Dust plume and particulate emission tracking from mine operations
Satellite sensors detect fugitive dust from haul roads, blasting events and tailings surfaces using aerosol optical depth retrievals and visible plume mapping. Daily synoptic coverage from MODIS and VIIRS tracks regional transport; Sentinel-2 and Planet resolve individual source areas.
Sensors
- MODIS MAIAC (Terra/Aqua): Retrieves aerosol optical depth at 1 km resolution using the Multi-Angle Implementation of Atmospheric Correction algorithm. Two overpasses daily per platform give up to four looks per day. Effective for regional plume transport and day-to-day AOD anomaly detection, but cannot resolve individual source areas within a mine footprint.
- VIIRS EDR (Suomi-NPP / NOAA-20): Aerosol optical depth Environmental Data Record at 6 km resolution, with one daily overpass per satellite. Complements MODIS for synoptic coverage and extends the archive beyond MODIS operational life. Useful for detecting persistent regional dust loading above background.
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at the equator with both satellites. The visible and SWIR bands allow qualitative plume delineation and surface disturbance mapping. Not a quantitative AOD instrument; plume detection relies on reflectance contrast against the background, which fails under thin or rapidly dispersing dust.
- Planet SuperDove: 3 m resolution, 8-band multispectral, with near-daily revisit over most land areas under commercial tasking. The coastal-blue and red-edge bands improve contrast for thin dust against bright mine surfaces. Temporal density is its main advantage; quantitative aerosol retrieval from SuperDove is not a standard operational product.
What the atmosphere actually records
Fugitive dust from a mine is not a single event. Haul roads generate continuous PM10 and PM2.5 as heavy vehicles disturb unpaved surfaces. Blasting releases short, dense plumes that can reach several hundred metres in height within seconds. Tailings surfaces, particularly dried pond margins, shed fine particles whenever wind speeds exceed the threshold for entrainment, typically around 6 to 9 metres per second for dry tailings depending on particle size and moisture content.
Each source type has a different spectral and temporal signature. A blasting plume is optically thick, short-lived and localised. A haul-road dust trail is semi-continuous, elongated and tied to shift patterns. Tailings wind erosion is diffuse, low-altitude and strongly correlated with surface dryness. Distinguishing them from space requires combining AOD magnitude, plume geometry, timing and surface context.
How aerosol optical depth turns physics into a number
AOD is a dimensionless measure of how much sunlight a column of atmosphere scatters or absorbs before reaching the surface. A clean atmosphere over arid terrain might show AOD values around 0.05 to 0.1 at 550 nm. A dense dust plume from a large open-pit operation can push column AOD above 1.0 locally. MODIS MAIAC retrieves this at 1 km spatial resolution by exploiting multi-temporal surface reflectance modelling, which makes it substantially more reliable over bright, heterogeneous mine surfaces than earlier MODIS Dark Target or Deep Blue products alone.
The honest limit here is significant. AOD is a column-integrated quantity. It does not directly give you a ground-level PM concentration. Converting AOD to surface PM2.5 requires an assumed or modelled aerosol vertical profile, particle size distribution and density. Published studies using MAIAC over industrial sites report uncertainties in PM2.5 estimates of 30 to 50 percent even with co-located ground sensors. Without in-situ validation, mass-loading numbers derived from satellite AOD alone should be treated as order-of-magnitude indicators, not regulatory-grade measurements.
Resolving the source: what 10 m and 3 m imagery add
MODIS and VIIRS tell you that dust is present and roughly where the column is thickest. They cannot tell you whether the source is the eastern haul road, the crusher feed conveyor or the tailings pond margin 800 metres away. Sentinel-2 at 10 m resolution can delineate plume footprints well enough to attribute emissions to specific infrastructure zones, provided the plume is optically thick enough to register as a reflectance anomaly against the background.
Planet SuperDove at 3 m with near-daily revisit adds temporal resolution that Sentinel-2 cannot match. If a mine operates a 24-hour haul cycle, a single Sentinel-2 pass may catch dust on one day and clear skies the next, missing the pattern entirely. SuperDove's revisit density allows analysts to build frequency maps of dust occurrence by zone across a month, which is far more useful for identifying chronic emitters than any single overpass.
Neither sensor provides quantitative mass flux. What they provide is spatial attribution and temporal frequency, which are the inputs regulators and mine operators actually need to prioritise mitigation investment.
Dispersion modelling: where satellite data ends and physics begins
To estimate downwind concentrations at a community receptor, satellite-derived plume extents and AOD anomalies must be coupled with an atmospheric dispersion model such as HYSPLIT or AERMOD. The satellite data constrains the source location, approximate emission timing and, loosely, relative emission intensity. The dispersion model propagates that forward given wind fields from reanalysis products such as ERA5 or MERRA-2.
The combined uncertainty is substantial. ERA5 wind fields at 31 km resolution do not capture terrain-channelled flows in complex topography. AERMOD requires source-term emission rates in grams per second, which satellite data cannot supply directly. The practical output is a probabilistic plume footprint showing which downwind areas are most frequently exposed, not a certified PM2.5 isopleth map. That distinction matters enormously in any regulatory or legal context.
