EO·Atlas
Preview build / EO·Atlas v0.9, content still landing
consortium · companies

MapBiomas Network

Compiled from public sources on 2026-05-24. Not independently verified by MapBiomas Network.

Does MapBiomas Network own this listing? Claim and verify it →

Multi-institution consortium coordinating annual land-cover mapping and alert systems across Brazil and Latin America. Members include IPAM Amazonia, Imazon, ISA, SOS Mata Atlantica, UFRGS, LAPIG, Geodatin, ArcPlan, The Nature Conservancy, Imaflora, and approximately 30 additional research and NGO partners. INPE contributes DETER and PRODES data feeds and participates in the Scientific Advisory Committee but is not an operating institution. The consortium has no single institutional operator or legal registration; coordination is led by an independent scientific team (general lead Tasso Azevedo, scientific lead Julia Shimbo, technical lead Marcos Rosa). Administrated through a consortium coordination secretariat funded by grants and philanthropic donors.

MapBiomas is an open-science consortium founded in 2015 from a seminar convened by Brazil's Climate Observatory to address how annual land use and land cover (LULC) maps could be produced across Brazil at scale, speed, and low cost.[1] The network operates as a distributed collaboration of universities, NGOs, and technology companies with no single legal entity or equity structure. Coordination is led by an independent scientific team including Tasso Azevedo (General Coordination), Julia Shimbo (Scientific Coordination), Marcos Rosa (Technical Coordination), and Carlos Souza Jr. (Technical-Scientific Coordination).[2]

The core technical infrastructure is Google Earth Engine, which provides cloud compute and access to the full Landsat archive from 1985 onward.[1] Annual LULC maps are produced at 30-metre resolution using a Random Forest classifier trained per biome by domain expert teams.[3] Up to 105 spectral metrics per pixel are derived annually from Landsat multispectral bands and calculated indices. Brazil's territory requires approximately 380 Landsat scenes per year; the full national dataset covers more than 9 billion pixels.[3] The latest annual LULC release, Collection 10.1 (February 2026), covers the period 1985 to 2024.[4]

Beyond the core annual LULC series, MapBiomas Brasil operates a suite of thematic monitoring systems: MapBiomas Alerta (deforestation alert validation at 3.7 m using Planet imagery, covering all six Brazilian biomes)[5], a Water module, Fire module (annual and monthly burned areas), Soil, Mining, Pasture, Agriculture, Atmosphere, and Climate Risk layers. A Rural Credit Monitor cross-references land cover with agricultural financing data to flag environmental compliance issues.

The MapBiomas model has been replicated across South America and Southeast Asia. Active national networks span Argentina, Bolivia, Chile, Colombia, Ecuador, Paraguay, Peru, Uruguay, and Venezuela, plus regional biome initiatives covering the Pan-Amazon, Atlantic Forest, and Chaco. MapBiomas Indonesia operates LANDY (LULC mapping), FIRE (burned area monitoring), and ALERTA (deforestation validation) for the Indonesian archipelago, operated by a coalition of Indonesian civil society organizations including Auriga, HAKA, Hutan Institute, and others.[6] The network spans South America and Indonesia.

All MapBiomas data, maps, methods, and code are publicly available at no cost via the MapBiomas Platform, Google Earth Engine toolkit, GEE data assets, and direct GeoTIFF download. Users include researchers, policymakers, civil society organizations, and companies engaged in environmental compliance and supply chain due diligence.

Manufactures

  • MapBiomas AlertaAnalysis serviceAlert aggregation from DETER (INPE), SAD-Imazon, SAD-Geodatin, SAD-SOS Mata Atlantica/ArcPlan, SAD-GeoKarten/UFRGS, SAD-IPAM/LAPIG, SIRAD-X radar (ISA/Xingu+), and GLAD (University of Maryland) across all six Brazilian biomes, PlanetScope 3.7 m resolution visual validation and false-positive rejection by trained analysts, Supervised Random Forest boundary refinement via Google Earth Engine, with training samples drawn from high-resolution before-and-after image pairs +3

    operational

    Operator pricingPricing not publicly listed by operator
  • MapBiomasAnalysis serviceAnnual land cover and land use classification (random forest per biome, Landsat 30 m, 1985 to present in Brazil), Sentinel-2 10 m Collection 2 beta product (parallel finer-resolution track), Regional initiatives across 11 countries plus 4 multinational regions (Pan-Amazonia, Gran Chaco, Atlantic Forest, Pampa) +2

    operational

    Operator pricingPricing not publicly listed by operator
Sources
Cite https://eo-atlas.org/companies/mapbiomas Markdown twin → Field definitions →