sites can support several concrete intelligence products. Operational tempo assessment: is the radar on for its normal fraction of each day, or has duty cycle increased, suggesting a heightened tracking campaign? Frequency drift monitoring: a radar whose carrier frequency shifts by even tens of kilohertz may be undergoing hardware changes or experiencing component degradation. Site activation status: for a network of Voronezh installations, orbital passes can confirm which sites are radiating on a given day, relevant to understanding the coverage geometry of an adversary's space surveillance network.

None of these inferences is clean or automatic. A duty cycle increase could mean a satellite conjunction event rather than a deliberate campaign. A frequency shift could be a calibration adjustment. Confidence requires corroboration with other sources, including open-source reporting, orbital conjunction data from Space-Track, and historical RF baselines built over months of collection. Satellize's approach to this type of programme is to be explicit about the confidence level attached to each inference, distinguishing 'observed change in emission pattern' from 'assessed operational significance'.

Governments procuring this capability should also understand that the most sensitive analytical products in this domain are almost certainly already held by national signals intelligence agencies. Commercial orbital RF monitoring fills a different niche: it is available to allies and partners who lack organic SIGINT collection, it is non-classified and therefore shareable across coalition partners, and it provides a persistent baseline that can be compared against classified reporting rather than replacing it.

The classification ceiling and what sits below it

The most important honesty note for any buyer: the detailed waveform parameters of Space Fence and the Voronezh systems are classified. This means that a commercial RF monitoring programme cannot produce a validated mode library for these systems from open sources alone. What it can produce is a change-detection product, flagging deviations from a self-generated baseline without necessarily knowing what the baseline represents in operational terms.

GRAVES is the outlier. Because its frequency (143.050 MHz) and general waveform type (CW-FM) are published in open ONERA and academic literature, it can serve as a genuine calibration reference for orbital RF receivers. A satellite that correctly characterises GRAVES against its published parameters has demonstrated its measurement chain is working. That validation matters when the same receiver is then turned toward less well-documented emitters. The bistatic use of GRAVES signals by amateur radio operators and university researchers, documented in the open literature, provides additional independent validation of what an orbital receiver should see.

Typical figures

Frequency coverage (commercial RF monitoring)	HF through Ku-band for HawkEye 360; GRAVES target at 143.050 MHz (VHF); Space Fence at S-band (specific frequency classified)
Geolocation accuracy (cluster TDOA/FDOA)	1–5 km CEP (HawkEye 360 published figures); adequate for site discrimination, not required for known fixed installations
Revisit at a fixed mid-latitude site	Several passes per day for a three-satellite cluster; higher for large constellations such as Spire LEMUR-2, but individual pass duration over a fixed point is typically 5–10 minutes
Minimum detectable signal	Not publicly specified for commercial systems; megawatt-class S-band emitters at 600 km are well above the detection threshold of any competent LEO RF receiver
Pulse timing resolution	Dependent on receiver sample rate; microsecond-class timing resolution is achievable with commodity SDR hardware at relevant bandwidths
Frequency measurement accuracy	Sub-kilohertz carrier frequency accuracy achievable with GPS-disciplined oscillators on LEO platforms; adequate for drift monitoring
Archive depth (commercial)	HawkEye 360 has operated since 2018; Spire LEMUR-2 since 2015. Baseline depth for change detection depends on tasking history for specific emitters
Classification ceiling	Detailed waveform parameters of Space Fence and Voronezh systems are classified; commercial products are limited to change detection against self-generated baselines
Delivery formats	Spectrogram time series, emitter activity reports, geolocation shapefiles, duty-cycle trend charts

Analytics Satellize can run

Operational tempo index	Duty-cycle estimation from pulse counting across successive orbital passes, compared against a rolling 90-day baseline	Weekly trend report with flagged deviations above a client-defined threshold
Carrier frequency drift log	GPS-disciplined frequency measurement on each pass, time-stamped and plotted against historical values; anomalies flagged at configurable tolerance	Time-series chart and alert feed when drift exceeds ±10 kHz from baseline
Site activation status map	Presence/absence of expected emission on each orbital pass over known Voronezh site coordinates, aggregated daily	Daily GIS layer showing active/inactive status per site, exportable to client GIS platform
GRAVES calibration validation report	Comparison of measured frequency, signal level, and waveform type against published ONERA parameters; used to validate receiver chain performance	Quarterly calibration certificate with measured versus published parameter table
Waveform anomaly alert	Spectrogram change detection using baseline subtraction; flags unexpected intra-pulse modulation changes or new spectral components	Event alert within 24 hours of the triggering pass, with annotated spectrogram attachment
Network coverage geometry model	Combination of confirmed active-site data with published or estimated beam parameters to model instantaneous space surveillance coverage, cross-referenced against Space-Track conjunction data	Monthly briefing slide deck with coverage gap analysis

Sources

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.