Geospatial machine learning in forestry & carbon markets

Trees are essential components of carbon in and off-setting programs. With help of geospatial imagery and machine learning analysis, monitor the land where trees are planted, follow their growth, and count them to get transparent and verifiable documentation to support carbon certifications and other sustainable practices KPIs.

From harvesting, soil preparation, plantation lines planning to saplings survival analysis – forest experts and land managers use geospatial machine learning to monitor forest cycles and map thousands of hectares of woodland remotely. It gives them the ability to analyze inaccessible locations where field research is impossible or hazardous. Check out how Skytec utilized Picterra to build Ranger – their cutting-edge remote sensing application enriching forestry management with geospatial technology.

Illegal logging is causing severe damage not only to the environment by accelerating deforestation and biodiversity loss. It also impacts the economy, resulting in lost revenues, and harms local societies by leading to conflicts over land, resources, or corruption. The use of geospatial imagery and Picterra’s geospatial machine learning platform allows for persistent monitoring of large regions of forest to spot deforestation patterns or illegal logging in a fraction of the time compared to using on-the-ground patrols. It also has the added benefit of not disrupting wildlife. Discover how using Picterra, SGS monitors the Liberian forests and alerts the authorities at any sign of illegal activity so they can take immediate action.

Detecting pest infestations before they spread, prevent major woodland loss and salvage timber value. Climate change and warming temperatures are fueling at the alarming rate the expansion of the destructive pests such as spine and spruce beetle outbreaks across world’s forests, ravaging tens of thousands of square kilometers of woodlands. Machine learning detection on geospatial imagery allows to spot early outbreaks of bark beetle infestation – when trees are infected, the tree crowns turn pink, making them possible to spot from above in satellite or aerial images. Check out how forestry experts at Waldstolz used Picterra detectors to recognize inklings of infestations in the images that would be impossible for the human eye to spot and prevented bark beetle outbreaks from spreading.

Oil palm produces about 35% of all vegetable oil, which has a massive impact on the ecosystem and leads to deforestation. With Picterra, it’s possible to detect palm trees, which is crucial for any move towards sustainability.

Thanks to geospatial machine learning, what used to take foresters many hours of work – like counting trees in drone images – can now become automated tasks achieved in minutes or a few hours. Use of ML detectors trained to infer and extract the right combination of information for example locating and extracting individual treetops combined with polygons of the plantation lines, generates more precise and refined results giving forest operators the ability to accurately calculate the percentages of effective forestry and its failures, in order to improve their production processes and decision-making. Discover how ICA, a leading GIS company in Uruguay, counted thousands of trees with 98% accuracy.

With Picterra, it’s possible to detect animals that are in danger or threatened with extinction. Thanks to the drone and satellite imagery combined with state-of-the-art machine learning algorithms, it only takes hours to assess the animal population across a large area. Researchers from the University of Santa Cruz, California, have used Picterra to detect and count sea lions and seals living on the Año Nuevo Island. The result: 140k seals and sea lions detected on 50 orthomosaics in 8 hours, comparing to 35 days when done manually.