Researchers have unveiled a simple yet effective satellite-based tool to track Spartina alterniflora, one of the most aggressive invasive plant species threatening coastal wetlands. Using freely accessible Sentinel-2 imagery, the new Spartina alterniflora Index (SAI) enables precise, large-scale mapping of this fast-spreading species. Outperforming traditional vegetation indices and rivaling machine learning models in accuracy, SAI offers a scalable and practical solution for environmental monitoring and wetland protection.
Originally introduced to China in the 1970s to control erosion and support land reclamation, Spartina alterniflora has since run rampant along the nation's coastline. Its dense, fast-growing stands are outcompeting native vegetation, disrupting local ecosystems, and altering sediment patterns. Monitoring its spread has proven difficult—traditional remote sensing methods often rely on large labeled datasets and struggle to scale across diverse landscapes. While hyperspectral imaging can offer detailed insights, its high cost and limited availability restrict widespread use. With China aiming to eliminate over 90% of S. alterniflora by 2025, there is an urgent demand for a monitoring tool that is accurate, low-cost, and widely applicable. Against this backdrop, developing a simple yet robust method for large-scale detection became a pressing need.
Published (DOI: 10.34133/remotesensing.0510) on April 29, 2025, in the Journal of Remote Sensing , the new study was conducted by scientists from Ningbo University and the Chinese Academy of Sciences. Their work introduces the Spartina alterniflora Index (SAI), a novel metric designed specifically for identifying the invasive species from satellite data. Unlike many existing approaches that falter when applied across different regions, the SAI method is both adaptable and efficient, offering a timely solution to support China's coastal wetland restoration efforts.
To construct the index, the researchers drew on two spectral bands—Red and Near-Infrared (NIR)—from Sentinel-2 images. The formula (Red - NIR)/NIR enhances the reflectance contrast between S. alterniflora and surrounding vegetation. When tested across six ecologically varied coastal zones in China, the SAI achieved high classification accuracy, exceeding 96% in some locations. Remarkably, the index matched the performance of sophisticated support vector machine (SVM) models, all while avoiding the need for extensive training data. Its effectiveness was further validated on Landsat-8 images and in international regions such as Argentina, the United States, and Australia, confirming the index's versatility and global relevance.
The team compiled a dataset of 3,672 manually labeled points across six diverse coastal zones, encompassing tidal flats and native plant communities. They processed cloud-free, low-tide Sentinel-2 imagery from 2020 via the Google Earth Engine platform. Spectral reflectance profiles were analyzed to identify optimal band pairings for detecting S. alterniflora. Applying a threshold-based classification approach, SAI achieved accuracies ranging from 87.70% to 97.35%, outperforming standard indices like NDVI and EVI. While SVM models offered slightly higher precision, they require complex training and struggle with regional transferability. The simplicity of SAI makes it well-suited for national-scale monitoring. When applied to map S. alterniflora across China in 2020, the index estimated a total coverage of 626 km²—closely aligning with official figures and demonstrating its operational promise.
"Our goal was to create a reliable, easy-to-use tool for large-scale monitoring of coastal invaders," said Dr. Gang Yang, corresponding author of the study. "The SAI method bridges the gap between overly complex machine learning models and overly simple vegetation indices. Its scalability and accuracy make it a valuable asset for ecological restoration and invasive species control."
To maximize contrast with surrounding vegetation, the study focused on images captured during the senescence period of S. alterniflora. Preprocessing included cloud filtering and low-tide selection using Google Earth Engine. Sample categories—S. alterniflora, native plants, and non-vegetation—were derived from fieldwork and high-resolution image interpretation. The SAI was applied on a pixel-by-pixel basis, and statistical methods were used to define classification thresholds. Performance was benchmarked against NDVI, EVI, and SVM methods using the same labeled dataset.
As coastal degradation intensifies under climate and development pressures, the SAI index offers a powerful and timely solution for tracking ecological threats. Its successful application in multiple countries suggests it could become a global standard for monitoring not only S. alterniflora but also other invasive wetland plants. As satellite imagery becomes increasingly accessible, SAI could support real-time tracking systems that enhance conservation, guide policy, and contribute to blue carbon initiatives. Future work will focus on automating threshold selection and broadening applicability to further support ecosystem management at scale.
