HF ionospheric propagation for SIGINT: modeling, prediction and collection planning

The HF band — 3 to 30 MHz — is the only portion of the radio spectrum where signals can travel thousands of kilometers using nothing but the ionosphere as a reflector. That capability makes HF indispensable for long-range military communications, over-the-horizon radar, and maritime messaging. It also makes the ionosphere the most consequential variable in HF SIGINT collection planning: a sensor that is perfectly positioned for the target area at noon may be blind to it by midnight, not because the target has moved, but because the ionosphere has. Understanding ionospheric propagation is not an academic exercise for SIGINT analysts — it determines which sensors can hear which targets, on which frequencies, during which hours. This article explains the propagation physics, the modeling tools that quantify it, and how collection planners integrate model output into operationally useful frequency recommendations and sensor tasking decisions.

HF propagation fundamentals for SIGINT analysts

The ionosphere is a region of the upper atmosphere, extending from roughly 60 km to 1,000 km altitude, where solar ultraviolet and X-ray radiation ionizes gas molecules and creates a plasma of free electrons. The electron density profile has distinct layered structure, each layer reflecting different aspects of the ionization balance between solar input and electron-ion recombination.

The D layer (60–90 km) exists only during daylight hours and is responsible for absorption of lower-frequency HF signals. It does not reflect signals — it attenuates them. Frequencies below about 7 MHz at mid-latitudes are heavily absorbed by the D layer during daylight, which effectively sets the lowest usable frequency (LUF) for long-range collection during daytime hours. The D layer disappears at night, making those lower frequencies viable again.

The E layer (90–150 km) reflects signals at lower HF frequencies and also hosts the phenomenon of sporadic-E (Es) — intense, localized ionization patches that can briefly reflect signals at frequencies well above the normal F-layer MUF. Sporadic-E is unpredictable and can create unexpected short-range propagation that both enables and complicates SIGINT collection.

The F layer (150–400 km) is the primary reflection layer for long-range HF propagation. During daylight it splits into F1 (150–200 km) and F2 (200–400 km) sub-layers. At night, F1 merges back into F2. The F2 layer peak electron density — parameterized as the critical frequency foF2 — is the most important single ionospheric parameter for HF SIGINT planning. It represents the highest frequency at which a vertically incident signal is reflected rather than transmitted through the layer.

The maximum usable frequency (MUF) for an oblique path is related to foF2 by the geometry of the propagation path. For a single-hop F2 path, the MUF is approximately:

MUF(path) = foF2 / sin(elevation_angle)
          = foF2 × MUF_factor

Where MUF_factor for typical single-hop F2 (range 1,000–2,000 km):
  - 1,000 km path:  MUF_factor ≈ 3.0–3.5
  - 2,000 km path:  MUF_factor ≈ 4.0–5.0
  - 3,000 km path:  MUF_factor ≈ 5.0–6.0 (approaching 2-hop geometry)

Example: foF2 = 8 MHz, 1,500 km path → MUF ≈ 28 MHz
         foF2 = 5 MHz, 1,500 km path → MUF ≈ 18 MHz

The lowest usable frequency (LUF) is set by the combined absorption of the D and E layers, which increases steeply as frequency decreases. The LUF is typically 2–4 MHz higher than the groundwave range limit during daytime, and may fall below 3 MHz at night when absorption vanishes. Collection planners operating below the LUF will find signals too weak against the galactic and atmospheric noise background to be useful, regardless of whether the sky-wave path geometry is otherwise favorable.

Skip zone and dead zone

The skip zone — also called the dead zone or silent zone — is the annular area on the earth's surface between the outer limit of ground-wave propagation and the near edge of the first sky-wave return. Inside this zone, no signal energy from the target reaches the sensor: ground wave has attenuated below the noise floor and sky wave has not yet completed its first hop. For SIGINT collection planning this zone is a fundamental constraint that must be mapped for every sensor-frequency combination.