Chronic exposure versus acute events: two different monitoring problems
Regulators and affected communities typically care about two things: the acute blast event that produces a visible, dense plume crossing a fence line, and the chronic low-level dust loading from haul roads that accumulates over months. Satellite monitoring is better suited to the chronic problem than the acute one.
A blasting event lasts minutes. The chance that a satellite overpass coincides with peak plume density is low unless the operator has commercial tasking capability and advance notice of blast schedules. MODIS and VIIRS may capture the residual haze an hour after a blast, but the dense initial plume will almost certainly be missed. For chronic haul-road and tailings emissions, monthly composite AOD anomaly maps and Sentinel-2 plume frequency layers built from a full archive provide a defensible picture of spatial patterns and seasonal variation. This is where satellite monitoring genuinely earns its place in an environmental management system.
Satellize runs this kind of multi-sensor composite analysis operationally. The workflow used for crop-condition mapping in the Tonga programme, combining open-constellation data with atmospheric correction pipelines, applies directly to AOD anomaly detection over mine sites.
What an operator or regulator should ask before commissioning this analysis
The most common mistake is treating satellite dust monitoring as a substitute for a ground-based PM network. It is not. The two are complementary. Satellite data provides spatial coverage that no ground network can match at reasonable cost; ground sensors provide the calibration data that makes satellite-derived estimates defensible.
Before commissioning, ask whether the mine is in a region with frequent cloud cover. Central African and tropical South-east Asian operations may have fewer than 30 clear Sentinel-2 acquisitions per year, which limits monthly composite quality severely. Ask whether the tailings surface is wet or dry, because wet tailings suppress AOD signal and dry tailings amplify it, creating seasonal bias. Ask what the regulatory threshold is and whether the analysis output needs to meet an evidentiary standard or is purely for internal operational management. The answers determine which sensors, which methods and which level of uncertainty disclosure are appropriate.
Typical figures
| Spatial resolution (AOD retrieval) | 1 km (MODIS MAIAC), 6 km (VIIRS EDR) |
| Spatial resolution (plume delineation) | 10 m (Sentinel-2 MSI), 3 m (Planet SuperDove) |
| Revisit frequency | Up to 4 times daily (MODIS Terra + Aqua); 1 per day (VIIRS); 5 days (Sentinel-2 dual satellite); near-daily (Planet SuperDove, commercially tasked) |
| Key spectral bands | Visible (443–665 nm) and SWIR for AOD; coastal blue, red-edge and SWIR for plume contrast in Sentinel-2 and SuperDove |
| AOD detection sensitivity | MAIAC uncertainty approximately ±0.05 + 15% over land; anomalies above ~0.2 above local background are reliably detectable |
| Minimum detectable plume (optical) | Optically thick plumes (AOD > 0.3 locally) at Sentinel-2 resolution; thin haul-road dust may be below detection threshold in a single pass |
| PM2.5 mass-loading uncertainty | 30–50% without co-located ground validation; order-of-magnitude only without in-situ calibration |
| Archive depth | MODIS from 2000; VIIRS from 2012; Sentinel-2 from 2015; Planet SuperDove from approximately 2021 |
| Cloud limitation | All optical sensors blind under cloud; tropical and equatorial mine sites may have fewer than 30 usable Sentinel-2 acquisitions per year |
| Delivery formats | GeoTIFF AOD anomaly rasters, GeoJSON plume polygons, CSV time-series by zone, PDF monthly summary report |
Analytics Satellize can run
| Monthly AOD anomaly map | MODIS MAIAC time-series differencing against multi-year baseline; anomalies flagged where monthly mean AOD exceeds baseline by more than 1.5 standard deviations | GeoTIFF raster layer with anomaly magnitude and significance flag, delivered monthly |
| Plume frequency map by source zone | Sentinel-2 and SuperDove visible-band reflectance anomaly detection across all clear acquisitions in a period; frequency of plume presence calculated per 10 m pixel | GeoTIFF frequency layer and GeoJSON zone attribution polygons, quarterly |
| Blasting event plume record | Optical plume delineation from Planet SuperDove or Sentinel-2 passes coinciding with known or inferred blast windows; plume area and approximate downwind extent measured | Per-event GeoJSON polygon with timestamp, estimated plume area and wind direction annotation |
| Tailings surface dryness index | Normalised Difference Moisture Index (NDMI) from Sentinel-2 SWIR and NIR bands; dry surface flag correlated with wind speed from ERA5 reanalysis to identify high-entrainment-risk periods | Weekly GeoTIFF surface moisture layer with risk-period annotations |
| Downwind exposure frequency report | Plume polygon stack overlaid with community and infrastructure receptor locations; frequency of plume overlap calculated per receptor over the analysis period | PDF report with receptor-level exposure frequency table and map, suitable for community consultation |
| Multi-year dust trend analysis | MODIS MAIAC and Sentinel-2 archive analysis from 2015 onward; Mann-Kendall trend test on annual mean AOD anomaly by source zone to identify worsening or improving emission patterns | PDF report with trend statistics, confidence intervals and annotated time-series charts |
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.