How MapBiomas and Google Earth Engine can support environmental regularisation in rural Brazil

The compliance problem is, at root, a data problem

Brazil's Cadastro Ambiental Rural (CAR) and Programa de Regularização Ambiental (PRA) require every rural property to document its land-cover status: the extent of native vegetation, the condition of legally protected areas, river-margin buffers, hilltop zones, steep slopes, and, where deforestation has occurred, its date and scale relative to the legal baseline.

In practice, the spatial evidence layer is the hardest part of the regularisation process to produce, and the part most vulnerable to challenge if produced carelessly. A property owner who can demonstrate only that trees were cleared, without showing when, or how much remains, is in a far weaker position than one who can present a time-stamped land-cover history anchored to official satellite data. The gap between those two situations is, increasingly, a matter of freely available tools used well or not used at all.

What is MapBiomas?

MapBiomas is Brazil's national land-cover classification platform, produced by a network of universities, NGOs and technology partners. It generates annual land-cover maps for every year from 1985 to the present at 30-metre resolution, covering the entire country across all biomes, including the Amazon, Cerrado, Pantanal, Caatinga, Atlantic Forest and Pampas.

The classification distinguishes approximately 30 land-cover classes: native forest, secondary vegetation at various stages of regeneration, pasture, agriculture, urban areas, water bodies, mining and others. Collection 8, the most recent release, covers 1985 through 2022 for the Brazilian Amazon and all biomes, validated against field data and independent reference samples.

The platform is free and publicly accessible at mapbiomas.org. For straightforward properties, users can draw a property boundary directly in the browser interface and download a clipped time-series dataset, annual land-cover maps, area statistics by class, and change summaries, without any GIS software.

What is Google Earth Engine?

Google Earth Engine (GEE) is a cloud-based geospatial analysis platform that hosts petabytes of satellite imagery, Landsat going back to 1972, Sentinel-2, MODIS, Sentinel-1 SAR and dozens of other datasets, alongside analysis tools that run on Google's infrastructure rather than a local computer.

For users without access to high-specification workstations, this matters enormously. Processing a 37-year time series for a 100,000-hectare property locally could take days; in GEE, the same computation runs in minutes. The platform is free for non-commercial and research use. Researchers, government agencies and environmental organisations worldwide, including Brazil's INPE, which runs the PRODES annual deforestation monitoring system on GEE, use it as standard infrastructure.

MapBiomas is itself built on Google Earth Engine, using Landsat and Sentinel imagery and machine-learning classifiers deployed across the GEE platform. Users can access MapBiomas collection assets directly within GEE scripts, enabling custom property-level analyses that go well beyond what the public download interface offers.

How they work together for regularisation

The combination of a defined property boundary, MapBiomas annual classifications and PRODES deforestation polygons gives a GIS analyst everything needed to build a spatial evidence package for CAR or PRA documentation. The workflow follows these steps:

1. Define the property boundary, from INCRA SIGEF cadastral data, a georeferenced field GPS track, or a professionally surveyed shapefile, and upload it as a GEE asset or load it directly in the platform.
2. Load MapBiomas Collection 8 annual land-cover classifications for the property extent, filtering to the full time series (1985–2022) or a targeted date range.
3. Filter to the key regulatory years: typically 2008 (the CAR legal baseline for the Amazon under the new Forest Code), 2012 (Forest Code enactment), 2016 and 2022, the most recent complete collection year.
4. Extract per-year area statistics by land-cover class: total native forest area, secondary vegetation, pasture, agriculture. These numbers become the quantitative foundation of the evidence package.
5. Identify native vegetation remnants relative to legal obligations: Área de Preservação Permanente (APP) buffers along watercourses, springs and hilltops; and the Reserva Legal fraction (20% of the property in the Amazon biome).
6. Load PRODES annual deforestation polygons as a verification layer and cross-reference with the MapBiomas classification to identify, date and measure deforestation events, especially any clearing after July 2008, which triggers PRA obligations under the Forest Code.
7. Export the outputs: georeferenced map plates (PDF or GeoTIFF), area-statistics tables (CSV or XLSX) and, where needed, a change-detection time series showing year-by-year land-cover transitions within the property boundary.

What this evidence can show

A well-constructed evidence package built from MapBiomas and PRODES data can demonstrate, with high specificity:

Whether deforestation occurred before or after the legal baseline. The Forest Code sets July 2008 as the cut-off date for the Brazilian Amazon. Clearing that predates that threshold is not subject to PRA obligations; clearing after it is. MapBiomas annual classifications, cross-referenced with PRODES, can place a deforestation event within a one-year window, sufficient precision for the vast majority of regularisation assessments.

The current extent of native vegetation relative to legal obligations. The analysis shows exactly how much native vegetation remains on the property, where it is located, and whether the APP and Reserva Legal areas are intact, partially cleared or fully cleared, the three conditions that determine the path through the PRA process.

Year-by-year change history. A time series from 1985 to 2022 gives both the producer and their legal or agronomic representative a defensible record that either corroborates or challenges assertions about when land was cleared and what for. In contested cases, this record, sourced from the same official satellite data IBAMA and state environmental agencies use, carries significant evidential weight.

The location and extent of Permanent Preservation Areas (APPs). Using the property boundary and hydrographic network data (from ANA or IBGE), buffer zones can be computed for every watercourse, spring, lake and hilltop on the property. Comparing those buffers against the land-cover history reveals how much of each APP is intact, when any clearing occurred, and what vegetation class now occupies it.

Limitations to be aware of

When to work directly with the platform, and when to bring in an analyst

For small, uncomplicated properties in well-mapped regions of Acre or Rondônia, where the land-cover history is straightforward and the boundaries are cleanly defined in INCRA SIGEF, a trained agronomist or rural technician can use the mapbiomas.org property-download interface directly. The platform now offers guided workflows for CAR-related queries, and the outputs are formatted for direct use in the regularisation process.

For complex properties, large areas with multiple vegetation classes, disputed or unclear boundaries, contested deforestation dates, properties that straddle biome or conservation-unit boundaries, or cases involving legal challenge, a GIS analyst working in Google Earth Engine can build a more rigorous evidence package. That package will include statistical validation, uncertainty quantification, cartographic outputs at appropriate scales, and metadata that documents every data source and processing step, the kind of defensibility that matters when a regularisation assessment is reviewed by IBAMA, a state environmental agency or a court.

ACS Consultorias has delivered this kind of evidence package for properties across Acre and Rondônia since 2021, supporting rural producers, indigenous communities and legal representatives through the regularisation process.

Conclusion

MapBiomas and Google Earth Engine make satellite-derived land-cover evidence accessible at a scale and cost that would have been unimaginable a decade ago. For environmental regularisation, the implications are significant: a producer who arrives at the regularisation process with a defensible spatial record, dated, sourced, verified, is in a fundamentally stronger position than one who does not.

That record does not require expensive proprietary software or specialist hardware. It requires knowing which tools exist, how to use them, and how to interpret their outputs with appropriate rigour.

At ACS Consultorias, we have used these pipelines in regularisation support across Acre and Rondônia. We are now building Envreg to systematise this workflow for agronomic and legal partners across the Amazon, making the spatial evidence layer a reliable, reproducible part of the regularisation process rather than a bottleneck. If you work with rural producers on regularisation and want to understand what an evidence package looks like in practice, we are open to a conversation.

Back to Insights