How Over-the-Horizon Radar Actually Works: Skywave Propagation and Thousand-Kilometre Detection

Conventional microwave radar has a hard ceiling on its useful range, and that ceiling is geometry, not power. The Earth curves away from any antenna at a predictable rate, and no amount of transmitter wattage will bend a 10 GHz beam around that curve. For an antenna mounted at 30 metres above sea level, the radio horizon sits at roughly 20 kilometres. Even airborne early warning platforms flying at 9,000 metres only push the horizon out to about 340 kilometres. Against a cruise missile skimming the surface at Mach 0.8, that translates to perhaps seven minutes of warning time. Against a ballistic missile in its boost phase 2,500 kilometres away, conventional radar sees nothing at all.

Over-the-horizon radar (OTH-R) exists to close that gap. By operating in the high-frequency (HF) band between 3 and 30 MHz, these systems exploit a phenomenon that microwave radars cannot: the refraction of radio waves by the ionosphere. An HF signal transmitted at the correct angle and frequency bends back toward the Earth's surface hundreds or thousands of kilometres from the transmitter, illuminates whatever is out there (aircraft, ships, missile launches, even weather formations), and the scattered return follows the same ionospheric path back to a receiver. The result is a radar that can detect targets at ranges from 1,000 to 3,500 kilometres, covering millions of square kilometres of airspace and ocean from a single fixed installation.

This is not theoretical. Operational OTH-R systems have been running continuous surveillance missions since the 1980s. The Australian JORN system watches the northern approaches from bases in Queensland and Western Australia. The French NOSTRADAMUS monitors air traffic across the Mediterranean and eastern Atlantic. Russia operates several OTH-R variants including the 29B6 Container system. The physics is well understood, but the engineering required to make it work reliably is extraordinary.

The Horizon Problem and the HF Solution

The radar horizon distance d for an antenna at height h above a spherical Earth of radius R follows a simple geometric relationship:

d = sqrt(2 * R * h)

With R approximately 6,371 kilometres, an antenna at 25 metres height has a horizon of about 17.8 kilometres. A target aircraft at 10,000 metres altitude extends this somewhat (you add the horizon distance from the target's own height), but the fundamental constraint remains. Microwave frequencies (1 to 40 GHz) propagate in straight lines through the troposphere. Atmospheric refraction bends them slightly, extending the effective horizon by roughly 15% compared to purely geometric calculations (the standard "4/3 Earth radius" model), but this is a minor correction.

HF frequencies behave differently because of the ionosphere. Between roughly 60 and 1,000 kilometres altitude, solar ultraviolet and X-ray radiation ionises atmospheric gases, creating layers of free electrons and ions. These free electrons interact with radio waves, and the interaction depends strongly on frequency. At HF frequencies, the electron density in the ionosphere is sufficient to refract signals back toward the ground. At microwave frequencies, the electron densities are far too low to have any meaningful effect; the signals pass straight through, which is exactly why satellite communications work at microwave bands.

The key parameter is the plasma frequency of the ionosphere, given by:

f_p = 9 * sqrt(N_e)

where N_e is the electron density in electrons per cubic metre and f_p is in Hz. When a radio wave enters a region where its frequency equals the local plasma frequency, it is reflected (or more precisely, refracted through a continuous gradient until it curves back downward). Typical peak electron densities in the F2 layer (the most useful layer for OTH-R) range from about 10^11 to 10^12 electrons per cubic metre, yielding critical frequencies (the maximum frequency reflected at vertical incidence) between roughly 3 and 10 MHz. By transmitting at oblique angles, Snell's law allows frequencies well above the critical frequency to be refracted back to Earth; the maximum usable frequency (MUF) for a given path can reach 30 MHz or more at low elevation angles.

This is the entire physical basis of OTH radar: choose a frequency below the MUF for the desired propagation path, transmit at the correct elevation angle, and the ionosphere acts as a curved mirror that bends the signal over the horizon.

Ionospheric Physics for Radar Engineers

The ionosphere is not a single reflective shell. It is a complex, dynamic, stratified medium that varies with time of day, season, solar activity, geomagnetic disturbances, and geographic location. Making OTH-R work operationally requires intimate, real-time knowledge of ionospheric conditions. This is arguably the hardest part of the whole enterprise.

Layer Structure

The ionosphere is conventionally divided into several layers, each with distinct characteristics:

D layer (60 to 90 km altitude): Present only during daytime, this layer has relatively low electron density but high collision frequency between electrons and neutral molecules. It does not reflect HF signals; instead, it absorbs them. D-layer absorption is inversely proportional to the square of the frequency, so lower HF frequencies (3 to 5 MHz) suffer far more attenuation than higher ones (20 to 30 MHz). At night, the D layer disappears, which is why AM broadcast stations can be heard at much greater distances after sunset.

E layer (90 to 150 km altitude): A daytime layer with moderate electron density. The E layer can reflect signals up to about 3 to 4 MHz at vertical incidence. Sporadic E (Es), patchy regions of unusually high electron density, can appear unpredictably and reflect much higher frequencies, sometimes causing unwanted multipath propagation that complicates OTH-R signal interpretation.

F1 layer (150 to 220 km altitude): Present during daytime, particularly in summer. Merges with the F2 layer at night. Moderate electron density, useful for some propagation paths but not the primary layer exploited by OTH-R.

