AI for Earth partners
Learn about our partners working on the front lines of sustainability.
Imazon uses artificial intelligence and Microsoft Azure services to prototype deforestation risk and prevent future losses to the Amazon rainforest.Learn about Imazon
The Amazon represents over half of our planet’s remaining rainforests, and its incredible biodiversity is an invaluable natural resource. Between 1985-2019, 336,681 square miles of native vegetation was lost to deforestation in Brazil alone - that’s a 10 percent destruction of Brazilian lands. Imazon, a Brazil-based organization, promotes sustainable development in the Amazon through studies, support for public policy formulation, broad dissemination of information, and capacity building. But without comprehensive machine learning, Imazon’s forest models were incomplete—giving them only half the picture.
Imazon, together with Fundo Vale and Microsoft are integrating an existing rainforest deforestation risk model algorithm into Azure. With a more robust image processing forest model, PrevisIA can identify unofficial roads, one of the leading indicators of future deforestation, simulating future scenarios to stop rainforest loss events, like forest fires, before they happen. The resulting risk maps, probability maps, and dashboards will not only allow Imazon to predict future deforestation, but also create risk alerts, and use AI to cover more areas at scale. The model will be more precise and allow for expanded data inputs. It is designed with policymakers in mind so they can make better decisions to protect our biodiversity, while conservationists can use the same dashboards to support their mission.
Using AI to help protect the Amazon rainforest
In collaboration with Microsoft, Imazon also announced an initiative to help protect the Amazon rainforest from deforestation and fires. Using advanced AI capabilities from Microsoft across a number of predictors, such as illegal roads opening in forest regions, and data from satellite images, Imazon will be able to predict the most likely deforestation zones. With the assistance of this technology, Imazon can identify critical areas to assist government agencies in protecting against deforestation and conserving Amazon rainforest.
How PrevisIA works
Imazon uses satellite images of Brazil’s Amazon rainforest and then stores them in the Azure cloud where Imazon AI algorithms detect unofficial roads and other risk factors of deforestation. The resulting output is visualized in an interactive map, highlighting the high-risk areas. This information is then used to support decision making for safeguarding the rainforest.
How Imazon uses Azure
- Azure Machine Learning and Azure Databricks for training and deploying deforestation models.
- Azure Storage for storing many types of critical data (e.g. data from satellites, image data).
- GPU optimized Virtual Machines for training, evaluating, and running deforestation models.
- Power BI for dashboards and data visualization.
- Azure SQL for storing processed data (e.g. predictions).
- App Service for scalable deployment of web apps and APIs.
- Event Hubs and Stream Analytics for event processing and data ingestion in near real-time.
- The Planetary Computer for accessing satellite imagery.