Li et al. (2026) A spectral-constrained HVCE framework for adaptive optical-SAR fusion to enhance soil moisture retrieval

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-08-04
• Authors: Yijie Li, Yifei Yao, Shikun Sun, Enhao Wang, Rong Liu, Jiale Ma, Pute Wu
• DOI: 10.1016/j.jhydrol.2026.136170

Research Groups

• Key Laboratory for Agricultural Soil and Water Engineering in Arid Area of Ministry of Education, Northwest A&F University
• Institute of Water Saving Agriculture in Arid Regions of China, Northwest A&F University
• Xinjiang Research Institute of Agriculture in Arid Areas, Northwest A&F University

Short Summary

This study proposes a spectral-constrained Helmert variance component estimation (HVCE) adaptive fusion framework for enhancing soil moisture retrieval from Sentinel-1 SAR and Sentinel-2 optical data. The framework outperforms conventional methods by preserving spatial-spectral fidelity while enhancing task-oriented information.

Objective

• Investigate the potential of adaptive fusion of optical-SAR imagery for improving soil moisture retrieval accuracy

Study Configuration

• Spatial Scale: Large-scale heterogeneous areas (alpine meadow, woodland-grassland, farmland) and small-scale irrigation districts
• Temporal Scale: Not specified

Methodology and Data

• Models used:
  • H-SVD (enhanced Multi-Resolution Singular Value Decomposition)
  • H-DWT (Discrete Wavelet Transform)
  • Gram-Schmidt
  • PCA
  • Hybrid-MSD
  • SVD
  • DWT
  • Entropy-Optimized SVD and DWT models, E-SVD and E-DWT
• Data sources:
  • Sentinel-1 SAR data
  • Sentinel-2 optical data
  • In-situ soil moisture measurements

Main Results

• The proposed framework outperformed conventional methods in terms of average correlation between fused data and in-situ SM (59% and 50% for H-SVD and H-DWT, respectively)
• Machine learning models trained on fused data achieved an average improvement of 26.3% in R2 and an 11.8% reduction in nRMSE for SM retrieval

Contributions

• This study establishes a physically constrained, transferable, and task-adaptive paradigm for pixel-level remote sensing data fusion
• The proposed framework provides a robust technical pathway for the synergistic retrieval of SM and key eco-hydrological parameters using multi-source remote sensing data

Funding

• Not specified in the paper

Citation

@article{Li2026spectralconstrained,
  author = {Li, Yijie and Yao, Yifei and Sun, Shikun and Wang, Enhao and Liu, Rong and Ma, Jiale and Wu, Pute},
  title = {A spectral-constrained HVCE framework for adaptive optical-SAR fusion to enhance soil moisture retrieval},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136170},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136170}
}