Project researchers Ben Strong, Science and Machine Learning Lead at Earth Genome; Lindsey Smart, Climate and Ocean Scientist for The Nature Conservancy; and Emily Landis, Global Climate and Ocean Director at The Nature Conservancy, joined eco magazine for an exclusive “topside talks” interview to share how the integration of ML and remote sensing is improving the efficiency and accessibility of seagrass mapping across the globe.
Seagrasses are crucial to marine environments around the world, providing key ecosystem services, including coastal resilience, carbon storage and sequestration, biodiversity, and habitat for juvenile fish populations.
Due to a wide range of factors, seagrass populations are reported to be declining by about 1.5% per year, with around 30% of seagrass meadows lost in the last century alone. And that rate may well rise given the limited understanding of the extent of seagrass systems and how they’ve changed over time.
“Seagrass are the unsung heroes of the sea. Capable of storing thousands of years’ worth of carbon by trapping organic matter in the soils beneath their roots, seagrasses shine as a potential climate mitigation solution. When compared, seagrass has been shown to rival and outcompete some tropical rainforests in terms of how much carbon they can store per hectare,” shared Emily Landis.
A new $2 million project funded by the Paul G. Allen Family Foundation is leveraging breakthrough AI and remote sensing science to create a scalable blueprint for seagrass ecosystem mapping with the goal of developing a more efficient and accessible tool to map seagrass across the globe.
The project is led by Earth Genome, providing the technical AI chops, and The Nature Conservancy—leading the coordination of on-the-ground research in four test sites. The project aims to address an extremely complex and technical problem when trying to map seagrass ecosystems from satellite data: turbidity and varying water depths—both of which differ from region to region.
Lindsey Smart expanded on the project, sharing, “One of the really exciting pieces of this project is that it is highly collaborative. We will be co-developing a lot of this work with seagrass and mapping experts. This will help us better understand the existing challenges and barriers to scaling seagrass mapping. We will work together to identify potential solution sets that leverage the AI tools that Earth Genome is at the forefront of.”
One of the key tools making this project possible is Earth Genome’s Earth Index platform, which they describe as “an early idea of what a ChatGPT can look like for satellite imagery.”
“One of the major challenges for investing in the protection of seagrass ecosystems is the process of obtaining the fundamental data vital to understanding where seagrass ecosystems exist and how they persist over time.
“The science of being able to map seagrass from remote sensing data through drones or satellite imagery is a convoluted and technical process. In addition to that technical process, seagrass ecosystems vary widely from location to location. So, developing a system that can map seagrasses across different regions is no easy feat. Through this project, we aim to create a scalable blueprint for others to map seagrass systems efficiently and accurately over time,” Ben Strong expanded.
The Paul G. Allen Family Foundation has a strong history in data and mapping projects that are focused on mapping the extent of change in the natural world. This partnership between Earth Genome and The Nature Conservancy will refine seagrass mapping methodologies and share findings with the global seagrass community.
To find out more and keep up with the project, visit:
Paul G. Allen Family Foundation – https://pgafamilyfoundation.org/programs/environment
The Nature Conservancy – nature.org
Earth Genome – www.earthgenome.org
eco magazine’s topside talks is an exclusive in-field look into the latest applications of applied marine science methods and technologies. To read more interviews, visit: https://ecomagazine.com/reporting/topside-talks/