F2 layer (220 to 800 km altitude): The most important layer for OTH-R. The F2 layer has the highest electron density of any ionospheric region, persists through the night (though at reduced density), and provides the refraction needed for long-range skywave propagation. Peak electron densities in the F2 layer vary from about 2 x 10^11 m^-3 at night to 10^12 m^-3 or more during daytime under high solar activity conditions.

Refraction Mechanics

The refraction of an HF wave in the ionosphere follows from Snell's law applied to a medium with a continuously varying refractive index. The refractive index n at a point in the ionosphere is approximately:

n = sqrt(1 - (f_p / f)^2)

where f is the radio frequency and f_p is the local plasma frequency. As the wave propagates upward into regions of increasing electron density, f_p increases, n decreases, and the wave bends. If the wave enters the ionosphere at an angle theta from vertical, it will be refracted back to Earth if the frequency satisfies:

f < f_c / cos(theta)

where f_c is the critical frequency (the plasma frequency at the peak of the layer). This is the secant law, and it defines the MUF for a given takeoff angle. At very low elevation angles (near grazing incidence), the MUF can be three to four times the critical frequency.

Skip Zones and Multi-Hop Propagation

For any given frequency and ionospheric condition, there is a minimum range below which the skywave signal cannot reach the ground. This is the skip zone. Signals transmitted at steep angles may penetrate through the ionosphere entirely; signals at shallower angles refract back to Earth at some minimum distance. For typical OTH-R operations in the 5 to 28 MHz band, the minimum range is usually 500 to 1,000 kilometres.

The maximum single-hop range depends on the geometry and the layer height. For F2 layer reflection at 300 km altitude, a single hop can reach roughly 3,000 to 3,500 kilometres. Beyond that, multi-hop propagation (where the signal bounces off the ground and returns to the ionosphere for a second refraction) can extend coverage further, though each hop introduces additional loss and the signal becomes progressively weaker and more distorted.

Temporal Variability

OTH-R operators must contend with ionospheric changes on multiple timescales:

Diurnal: The ionosphere changes dramatically between day and night. The D layer (which absorbs HF energy) disappears after sunset, reducing path loss but also changing the optimal operating frequency. The F2 layer electron density drops at night, lowering the MUF by 30 to 50%. An OTH-R system that operates at 20 MHz during the day may need to shift down to 8 MHz at night.

Seasonal: Summer brings higher D-layer absorption and sporadic E occurrence. Winter generally provides more stable F2 propagation at middle latitudes. Equatorial regions experience equatorial spread-F, a turbulent condition that can scatter and distort HF signals badly.

Solar cycle: The 11-year solar cycle profoundly affects the ionosphere. At solar maximum, F2 critical frequencies can exceed 12 MHz, allowing OTH-R systems to operate at higher frequencies with better resolution and less interference. At solar minimum, critical frequencies may drop below 5 MHz, severely limiting OTH-R performance and forcing operation in the most congested part of the HF band.

Geomagnetic storms: Solar flares and coronal mass ejections can cause ionospheric storms lasting hours to days. During the initial phase, enhanced solar X-ray flux increases D-layer absorption (a "short-wave fadeout"), potentially blacking out HF propagation entirely. Subsequent geomagnetic effects can distort the F layer, create irregularities, and make the ionosphere unpredictable for hours.

Real-Time Ionospheric Sounding

No OTH-R system can function without continuous, real-time knowledge of ionospheric conditions along its propagation paths. There are several techniques used in practice:

Vertical incidence sounders (ionosondes): These transmit a swept-frequency pulse straight up and measure the echo return time versus frequency, producing an ionogram that shows the critical frequencies and virtual heights of each ionospheric layer. Networks of ionosondes (such as the Global Ionospheric Radio Observatory) provide regional ionospheric data, but an OTH-R installation typically operates its own dedicated ionosondes at or near the transmitter and receiver sites.

Oblique incidence sounders: These transmit swept-frequency pulses along the actual propagation paths used by the OTH-R system. The resulting oblique ionograms directly show which frequencies will propagate to which ranges, accounting for the actual path geometry rather than relying on models to convert vertical-incidence data.

Backscatter sounding: The OTH-R system itself can be used as a sounder. By transmitting a frequency-swept pulse and analysing the backscattered returns from the ground or sea surface at various ranges, the system can map out the propagation conditions in near-real time. Strong ground clutter returns at a particular range and frequency confirm that the ionospheric path is open; absence of returns indicates the frequency is too high (signal penetrates the ionosphere) or too low (excessive absorption).

Ionospheric models: Systems like the International Reference Ionosphere (IRI) provide statistical predictions of ionospheric parameters based on location, time, season, and solar activity indices. These models are useful for planning and initial frequency selection, but they cannot capture the real-time variability needed for operational radar. They serve as a baseline that real-time sounding data corrects and refines.

Modern OTH-R systems combine all of these data sources through adaptive frequency management algorithms that continuously select the optimal operating frequency (or frequencies) and adjust beam elevation angles to maintain coverage. The JORN system, for instance, can change its operating frequency every few seconds if conditions warrant it.

Skywave OTH-R System Architecture

A skywave OTH-R installation is physically enormous. The transmitter and receiver are separated by 100 to 200 kilometres (a bistatic configuration), which is necessary to prevent the powerful transmitted signal from overwhelming the sensitive receiver. Each site covers a substantial area of land.

