Freely available satellite imagery has enormous potential to deliver low-cost data to support forest management. Yet forest management has been slow to adopt satellite derived maps for management planning, arguably because satellite remote sensing has not yet delivered sufficient value over traditional sources of information. Researchers at UMaine have been working to develop scalable remote sensing and machine learning workflows to map tree species, aboveground carbon, and other forest attributes at high resolution over large areas. Workflows rely on satellite imagery collected throughout the spring, summer, and fall, typically over multiple years. Cloud cover and other considerations make it difficult to select the best available images and to coordinate image processing tasks over large areas. Additional difficulty arises when projects span multiple sites or longer periods of time, and when processed imagery needs to be shared across multiple projects.
This project is funded through the NSF Center for Advanced Forestry Systems with the goal of creating and implementing improved procedures and tools for managing, visualizing, and utilizing satellite imagery and other remote sensing data for large forest mapping projects. Project contributions will include the design and implementation of a spatial database containing large quantities of remote sensing imagery and related metadata. Data will be made accessible through a server and API for on-demand identification and delivery of best-available data to forest mapping applications.