Illegal landfill and open waste-dumping site detection
Unpermitted dump sites leave measurable signatures in thermal, shortwave-infrared and visible imagery. Satellite monitoring can detect active decomposition, leachate staining and site expansion before ground inspectors arrive.
Sensors
- Landsat 8/9 TIRS: Thermal infrared bands 10 and 11 at 100 m native resolution (resampled to 30 m in products). Detects surface temperature anomalies from microbial decomposition heat; revisit 8 days per satellite, 4 days combined. Sensitivity to temperature differences of roughly 0.1 K under clear sky.
- Sentinel-2 MSI: 13 spectral bands from visible to shortwave infrared at 10–20 m resolution. Red-edge and SWIR bands are sensitive to bare disturbed soil, leachate-stained ground and anomalous moisture. 5-day revisit at the equator, shorter at mid-latitudes. Free and open archive from 2015.
- Maxar WorldView-3: 30 cm panchromatic, 1.24 m multispectral, 3.7 m SWIR. Resolves individual waste mounds, vehicle tracks and bund walls. On-demand tasking; no fixed revisit but retasking within 1–3 days is typical for priority targets.
- Planet SuperDove: 3 m resolution, 8 spectral bands including red-edge. Near-daily revisit globally. Well suited to tracking the lateral growth of a dump site week by week, though thermal detection is not available on this sensor.
What decomposing waste looks like from orbit
Active organic decomposition is exothermic. A large open dump processing food waste, green waste or mixed municipal refuse generates measurable surface heat. Landsat 8 TIRS has a noise-equivalent temperature difference of roughly 0.4 K and can resolve thermal anomalies at 100 m native resolution. A site covering even two or three hectares of actively decomposing material will typically register 2–5 K above the ambient surface temperature of surrounding soil or grassland on a clear-sky overpass. That signal is modest but consistent, and it persists across seasons in ways that seasonal agricultural warmth does not.
Optical bands add a second, independent line of evidence. Bare disturbed soil has a distinctive spectral profile in Sentinel-2's SWIR bands (bands 11 and 12, centred at 1610 nm and 2190 nm). Leachate, the liquid that percolates through decomposing waste, stains surrounding soil with elevated iron oxides and organic matter. The Bare Soil Index and the Modified Normalised Difference Water Index both respond to these changes. Neither signal is unique to waste sites, but the combination of thermal anomaly, disturbed-soil spectral signature and irregular geometry is highly discriminating.
Distinguishing a dump from a building site or quarry
Disturbed soil is common. Construction sites, quarries and agricultural land-clearing all produce bare-soil spectral signatures. The discriminating features for waste sites are temporal behaviour, spatial texture and the presence of leachate staining.
A construction site expands in a planned geometry and eventually develops hard surfaces or structures. A quarry produces a pit with vertical walls and regular extraction patterns. An illegal dump grows irregularly, often in pulses corresponding to tipping events, and it does not develop hard infrastructure. Planet SuperDove's near-daily cadence is particularly useful here: a time-series of 3 m imagery can show whether a disturbed patch is growing at the edges, whether vehicle tracks lead to a central tipping point, and whether the surface texture is consistent with loose mixed refuse rather than graded soil. Thermal persistence across multiple Landsat overpasses further separates active decomposition from a one-time disturbance.
Very high resolution imagery from WorldView-3 resolves the surface directly: waste mounds, informal bunding, pooled leachate and access tracks are all visible at 30 cm. This is the confirmation layer once a candidate site has been flagged by medium-resolution analysis.
Spectral indices that do the heavy lifting
Several published index combinations are directly applicable. The Normalised Difference Vegetation Index, used in reverse, identifies areas of vegetation suppression consistent with leachate toxicity around a dump perimeter. The Normalised Difference Built-up Index using SWIR and NIR separates bare waste material from vegetated or water surfaces. The Iron Oxide Ratio, computed from Sentinel-2 bands 4 and 2, highlights leachate-stained soils where iron mobilisation has occurred. None of these is a dedicated waste-detection index, but stacking them as a multi-criteria classifier produces a detection layer with meaningfully low false-positive rates when combined with thermal data.
Cloud cover is the persistent constraint. Sentinel-2 carries no synthetic aperture radar, so a persistently cloudy region can go weeks without a usable optical overpass. Sentinel-1 SAR provides all-weather backscatter imagery at 5–20 m resolution, and while it does not directly detect waste composition, it does detect surface roughness changes and moisture anomalies consistent with leachate spread. A combined optical-SAR workflow is more reliable in tropical or monsoon climates than optical alone.
Growth-rate monitoring and enforcement prioritisation
Regulators rarely have the field capacity to inspect every complaint. Satellite monitoring reframes the problem: instead of reacting to complaints, an enforcement agency can maintain a ranked watchlist of sites ordered by growth rate and thermal activity.
Change detection on Sentinel-2 at 10 m resolution can resolve lateral expansion of roughly 100 m² between overpasses. A site growing by 500 m² per month is visibly different from one that has stabilised. Combining area-change rate with the thermal anomaly magnitude gives a proxy for active versus inactive dumping, which matters for enforcement timing. Prosecuting an operator while tipping is actively occurring is evidentially stronger than arriving weeks after activity has ceased.
