Saharan and Asian dust plume transport tracking
Mineral dust lofted from the Sahara and Asian deserts can travel thousands of kilometres, disrupting aviation, agriculture and public health. Satellites detect and track these plumes using UV absorbing aerosol index, thermal infrared brightness-temperature differences, and lidar backscatter profiles.
Sensors
- OMPS (Ozone Mapping and Profiler Suite) on Suomi-NPP: Backscatter UV sensor producing the UV Absorbing Aerosol Index (AAI) at roughly 50 km nadir resolution, daily global coverage. Sensitive to elevated absorbing aerosols; dust altitude above roughly 1–2 km amplifies the AAI signal significantly.
- SEVIRI on Meteosat-11: Spinning Enhanced Visible and InfraRed Imager providing full-disc imagery every 15 minutes (rapid-scan mode every 5 minutes over Europe and North Africa) at 3 km resolution in thermal infrared channels. The 8.7–10.8–12.0 µm brightness-temperature difference (BTD) triplet is the standard operational method for separating dust from cloud over warm desert surfaces.
- CALIOP lidar on CALIPSO: Two-wavelength (532 nm and 1064 nm) polarisation lidar providing vertical backscatter profiles with 30–60 m vertical resolution and roughly 333 m horizontal sampling along a narrow curtain swath. Depolarisation ratio distinguishes non-spherical dust particles from spherical water droplets. Revisit at any given point is roughly 16 days, so CALIOP is used for validation and case-study depth, not operational daily tracking.
- MODIS Deep Blue AOD: The Deep Blue algorithm retrieves aerosol optical depth (AOD) at 550 nm over bright desert surfaces where standard dark-target retrievals fail. Spatial resolution 10 km, twice-daily coverage per instrument (Terra and Aqua combined). Provides quantitative dust loading but cannot separate dust altitude or distinguish dust from smoke without ancillary data.
What a brightness-temperature triplet gives away
Over a warm desert surface, mineral dust and water-ice cloud produce opposite signs in the 10.8–12.0 µm brightness-temperature difference. Dust depresses the 10.8 µm channel more than the 12.0 µm channel, yielding a negative BTD, while ice cloud does the opposite. EUMETSAT's operational Dust RGB product, derived from SEVIRI, exploits this alongside the 8.7 µm channel to colour-code dust pink or magenta against a background of white or cyan cloud. The result is unambiguous for dense, elevated plumes over hot land surfaces.
The method has real limits. Over cold sea surfaces or at night when the land–surface temperature contrast collapses, the BTD signal weakens and thin dust layers can vanish into the noise. Mixed scenes, where a dust layer sits beneath or above cloud, are genuinely ambiguous in the thermal infrared alone. SEVIRI's 15-minute repeat is its main advantage: it resolves the diurnal cycle of dust emission driven by low-level jets over the Sahara and captures rapid plume evolution that a polar-orbiting sensor misses entirely.
Why the UV absorbing aerosol index responds to altitude, not just loading
The UV Absorbing Aerosol Index from OMPS (and its predecessors TOMS and OMI) is not a straightforward measure of column aerosol mass. It is a residual: the difference between observed backscattered UV radiance and what a pure Rayleigh atmosphere would produce. Absorbing aerosols above the boundary layer reduce backscatter below the Rayleigh baseline, producing a positive AAI. The higher the dust layer, the larger the signal for the same optical depth, because more of the Rayleigh-scattering column lies beneath the absorbing layer.
This altitude sensitivity is useful. A positive AAI over the Atlantic almost certainly indicates lofted Saharan dust rather than low marine aerosol. But it creates an interpretive trap: a thin dust layer at 4 km altitude can produce a larger AAI than a denser layer at 500 m. Quantitative mass-loading estimates from AAI alone carry substantial uncertainty. When dust is below cloud, the UV signal is essentially blocked and the AAI underestimates or misses the event entirely. OMPS's 50 km footprint also means that sub-pixel structure, such as filamentary dust streaks or the leading edge of a plume, is smoothed out.
CALIOP: the curtain that settles arguments
When SEVIRI and OMPS disagree, or when a dust–smoke mixture is suspected, CALIOP's vertical backscatter curtain is the arbitrating dataset. The 532 nm depolarisation ratio is the key discriminant: mineral dust particles are non-spherical and depolarise strongly (ratios typically 0.25–0.35), while marine aerosol and smoke are more spherical and depolarise weakly. Published CALIPSO studies have used this to map the vertical structure of Saharan Air Layer transport across the Atlantic at altitudes of 2–5 km, well above the marine boundary layer.
The operational constraint is severe. CALIOP's swath is effectively a line, roughly 70 m wide at the surface. A given location is overflown on a 16-day repeat cycle, so the lidar cannot track a fast-moving plume in real time. Its role in an operational dust-monitoring workflow is to provide high-quality vertical profiles for model initialisation and to validate the column retrievals from SEVIRI and OMPS on the days it happens to overfly the plume of interest.
Trajectory analysis: where sensors stop and models begin
No single sensor tracks a dust parcel continuously from source to deposition. Operational dust trajectory analysis stitches together geostationary imagery, polar-orbiter AOD retrievals, and numerical transport models. NOAA's HYSPLIT model and the Copernicus Atmosphere Monitoring Service (CAMS) dust forecasts are the standard public frameworks. They ingest satellite-derived source-emission estimates and AOD fields as constraints, then propagate dust parcels forward or backward in time using reanalysis wind fields.
