DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación
Artículo científico (JournalArticle)
A generalizable deep learning framework for large-scale mapping of seagrass habitats
Seagrass meadows play a vital role in supporting coastal communities by promoting biodiversity, mitigating coastal erosion, and contributing to the local economy. These ecosystems face significant threats, including habitat loss, degradation, and climate change. This has led the United Nations to recognize the urgency of conserving marine ecosystems, highlighting the need for evidence-based conservation strategies and high-quality monitoring methods. However, traditional monitoring approaches are often time-consuming, labor-intensive, and costly, limiting their scalability and effectiveness in large-scale applications. Here, we present a deep learning framework based on convolutional neural networks to identify Posidonia oceanica meadows in the Mediterranean Sea using satellite imagery. We demonstrate the generalization capability and robustness of the model by introducing appropriate metrics that overcome the limitations of current approaches. We show that our model is capable of providing reliable estimates of the distribution of the considered habitats and accurate measures of their extension areas. Our study contributes to the development of a reliable map of the distribution of Posidonia oceanica meadows in the Mediterranean Sea, showcasing the transformative potential of remote sensing and machine learning technologies for marine habitat monitoring., This work was supported through grants TED2021-131836B-I00 (SEDIMENT) funded by Spanish Ministry of Science and Innovation MICIU/AEI/10.13039/501100011033 and by the European Union NextGenerationEU/PRTR Program; PID2021-123723OB-C22 (CYCLE) and PID2024-156062OB-I00 (CHANGE-ME) funded by MICIU/AEI/10.13039/501100011033 and by ERDF, EU; CEX2021-001164-M (María de Maeztu Program for Units of Excellence in R&D) funded by MICIU/AEI /10.13039/501100011033. We acknowledge Nuria Marbà for her valuable initial comments and suggestions. We acknowledge Marcial Bardolet and Jorge Moreno, from the Species Protection Service of the Government of the Balearic Islands, who provided the habitat data used in this study. We also acknowledge Planet Labs for access to satellite imagery data under their Education and Research Program. We express our heartfelt gratitude to José Luis Cantero Rada for being a profound source of inspiration., With funding from the Spanish government through the "María de Maeztu Unit of Excellence" accreditation (CEX2021-001164-M), Peer reviewed
Proyecto:
AEI / Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023 TED2021-131836B-I00
,
AEI / Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023 PID2021-123723OB-C22
,
AEI PID2024-156062OB-I00
,
AEI CEX2021-001164-M
DOI: http://hdl.handle.net/10261/410658, https://api.elsevier.com/content/abstract/scopus_id/105020918463
HANDLE: http://hdl.handle.net/10261/410658, https://api.elsevier.com/content/abstract/scopus_id/105020918463