Transmitter

The transmitter array typically consists of a row of vertically polarised antennas (often log-periodic or wide-band monopole/dipole elements) spanning 1 to 2.5 kilometres in length. The array must operate across the full 5 to 28 MHz range used by the radar, which means the individual antenna elements must be broadband. Transmitter power ranges from 200 kilowatts to 2 megawatts, with effective isotropic radiated power (EIRP) reaching tens of megawatts when antenna gain is included. The array is phased to form a directional beam that can be electronically steered in azimuth and, to some extent, in elevation.

The JORN transmitter at Laverton, Western Australia, uses an array approximately 2.8 kilometres long with 28 transmitting elements, each driven by a 20 kW solid-state amplifier. The total radiated power is in the range of 560 kW. The French NOSTRADAMUS transmitter is reported to use a similar scale of installation.

Receiver

The receiver array is typically even larger than the transmitter. The JORN receiver at Alice Springs uses a receiving array about 3.2 kilometres in length, consisting of 480 elements arranged in a linear or slightly curved configuration. The large aperture is necessary to achieve adequate angular resolution; at HF frequencies, where wavelengths range from 10 to 100 metres, enormous arrays are needed to achieve even modest beamwidths. A 3 km aperture at 15 MHz (20 m wavelength) yields a beamwidth of roughly 0.4 degrees, which at 2,000 km range corresponds to a cross-range resolution of about 14 kilometres.

The receiver must handle an extremely challenging dynamic range problem. The desired target echoes (from aircraft with radar cross sections of perhaps 10 to 100 square metres) are extraordinarily weak after traversing 4,000 km of round-trip ionospheric propagation. Meanwhile, the receiver is illuminated by ground clutter returns that are 60 to 100 dB stronger, by atmospheric noise from worldwide lightning activity, by deliberate and unintentional HF interference from the thousands of users in the HF band, and by shortwave broadcast stations that can be 120 dB above the noise floor. The receiver front-end must handle all of this without saturating, while preserving the faint target echoes buried in the noise.

Engineering the Antenna Arrays

Building and maintaining a multi-kilometre HF antenna array is an engineering undertaking on a scale more comparable to civil infrastructure than to typical radar installations. The antenna elements themselves (log-periodic dipoles, wide-band monopoles, or folded dipole arrays) must each be 10 to 30 metres tall to resonate efficiently across the 5 to 28 MHz operating band. These structures are mounted on steel masts or lattice towers, anchored with concrete foundations and supported by guy wires, and must withstand decades of exposure to wind, heat, ice, and corrosion. In Australia, the JORN sites endure desert temperatures exceeding 45 degrees Celsius; in Russia, the Container arrays face sub-Arctic winters with heavy ice loading.

The ground plane beneath each element is critical to antenna performance. Vertically polarised monopole arrays require an extensive ground screen of buried copper radials or wire mesh extending 50 to 100 metres from each element. For a transmitter array with 28 elements stretched across nearly 3 kilometres, the total length of buried copper wire runs to hundreds of kilometres. Soil conductivity varies across the site, and ground treatment (including chemical soil enhancement with sodium chloride or bentonite slurry) is sometimes applied to improve the effective ground plane, particularly in arid environments with poor natural conductivity.

Power supply is a significant concern. A transmitter site radiating 500 kW to 2 MW requires dedicated high-voltage electrical supply, often from purpose-built substations connected to the regional power grid. Remote sites may require redundant supply lines or on-site diesel generation for backup. The receiver site draws less power, but the digital signal processing infrastructure (hundreds of analogue-to-digital converters, beamforming processors, and real-time computing systems) still demands a reliable supply of several hundred kilowatts, along with extensive cooling systems for the electronics shelters.

Maintenance of these arrays is continuous. Each antenna element and its feed network must be periodically inspected and tested. Corrosion of connections, degradation of coaxial cables and baluns, damage from lightning strikes, and mechanical wear from wind loading all require ongoing attention. The JORN programme employs dedicated field maintenance teams who travel across the array on a regular schedule. Vegetation management is also necessary; in tropical or semi-arid locations, scrub growth must be controlled to prevent interference with the ground screen and to maintain access roads along the array.

The cost of constructing and operating an OTH-R system reflects this scale. JORN's total programme cost (development, construction, and initial upgrades) has been reported at over 1.8 billion Australian dollars. The ongoing annual operating cost, including maintenance, manpower, power, and ionospheric monitoring, runs to tens of millions of dollars per year. These figures are modest compared to major platforms like fighter aircraft or warships, but they are substantial for a fixed sensor installation. The tradeoff is coverage: a single OTH-R node surveys millions of square kilometres continuously, a task that would require dozens of conventional radars or airborne platforms to replicate.

Waveform and Bandwidth

OTH-R systems typically use frequency-modulated continuous wave (FMCW) or similar waveforms rather than the short, high-peak-power pulses used by microwave radars. The available bandwidth in the HF band is severely constrained by the need to avoid interference with other users and by the coherent bandwidth of the ionospheric channel (typically 10 to 50 kHz). This limited bandwidth directly constrains range resolution. A 10 kHz bandwidth yields a range resolution of about 15 kilometres; 50 kHz gives roughly 3 kilometres. In practice, most OTH-R systems achieve range resolution cells of 10 to 40 kilometres, which is orders of magnitude coarser than microwave radar.

