EUDR
Deforestation-free supply chains now demand continental evidence
Sentinel-2 + LiDAR
AI trained on airborne truth, applied continent-wide
VMAP
Powered by Terrak.io, delivered by WWF, ANU and Haizea
The Challenge
Monitoring forests underpins a growing list of policy and commercial decisions — national greenhouse accounting, biodiversity conservation, land-use planning, and increasingly, deforestation-free supply chains. The EU Deforestation Regulation now requires companies placing beef, leather, soy, timber and other commodities on the EU market to prove their supply chains are not linked to land that was deforested after 2020.
For Australian exporters that obligation is real and continental in scale. Evidence has to be produced, defended and refreshed for every relevant property across 7.7 million square kilometres — a landmass larger than the entire EU — and it has to hold up to the scrutiny of regulators, auditors and trade partners.
Satellite imagery is the only viable source at that scale, but using it well is not easy: high-resolution, up-to-date, cloud-free, radiometrically consistent, analysis-ready observations have to be assembled from a firehose of raw scenes. The resulting archives reach tens of petabytes, so any operational product must also solve the engineering problem of storing, processing and repeatedly rerunning that archive at national scale.
The Solution
WWF, the Australian National University and Haizea Analytics have teamed up to close that gap. VMAP is the result: an accurate, annually updated, high-quality product that maps forest extent and change across the whole of Australia since 2020.
Analysis-Ready Sentinel-2 on Terrak.io
Terrak.io hosts the full Sentinel-2 archive as a single queryable resource. Users — and models — describe the analysis they want, not the plumbing: no tile downloads, no grid alignment, no cloud masking gymnastics.
Airborne LiDAR as Ground Truth
Publicly available airborne LiDAR surveys give centimetre-scale vertical structure for slices of the continent. Those surveys are used to label what forest actually looks like from space — a training signal far richer than photo-interpreted polygons.
AI Trained and Scaled on One Engine
AI models are trained and optimised directly against Terrak.io's training-sample and batch pipelines, then run annually at national scale. The same engine serves the model at an interactive tile, a national GeoTIFF or a per-property zonal statistic.
The result is a forest map with continental coverage, annual updates and a reproducible, LiDAR-grounded methodology — produced on the same platform that serves it, so every tile on the map is traceable back to the pixels and the model that generated it.
How It Works
Analysis-ready Sentinel-2, airborne LiDAR labels and AI models — trained, run and delivered on the same engine, once a year, for the whole continent
Analysis-Ready Sentinel-2
The full Sentinel-2 archive is hosted on Terrak.io and exposed as a single queryable resource — clean, cloud-masked and time-aligned — so downstream steps see a consistent continental raster rather than a pile of raw scenes.
LiDAR-Derived Training Labels
Airborne LiDAR surveys give centimetre-scale vertical structure for slices of the continent. Those surveys are turned into high-quality labels of what forest actually is — a far richer training signal than photo-interpreted polygons.
AI Model Training on Terrak.io
AI models are trained and optimised directly against Terrak.io's training-sample pipeline, which materialises pixel values at labelled points at continental scale — so the model sees the same data representation it will run on in production.
Annual National Inference
Once a year the trained model is run over the whole of Australia through Terrak.io's batch orchestrator, producing a fresh forest extent and change layer for the full continent, versioned against previous years.
Standards-Compliant Delivery
The output layers are served back through the same engine as OGC WMS, XYZ web-map tiles, GeoTIFF or NetCDF over any polygon, and CSV zonal statistics — so VMAP drops straight into GIS, web-mapping libraries and compliance workflows.
VMAP is now used by:
Informs government policy: VMAP gives Commonwealth and state agencies an independent, continentally consistent reference for forest extent and change — useable in greenhouse accounting, biodiversity reporting and land-use policy without waiting on bespoke ad-hoc studies.
Supports NGO advocacy and landscape programs: WWF and partner organisations use VMAP to quantify forest loss in priority landscapes, target restoration programs and hold commitments to account with the same evidence base everyone else is looking at.
Enables EUDR-aligned commercial due diligence: Exporters, traders and their auditors use VMAP to check whether commodity supply chains touch land deforested after 2020 — a baseline the EU Deforestation Regulation explicitly requires and that the product is built around.
Equips researchers with reproducible science: The same analysis-ready Sentinel-2 stack, LiDAR labels and Terrak.io pipelines that produce VMAP are available to university and CSIRO teams — so follow-on science builds on the same foundations rather than starting from raw scenes.
Explore Our Case Studies
Evidence-based environmental policy in the ACT
ACT Government Sustainability
Annual environmental reporting across Australia
TERN & ANU Environmental Monitoring
Enabling Australia's Nature Repair Market
Australian Government
Continental bushfire intelligence with an AI agent
OzHazard — Terrakio Agent in production
A continental, EUDR-ready map of Australia's forests
VMAP — WWF, ANU & Haizea Analytics
Take VMAP to your supply chain, landscape or jurisdiction
VMAP is already used by government, NGOs, researchers and commercial companies to inform decisions and better understand Australia's forests. The same Terrak.io pipeline — analysis-ready Sentinel-2, LiDAR-grounded AI models, annual national runs, standards-compliant delivery — can be re-targeted to your jurisdiction or supply chain.
