Jia et al. (2025) Fusing SAR image and CYGNSS data for monitoring river water level changes by machine learning

Identification

• Journal: Remote Sensing of Environment
• Year: 2025
• Date: 2025-07-25
• Authors: Yan Jia, Quan Liu, Chunqiao Song, Zhiyu Xiao, Qiang Dai, Shuanggen Jin, Patrizia Savi
• DOI: 10.1016/j.rse.2025.114927

Research Groups

• Department of Surveying and Geoinformatics, Nanjing University of Posts and Telecommunications, China
• Key Laboratory of Watershed Geographic Sciences, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, China
• University of Chinese Academy of Sciences (UCASNJ), China
• Key Laboratory of VGE of Ministry of Education, Nanjing Normal University, China
• School of Surveying and Land Information Engineering, Henan Polytechnic University, China
• Shanghai Astronomical Observatory, Chinese Academy of Sciences, China
• School of Artificial Intelligence, Anhui University, China
• Department of Electronics and Telecommunications, Politecnico di Torino, Italy

Short Summary

The study proposes a machine learning-based fusion of Sentinel-1 SAR imagery and CYGNSS GNSS-R data to improve the accuracy and temporal resolution of river water level estimation. The fusion approach significantly reduced estimation errors compared to using single-source data.

Objective

• To enhance the precision and temporal frequency of river water level monitoring by integrating high-resolution SAR backscattering coefficients with high-frequency CYGNSS observations using machine learning.

Study Configuration

• Spatial Scale: Regional river networks (validated across 15 training/cross-validation gauge sites and 8 independent validation stations).
• Temporal Scale: Daily temporal resolution.

Methodology and Data

• Models used: Advanced Machine Learning (ML) fusion algorithm with ten-fold cross-validation (CV).
• Data sources:
  • Sentinel-1 Synthetic Aperture Radar (SAR) imagery (backscattering coefficients).
  • Cyclone Global Navigation Satellite System (CYGNSS) GNSS-Reflectometry data.
  • In-situ river gauge observations.

Main Results

• Precision Improvement: The fusion algorithm reduced the Root Mean Square Error (RMSE) from 0.341 m to 0.168 m (a 50.74% reduction).
• Correlation Improvement: The correlation coefficient ($R$) increased from 0.876 to 0.936.
• Station-level Performance: Over 35% improvement in RMSE was observed at 8 out of 15 stations.
• Generalizability: Testing on 8 independent hydrological stations showed a reduction in RMSE from 0.479 m to 0.202 m and an increase in $R$ from 0.848 to 0.927.

Contributions

• Demonstrates the complementary nature of SAR (high spatial resolution/all-weather) and CYGNSS (high temporal frequency/wide coverage) data.
• Provides a robust machine learning framework that overcomes the accuracy and resolution limitations of single-satellite observations for river water level monitoring.

Funding

• Not specified in the provided text.

Citation

@article{Jia2025Fusing,
  author = {Jia, Yan and Liu, Quan and Song, Chunqiao and Xiao, Zhiyu and Dai, Qiang and Jin, Shuanggen and Savi, Patrizia},
  title = {Fusing SAR image and CYGNSS data for monitoring river water level changes by machine learning},
  journal = {Remote Sensing of Environment},
  year = {2025},
  doi = {10.1016/j.rse.2025.114927},
  url = {https://doi.org/10.1016/j.rse.2025.114927}
}