This coarse resolution means OTH-R is not a fire-control or targeting radar. It is a wide-area surveillance sensor that detects and tracks targets at strategic distances, providing early warning and cueing information to other, higher-resolution sensors and weapon systems.

Signal Processing: Finding Targets in an Ocean of Clutter

The signal processing challenge in OTH-R is formidable. The raw received data is dominated by clutter (ground, sea, and ionospheric returns), interference (other HF users), and noise (atmospheric and man-made). Targets are typically 40 to 80 dB below the clutter level. Extracting target detections requires exploiting the one characteristic that separates moving targets from stationary clutter: Doppler shift.

Doppler Processing

A moving target imparts a Doppler shift to the returned signal proportional to its radial velocity relative to the radar propagation path. For an aircraft moving at 250 m/s at 15 MHz, the Doppler shift is:

f_d = 2 * v * f / c = 2 * 250 * 15e6 / 3e8 = 25 Hz

This is a small shift, but it is measurable with long coherent integration times. OTH-R systems typically use coherent integration periods (CIT) of 10 to 60 seconds. A 20-second CIT provides a Doppler resolution of 0.05 Hz, which at 15 MHz corresponds to a velocity resolution of about 0.5 m/s. This fine Doppler resolution is what allows the separation of slow-moving targets from the ground/sea clutter spectrum.

The processing chain generally follows these steps:

1. Range gating: The received signal is divided into range cells (typically 15 to 40 km wide) based on time delay.

2. Beamforming: The multi-element receiver array digitally forms narrow beams in azimuth, dividing the surveillance region into angular cells.

3. Doppler filtering: Within each range-azimuth cell, a coherent FFT over the integration period separates returns by Doppler frequency. Ground clutter concentrates around zero Doppler (with some spread due to ionospheric motion). Target echoes appear at their characteristic Doppler offsets.

4. CFAR detection: Constant false alarm rate algorithms establish adaptive thresholds for each Doppler-range-azimuth cell, accounting for the varying clutter and noise environment.

5. Track formation: Detections from successive integration periods are associated and formed into tracks, using predictive filtering (typically Kalman filters or interacting multiple model estimators) to maintain continuity and refine position estimates.

Ionospheric Clutter and Contamination

One of the most challenging aspects of OTH-R signal processing is dealing with ionospheric clutter. The ionosphere is not a perfect, stationary mirror. It contains irregularities that move, plasma instabilities that grow and decay, and travelling ionospheric disturbances (TIDs), which are gravity waves propagating through the ionosphere that modulate the electron density. These phenomena create clutter that is spread in both range and Doppler, contaminating the regions of the Doppler spectrum where targets would appear.

TIDs are particularly problematic. They manifest as quasi-periodic modulations of the ionospheric reflection height, travelling horizontally at speeds of 50 to 300 m/s. The resulting Doppler contamination can mimic aircraft signatures, and sophisticated algorithms are needed to distinguish TID-induced artefacts from real targets. Techniques include multi-frequency processing (TID effects vary with frequency in a predictable way, while target Doppler does not), spatial filtering exploiting the directional nature of TIDs, and adaptive cancellation using reference beams pointed at known clutter-only regions.

Bragg Scattering from the Sea Surface

When OTH-R illuminates the ocean surface, the dominant scattering mechanism is Bragg scattering from ocean waves whose wavelength is half the radar wavelength. At 15 MHz (wavelength 20 m), the resonant ocean waves have a 10 m wavelength, which corresponds to moderate sea states. The Bragg-scattered returns produce two distinct spectral peaks at Doppler frequencies corresponding to the phase velocities of the resonant ocean waves moving toward and away from the radar.

These Bragg lines are both a challenge and an opportunity. They constitute strong clutter that must be accounted for in target detection over the ocean. But they also carry information about the sea state, surface currents, and wind direction, making OTH-R systems useful for oceanographic surveillance as well. Several nations exploit this dual capability.

Coordinate Registration and Accuracy

Converting OTH-R detections into accurate geographic coordinates is a non-trivial problem. The radar measures range (from time delay), azimuth (from beamforming), and possibly Doppler (from which radial velocity can be inferred). But the range measurement is the group path length through the ionosphere, not the straight-line ground range. The signal travels up to the ionosphere (at perhaps 250 to 350 km altitude), refracts back down, and arrives at the target at some ground range that depends on the ionospheric conditions at the time.

To convert slant range (the measured delay) into ground range, the system must know (or estimate) the ionospheric profile along the propagation path. Errors in this profile translate directly into range errors in the coordinate registration. Typical registration accuracy for a well-calibrated OTH-R system is 10 to 30 kilometres, with occasional excursions to 50 km or more during disturbed ionospheric conditions.

Several techniques improve coordinate registration:

Multi-frequency registration: By transmitting at multiple frequencies simultaneously or in rapid succession, the system can exploit the frequency-dependent refraction to triangulate the actual ionospheric conditions and correct for path errors.

Known target registration: If targets of known position (transponder-equipped cooperative aircraft, ships reporting via AIS, fixed landmarks) are within the radar's field of view, their known positions can be used to calibrate the coordinate transformation in real time.

