Illegal cattle ranch expansion into protected forest
Multispectral time series at 3–10 m resolution can detect pasture clearance inside protected forest zones within weeks, mapping bare soil, grass and corral infrastructure against cadastral boundaries to identify unlicensed conversion.
Sensors
- Planet SuperDove (PlanetScope): 3 m resolution, 8 spectral bands including red-edge and near-infrared, near-daily global revisit. The combination of fine spatial detail and high cadence makes it the primary tool for catching clearance events within days to weeks of occurrence.
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 20 m in red-edge and shortwave infrared, 5-day revisit at the equator with two satellites. Free and open; the shortwave infrared bands (SWIR, 1610 nm and 2190 nm) sharpen the contrast between bare soil, dry grass and green canopy. Cloud cover in humid tropics can reduce effective revisit to 2–6 weeks per clear observation.
- Landsat 8/9 OLI: 30 m resolution, 16-day single-satellite revisit (8-day combined). Coarser than SuperDove or Sentinel-2 for small clearances, but the 40-year Landsat archive establishes long-term baseline land cover, and the SWIR bands discriminate active burn scars from older pasture.
- PRODES / DETER (INPE): Brazil's National Institute for Space Research publishes annual PRODES deforestation polygons and near-real-time DETER alerts, typically flagging clearances above 3–6 hectares within days to weeks. Publicly available for the Legal Amazon; useful as a cross-validation layer and for jurisdictional context.
What a pasture boundary gives away
Closed-canopy tropical forest and cattle pasture occupy opposite ends of the near-infrared reflectance spectrum. Forest canopies absorb red light heavily and reflect strongly in the near-infrared, producing high Normalised Difference Vegetation Index (NDVI) values, typically above 0.7. Freshly cleared bare soil and dry grass fall well below 0.4. That contrast is detectable even at Sentinel-2's 10 m resolution. At Planet SuperDove's 3 m, individual corral fences, watering troughs and access tracks become visible as geometric features that no natural disturbance produces.
The spectral signature evolves predictably after clearance. Burn scars show elevated shortwave infrared reflectance in the weeks after fire. Pioneer grass species then colonise, shifting the NDVI upward but keeping the red-edge response lower than intact forest. Tracking that trajectory through a time series distinguishes a genuine conversion from a temporary agricultural disturbance. It also helps analysts date the event, which matters enormously for legal proceedings.
Why protected zones are still being cleared
Brazil's Forest Code and the Amazon Soy Moratorium have reduced deforestation rates in some commodity supply chains, but cattle ranching operates under weaker traceability pressure than soy. Ranchers can graze cleared land for years before cattle enter a monitored slaughterhouse supply chain. The economic incentive is land tenure: clearing and occupying forest, even illegally, has historically established de facto property claims in parts of the Brazilian frontier.
Protected areas and Indigenous territories are not immune. INPE's PRODES data show that a significant share of annual Amazon deforestation occurs inside or immediately adjacent to protected zones, often in small increments designed to stay below the thresholds of coarser monitoring systems. That is precisely where daily 3 m imagery changes the calculus: incremental clearance that would accumulate undetected over a 16-day Landsat cycle is visible as a sequence of individual events at daily revisit.
Mapping infrastructure, not just canopy loss
Canopy-loss detection is necessary but not sufficient for enforcement. A polygon showing forest loss does not, on its own, prove cattle ranching rather than subsistence agriculture or even legal timber harvest. Infrastructure mapping closes that gap. At 3 m resolution, cattle-specific features are distinguishable: rectangular corrals with hard geometric edges, circular watering points, salt-lick clearings, and the radial track patterns that cattle create moving between pasture and water. These are structurally different from the linear clearance patterns of crop agriculture or the irregular gaps of selective logging.
Overlaying detected infrastructure against cadastral boundaries, legal concession polygons and existing moratorium layers converts a remote-sensing observation into an enforcement-relevant finding. A corral inside a legally gazetted protected area, with no valid concession on record, is a different legal object from a corral in an ambiguous tenure zone. That overlay work is where satellite data becomes actionable rather than merely interesting.
Honest limits of the method
Cloud cover is the dominant operational constraint in humid tropical regions. The Amazon basin can sustain weeks of persistent cloud during the wet season, reducing the frequency of usable optical observations regardless of satellite revisit rate. Synthetic aperture radar (SAR) penetrates cloud and can detect canopy disturbance through backscatter change, but SAR at freely available resolutions (Sentinel-1 at 10 m) struggles to distinguish cattle pasture from other open land uses without a supporting optical observation to confirm vegetation type.
Minimum detectable clearance size is a practical limit. DETER's published threshold for near-real-time alerts is around 3–6 hectares. Planet SuperDove can in principle detect smaller clearances, but analyst workload and false-positive rates rise sharply below roughly 1 hectare, particularly near existing forest edges where natural canopy gaps occur. Ranchers who clear in small, scattered patches are harder to catch than those who clear large contiguous areas. And satellite imagery cannot directly identify who owns or operates a property; that link requires ground records, which are often incomplete or falsified in frontier zones.