First-hop skip distance calculation. The minimum skip distance for a single-hop F2 path is determined by the virtual height of the F2 layer (h'F) and the critical frequency foF2. The approximate formula for the minimum skip distance D_skip is:

D_skip ≈ 2 × sqrt(h'F² + (c / (2π × foF2))²) × (MUF_factor at limit)

Simplified operational approximation:
  D_skip(km) ≈ 2 × h'F × tan(max_elevation_angle)

Typical values (h'F = 300 km, foF2 = 7 MHz):
  At 7 MHz  (near MUF for vertical):   D_skip ≈ 150 km  (very short)
  At 14 MHz (MUF_factor ~2):           D_skip ≈ 500 km
  At 21 MHz (MUF_factor ~3):           D_skip ≈ 900 km
  At 28 MHz (MUF_factor ~4):           D_skip ≈ 1,400 km

The key operational implication is that higher frequencies produce larger skip zones. A sensor listening at 28 MHz has a dead zone extending to over 1,400 km around it — any target within that ring is invisible to that sensor on that frequency. Conversely, lower frequencies near the LUF have very short skip distances, making them effective for collection on nearby targets.

Coverage gap between ground wave and sky wave. The ground-wave range limit depends strongly on frequency and ground conductivity. Over typical land conductivity, groundwave at 5 MHz reaches perhaps 200–300 km; at 15 MHz it barely reaches 50 km. The gap between groundwave range and first-hop skip distance defines the dead zone width. At 15 MHz with a 50 km groundwave range and a 700 km skip distance, the dead zone spans 650 km in radius — a circle nearly 1,300 km across within which no collection is possible from a sensor on that frequency.

Multi-hop fill-in strategies. Two-hop and three-hop propagation create additional coverage zones at multiples of the single-hop skip distance. A target inside the first-hop dead zone of a nearby sensor may fall within the illuminated zone of a more distant sensor whose first-hop lands on the target. The practical approach for closing dead-zone gaps is multi-sensor geometry: position additional sensors so that each sensor's illuminated sky-wave zones cover the dead zones of others. Propagation model coverage maps, generated for each candidate sensor location and frequency, make these complementary positions visible and allow the collection planner to verify that the combined multi-sensor coverage leaves no persistent gaps over the target area.

Propagation models: VOACAP, IRI, ITU-R P.533

Three propagation modeling systems dominate operational HF SIGINT planning. Each serves a different function, and collection planning software typically integrates all three rather than relying on any single model alone.

VOACAP (Voice of America Coverage Analysis Program) is the most widely used operational HF circuit prediction tool. It predicts signal strength, signal-to-noise ratio, required power, and monthly circuit reliability for a specific transmitter-receiver path, date, time, frequency, and antenna configuration. VOACAP operates on monthly-median ionospheric statistics derived from the URSI/CCIR global ionospheric coefficient databases, adjusted by the solar flux index (F10.7). Its output is probabilistic: a reliability figure representing the percentage of days in the specified month when the path is predicted to be usable. For collection planning, VOACAP reliability figures above 75% indicate consistently usable paths; figures below 25% indicate paths that should not be depended upon for primary collection.

VOACAP requires the following primary inputs:

• Transmitter and receiver geographic coordinates (decimal degrees)
• Month and year (for seasonal ionospheric statistics)
• UTC hour range for the prediction window
• Frequency or frequency sweep range
• Solar flux index (F10.7) — use observed 81-day smoothed value
• Transmit and receive antenna models (gain pattern vs. elevation angle)
• Required SNR threshold for the signal type being collected

IRI (International Reference Ionosphere) is the international standard empirical model of the ionosphere, maintained by COSPAR and URSI. Unlike VOACAP, which predicts circuit performance, IRI models the ionospheric state itself — electron density profiles, ion composition, foF2, virtual height, TEC — as a function of geographic location, altitude, date, and time. IRI-2020 is the current recommended version. Collection planning systems use IRI to generate the ionospheric parameter fields that feed into both VOACAP-style circuit predictions and real-time geolocation correction engines. IRI is more physically complete than VOACAP's internal ionospheric representation, particularly for electron density profile shape, but it shares the limitation of being a climatological model that does not account for real-time disturbances.