Ionospheric tomography: Using data from the radar's own backscatter sounding, ionosondes, and GPS total electron content measurements, a three-dimensional model of the ionosphere can be constructed and used for ray tracing to convert measured delays into ground coordinates.

Track-level registration: Over time, as a target track develops, the consistency of the track can be used to refine the coordinate registration. An aircraft following a known airway, for instance, provides continuous calibration data.

Despite these techniques, OTH-R will never match the positional accuracy of microwave radar. This is acceptable given its role: strategic early warning and wide-area surveillance, where knowing that an aircraft formation is approaching at 2,000 km range with an uncertainty of 20 km is vastly more useful than having no detection at all.

Operational OTH-R Systems

JORN (Australia)

The Jindalee Operational Radar Network is perhaps the most capable and best-documented OTH-R system in the Western world. Developed by the Australian Defence Science and Technology Organisation (DSTO, now DST Group) and built by RLM Systems (later Lockheed Martin Australia, now Leidos Australia), JORN became fully operational in 2003 and has been continuously upgraded since.

JORN consists of three radar nodes: Radar 1 at Longreach, Queensland; Radar 2 at Laverton, Western Australia; and Radar 3 at Alice Springs, Northern Territory (this is the receiver site; the corresponding transmitter is at Harts Range, about 180 km away). A coordination centre at RAAF Base Edinburgh near Adelaide integrates data from all three nodes.

Each radar covers a sector of approximately 90 degrees in azimuth with a range coverage of roughly 1,000 to 3,000 kilometres. Together, the three radars provide overlapping coverage across the entire northern approaches to Australia, from the Indian Ocean through Southeast Asia to the western Pacific. The system can detect aircraft with radar cross sections down to approximately 2 to 5 square metres at ranges exceeding 2,000 kilometres.

JORN has been described by RAAF officials as the most important single intelligence, surveillance, and reconnaissance asset in Australia's defence inventory. It provides continuous air and maritime surveillance across an enormous area, detecting and tracking not only aircraft but also ships, weather fronts, and certain types of ionospheric disturbances.

A major upgrade programme (JORN Phase 6) has been underway, with BAE Systems Australia contracted to modernise the system's signal processing, display systems, and frequency management algorithms.

NOSTRADAMUS (France)

France's OTH-R system, NOSTRADAMUS (Nouveau Systeme Trans-horizon pour la Detection, l'Analyse et la Mesure des Unites Speciales), was developed by ONERA (the French national aerospace research centre) and Thales. The system became operational with the French Air Force (Armee de l'Air et de l'Espace) in the early 2010s.

NOSTRADAMUS is located in southern France, with its transmitter and receiver sites separated by roughly 100 kilometres. The system provides coverage over the Mediterranean Sea, parts of North Africa, and the eastern Atlantic, monitoring air traffic and maritime activity across an area of strategic importance to France and NATO.

The system uses a bistatic configuration with a linear transmitter array and a separate, larger receiver array. The receiver reportedly incorporates several hundred individual antenna elements spanning over a kilometre, providing azimuthal beamwidths narrow enough to separate closely spaced targets at ranges beyond 1,500 kilometres. NOSTRADAMUS operates across the 5 to 28 MHz band and employs sophisticated adaptive frequency management that accounts for the particular ionospheric conditions of the mid-latitude European sector, including seasonal sporadic-E activity that is especially prevalent over the Mediterranean basin during summer months.

One of the system's notable technical features is its advanced clutter mitigation architecture. The Mediterranean environment presents challenges that differ from the open-ocean scenarios faced by JORN or ROTHR. Dense commercial air traffic across southern European airspace, heavy maritime shipping in confined sea lanes, and significant HF spectrum congestion from European broadcasters and communication services all complicate target detection. NOSTRADAMUS addresses this through multi-frequency simultaneous operation, transmitting and receiving on several HF channels in parallel to exploit frequency diversity for both clutter rejection and ionospheric path redundancy. If propagation conditions degrade on one frequency, other channels may still provide usable returns.

French authorities have described the system's detection capability as extending to about 2,000 to 3,000 kilometres, with the ability to detect and track both aircraft and ships. The system feeds into the French military's recognised air picture and contributes to NATO's integrated air defence architecture for the southern flank.

Container / 29B6 (Russia)

Russia has long invested heavily in OTH-R technology. The most modern system is the 29B6 "Container," which entered service with the Russian Aerospace Forces in 2019. The first operational system is located near Kovylkino in Mordovia, with its receiver site near Gorodets in Nizhny Novgorod Oblast, approximately 300 km from the transmitter.

The Container system is reported to cover a sector of roughly 180 degrees in azimuth with a range of up to 3,000 kilometres, providing coverage over much of Europe and the North Atlantic. Russian defence officials have stated that the system can detect and track aircraft, including cruise missiles, and provide early warning of massed air attacks. The transmitter array reportedly spans about 440 metres, and the receiver array is approximately 1,300 metres long.

Russia reportedly plans to deploy multiple Container systems around its borders to provide complete 360-degree OTH-R coverage, a concept of operations similar in ambition to JORN's multi-node approach.

Duga (Soviet Union, Historical)

No discussion of OTH-R is complete without mentioning the Duga system, one of the most extraordinary (and infamous) radar systems ever built. The Duga was a Soviet OTH-R system designed primarily for early warning of intercontinental ballistic missile launches from the United States.

