New technologies to map environmental crime in the Amazon Basin (commentary)

Environmental crime slows climate action, deters investment in nature-based solutions, and undermines the green transition. Around the world, land grabbing, illegal deforestation, illicit mining, poaching and a rash of other crimes are ravaging tropical forests, eroding biodiversity, and reversing sustainable development. Despite growing awareness of the problem, an ecosystem of criminality persists in most major biomes, from the Amazon Basin to the African Great Lakes region and the rainforests of Southeast Asia. Preventing and reducing environmental crime is challenging due to the logistical constraints posed by policing vast geographic territories. Dismantling criminal markets is frustrated by the weak rule of law in areas affected by metastasizing organized crime and meager economic alternatives to the unsustainable extraction of natural resources.

Notwithstanding these and other obstacles, new technologies are helping to optimize detection, deterrence and disruption of environmental crime, including in the Amazon Basin. Governments, companies and non-governmental organizations are starting to leverage geospatial and predictive analytics to anticipate where certain types of environmental crime are more likely to occur. In Brazil, for example, several promising initiatives are combining remote sensing data with machine learning tools to monitor and predict everything from selective logging to wildcat gold mining. Many of these innovations are being designed and developed through public-private partnerships, including with state agencies, technology firms, universities, and specialized non-governmental organizations.

Take the case of MapBiomas, a multi-partner initiative tracking phenomena such as land use, fire scars, soil carbon stocks, industrial mining, and deforestation. Launched in 2015, it receives funds from multiple sources to power a collaborative network of universities, non-governmental organizations and technology companies, including Google, EcoStage and Terras App Solutions. In 2019, MapBiomas debuted an alert system to track deforestation at multiple scales (minimum of 0.3 hectares) using maps, charts and tables. Since its inception, the initiative has issued over 392,000 alerts and it is used by a wide range of public, private and activist groups.

Another powerful mapping tool is Amazônia Protege, launched in 2017 by the Federal Public Prosecutor’s Office (MPF), the Brazilian Institute of Environment and Renewable Natural Resources (IBAMA), the Chico Mendes Institute for Biodiversity Conservation (ICMBio) and the Federal University of Lavras (UNFLA). The ultimate goal of Amazonia Protege is to identify infractions and mobilize sanctions against individuals and organizations responsible for deforestation of at least 60 hectares. It does this by combining satellite imagery from the National Institute for Space Research (INPE) with public datasets such as the Rural Land Registry (CAR) and Land Management System (SIGEF). The platform generates real-time information that can be used by supermarkets, slaughterhouses and companies to undertake enhanced due diligence prior to purchasing Amazon-sourced products.

The MPF and UNFLA also teamed up to develop GeoRadar, which was officially launched in 2022. The tool integrates information from over 450 separate data sources and focuses primarily on protected and traditional territories in Brazil’s Legal Amazon. Designed for internal use, Brazilian public prosecutors are using the platform to cross-reference geographical, social, and economic data involving environmental crime and generate graphs and reports that can be included as evidence in investigations and legal proceedings against corrupt officials and organized crime. While still in its early days, the tool is credited with improving the speed and scale of operations for an agency that is already facing a deluge of cases.

Meanwhile, Brazil’s Ministry of Justice and Public Security (MJ), together with the Brazilian data analytics company Santiago & Cintra Consultoria (SCCON), launched the Brazil MAIS program in 2021. The information management system leverages satellite data generated by Planet in order to facilitate government officials’ access to geospatial insight on deforestation. Launched in 2021, the tool provides over 390 public agencies with access to high-resolution data generated by Planet and other sources, and also offers online distance learning and training to its more than 14,000 users. The platform not only facilitates access to data, but also provides online distance learning and training. The intervention has contributed to fines, seizures, and frozen assets totaling over US$ 3 billion and boasts an estimated return of investment of 8,490%.