ITU-R P.533 is the International Telecommunication Union's formal recommendation for HF propagation prediction, implemented in the freely available ITURHFPROP software. It incorporates physical propagation mode analysis — separating E-layer modes, F1-layer modes, and F2-layer modes by path geometry — and applies separate signal strength and noise predictions for each mode before combining them. For SIGINT applications, the multi-mode output of P.533 is particularly valuable because it allows the collection system to determine which propagation mode dominates on a given path and hour, which in turn allows the geolocation engine to apply mode-appropriate ionospheric corrections.

A practical accuracy comparison for mid-latitude paths under quiet geomagnetic conditions:

Diurnal and seasonal propagation effects

Ionospheric propagation is never static. The electron density that governs foF2, MUF, skip distance, and absorption varies continuously with solar illumination, season, geographic latitude, and the 11-year solar cycle. SIGINT collection planning that ignores this variability produces frequency schedules that work at some times and fail entirely at others.

Day versus night F-layer behavior. During daylight hours, solar UV drives foF2 to its daily peak, typically in the early afternoon local time. MUF for mid-range paths peaks at 15–30 MHz. The D layer is fully ionized and absorbs lower-frequency signals. As local sunset approaches, the D layer begins to dissipate, absorption falls rapidly, and foF2 starts its nocturnal decline — but much more slowly than D-layer dissipation. The period from local sunset to approximately 2–3 hours after local midnight is characterized by declining MUF (falling from its daytime peak toward a nighttime value perhaps 30–50% lower) but also by vanishing D-layer absorption. The net effect is that lower-frequency bands (4–8 MHz) that were blocked during daylight become usable for long-range collection in the early evening hours, often with excellent signal-to-noise ratios.

Seasonal MUF variation. foF2 varies with season differently in summer and winter hemispheres, and the effect is complicated by the so-called winter anomaly: at many mid-latitude locations, foF2 is actually higher in winter daytime than in summer daytime, despite the lower solar elevation angle. This counter-intuitive effect is due to reduced chemical loss rates in the winter F layer. The practical result is that winter midday collection windows at higher frequencies (20–28 MHz) are often viable when summer collection at the same time would be constrained by a lower MUF ceiling. Collection plans that are finalized in summer and applied in winter (or vice versa) without re-running the propagation model for the new season will have systematic frequency errors.

Equinox enhancement effects. Around the March and September equinoxes, foF2 peaks at higher values than typical summer or winter solstice conditions at many locations. This equinox enhancement — thought to arise from reduced recombination rates driven by atmospheric circulation changes — extends the usable frequency range to higher frequencies and improves collection reliability on paths that are marginal during solstice periods. Equinox periods are often the best time of year for very long-range HF collection (3,000 km or more), when the high MUF supports multi-hop paths that are otherwise too unreliable to depend upon.

The gray line. The terminator — the boundary between daylit and dark hemispheres as the Earth rotates — creates a brief propagation enhancement as it passes over a sensor or target. Along the gray line, the F layer remains ionized from preceding daylight while the D layer has dissipated on the night side. This creates enhanced propagation on lower-frequency HF paths (3–8 MHz) that are otherwise blocked by daytime D-layer absorption. Collection planners must model gray line timing across the target area and pre-schedule sensor frequency changes to exploit these enhancement windows, which typically last 20–40 minutes for any given path.

Optimal frequency selection for SIGINT collection

Selecting the right frequency is the most operationally consequential decision in HF collection planning. The wrong frequency means no intercept; the right frequency — precisely timed — means reliable coverage throughout the collection window. Propagation models support frequency selection through the calculation of three key parameters for each target-sensor path and each hour of the day: the MUF (ceiling), the LUF (floor), and the optimum working frequency (OWF).

