Using AI to scale access to forest carbon markets
THE QUESTION
Can AI successfully analyse satellite images of forest cover and provide a more accurate calculation of carbon levels in Tanzanian forests?
LOCATION: Tanzania
SECTOR: Climate and Environment
TECH: AI
TIMELINE: September 2022 - Present
PIONEER: Tom Ratsakatika
PARTNERS: Omdena
The Challenge
Tanzania has Africa’s third largest forest cover yet the fifth highest rate of deforestation globally. Deforestation accounts for 73% of Tanzania’s greenhouse gas emissions, primarily driven by farming and charcoal production. There is a significant opportunity for Tanzania to tap into global carbon markets, increasing access to much-needed climate finance whilst tackling deforestation. A key challenge local authorities and project developers is the cost, accuracy and timeliness of forest carbon data. Access to cheaper, better data will support the government and developers to sustainably grow the forest carbon sector in Tanzania.
The Idea
The pilot aims to train machine learning models on satellite imagery of Tanzanian forests to calculate changes in carbon density. The pilot will determine if recent advances in machine learning can provide actionable data to support the growth of the forest carbon sector in Tanzania. Machine learning could increase data accuracy and reduce the cost of current labour-intensive methods. However, the models must be tailored to Tanzania’s local biomes such as Miombo woodland, cashew nut trees and indigenous grassland. If successful, the evidence from this pilot could support replication in similar biomes across East Africa.