The better-known installation, often called the "Chernobyl-2" system, was located near Chernobyl in Ukraine and included a receiver antenna array that still stands today: a massive steel structure approximately 150 metres tall and 500 metres wide. The corresponding transmitter was located near Liubech, about 60 km away.

The Duga system operated in the 5 to 28 MHz range and emitted a distinctive repetitive tapping signal at about 10 Hz that was audible on shortwave receivers worldwide, earning it the nickname "Russian Woodpecker." The signal caused significant interference to HF communications and broadcasting across the globe, and its identification and characterisation became a minor cause of international controversy during the late 1970s and 1980s.

The Chernobyl-2 Duga site was abandoned following the Chernobyl nuclear disaster in 1986, as it fell within the exclusion zone. The remaining Duga antennas near Chernobyl have become a well-known industrial ruin and a draw for tourists visiting the exclusion zone.

ROTHR (United States)

The AN/TPS-71 Relocatable Over-the-Horizon Radar (ROTHR) was developed by Raytheon for the US Navy, primarily for counter-narcotics maritime surveillance. Three ROTHR systems were built, with installations in Virginia (covering the Caribbean), Texas (covering the eastern Pacific approaches to Central America), and Puerto Rico.

ROTHR operates in the 5 to 28 MHz band with a transmitter-receiver separation of about 160 km. Each system covers a 64-degree azimuth sector at ranges of 500 to 3,000 km. The system is optimised for detecting and tracking surface vessels and low-flying aircraft. The transmitter antenna array is approximately 2.4 km long, and the receiver uses a similar-scale array.

Surface Wave OTH Radar

Not all OTH radars rely on ionospheric skywave propagation. An alternative approach exploits the ground wave (more precisely, the surface wave), where the radio signal follows the curvature of the Earth's surface rather than refracting through the ionosphere. Surface wave propagation works best over seawater, which has high conductivity, and is effective at the lower end of the HF band and into the upper MF band (2 to 15 MHz).

Surface wave OTH radars (SW-OTH) have shorter range than skywave systems, typically 200 to 400 kilometres, but they offer several important advantages:

No skip zone: A skywave OTH-R cannot see targets closer than about 500 to 1,000 km. A surface wave system provides continuous coverage from the coastline outward.

Better accuracy: Since the signal follows the Earth's surface rather than refracting through the variable ionosphere, range measurements correspond more directly to ground range. Position accuracy of 1 to 5 km is typical.

Smaller installation: The antenna arrays, while still large by microwave radar standards, are much smaller than skywave systems. A typical SW-OTH antenna might span 200 to 400 metres rather than 2 to 3 kilometres.

Simpler operation: There is no need for real-time ionospheric sounding or adaptive frequency management driven by ionospheric conditions.

The primary applications of SW-OTH are exclusive economic zone (EEZ) surveillance, coastal defence, and maritime domain awareness. Several operational systems exist:

Thales Coastwatcher 100: A widely exported SW-OTH system operating in the HF band, designed for coastal surveillance out to roughly 200 nautical miles (370 km). It can detect and track ships, aircraft, and even icebergs, and has been sold to multiple nations for EEZ monitoring.

Leonardo TPS-740: An Italian-developed SW-OTH radar for coastal and maritime surveillance, designed to detect surface vessels and low-flying aircraft at ranges up to 200 km.

Podsolnukh / Sunflower (Russia): This is a Russian SW-OTH system designed for coastal surveillance. The Sunflower-E variant is deployed at multiple locations along the Russian coastline and can detect surface targets at ranges up to 450 km and air targets up to 500 km, according to Russian specifications. The system reportedly uses a transmit array of about 100 metres aperture.

WERA (Germany): Developed by Helzel Messtechnik, WERA is a ground-wave radar system primarily designed for oceanographic applications (measuring surface currents, wave heights, and wind direction using Bragg scattering), but it has been adapted for vessel detection and tracking as well. Multiple WERA installations operate along European coastlines for both scientific and security purposes.

Maritime Surveillance and Oceanography

OTH radar, both skywave and surface wave, has proven particularly valuable for maritime surveillance. Ships present relatively large radar cross sections (hundreds to thousands of square metres for cargo vessels and warships), move slowly (10 to 30 knots, producing Doppler shifts of a few Hz at HF frequencies), and operate in an ocean environment where Bragg scattering from sea waves provides useful oceanographic data as a secondary product.

The Bragg scattering phenomenon is especially relevant. When an HF radar illuminates the sea surface, the dominant backscatter comes from ocean waves with a wavelength equal to half the radar wavelength (this is the first-order Bragg condition). At 15 MHz (radar wavelength 20 m), the resonant ocean waves are 10 m long, which falls in the ocean wave spectral peak for typical sea states. The backscattered power is proportional to the spectral density of the ocean waves at the Bragg wavelength, and the Doppler shift of the Bragg return is determined by the phase velocity of these waves (modified by surface currents).

The result is a Doppler spectrum from each ocean range-azimuth cell that contains:

1. Two strong Bragg peaks at positive and negative Doppler, corresponding to ocean waves travelling toward and away from the radar.
2. A broader second-order continuum around the Bragg peaks, arising from wave-wave interactions and containing information about the full directional ocean wave spectrum.
3. Any ship echoes, which appear as discrete Doppler lines usually between the two Bragg peaks (since most ships move slower than the phase velocity of 10 m ocean waves).