The OWF is conventionally defined as 0.85 × MUF, providing a 15% margin below the MUF ceiling. This margin accounts for short-term ionospheric variability not captured by the monthly-median model: foF2 can vary by ±10–15% from its median value on any given day, and operating at the MUF means a 50% probability of being above it at any given moment. Operating at OWF reduces that risk to roughly 15–20%, acceptable for primary collection. For high-priority targets where gaps are unacceptable, planners should designate both an OWF primary frequency and a fallback frequency at 0.65–0.70 × MUF.

Automated frequency recommendation from propagation model output works as follows:

For each hour h in [00:00 UTC .. 23:00 UTC]:
  1. Run VOACAP for path(sensor_lat, sensor_lon → target_lat, target_lon)
     at month M, F10.7=SSN, hour h
  2. Extract:
       MUF(h)   — maximum usable frequency
       LUF(h)   — lowest usable frequency
       OWF(h)   = 0.85 × MUF(h)
  3. If OWF(h) < LUF(h): → mark hour h as "BLACKOUT" (no usable path)
  4. Else: → recommend collection_freq = OWF(h)
            primary_band   = floor(OWF(h) / 1e6) MHz band
            fallback_band  = floor(0.68 × MUF(h) / 1e6) MHz band
  5. Quantize to available receiver tuning grid
     (typically 5 kHz steps in military HF receivers)

Output: 24-element frequency schedule + blackout hour flags

Across the 2–30 MHz band, the frequency management challenge is that an HF sensor cannot simultaneously cover all frequencies. A wideband recording receiver with sufficient bandwidth can capture a contiguous segment — a typical architecture records 500 kHz to 3 MHz instantaneous bandwidth — but does not span the full HF band at once. For a specific target-sensor path, the propagation model narrows the search: at any given hour, only a 2–4 MHz window around the OWF is worth monitoring, because signals outside that window are either above the MUF (not arriving) or below the LUF (too weak). Collection managers use the propagation-derived frequency schedule to pre-program receiver center frequencies and ensure that the right band is monitored during each window.

Sensor placement optimization using propagation coverage

Propagation models are most powerful when used not just to characterize existing sensor coverage but to drive sensor placement decisions for new sites. Coverage map generation — running the propagation model for a grid of potential target locations and plotting the results geographically — reveals exactly which areas are well covered, which are marginal, and which are inside persistent dead zones from any candidate sensor position.

Coverage map generation. To generate a coverage map for a candidate sensor site at a given frequency and hour, the collection planner runs the propagation model in reverse: the sensor is treated as a receiver, and the model is run for a regular geographic grid of potential transmitter (target) locations, producing a predicted received power or reliability value at each grid point. Grid spacing of 50–100 km is adequate for strategic collection planning; 10–25 km may be needed for tactical applications. The resulting map shows contour lines of constant collection probability across the geographic area of interest.

Gap analysis for target areas. Overlaying coverage maps from multiple sensors and frequencies reveals composite coverage across the collection system. A gap analysis identifies regions where no sensor achieves the minimum reliability threshold (typically 50% or 75% for primary collection requirements) at any available frequency during any hour of the collection window. These gaps are either addressed by adding sensors, repositioning existing sensors, or accepting reduced coverage as a known limitation. Gap analysis output feeds directly into the HF direction finding network architecture planning process, where sensor geometry also drives geolocation accuracy.

Multi-site collection geometry. The interaction between propagation coverage and geolocation geometry is critical for network design. A sensor placement that provides excellent single-sensor collection coverage may be geometrically unfavorable for multi-station bearing intersection — for example, if all available sensors happen to be in a roughly collinear arrangement relative to the target area, the geolocation accuracy for targets on the line extended through the network will be poor regardless of signal quality. The combined optimization — maximize coverage probability while minimizing dilution of precision across the target area — requires running both a propagation model and a geolocation geometry simulator simultaneously for candidate site configurations. The result is a Pareto-optimal set of site configurations that trade off coverage completeness against geolocation accuracy.

