Zhan et al. (2026) Integrating SWOT data and machine learning for high-resolution river surface velocity retrieval in data-scarce headwater regions

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-09-11
• Authors: Pengfei Zhan, Kai Liu, Jiaming Na, Tan Chen, Jinkai Guo, Chunqiao Song
• DOI: 10.1016/j.jhydrol.2026.136405

Research Groups

• State Key Laboratory of Lake and Watershed Science for Water Security, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences
• University of Chinese Academy of Sciences
• College of Civil Engineering, Nanjing Forestry University

Short Summary

This study integrates SWOT data and machine learning to retrieve high-resolution river surface velocity in data-scarce headwater regions of the Tibetan Plateau. The developed framework achieved strong performance in retrieving spatially continuous river velocity fields.

Objective

• Investigate the feasibility of using SWOT data and machine learning to retrieve river surface velocity in ungauged high-altitude river systems.

Study Configuration

• Spatial Scale: 2-km river reaches across the eastern Tibetan Plateau.
• Temporal Scale: Not specified.

Methodology and Data

• Models used: XGBoost model with longitudinal water surface slope, river width, and discharge as mandatory hydraulic predictors.
• Data sources: SWOT Level-2 Pixel cloud data, in-situ velocity measurements, temperature, NDVI, and other hydroclimatic and riparian context proxies.

Main Results

• The optimal model achieved strong independent testing skill (R2 = 0.80, MAE = 0.09 m/s, RMSE = 0.17 m/s, MAPE = 8.44%).
• Reach-aggregated surface velocities for 5,434 river reaches across the eastern TP ranged from 0.87–1.70 m/s (median 0.95 m/s; IQR 0.92–1.46 m/s).

Contributions

• This study demonstrates that integrating SWOT altimetry with machine learning models offers a scalable pathway for monitoring river hydrodynamics in ungauged high-altitude river systems.

Funding

• Not specified in the provided text.

Citation

@article{Zhan2026Integrating,
  author = {Zhan, Pengfei and Liu, Kai and Na, Jiaming and Chen, Tan and Guo, Jinkai and Song, Chunqiao},
  title = {Integrating SWOT data and machine learning for high-resolution river surface velocity retrieval in data-scarce headwater regions},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136405},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136405}
}