Sensors
- MODIS (Terra/Aqua): Sea-surface temperature at 1 km resolution, daily global coverage from two satellites. The MODIS SST product (MOD28/MYD28) provides the skin temperature input used to estimate evaporation duct height via the Paulus-Jeske and related empirical models. Cloud contamination is the principal gap; clear-sky compositing over 8-day windows is standard practice.
- Sentinel-3 SLSTR: Dual-view sea-surface temperature at 1 km nadir resolution, with a 27-day exact repeat but effective daily revisit at mid-latitudes from two satellites (Sentinel-3A and 3B). SLSTR's dual-view geometry reduces aerosol contamination bias relative to single-view sensors, which matters when SST accuracy of better than 0.3 K is needed for duct-height estimation.
- ERA5 reanalysis (ECMWF): Hourly global atmospheric profiles at roughly 31 km horizontal resolution and 137 pressure levels, extending back to 1940. ERA5 supplies the specific humidity, temperature and pressure profiles from which modified refractivity (M-unit) gradients are computed. It is not a real-time product; typical data latency is five days behind real time for the high-resolution stream.
- AIRS (Aqua): Atmospheric Infrared Sounder, providing temperature and water vapour profiles at roughly 45 km footprint resolution, twice daily. AIRS retrievals extend from the surface to 1 hPa, but boundary-layer accuracy degrades in the lowest 1-2 km precisely where evaporation and surface-based ducts form. Useful for elevated duct identification above 500 m; less reliable for shallow evaporation ducts.
Why a thin layer of moist air becomes a telecommunications liability
Tropospheric ducting is not a rare anomaly. In coastal and maritime environments it is a routine feature of the lower atmosphere, occurring whenever a sharp negative gradient in modified atmospheric refractivity (M) forms within a layer a few tens of metres to a few hundred metres thick. The ITU-R P.453 standard defines the critical threshold as a refractivity gradient steeper than minus 100 M-units per kilometre. Below that gradient, microwave energy at frequencies from roughly 300 MHz to 30 GHz can be trapped and guided horizontally, bypassing the geometric horizon entirely.
Two duct types matter most to interference planners. The evaporation duct forms almost continuously over open water wherever the sea surface is warmer than the air immediately above it, typically reaching heights of 5 to 40 m. The elevated duct, by contrast, forms at the top of the marine boundary layer or beneath a temperature inversion and can extend to several hundred metres. Both can carry interference from a transmitter well beyond its licensed coordination zone, sometimes by 300 to 500 km over calm tropical seas. For a spectrum regulator or a licensed operator sharing a frequency band with a neighbour across open water, that geometry is a standing problem, not an edge case.
What satellite data actually contributes, and what it cannot
The honest answer is that satellite inputs produce statistical likelihood maps, not real-time duct forecasts. No current satellite sensor directly measures refractivity gradients in the lowest 100 m of the atmosphere with the vertical resolution the physics demands. What sensors do provide is the boundary condition data from which duct probability can be inferred.
Sea-surface temperature from MODIS or Sentinel-3 SLSTR feeds empirical models (Paulus-Jeske, Babin-Young-Carton) that estimate evaporation duct height from the air-sea temperature difference and near-surface humidity. The accuracy of those estimates depends heavily on the quality of the near-surface meteorological inputs, which satellite SST alone cannot supply. ERA5 provides the humidity and temperature profiles needed to compute M-unit gradients through the full boundary layer, but its 31 km horizontal resolution smooths out the mesoscale variability that produces the most intense duct corridors. AIRS adds vertical profile information but with a footprint too coarse for anything below roughly 500 m.
The practical output is a multi-year climatology: for a given geographic corridor and month, what is the probability that a duct of a given height and strength will be present? That is exactly what ITU-R P.453 refractivity gradient maps encode at a global scale. Satellite-derived products can regionalise and sharpen those maps where local SST gradients, coastal upwelling or orographic effects create systematic deviations from the global climatology.
Building a corridor likelihood map: the method in plain terms
The workflow starts with a multi-year SST archive, typically ten or more years of MODIS or Sentinel-3 monthly composites, co-registered to an ERA5 grid. For each grid cell and month, the air-sea temperature difference and the ERA5 surface specific humidity are used to compute a modelled evaporation duct height distribution. The fraction of time the duct height exceeds a threshold relevant to the frequency band under study (say, 20 m for a 10 GHz link) becomes the first probability layer.
For elevated ducts, ERA5 profiles are interrogated for the presence of a temperature inversion and a sharp humidity decrease with altitude, the two conditions that together produce a negative M-gradient above the surface. The height, thickness and M-unit gradient of candidate elevated duct layers are extracted and aggregated into seasonal probability maps. Both layers are then combined with a simple two-dimensional propagation model to estimate, for a given transmitter location, the geographic footprint of elevated field strength that could cause interference. The result is a set of corridor polygons, each labelled with a monthly exceedance probability, which a frequency coordinator can overlay on a licence database.
Cloud contamination in the SST input is the most significant data-quality issue. A single cloud-free observation can be biased by residual aerosol or sun glint. Eight-day and monthly composites reduce random noise but can mask rapid SST changes associated with upwelling events, which are precisely the conditions that suppress evaporation ducts in otherwise duct-prone regions. Any honest deliverable should carry an uncertainty layer alongside the probability estimate.