Geomagnetic disturbance and propagation disruption

The ionospheric propagation described above assumes quiet geomagnetic conditions. Solar and geomagnetic disturbances can degrade, disrupt, or completely eliminate HF propagation, and the impact on SIGINT collection can range from inconvenient to operationally decisive. Collection planners must integrate space-weather monitoring into the collection management workflow, not treat it as an occasional concern.

K-index monitoring and alert thresholds. The planetary geomagnetic Kp index (0–9 scale, updated every 3 hours by GFZ Potsdam) is the standard operational indicator of geomagnetic disturbance. Collection planning systems should monitor Kp continuously and apply tiered response protocols:

• Kp 0–3 (quiet): Normal propagation. Propagation model predictions apply without correction. Standard frequency schedules remain valid.
• Kp 4 (unsettled): Minor ionospheric irregularity. Reduce OWF by 10%. Widen skip zone estimates by 10–15%. Flag high-latitude paths as potentially unreliable.
• Kp 5 (minor storm): Noticeable MUF depression. Reduce collection frequencies toward the lower end of the modeled LUF-MUF window. Activate fallback frequencies. Alert collection manager to verify sensor performance on all long-range paths.
• Kp 6–7 (moderate storm): Significant degradation at high latitudes; moderate impact at mid-latitudes. MUF may drop 20–40%. Long-range (2,000+ km) paths likely unreliable. Shift collection priority to closer targets and shorter paths. Activate NVIS HF radio intelligence mode for targets within 500 km.
• Kp 8–9 (severe/extreme storm): Near-total disruption at high latitudes. HF blackout possible at all latitudes for extended periods. Activate alternate collection modes (satellite intercept, VHF/UHF if in range, NVIS at very short range). Document collection gap in record.

Blackout prediction using space-weather alerts. NOAA Space Weather Prediction Center (SWPC) issues geomagnetic storm watches (24–72 hours in advance), warnings (within 24 hours), and alerts (storm has commenced). Collection management systems should ingest SWPC alerts via NOAA's machine-readable product feeds (JSON) and automatically propagate their implications into the frequency recommendation and sensor tasking systems. A storm watch should trigger pre-computation of degraded-condition coverage maps and pre-loading of fallback frequency schedules so that operators can activate them with a single command rather than computing them under pressure during an active disturbance.

Polar cap absorption. Following a major solar proton event (SPE), particles penetrate to the polar ionosphere and dramatically increase D-layer absorption at high latitudes — an effect called polar cap absorption (PCA). PCA can produce near-total HF blackout at latitudes above 60° lasting from hours to days. Trans-polar HF paths — which connect Europe to North America, or Russia to Alaska — are particularly vulnerable. Collection planners who depend on trans-polar paths for coverage of Arctic or northern Russian targets must have pre-planned alternative routing via non-polar paths or accept collection gaps during PCA events.

Alternative collection strategies during disturbances. The primary mitigation for geomagnetic disturbance is frequency agility and collection mode diversity. Concretely: maintain a NVIS-capable collection element for each theater that provides coverage of targets within 200–500 km regardless of long-range ionospheric conditions; ensure all long-range collection sensors have sufficient frequency agility to step down to lower frequencies as MUF depresses; integrate real-time ionosonde data (from GIRO network stations, where available) into the frequency recommendation engine to replace climatological model assumptions with measured ionospheric parameters. The collection system that can ingest a real-time foF2 measurement from the nearest ionosonde and automatically recompute OWF recommendations within seconds of an ionospheric change is operationally far more resilient than one that updates its frequency schedule only when a human operator notices degraded performance.