Pachama trains machine learning models using satellite imagery, a vast network of field plots, LiDAR imaging, and other remote sensing data to identify key forest characteristics that are used to estimate carbon.

We are developing machine learning models for business-as-usual emissions of carbon projects. We do this by matching pixels within the surrounding region to each and every pixel within the project boundary, based on attributes such as distance to roads, topography, and forest structure. Unlike status quo static baselines, Pachama’s dynamic baseline is updated annually to observe what actually happened in forests without carbon projects, and capture shifts in background land use that are impossible to predict due to elections or commodity prices.

We train additional machine learning models to estimate forest carbon storage. Our vision is to apply these carbon models over large regions, eliminating the need for expensive and time-consuming manual measurements for individual carbon projects.

Carbon offsetters wish to be notified of any significant changes to a project’s health, such as from illegal logging, as it can impact their net-zero targets. Pachama is developing a forest monitoring system to detect changes in carbon projects over time, and to share these insights and updates directly to offsetters and project developers alike.

Pachama’s vision is to develop these tools into a platform that organizations and individuals can use to create new forest carbon projects, conserving and restoring millions of hectares of forest worldwide.