Global deforestation continues at an alarming rate: the FAO reports an average loss of 10 million hectares of forest per year between 2015 and 2020. In Latin America, the Peruvian Amazon, Brazil's Cerrado, and Andean cloud forests face growing pressure. Satellite forest monitoring is now the only tool that can detect this loss at the scale and speed needed to act before damage becomes irreversible.
How satellite forest monitoring works
Earth observation satellites capture multispectral images that identify land cover types — forest, agriculture, water, bare soil — with sub-metric precision. Forest monitoring is based on comparing images of the same area at different moments to detect changes in forest cover.
The technical process involves:
- Spectral indices: NDVI (Normalized Difference Vegetation Index) measures vegetation density and health. Values above 0.6 indicate dense forest; sharp drops signal disturbance or loss.
- Change detection: comparison of pre- and post-event images to identify pixels that changed from forest to another category.
- Cover classification: machine learning algorithms assign categories to each pixel (primary forest, secondary forest, pasture, cropland, etc.).
- Alert generation: when a change exceeding a defined threshold is detected (e.g. loss of ≥0.5 ha), a georeferenced alert is generated.
Sentinel-2, Landsat, or Planet: which to use?
There is no single ideal satellite. The choice depends on the balance between resolution, revisit frequency, and cost:
Sentinel-2 (ESA)
- Resolution: 10 meters per pixel in visible and near-infrared bands
- Revisit: every 5 days over the same area (combining Sentinel-2A and 2B)
- Cost: free — open-access data from ESA's Copernicus programme
- Best for: operational monitoring of parcels ≥1 ha, forest cover analysis, EUDR compliance
- Limitation: tropical cloud cover can obstruct views for weeks during the rainy season
Landsat 8/9 (USGS/NASA)
- Resolution: 30 meters per pixel
- Revisit: every 16 days
- Cost: free
- Best for: long time-series (Landsat data goes back to 1972), long-term trend analysis, historical reporting
- Limitation: insufficient resolution to reliably detect losses under ~1 ha
Planet Labs (SkySat / PlanetScope)
- Resolution: 3–5 meters per pixel (SkySat: down to 0.5m)
- Revisit: daily or near-daily
- Cost: commercial — subscription or per-area pricing
- Best for: forensic high-resolution evidence, small plots (<1 ha), alert verification, legal proceedings
- Limitation: high cost for large coverage areas
Types of deforestation alerts
Not all alerts carry the same urgency or meaning. Modern forest monitoring systems generate several types:
Confirmed deforestation alerts
Forest cover loss verified with high-quality, cloud-free imagery. These form the basis for legal action or compliance decisions.
Disturbance (degradation) alerts
Partial cover reduction that doesn't reach full deforestation — selective logging, partial fires, trail opening. These are early warning signals requiring follow-up.
Preliminary alerts
Systems like GLAD (Global Land Analysis and Discovery) or DETER (Brazil/INPE) generate alerts within 16 days before confirmation with clean imagery. Useful for rapid response but have higher false positive rates.
Supply chain traceability: from tree to exporter
Forest monitoring is not limited to protected areas. For EUDR compliance and certifications like FSC, you need to trace each product lot back to its origin parcel and verify that parcel has no post-2020 deforestation.
This requires:
- Georeferenced database of suppliers and producers
- Polygons for each production parcel
- Historical forest cover analysis for each polygon
- Lot-to-parcel linkage (with weights and dates)
- Transformation traceability: from cherry coffee to parchment to export container
Field forest census: when satellites aren't enough
Satellites have real limitations: tropical cloud cover, insufficient resolution for individual trees, inability to distinguish species. For forest management plans, concession inventories, and SERNANP/SERFOR assessments, field verification is required.
A field forest census records:
- DBH (Diameter at Breast Height) for each individual tree
- Total height and commercial height
- Species (scientific name and common name)
- Phytosanitary condition
- GPS coordinates of the tree
These measurements compute timber volume: V = π/4 × DBH² × H × 0.7 (standard form factor for tropical forests), which is the basis for the Forest Management Plan.
How Terralyr integrates monitoring into the workflow
Terralyr centralises the entire forest monitoring cycle in a single platform:
- Satellite alert layer: deforestation alerts with date, area, and severity on the satellite base map
- Observation module: alert registration, classification, and resolution workflow with audit trail
- Offline field app: forest census with GPS, photos, and dendrometric data without connectivity
- Supply chain traceability: supplier → parcel → lot → export linkage
- Post-approval monitoring: continuous monitoring of approved parcels with automatic alerts on change
- Monitoring report: SERNANP/SERFOR report generation with updated data
🌲 Terralyr — Forest Management Module
Real-time deforestation alerts, parcel-level supply chain traceability, offline GPS field census, and observation/incident workflow. Fully integrated with SERNANP and SERFOR official instruments.