By analysing the Bragg peaks and the second-order spectrum, OTH radar can extract measurements of significant wave height, dominant wave period and direction, surface current velocity, and wind direction. This capability has led to the deployment of dedicated HF surface-wave radars for oceanographic and coastal monitoring, separate from military surveillance applications.

Ballistic Missile Detection

The original motivation for OTH-R development during the Cold War was the detection of ballistic missile launches. This is perhaps the most challenging application of OTH-R technology, for several reasons:

Speed: An ICBM in its boost phase accelerates rapidly, reaching velocities of several km/s within minutes. The associated Doppler shifts are large and distinctive, but the target is only in the boost phase (where the rocket exhaust creates a large radar and infrared signature) for a few minutes.

Radar cross section: The missile body itself has a relatively small RCS (a few square metres), but the rocket exhaust plume creates a much larger effective target due to ionisation of the exhaust gases. This ionised plume can have an effective RCS of hundreds or thousands of square metres at HF frequencies, making it detectable at extreme range.

Boost phase signatures: During the boost phase, the rocket motor produces an exhaust plume containing aluminium oxide particles, ionised gases, and superheated water vapour. At HF frequencies, the ionised plume column can extend several kilometres behind the missile, creating a radar target with an effective cross section vastly exceeding the missile airframe alone. The plume's radar signature is also spectrally distinctive: the rapid acceleration of the missile (from zero to several km/s within 60 to 180 seconds, depending on the missile type) produces a characteristic Doppler chirp that sweeps through a wide frequency range. This chirp signature is difficult to mimic with conventional aircraft or atmospheric phenomena, making it a useful discriminant for reducing false alarm rates.

Ionospheric disturbance: A large rocket launch creates localised ionospheric disturbances as the exhaust plume rises through the ionosphere. The chemical species in the exhaust (particularly water vapour and carbon dioxide) react with the ambient ionospheric plasma, causing localised electron depletion known as an "ionospheric hole." These depletion zones can extend hundreds of kilometres laterally and persist for tens of minutes to hours. OTH-R systems can detect these disturbances as transient changes in the propagation characteristics along specific paths: a backscatter sounding pulse that previously returned a strong ground echo from a certain range may suddenly show a weakened or absent return, indicating that the ionospheric refraction path has been disrupted. This provides an additional detection mechanism independent of the direct radar return from the missile. Soviet-era Duga systems were specifically designed to exploit this phenomenon, monitoring for sudden changes in HF propagation over known ICBM launch areas in the central United States.

Timeliness: For ballistic missile warning, detection must be rapid (within minutes of launch) and reliable. The consequences of both missed detections and false alarms are severe. The inherent variability of ionospheric propagation makes guaranteeing reliable detection performance extremely difficult.

Modern OTH-R systems retain this capability as one of several missions, though satellite-based infrared early warning systems (such as the US SBIRS constellation and the Russian EKS Tundra satellites) have largely taken over the primary missile warning role due to their global coverage and all-weather, all-ionosphere reliability. That said, OTH-R retains value as a complementary and redundant layer in missile warning architectures. Unlike satellites, OTH-R systems are ground-based and difficult to destroy with anti-satellite weapons.

Limitations and Vulnerabilities

OTH-R is a powerful capability, but it operates within significant constraints:

Ionospheric dependence (skywave): The fundamental limitation. The ionosphere is a natural phenomenon beyond human control, and its variability directly impacts OTH-R performance. During severe geomagnetic storms, skywave OTH-R may be completely unavailable for hours. Even under normal conditions, performance varies significantly between day and night, between solar maximum and solar minimum, and between different geographic regions.

Coarse resolution: Range resolution of 10 to 40 km and angular resolution of 0.5 to 2 degrees means that individual targets cannot be identified or classified in detail. OTH-R provides detection and tracking, not identification.

Coordinate uncertainty: The 10 to 30 km position accuracy (sometimes worse) limits the tactical utility of OTH-R detections. Targets detected by OTH-R must be re-acquired and identified by higher-resolution sensors before tactical decisions can be made.

Enormous physical footprint: The antenna arrays span kilometres and cannot be concealed or relocated quickly. They are vulnerable to physical attack (though their inland locations, typically hundreds of kilometres from the coast, provide some protection) and their locations are well known.

HF spectrum congestion: The HF band is shared with many other users: shortwave broadcasters, maritime and aviation communications, amateur radio, military communications networks, and various data services. OTH-R must operate around these other signals, and interference management is a continuous operational challenge. A deliberate jamming campaign in the HF band, using multiple transmitters across a range of frequencies, could significantly degrade OTH-R performance over a wide area.

High-altitude electromagnetic pulse (HEMP): A nuclear detonation at high altitude (typically 30 to 400 km) generates an electromagnetic pulse through three distinct mechanisms, designated E1, E2, and E3. The E1 component is a brief, intense pulse (nanosecond rise time, lasting microseconds) caused by Compton scattering of gamma rays off atmospheric molecules, producing high-energy electrons that radiate broadband electromagnetic energy. The E1 pulse can damage unshielded electronics at the receiver site, including low-noise amplifiers and analogue-to-digital converters, unless adequate electromagnetic shielding (Faraday cages, surge arrestors on all antenna feeds, fibre-optic signal isolation) is incorporated into the system design. The E3 component, a slower geomagnetically induced current lasting seconds to minutes, can disrupt the power grid supplying the radar, potentially causing transformer damage and extended power outages.