From alert to evidence package
An enforcement agency receiving a deforestation alert needs more than a polygon. It needs a change-detection time series showing the progression of clearance, georeferenced imagery at a resolution courts will accept, a comparison against the legal land-status layer, and an estimate of the area converted. Producing that package from raw imagery requires consistent pre-processing: atmospheric correction, cloud masking, co-registration across sensors and dates, and change-detection algorithms calibrated to local vegetation phenology.
Satellize structures analytics outputs for this use case as GIS-ready layers with accompanying PDF evidence summaries, formatted to the evidentiary standards that environmental prosecutors in Brazil, Colombia and Indonesia have documented in published enforcement guidance. The Tonga crop-estimation programme demonstrated that the same multispectral time-series pipeline that estimates crop area can be adapted to detect land-use change; the underlying spectral-change methods are the same. Agencies that want to run this analysis continuously, rather than on request, can subscribe to a monitoring feed with configurable alert thresholds by jurisdiction.
Integrating with existing national systems
INPE's PRODES and DETER systems are the authoritative public benchmarks for Amazon deforestation monitoring. Any commercial analytics layer should complement them, not duplicate them. The value of a commercial layer lies in three areas PRODES does not cover: finer spatial resolution for small-clearance detection, coverage outside the Legal Amazon (including the Cerrado and Atlantic Forest biomes, which have weaker public monitoring), and the ability to cross-reference against supply-chain and cadastral data that a national agency may not have access to.
For governments outside Brazil, no equivalent public near-real-time system exists. Countries with significant forest cover but limited monitoring capacity, across Central Africa, Southeast Asia and parts of Central America, are relying on Sentinel-2 and Landsat data processed through global platforms such as Global Forest Watch, which provides annual or near-annual canopy-loss estimates. Weekly or fortnightly commercial alert cadence, tied to a national cadastral layer and a specific enforcement mandate, is a material improvement on that baseline.
Typical figures
| Primary spatial resolution | 3 m (Planet SuperDove), 10–20 m (Sentinel-2 MSI), 30 m (Landsat 8/9 OLI) |
| Revisit cadence | Near-daily (Planet SuperDove); 5-day at equator (Sentinel-2 two-satellite); 8-day combined (Landsat 8+9) |
| Effective tropical revisit (cloud-limited) | 2–6 weeks for clear optical observation in wet season; SAR (Sentinel-1) unaffected by cloud |
| Key spectral bands | Red (665 nm), near-infrared (842 nm), red-edge (705 nm, 740 nm), SWIR (1610 nm, 2190 nm) |
| Minimum detectable clearance | ~1 ha reliably at 3 m resolution; ~3–6 ha for near-real-time public DETER alerts |
| Change-detection latency | Days to weeks from clearance event to alert, depending on cloud cover and revisit |
| Archive depth | Landsat: from 1972; Sentinel-2: from 2015; Planet SuperDove: from approximately 2017 |
| Ancillary data layers | INPE PRODES/DETER polygons, national cadastral boundaries, legal protected-area and concession shapefiles |
| Delivery formats | GeoTIFF change-detection rasters, GeoJSON/Shapefile alert polygons, PDF evidence summary reports |
Analytics Satellize can run
| Canopy-loss alert polygons | Bi-temporal NDVI and SWIR differencing on Sentinel-2 or SuperDove time series, thresholded against local forest baseline | Weekly GeoJSON alert feed with date-of-detection and estimated area, clipped to client jurisdiction |
| Clearance progression time series | Dense time-stack analysis using all available clear observations; change trajectory fitted to distinguish rapid clearance from gradual degradation | Per-polygon animation and tabular area-by-date record for evidentiary use |
| Cattle infrastructure classification | Object-based image analysis (OBIA) at 3 m resolution to detect geometric corral, track and watering-point signatures distinct from crop or logging patterns | GIS point and polygon layer of detected infrastructure features with confidence scores |
| Legal-status overlay and violation flag | Spatial join of detected clearance and infrastructure polygons against protected-area, Indigenous territory and concession cadastral layers | Flagged violation shapefile with legal zone attribution, for direct import into enforcement case-management systems |
| Historical baseline land-cover map | Multi-year Landsat and Sentinel-2 composite classification using supervised random-forest classifier, trained on published land-cover reference data | Raster land-cover map by year, establishing pre-clearance forest extent for legal area calculation |
| Supply-chain linkage risk score | Spatial proximity analysis between detected clearance events and registered slaughterhouse catchment areas, using published supply-chain methodology from Trase and similar open datasets | Tabular risk report linking deforestation polygons to proximate processing facilities, for due-diligence use |
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.