Fugitive dust emission mapping at open-pit mines and quarries

Open-pit mines generate fugitive dust plumes detectable in multispectral and aerosol-retrieval data. This page explains how MODIS MAIAC AOD, Sentinel-2 surface reflectance, and TROPOMI aerosol index combine to map plume extent, frequency, and emission loading at active mine sites.

Sensors

MODIS Terra and Aqua (MAIAC AOD product): 1 km spatial resolution AOD retrievals at 550 nm, twice-daily overpass (Terra ~10:30 LT, Aqua ~13:30 LT). The MAIAC algorithm resolves sub-pixel surface anisotropy and produces per-pixel AOD uncertainty estimates, making it the standard reference for quantifying dust loading over heterogeneous mine terrain. Archive extends to 2000 (Terra) and 2002 (Aqua).
Sentinel-2 MSI: 10 m (visible/NIR) and 20 m (SWIR) resolution, 5-day revisit at the equator with both satellites. Band 4 (665 nm) and Band 11 (1610 nm) ratios capture surface albedo changes from dust deposition. Plume edges are visible when optical thickness is sufficient, but thin or semi-transparent plumes below roughly AOD 0.2 are difficult to distinguish from haze without ancillary data.
Sentinel-5P TROPOMI: UV Aerosol Index (UVAI) at 3.5 x 5.5 km pixel size (post-August 2019 upgrade from 7 x 3.5 km), daily global coverage. UVAI is sensitive to absorbing aerosols including mineral dust; it does not give mass concentration directly but flags plume presence and transport direction reliably. Useful for connecting mine-site emissions to regional dust events.
MODIS Terra/Aqua surface reflectance (MOD09/MYD09): 500 m daily composites in seven bands (459–2155 nm). Band 3 (459–479 nm) and Band 7 (2105–2155 nm) ratios help distinguish fresh dust deposition on darker overburden from background soil. Useful for detecting albedo anomalies on spoil heaps and haul roads between major plume events.
Landsat 8/9 OLI: 30 m multispectral resolution, 16-day revisit per satellite (8-day combined). Coastal aerosol band (443 nm) and Band 2 (482 nm) are sensitive to scattering by fine particulate. Useful for high-spatial-resolution plume boundary mapping on cloud-free days, and for cross-calibrating Sentinel-2 albedo change estimates.

What a dust plume actually looks like in the data

Fugitive dust from open-pit mining enters the atmosphere as a mixture of coarse particles (PM10 and larger) that settle within a few kilometres, and fine particles (PM2.5) that can travel tens to hundreds of kilometres downwind. In multispectral imagery, an active plume appears as an elevated surface reflectance in blue and green bands overlying the mine footprint, with the plume axis tracking prevailing wind. At Sentinel-2 resolution, individual haul-road dust trails from heavy vehicles are sometimes resolved on large sites, though this requires near-zero cloud cover and a plume dense enough to raise band-4 reflectance above the sensor noise floor.

MAIAC AOD tells a different story: it integrates the full atmospheric column, so even a diffuse plume that looks faint in RGB imagery can show AOD enhancements of 0.3 to 0.8 above local background. Published studies of large copper and iron-ore mines in arid regions have documented persistent AOD anomalies detectable on more than 60 percent of clear-sky MODIS overpasses during dry seasons. The 1 km MAIAC pixel is coarse relative to the mine footprint, but the signal is quantitative in a way that image-based plume tracing is not.

Separating mine dust from the regional background

This is the central analytical difficulty, and any method that glosses over it is overselling. Arid and semi-arid regions, which host most large open-pit operations, also produce windblown mineral dust that is spectrally nearly identical to mining-derived material. Both are silicate-dominated, both elevate MAIAC AOD, and both show similar TROPOMI UVAI signatures.

The practical separation strategy combines three lines of evidence. First, spatial coincidence: a persistent AOD enhancement centred on the mine pit, present across multiple overpasses and wind directions, is unlikely to be purely natural. Second, temporal pattern: mining dust is correlated with operational schedules, so weekday-versus-weekend differences in AOD or plume frequency can be diagnostic, a method published in peer-reviewed remote sensing literature for coal mines in northern China. Third, back-trajectory analysis using publicly available HYSPLIT model outputs links downwind receptor measurements to source regions, distinguishing a mine site from a regional dust source 200 km upwind. None of these alone is conclusive; the combination reduces ambiguity to manageable levels for regulatory purposes.

Honest limits of the aerosol approach

Cloud cover is the bluntest constraint. In tropical and sub-tropical mining regions, wet seasons can reduce usable MODIS clear-sky observations to fewer than 30 percent of days. Sentinel-2's higher resolution does not help if the scene is cloud-contaminated. This means emission frequency estimates based on optical and aerosol data are inherently dry-season biased, and any annual loading estimate must carry a corresponding uncertainty band.

