Deep Learning‐Based Fusion of Hydrodynamic Model and Remote Sensing Observation for Mapping the Inundation Dynamics in Large Wetlands Under Changing Environment
Water Resources Research · 2026-10 · 62 (10)
DOI: 10.1029/2025wr040618 ↗Kurz gefasst
Fachartikel in der Zeitschrift Water Resources Research, online erschienen am 13.10.2026. Verfasst von Siyang Yao und 6 weiteren Autor:innen. Die Arbeit nennt 87 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: China Postdoctoral Science Foundation.
Abstract (Originalsprache)
Wetlands play a crucial role in maintaining the global ecosystem balance, and determining the inundation dynamics is of great significance for understanding wetland ecosystems. Hydrodynamic models are effective tools for mapping the inundation dynamics; however, they are affected by uncertainties arising from environmental changes, including anthropogenic pressures and land cover modifications, especially in large wetlands. To overcome hydrodynamic model limitations, this study integrated a hydrodynamic module, a remote sensing module, and a deep learning fusion module to develop an efficient approach for mapping the inundation dynamics. This approach was applied in the Poyang Lake Wetland in China, a vital conservation wetland in the Ramsar Convention. The effectiveness of this approach was verified by comparing the simulated inundation dynamics of the hydrodynamic model during the dry season. We found that the performance of the hydrodynamic model was relatively poor. In particular, the performance in the sub‐lake area was significantly inferior than the other zone, with an overall accuracy (OA, dimensionless, ranging from 0 to 1) of only 0.63–0.66. The proposed approach in this study exhibited better performance, especially in the sub‐lake area, with an OA increasing to 0.82–0.92. Furthermore, the hydrodynamic model tended to overestimate the inundation time, with the median overestimated time reaching 42–50 days. Our study provides a transferable framework that can improve the accuracy of wetland hydrological monitoring under changing environment, contribute to maintaining the ecological integrity and stability.
Förderung
- China Postdoctoral Science Foundation
Zitieren
APA
Yao, S., Liu, J., Xu, L., Le, Y., Liao, D., Wu, J., & Wan, S. (2026). Deep Learning‐Based Fusion of Hydrodynamic Model and Remote Sensing Observation for Mapping the Inundation Dynamics in Large Wetlands Under Changing Environment. Water Resources Research, 62(10). https://doi.org/10.1029/2025wr040618
Vancouver
Yao S, Liu J, Xu L, Le Y, Liao D, Wu J, et al. Deep Learning‐Based Fusion of Hydrodynamic Model and Remote Sensing Observation for Mapping the Inundation Dynamics in Large Wetlands Under Changing Environment. Water Resources Research. 2026;62(10). doi:10.1029/2025wr040618
BibTeX
@article{Yao2026,
author = {Yao, Siyang and Liu, Jutao and Xu, Ligang and Le, Ying and Liao, Dong and Wu, Jiang and Wan, Siyu},
title = {Deep Learning‐Based Fusion of Hydrodynamic Model and Remote Sensing Observation for Mapping the Inundation Dynamics in Large Wetlands Under Changing Environment},
journal = {Water Resources Research},
year = {2026},
volume = {62},
number = {10},
doi = {10.1029/2025wr040618}
}
EUVIMED