The honest uncertainty budget is large. Emission flux depends on surface wind speed to roughly the third power, so small wind errors compound quickly. Dry and wet deposition parameterisations are poorly constrained over ocean. Plume trajectories from HYSPLIT for Saharan events crossing the Atlantic can diverge by several hundred kilometres over a 5-day forecast horizon. For operational decisions, such as aviation rerouting or public-health alerts in the Caribbean during a major Godzilla dust event, the satellite imagery provides the ground truth against which model skill is assessed, not the other way around.
Where the method fails and what that means operationally
Three failure modes matter most. First, dust below persistent cloud is invisible to both UV and thermal infrared methods; CALIOP can sometimes penetrate thin cloud but not deep convective systems. Second, dust mixed with smoke, common over West Africa during biomass-burning season, produces ambiguous AAI and BTD signals; the two aerosol types absorb UV similarly but have different size distributions and depolarisation ratios, requiring CALIOP or spectral inversion to separate them. Third, Asian dust events originating from the Taklamakan and Gobi deserts often travel at high altitude (4–8 km) over the Pacific, where the BTD method is less reliable over cold ocean surfaces and OMPS AAI is the primary detection tool.
These limits are not reasons to avoid satellite-based dust monitoring. They are reasons to build a workflow that routes each sensor's output to the question it can actually answer, and to be explicit about which parts of a given event are well-observed and which are inferred.
Applied analysis and what Satellize does with this
Governments and infrastructure operators with exposure to dust, whether airports, solar-farm operators whose panel efficiency drops sharply during high-AOD events, or health ministries issuing PM10 advisories, need daily products rather than academic case studies. That means automated ingestion of SEVIRI rapid-scan composites, OMPS AAI swaths, and CAMS ensemble dust fields, fused into a single plume-state estimate with an honest confidence flag attached.
Satellize runs analytics on open constellations including SEVIRI and MODIS Deep Blue, and can incorporate CALIOP validation passes for clients who need vertical-structure confirmation. The Overhead column has covered Saharan transport events as they develop. For governments building sovereign early-warning capacity, the more important question is often not which sensor to use but how to operationalise the data pipeline so that a duty forecaster receives an actionable product within two hours of a SEVIRI scan, not two days after a researcher processes the archive. That is the design question worth discussing first.
Typical figures
| SEVIRI spatial resolution | 3 km at sub-satellite point (Meteosat-11 positioned at 0° longitude) |
| SEVIRI repeat cycle | 15 minutes full disc; 5 minutes in rapid-scan mode over Europe and North Africa |
| OMPS AAI spatial resolution | Approximately 50 × 50 km nadir footprint (Suomi-NPP) |
| OMPS daily coverage | Global, once per day (polar orbit, ~101-minute repeat) |
| CALIOP vertical resolution | 30–60 m in the troposphere; ~333 m horizontal along-track sampling |
| CALIOP revisit | ~16-day repeat at any fixed location; near-real-time swath available within hours of overpass |
| MODIS Deep Blue AOD resolution | 10 km; twice-daily combined Terra + Aqua coverage |
| Minimum detectable dust AOD (MODIS Deep Blue) | Approximately 0.05–0.10 over bright surfaces; retrieval uncertainty ±0.05 + 15% of AOD |
| Archive depth | SEVIRI from 2004; OMPS/Suomi-NPP from 2012; CALIOP from 2006; TOMS/OMI AAI from 1978 |
| Latency (operational products) | SEVIRI Dust RGB: near-real-time (~15 min delay); CAMS dust forecast: 4× daily, ~6 h latency |
Analytics Satellize can run
| Daily dust plume extent map | SEVIRI BTD triplet (8.7, 10.8, 12.0 µm) thresholded and vectorised; OMPS AAI overlay for elevated-layer confirmation | GeoTIFF and GeoJSON polygon layer, updated every 15 minutes during active events |
| Plume trajectory forecast | HYSPLIT forward-trajectory ensemble initialised from satellite-derived source coordinates and CAMS wind fields | 72-hour trajectory ensemble shapefile with percentile-spread envelope; PDF briefing note |
| Aerosol optical depth time series | MODIS Deep Blue AOD aggregated over client-defined area of interest; gap-filled with CAMS reanalysis on cloudy days | CSV time series with daily and 7-day rolling mean; threshold-breach alerts via API |
| Vertical structure profile (event validation) | CALIOP 532 nm backscatter and depolarisation ratio extracted for overpasses intersecting the plume; dust layer top and base heights reported | Curtain-plot PDF and NetCDF profile for model validation or insurance documentation |
| Dust–smoke discrimination flag | Combined CALIOP depolarisation ratio and OMPS AAI spectral slope; flags mixed aerosol scenes for analyst review | Raster confidence layer appended to daily plume map; flagged scenes escalated to analyst commentary |
| Event impact report | Plume AOD correlated with surface PM10 station data where available; solar-irradiance reduction estimated from AOD using published direct-normal-irradiance attenuation coefficients | Post-event PDF report with timeline, peak loading, affected area, and estimated irradiance deficit for solar-farm operators |
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.