MAIAC AOD retrieval quality degrades over bright desert surfaces because the algorithm relies on assumptions about surface reflectance that break down when the ground itself is highly reflective. Mine spoil heaps and bare rock can push surface reflectance in the blue band above 0.15, where MAIAC quality flags indicate reduced confidence. Users should filter to high-quality retrievals (QA flag = 0 or 1 in the MCD19A2 product) before computing statistics. Even then, absolute AOD-to-mass-concentration conversion requires site-specific aerosol optical properties, which are not available from orbit alone. The satellite record constrains plume extent and relative loading; it does not replace ground-based PM monitors for regulatory compliance.

Building a plume frequency climatology

A useful regulatory product is a multi-year plume frequency map: for each 1 km MAIAC pixel surrounding a mine, the fraction of clear-sky overpasses on which AOD exceeds a threshold above the seasonal background. Thresholds in the literature typically range from 0.2 to 0.5 AOD units above a rolling 30-day background, depending on regional dust climatology. Applied to the full MODIS archive (2000 onwards for Terra), this produces a 20-plus-year record of emission frequency that captures changes in mine expansion, operational intensity, and dust suppression effectiveness.

Sentinel-2 surface reflectance time series complement this by mapping albedo changes on the mine surface itself. Persistent dust deposition on haul roads and overburden dumps increases shortwave albedo measurably over months to years, and the spatial pattern of albedo change identifies which parts of the operation are the dominant emission sources. This is actionable: haul roads account for a disproportionate share of fugitive dust at most large mines, and knowing which roads are the worst offenders has direct operational value.

What the method cannot do, and what comes next

Orbit-based dust mapping does not provide particle size distribution, chemical speciation, or health-relevant PM2.5 mass concentration without ground truth. It also cannot detect indoor or enclosed processing emissions, and it struggles with small quarries whose footprints fall within a single MAIAC pixel where the plume signal is diluted below detection.

The next analytical step for sites where satellite evidence suggests significant emission is source apportionment at the sub-site level, combining Sentinel-2 plume trajectory mapping with wind-field data to identify which operational units (blasting areas, crushers, haul roads, stockpiles) contribute most. Satellize runs this kind of multi-source fusion analysis on open constellations, with commercial tasking added where higher revisit or finer resolution is needed.

Typical figures

AOD spatial resolution (MAIAC)	1 km per pixel (MCD19A2 product)
Multispectral resolution (Sentinel-2)	10 m (visible/NIR bands), 20 m (SWIR bands)
Revisit frequency	Twice daily (MODIS Terra + Aqua); 5-day (Sentinel-2 dual satellite); daily (TROPOMI UVAI)
TROPOMI aerosol index pixel size	3.5 x 5.5 km (post-August 2019)
Minimum detectable AOD enhancement	Approximately 0.1–0.2 AOD units above background (MAIAC, high-QA pixels); thinner plumes are below reliable detection
Usable archive depth	MODIS: 2000 (Terra) / 2002 (Aqua) to present; Sentinel-2: 2015 to present; Landsat: 1972 to present (coarser pre-OLI)
Cloud-cover constraint	Optical and aerosol methods require clear-sky; wet-season clear-sky fractions can fall below 30% in tropical regions
Key spectral bands for dust detection	Blue (443–480 nm), green (560 nm), SWIR (1610 nm, 2190 nm); UV aerosol index (340–380 nm, TROPOMI)
Delivery formats	GeoTIFF AOD grids, vector plume polygons, CSV frequency statistics, PDF regulatory summary

Analytics Satellize can run

Multi-year AOD anomaly climatology	MAIAC MCD19A2 time-series analysis with seasonal background subtraction and QA filtering	Annual plume frequency map (GeoTIFF + PDF report) showing per-pixel exceedance fraction across the full MODIS archive
Plume extent and trajectory mapping	Sentinel-2 band-ratio thresholding (B4/B2 and B11) combined with HYSPLIT back-trajectory wind attribution	Per-event vector polygons of plume extent with attributed wind direction and estimated transport distance
Surface albedo change detection on mine infrastructure	Sentinel-2 and Landsat OLI shortwave albedo time series, change detection against pre-mining or early-operational baseline	GIS layer identifying haul roads, stockpiles and overburden areas with statistically significant albedo increase
Weekday/weekend emission pattern analysis	MODIS AOD stratified by day-of-week over multi-year record to test operational versus natural dust contributions	Statistical summary table and chart showing operational signal strength; included in regulatory evidence package
Seasonal emission loading estimate	Integration of MAIAC AOD enhancements over mine footprint buffer, scaled by published AOD-to-column-mass relationships for mineral dust	Seasonal and annual relative loading index (not absolute mass; uncertainty range stated explicitly in report)
Compliance monitoring alert feed	Near-real-time MODIS and Sentinel-5P ingestion with threshold exceedance flagging against site-specific AOD baseline	Automated alert (email or API) within 24 hours of a qualifying plume event, with supporting imagery attachment

Sources

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.