Wetl-SAM2YNet: A foundation model driven framework for fine-scale wetland mapping
6 affiliations
- 1Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, 610213, China; College of Computer Science and Cyber Security, Chengdu University of Technology, Chengdu, 610059, China
- 2Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, 610213, China; Corresponding authors at: Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, No. 189, QunXianNan Street, Tianfu New Area, Chengdu, Sichuan, China.
- 3Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, 610213, China
- 4Institute of Geography and Resources Science, Sichuan Normal University, Chengdu, 610101, China
- 5Geological Bureau, Xinjiang Uyghur Autonomous Region, Urumqi, 830000, China; School of Ecology and Environment, Xinjiang University, Urumqi, 830046, China
- 6College of Computer Science and Cyber Security, Chengdu University of Technology, Chengdu, 610059, China; Corresponding authors at: Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, No. 189, QunXianNan Street, Tianfu New Area, Chengdu, Sichuan, China.
Abstract
Fine-scale wetland mapping is challenged by fragmented patch shapes, spectral confusion between wet and dry surfaces, and the scarcity of annotated high-resolution training samples. This study proposes Wetl-SAM2YNet, a framework based on Segment Anything Model 2 (SAM2, a vision foundation model), designed for community level wetland classification in arid regions. The model couples a frozen SAM2 Hiera backbone with a lightweight auxiliary encoder that takes multispectral bands and spectral indices sensitive to wetlands (NDWI, SAVI, DVI, RVI) as input to resolve spectral ambiguities in the visual of arid region wetlands. An Adaptive Cross-Gated Fusion (ACGF) module reconciles generic visual priors with non-visual wetland features, combining the two feature types at multiple scales as the model processes wetland landscapes. Gaussian weighted sliding window inference preserves the continuity of narrow riverside corridors and small fragmented patches. Evaluated on a GF-1/2 dataset comprising only 226 training patches, Wetl-SAM2YNet achieves an overall accuracy of 84.89%, an IoU of 74.40%, and a Kappa coefficient of 84.37%, substantially outperforming CNN and Transformer baselines. The resulting 2 m resolution map resolves seven vegetation communities along hydrological and salinity gradients, providing finer thematic and spatial detail than the global wetland products evaluated in this study. These results demonstrate effective, data efficient, community-level mapping of fragmented wetlands under limited training samples, supporting ecological assessment and conservation planning along hydrological and salinity gradients.
Related
- An integrated data-driven framework for mapping grapevine berry valorization.
- Mapping the soil microbiome functions shaping wetland methane emissions
- CellPatch: a flexible and efficient framework for single-cell foundation models empowered by heuristic gene patching
- Spatially guided translation from histology images to transcriptomic profiles using foundation model-driven contrastive learning
- A Framework for Clustering and Zero-Shot Labeling of Ecological Images Using Foundation Models
BibTeX
@article{zhang2026wetlsam2ynet,
title = {{Wetl-SAM2YNet: A foundation model driven framework for fine-scale wetland mapping}},
author = {Zhaoyi Zhang and Hui Zhao and Huaan Jin and Jing Yang and Quanzhi Yuan and Zhaojie Chen and Yaohua Luo},
journal = {Ecological Informatics},
year = {2026},
month = oct,
doi = {10.1016/j.ecoinf.2026.104119},
url = {https://doi.org/10.1016/j.ecoinf.2026.104119},
note = {CC BY},
}
RIS
TY - JOUR
TI - Wetl-SAM2YNet: A foundation model driven framework for fine-scale wetland mapping
AU - Zhaoyi Zhang
AU - Hui Zhao
AU - Huaan Jin
AU - Jing Yang
AU - Quanzhi Yuan
AU - Zhaojie Chen
AU - Yaohua Luo
PY - 2026
DA - 2026/10/31
JO - Ecological Informatics
DO - 10.1016/j.ecoinf.2026.104119
UR - https://doi.org/10.1016/j.ecoinf.2026.104119
AB - Fine-scale wetland mapping is challenged by fragmented patch shapes, spectral confusion between wet and dry surfaces, and the scarcity of annotated high-resolution training samples. This study proposes Wetl-SAM2YNet, a framework based on Segment Anything Model 2 (SAM2, a vision foundation model), designed for community level wetland classification in arid regions. The model couples a frozen SAM2 Hiera backbone with a lightweight auxiliary encoder that takes multispectral bands and spectral indices sensitive to wetlands (NDWI, SAVI, DVI, RVI) as input to resolve spectral ambiguities in the visual of arid region wetlands. An Adaptive Cross-Gated Fusion (ACGF) module reconciles generic visual priors with non-visual wetland features, combining the two feature types at multiple scales as the model processes wetland landscapes. Gaussian weighted sliding window inference preserves the continuity of narrow riverside corridors and small fragmented patches. Evaluated on a GF-1/2 dataset comprising only 226 training patches, Wetl-SAM2YNet achieves an overall accuracy of 84.89%, an IoU of 74.40%, and a Kappa coefficient of 84.37%, substantially outperforming CNN and Transformer baselines. The resulting 2 m resolution map resolves seven vegetation communities along hydrological and salinity gradients, providing finer thematic and spatial detail than the global wetland products evaluated in this study. These results demonstrate effective, data efficient, community-level mapping of fragmented wetlands under limited training samples, supporting ecological assessment and conservation planning along hydrological and salinity gradients.
SN - 1574-9541
SN - 1878-0512
ER -