Launched in 2021 by Microsoft, IMAZON and Fundo Vale, PrevisIA draws on spatial analytics, machine learning algorithms, and the Azure cloud to predict and monitor the areas most at risk of deforestation. PrevisIA considers several risk factors – the extent of legal and illegal roads, land cover and urban infrastructure, and a host of socio-economic characteristics – to determine an area’s vulnerability to deforestation. It generates heat maps to indicate the conservation units, indigenous lands, quilombola territories and other settlements that are most exposed, while also providing rankings of states and municipalities. To date, PrevisIA has identified over 15,000 km2 of territory at “high” or “very high” risk of deforestation.

SeloVerde was launched with support from the Climate and Land Use Alliance (CLUA) and Norway’s International Climate and Forest Initiative (NICFI), and in partnership with the Pará State Secretary for Environment and Sustainability in Para (SEMAS-PA), the Center for Territorial Intelligence (CIT) and the Federal University of Minas Gerais (UFMG). SeloVerde provides real-time monitoring of over 20 million hectares of forested areas. It aims to reduce incentives for deforestation by expanding transparency and traceability at the property level (CAR). The tool uses Amazon Web Services to efficiently process geospatial and public data and ensure that the owners of over 300,000 rural properties comply with relevant legislation and regulations such as the Forest Code.

The Igarapé Institute developed Amazônia in Loco in 2023 to track risks and opportunities in the Brazilian Legal Amazon. The tool features more than 80 environmental, social, and economic indicators from 25 public data sources across all 772 municipalities. Developed using ArcGIS, the interface leverages remote sensing and public data to support decision-making and due diligence in the public and private sectors. A key objective is to “de-risk territories” by improving the accessibility and useability of data on land use, land occupation, socio-economic characteristics, and security-related challenges.

Outside the Amazon, FIP-Cerrado  provides critical real-time information on the outbreak of fires in the Cerrado biome. Launched by the Federal University of Minas Gerais (UFMG), the University of Brasilia (UnB), ICMBio and INPE, it combines high-resolution remote sensing imagery (up to 0.04 hectares in some areas) with climate data and other sources to simulate the spread of fire. The platform simulates fire risks at least three times a day and identifies potential sources of ignition, the abundance of combustible material, the humidity of vegetation and the risk of burning. It can also estimate the potential CO2 emissions and wider effects, taking into consideration anthropogenic factors (e.g. land zoning, roads, urban areas). The tool reportedly has an accuracy rate approaching 90 percent.

Meanwhile, the Radar Mining Monitoring Initiative (RAMI) rapidly detects emerging illegal mining operations in the southern Peruvian Amazon, particularly in areas near protected or vulnerable areas. It can distinguish between illegal, informal and formal mining activities in locations such as Madre de Dios (Peru) where over 300,000 hectares are being tracked. In 2023, SEMAS-PA announced that RAMI would be used to track illegal open-pit mining in the Brazilian state of Pará. An added value is its ability to generate cloud-free mosaic imagery, penetrating areas that are frequently hard to view from space.

Launched by the Interethnic Association for the Development of the Peruvian Forest (AIDESEP), the Sistema de Alertas e Ações Preventivas (SAAT) combines drone, GPS, phone and other technologies to rapidly identify and disseminate information on potential threats to eight of the most at-risk communities in northeast Peru. It has empowered indigenous communities to warn local authorities about the outbreak of forest fires, poaching of wildlife, risks to environmental defenders, encroaching farming and infrastructure settlements, and even the incidence of COVID-related infections.

Another regional initiative is Amazon Mining Watch, a project of the Rainforest Investigations Network and Earthrise Media. Launched in 2022, the platform detects legal and illegal mining activities, including open-pit mining, across the entire Amazon region. The underlying machine learning algorithms reportedly analyze more than 326 million high resolution images every four months. The training data was originally developed using “identification marathons” involving highschool students who helped train the algorithm to recognize characteristics of illegal mining at a resolution of 440 by 440 meters.

These and other geospatial and AI-enabled tools are potentially “game changing” in terms of expanding the capabilities of governments, companies, and civil society groups to slow environmental crime in real-time. But while they can potentially enhance monitoring, reporting and verification and due diligence, technology-enabled platforms are not a panacea. Investment is needed to scale up their quality and expand use-cases. Training and assistance is needed to empower local governments and communities to scale-up use on the front line. These and other efforts are needed to help convert insight into action.

Header image: AI generated image of a satellite over South America