New Study Demonstrates How NASA Hyperspectral Imaging and Machine Learning Can Transform Biodiversity Management in California

Invasive species move fast, and management tools must move faster. A new peer-reviewed study published in Ecosphere reveals how advanced hyperspectral remote sensing and machine learning can equip land managers with powerful, data-driven tools to prioritize restoration and monitor ecosystem health over time. 

The research stems from the NASA Catalyst Preserve Project, a collaborative effort led by NASA Jet Propulsion Laboratory (JPL) in collaboration with The Nature Conservancy, University of California Santa Barbara and Columbia University. The project focused on two California biodiversity hotspots: The Nature Conservancy’s Jack and Laura Dangermond Preserve and the Sedgwick Reserve managed by UC Santa Barbara. 

Using high-resolution airborne imagery from NASA’s AVIRIS-NG instrument combined with machine learning models, researchers were able to: 

• Accurately map invasive iceplant across complex coastal landscapes, helping identify where removal efforts can be prioritized and tracked over time. 
• Identify likely patches of native perennial grasses within heavily invaded grasslands, an important first step toward protecting and restoring remaining native habitat. 
• Show that timing matters: imagery collected later in the spring provided stronger signals for distinguishing species, driven by seasonal differences in plant growth and moisture.

Invasive species like iceplant form dense mats that crowd out native vegetation and alter entire ecosystems. At the same time, remnant native species are often fragmented and difficult to detect. Being able to map both, at scale and with high accuracy, enables land managers to decide where to act first, how to allocate limited restoration resources and whether their efforts are working.

“Identifying new ways to work with non-profit and state partners to make NASA Earth science useful to managers and planners is a critical part of our mission, and I am delighted to have had the chance to develop these new tools with this excellent team,” said Dr. Kimberley Miner, Earth scientist at JPL.

At Columbia University, the project also created hands-on training opportunities for students. “Our Environmental Science & Policy graduates are helping conservation managers with data-driven tools to prioritize restoration efforts and monitor their effectiveness over time,” said faculty advisor Sara Tjossem.

For The Nature Conservancy, the project represents more than a technological breakthrough; it reflects growing collaborations around conservation science.

“What excites me most is the incredible amount of data we now have and the ecosystem we’ve built around it,” said Kelly Easterday of The Nature Conservancy. “We don’t yet know all the directions this work will take, but we do know we have the tools, collaborations and momentum to ask new questions, train students, support land stewards and reach an incredible audience.

Beyond the scientific findings, the project fostered cross-institutional collaboration among scientists, graduate and undergraduate researchers, and on-the-ground stewardship teams. The resulting dataset and workflow provide a scalable model that can be adapted for biodiversity management efforts far beyond California.

“At the Sedgwick Reserve, we often need to point researchers to sites supporting native grasses where they can conduct experiments or sample native plant and animal species,” commented Frank Davis of University of California Santa Barbara. “The ability to track these rare sites reserve-wide using hyperspectral imagery and machine learning is a significant step forward and illustrates the potential for targeted monitoring of plant diversity across large landscapes.”

The full study, Applying machine learning to biodiversity management: Lessons from two California biodiversity hotspots, is available in Ecosphere.

This work was made possible through the collaboration and dedication of project team members: Kimberley Miner, Frank Davis, Kelly Easterday, Sarah Tjossem, Latha Baskaran, Mark Reynolds, Kristen Zumdahl, Radhika Ajayan, Saiarchana Darira, Karin Lin, Christina Van Dyke, Anna Veldman and David Schimel.