Tan et al. (2026) Balancing segmentation accuracy and hydrological plausibility in multi-source remote sensing flood inundation mapping with adaptive terrain-constrained learning

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-08-09
• Authors: Yuping Tan, Guoqiang Wang, A Yinglan, Baolin Xue, Jin Wu, Yuntao Wang, Qiao Wang
• DOI: 10.1016/j.jhydrol.2026.136209

Research Groups

• Advanced Interdisciplinary Institute of Satellite Applications, Beijing Normal University
• Faculty of Geographical Science, Beijing Normal University

Short Summary

This study proposes an adaptive terrain-constrained learning framework to balance segmentation accuracy with hydrological plausibility in multi-source remote sensing flood inundation mapping.

Objective

• Frame flood inundation mapping as a multi-objective problem that requires balancing image-based segmentation accuracy with terrain-hydrological plausibility

Study Configuration

• Spatial Scale: Regional scale (cross-regional 10-fold validation)
• Temporal Scale: Event-based (flood events)

Methodology and Data

• Models used: U-Net architecture with adaptive terrain-constrained learning framework
• Data sources: Multi-source remote sensing data (SAR, optical, DEM-derived terrain information) from Sen1Floods11 dataset

Main Results

• The proposed Adaptive-Dynamic strategy achieved the best overall segmentation performance, with an F1-score of 0.864 and an IoU of 0.761.
• Terrain-derived hydrological priors are most effective when introduced as adaptive soft constraints that regularize image-derived flood predictions.

Contributions

• This study provides a novel approach to balancing segmentation accuracy and hydrological plausibility in multi-source remote sensing flood inundation mapping, which can improve the reliability of flood maps for decision-making.

Funding

• This research was funded by the National Natural Science Foundation of China (Grant No. 42141003)

Citation

@article{Tan2026Balancing,
  author = {Tan, Yuping and Wang, Guoqiang and Yinglan, A and Xue, Baolin and Wu, Jin and Wang, Yuntao and Wang, Qiao},
  title = {Balancing segmentation accuracy and hydrological plausibility in multi-source remote sensing flood inundation mapping with adaptive terrain-constrained learning},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136209},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136209}
}