Resolution floors and the limits of what the archive can tell you
ERA5's 31 km grid is the binding constraint for elevated duct mapping. Coastal zones, island chains and regions with strong SST fronts can produce duct conditions that vary significantly over distances of 10 to 20 km, well below the reanalysis resolution. Downscaling using mesoscale model output or high-resolution SST (the GHRSST Level-4 blended products reach 1 km) can improve spatial specificity, but introduces additional uncertainty from the downscaling assumptions.
AIRS vertical resolution in the boundary layer is nominally around 1-2 km, which is too coarse to resolve the thin elevated duct layers (often 50 to 200 m thick) that cause the most persistent interference. Radio occultation data from GNSS constellations, processed by providers such as Spire Global, offers much finer vertical resolution (around 100 m in the lower troposphere) and is increasingly used alongside reanalysis to validate duct climatologies. That data stream is worth including in any serious corridor assessment, though it is not a satellite imagery product in the conventional sense.
Finally, the archive depth of ERA5 (back to 1940, with consistent quality from roughly 1979 when satellite-era observations begin to dominate the assimilation) is a genuine advantage. Decade-scale trends in duct frequency linked to sea-surface warming are detectable in the record, which matters for spectrum planning horizons of ten years or more.
From probability map to a decision a planner can use
A frequency coordinator needs to know three things: which corridors are prone to ducting, at what time of year, and with what probability of exceeding the interference threshold for the band in question. The satellite-derived climatology answers the first two questions with reasonable confidence. The third requires combining the duct probability with a propagation model calibrated to the frequency, polarisation and antenna heights of the specific link under study.
ITU-R P.453 provides the globally consistent refractivity gradient statistics that regulators treat as the reference. Satellite-derived maps are most valuable where the ITU-R global grid is too coarse to capture local effects, such as the strong SST gradient at the edge of a western boundary current, or the persistent temperature inversions over an upwelling zone off a desert coast. In those situations a regional climatology built from ten-plus years of MODIS and ERA5 data can shift the exceedance probability estimate by a factor of two or more relative to the global map, which is the difference between a coordination distance of 200 km and 400 km.
Satellize builds these regional climatologies as GIS-deliverable raster products with per-pixel uncertainty bounds, suitable for direct import into spectrum management platforms. The analytical approach is the same class of boundary-layer climatology work that underpins our crop-estimation programme for the Kingdom of Tonga, adapted from agronomic variables to atmospheric refractivity.
Typical figures
| SST spatial resolution | 1 km (MODIS, Sentinel-3 SLSTR nadir); GHRSST Level-4 blended products available at 1 km |
| SST revisit | Daily (two MODIS satellites, two Sentinel-3 satellites); effective clear-sky composite typically 8-day to monthly |
| ERA5 horizontal resolution | ~31 km (0.25° grid); 137 pressure levels, hourly timestep |
| ERA5 archive depth | 1940 to present; consistent satellite-era quality from ~1979 |
| ERA5 data latency | ~5 days behind real time (high-resolution stream); preliminary stream ~2 days |
| AIRS vertical resolution (boundary layer) | ~1-2 km in lowest troposphere; insufficient to resolve ducts thinner than ~200 m |
| Applicable frequency range | ~300 MHz to 30 GHz (microwave and UHF bands most affected by tropospheric ducting) |
| ITU-R reference standard | ITU-R P.453: refractivity gradient exceedance maps, threshold minus 100 M-units/km |
| Evaporation duct height range | Typically 5 to 40 m over open ocean; modelled from SST and near-surface humidity |
| Elevated duct height range | Typically 100 to 1000 m; associated with temperature inversions at marine boundary-layer top |
Analytics Satellize can run
| Evaporation duct height climatology | Paulus-Jeske or Babin-Young-Carton empirical model applied to MODIS/Sentinel-3 SST and ERA5 near-surface humidity; monthly percentile distributions per grid cell | GeoTIFF raster stack (monthly, 10-year climatology) with median and 90th-percentile duct height per pixel, plus per-pixel uncertainty layer |
| Elevated duct probability map | ERA5 profile interrogation for temperature inversion height, thickness and M-unit gradient; exceedance fraction aggregated by month and season | GIS polygon layer of seasonal elevated-duct probability corridors, labelled with exceedance probability and typical duct-layer altitude |
| Interference corridor footprint | Two-dimensional beyond-horizon propagation model (ITU-R P.452 framework) driven by duct climatology; transmitter location and antenna parameters supplied by client | Monthly corridor polygons showing geographic area of elevated field strength, with probability contours at 10%, 50% and 90% exceedance |
| Regional refractivity gradient deviation from ITU-R P.453 | Pixel-level comparison of satellite-derived gradient statistics against ITU-R P.453 global grid; identification of zones where local conditions systematically exceed global reference | Tabular and GIS report of deviation zones with recommended coordination distance adjustments, suitable for submission to national spectrum regulator |
| Decadal trend analysis in duct frequency | Linear trend fitting to ERA5 SST and boundary-layer humidity time series (1979 to present); Mann-Kendall significance testing on duct-exceedance fractions | PDF technical report with trend maps and confidence intervals, relevant to spectrum planning horizons of 10 or more years |
| Coastal upwelling SST anomaly flag | Detection of negative SST anomalies relative to climatological mean using MODIS or Sentinel-3 monthly composites; upwelling zones suppress evaporation ducts and reduce interference risk | Seasonal anomaly flag layer overlaid on corridor map, indicating months when duct probability is suppressed by upwelling and months when it is elevated |
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.