Beyond direct equipment damage, the nuclear detonation produces massive ionisation of the lower atmosphere. Beta particles (electrons) released by the weapon's fission products are trapped along geomagnetic field lines and deposited into the D-region of the ionosphere, dramatically increasing electron density and hence HF absorption. This enhanced D-layer absorption can black out HF propagation over continental-scale areas for minutes to hours, depending on weapon yield and burst altitude. A single high-yield detonation at 300 km altitude could suppress OTH-R capability across an entire hemisphere. This is a particular concern because the HEMP scenario (nuclear conflict) is exactly the scenario in which OTH-R's missile warning capability would be most needed. Hardening OTH-R installations against HEMP effects is technically feasible but adds significant cost and complexity to already expensive systems.

Processing complexity: The signal processing chain in an OTH-R system is among the most demanding of any operational radar. Each coherent integration period produces a three-dimensional data cube (range, azimuth, Doppler) that may contain thousands of range cells, hundreds of beam positions, and thousands of Doppler bins. Processing this cube in real time requires sustained computational throughput in the range of hundreds of billions of floating-point operations per second. The algorithms themselves must be adaptive: clutter statistics change with ionospheric conditions, interference patterns shift as HF band usage varies, and the coordinate registration must be updated continuously as the ionosphere evolves. False alarm management is particularly challenging because ionospheric irregularities, meteor trails, sporadic-E patches, and travelling ionospheric disturbances can all produce transient radar returns that resemble aircraft targets. Discriminating these from real targets requires multi-dimensional statistical analysis and, increasingly, machine-learning classifiers trained on large historical datasets.

Future Directions

OTH-R technology continues to evolve along several lines:

Digital beamforming and MIMO: Modern signal processing allows fully digital beamforming at the receiver, forming multiple simultaneous beams across the entire surveillance sector. Earlier OTH-R systems used analogue beamforming networks that could only form a limited number of beams at a time, requiring sequential scanning across the surveillance sector. Fully digital receivers (where each antenna element has its own analogue-to-digital converter) allow the entire sector to be covered simultaneously, eliminating revisit time penalties and enabling adaptive null steering to suppress directional interference sources. Multiple-input multiple-output (MIMO) techniques take this further: by having each transmitter element radiate an orthogonal waveform (using different frequency offsets, codes, or time slots), the system creates a virtual aperture much larger than the physical array. This promises improved angular resolution and the ability to resolve multiple targets within the same range cell, a significant limitation of current systems.

Cognitive and adaptive radar: Machine learning techniques are being applied to several aspects of OTH-R operation. For frequency management, neural networks trained on years of ionospheric observations can predict optimal operating frequencies 15 to 60 minutes into the future, allowing proactive rather than reactive frequency changes. For clutter classification, convolutional neural networks and recurrent architectures can learn to distinguish TID-induced artefacts from real targets based on their spatiotemporal evolution across successive coherent integration periods. For track management, probabilistic data association algorithms enhanced with learned models of target behaviour improve the system's ability to maintain tracks through periods of poor propagation, when detections may be intermittent or displaced by ionospheric distortion.

Multi-static configurations: Using multiple transmitters and receivers in a networked configuration can improve coverage, accuracy, and resilience. If one propagation path is disrupted by ionospheric disturbance, others may still be viable.

Integration with space-based sensors: Combining OTH-R detections with satellite-based surveillance (SAR imagery, AIS data, ELINT) creates a more complete maritime and air surveillance picture than either sensor type can provide alone. OTH-R provides continuous, wide-area detection of moving targets across millions of square kilometres, while satellite SAR provides high-resolution imagery of specific areas on a revisit schedule. Fusing these data sources allows OTH-R tracks to cue satellite tasking (directing a SAR satellite to image a suspicious contact detected by the radar) and satellite data to refine OTH-R coordinate registration (using known ship positions from satellite AIS or imagery to calibrate the ionospheric propagation model). Several nations are actively developing these fusion architectures, and JORN's Phase 6 upgrade includes improved interfaces for multi-source data integration.

Compact surface-wave systems: Advances in signal processing and antenna design are enabling smaller, lower-cost surface-wave OTH radars suitable for deployment by nations with limited defence budgets. Modern compact SW-OTH systems can achieve useful detection ranges (150 to 300 km for ships, 100 to 200 km for aircraft) with antenna arrays as short as 50 to 100 metres, using advanced digital beamforming and CFAR processing to compensate for the reduced aperture. Container-based systems that can be transported by truck and erected in days are under development, offering rapid deployability for expeditionary operations or disaster response. These systems sacrifice some range and angular resolution compared to large fixed installations but provide useful EEZ surveillance at a fraction of the cost, making wide-area maritime domain awareness accessible to smaller coastal nations.

OTH radar remains one of the few sensor technologies capable of providing continuous, wide-area surveillance across strategic distances. Its physics are dictated by the ionosphere and the geometry of the Earth, and those have not changed. What has changed is our ability to process signals, model the ionosphere, and integrate OTH-R data with other sensor networks. The result is that OTH-R, conceived during the Cold War as a single-purpose missile warning system, has become a versatile multi-mission surveillance capability that will remain operationally relevant for decades to come.