Archive depth is an asset here. Sentinel-2 data runs from 2015 and Landsat from the early 1970s. For contested enforcement cases, a retrospective time-series can establish exactly when a site appeared, how rapidly it grew, and whether it predates a permit application. That kind of documentary evidence is difficult to produce by any other means at reasonable cost.
Honest limits of the method
Thermal detection requires clear sky and adequate site size. A small fly-tip of a few hundred square metres will not produce a thermal anomaly detectable by TIRS at 100 m resolution. Sentinel-2 at 10 m can map the footprint of such a site if it has been present long enough to stain the soil, but fresh small deposits may simply be invisible until they grow. WorldView-3 tasking can close this gap for priority locations, but it is not cost-effective for blanket national surveillance.
Spectral indices are sensitive to soil type. Laterite-rich soils in tropical regions naturally show elevated iron oxide ratios, which can produce false positives for leachate staining. Calibration against local soil baselines is necessary before deploying a classifier in a new jurisdiction. Similarly, industrial sites with legal heat sources, such as composting facilities or biomass processing plants, can mimic the thermal signature of an illegal dump. Cross-referencing against permit registries is an essential step, not an optional refinement.
Satellize applies this multi-index, multi-sensor workflow operationally, drawing on the same open constellations used in its Tonga crop-estimation programme and adding commercial tasking where resolution demands it. The output is a ranked site register, not a definitive legal record. Ground verification remains the enforcement agency's responsibility.
From pixel to prosecution file
The analytic chain ends with a deliverable that an enforcement officer can act on. That means georeferenced polygons with area measurements, a time-series chart showing growth dates, a thermal anomaly summary, and, where VHR imagery has been tasked, annotated imagery showing site features. The coordinate precision of Sentinel-2 orthorectified products is better than 12 m CE90 under standard processing, sufficient for field navigation. WorldView-3 products are accurate to sub-3 m CE90 with rational polynomial coefficients.
Jurisdictions with limited enforcement capacity tend to benefit most from the watchlist model: a monthly update that flags new sites and ranks existing ones by activity level. That is a manageable operational rhythm for a small inspectorate, and it concentrates scarce field visits where satellite evidence already suggests active violation.
Typical figures
| Thermal spatial resolution | 100 m native (Landsat 8/9 TIRS), resampled to 30 m in standard products |
| Optical spatial resolution | 10 m (Sentinel-2 MSI), 3 m (Planet SuperDove), 1.24 m multispectral / 30 cm pan (WorldView-3) |
| Revisit cadence | 4–5 days optical (Sentinel-2 combined), near-daily (Planet), 8 days per Landsat satellite |
| Minimum detectable thermal anomaly | Approximately 2–5 K above ambient for sites larger than ~2 ha under clear sky (TIRS) |
| Minimum mappable site footprint | ~100 m² with WorldView-3; ~1,000 m² with Sentinel-2; ~2 ha for reliable thermal detection |
| Key spectral bands | SWIR 1610 nm and 2190 nm (soil disturbance); NIR 842 nm (vegetation suppression); TIR 10.9 µm (thermal anomaly); Red-edge 705 nm (leachate stress) |
| Archive depth | Sentinel-2 from 2015; Landsat from 1972 (thermal from Landsat 8, 2013) |
| Cloud cover constraint | Optical bands unusable under cloud; SAR (Sentinel-1, C-band, 5–20 m) provides all-weather surface-change proxy |
| Coordinate accuracy | Better than 12 m CE90 (Sentinel-2 standard); sub-3 m CE90 (WorldView-3 with RPC) |
| Delivery formats | GeoTIFF change layers, GeoJSON site polygons, PDF enforcement summary, time-series CSV |
Analytics Satellize can run
| New-site detection alert | Multi-index change detection on Sentinel-2 SWIR and NIR bands against baseline composite; threshold exceedance triggers alert | Automated GeoJSON alert with site centroid, estimated area and first-detection date |
| Site growth-rate time-series | Polygon area measurement on sequential Sentinel-2 or Planet imagery; monthly differencing | CSV and chart showing area in m² per overpass date; growth-rate ranking for enforcement prioritisation |
| Thermal anomaly assessment | Land surface temperature retrieval from Landsat 8/9 TIRS bands 10 and 11 using split-window algorithm; comparison against local ambient baseline | Raster LST layer and tabular anomaly magnitude per site per Landsat overpass |
| Leachate spread mapping | Iron Oxide Ratio and Modified NDWI computed from Sentinel-2; spatial extent of stained soil delineated by supervised classifier calibrated to local soil baseline | GeoTIFF leachate extent polygon with uncertainty buffer; updated monthly |
| Retrospective establishment date | Archive time-series analysis on Sentinel-2 (from 2015) and Landsat (from 2013 for thermal); first-detection date established by sequential change-point analysis | PDF enforcement summary with dated imagery strip and annotated change sequence |
| VHR site characterisation | On-demand WorldView-3 tasking; manual and semi-automated annotation of waste mounds, access tracks, leachate pools and bund structures at 30 cm resolution | Annotated GeoTIFF and feature GeoJSON suitable for inclusion in enforcement file |
| National watchlist register | Ranked list of candidate sites by composite score: growth rate, thermal anomaly magnitude, proximity to watercourses; updated on monthly cadence | Interactive GIS layer and PDF monthly report with top-ranked sites flagged for inspection |
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.