Han et al. (2025) Spatial series approach to estimate soil moisture over wheat fields from a single SAR image

Identification

• Journal: Agricultural Water Management
• Year: 2025
• Date: 2025-10-15
• Authors: Wentao Han, M.X. Wang, Yangyang Cao, Zhengdong Luo, Cui Zhou, Jianjun Zhu, Haiqiang Fu, Qinghua Xie
• DOI: 10.1016/j.agwat.2025.109883

Research Groups

• School of Civil Engineering, Xiangtan University, Xiangtan 411105, China
• School of Advanced Interdisciplinary Studies, Central South University of Forestry and Technology, Changsha 41004, China
• School of Geosciences and Info-Physics, Central South University, Changsha 410083, China
• School of Geography and Information Engineering, China University of Geoscience (Wuhan), Wuhan 430074, China

Short Summary

This study introduces a spatial series approach (SSA) to estimate soil moisture (SM) over wheat fields using a single Synthetic Aperture Radar (SAR) image, leveraging backscattering intensity ratios (BIRs) to implicitly account for vegetation and roughness effects. The method achieves stable SM retrieval with a root mean square error (RMSE) of approximately 10% and a correlation coefficient (R) exceeding 0.7, offering a robust solution for SM monitoring in scenarios with limited time-series SAR data.

Objective

• To decouple soil moisture (SM) from confounding factors such as vegetation coverage and surface roughness by proposing a novel spatial series approach (SSA) that utilizes backscattering intensity ratios (BIRs) from a single Synthetic Aperture Radar (SAR) image.
• To expand the observational dimensions for SM retrieval and mitigate the limitations imposed by the low temporal resolution of traditional time-series SAR data.

Study Configuration

• Spatial Scale: Agricultural fields in Winnipeg, Manitoba, Canada (49°53’N, 97°09’W), predominantly cultivated with wheat. UAVSAR data with a spatial resolution of 5.0 meters (range direction) × 7.2 meters (azimuth direction).
• Temporal Scale: Four distinct acquisition dates in 2012: June 17, June 23, June 25, and July 13, representing various phenological stages of wheat.

Methodology and Data

• Models used:
  • Antropov model (for volume scattering contribution and canopy orientation estimation)
  • Dubois empirical model (adapted for ground surface scattering, incorporating a vegetation attenuation term)
  • Water Cloud Model (WCM) (for modeling vegetation attenuation)
  • Topp model (for converting dielectric constant to soil moisture)
  • Kim and van Zyl linear function (used for comparative SSA derivation)
  • X-bragg model (for dual-polarization roughness estimation)
• Data sources:
  • L-band quad-polarimetric UAVSAR data (acquired at incidence angles ranging from 25° to 65°)
  • Concurrent in-situ field measurements of soil moisture (ranging from 0.10 to 0.70 cubic centimeters per cubic centimeter)

Main Results

• The proposed Spatial Series Approach (SSA) demonstrated consistent performance across all acquisition dates, yielding a Root Mean Square Error (RMSE) of approximately 10% (e.g., 9.96% to 10.67% for Dubois-based full polarization) and correlation coefficients (R) consistently above 0.7 (e.g., 0.72 to 0.77 for Dubois-based full polarization).
• The SSA significantly outperformed the conventional Dubois model, which exhibited RMSE values from 9.10% to 17.06% and R values from –0.64 to 0.72, often failing under specific conditions.
• The inversion success rate for PolSAR data using SSA was high, exceeding 94%, even in vegetated areas. Inversion failures were primarily localized at agricultural field edges due to mixed-pixel effects.
• The method showed stable performance across different polarimetric channels (HH, VV, and full polarization), with full polarization generally providing superior accuracy. VV polarization typically outperformed HH polarization due to reduced volume scattering and enhanced ground penetration.
• The stability of the SSA was confirmed under varying parameter thresholds and minimum spatial series point requirements, showing only minor fluctuations in accuracy. While increasing the number of spatial series points from 5 to 20 generally improved accuracy, it also led to a reduction in the retrieval success rate.

Contributions

• Introduces a novel spatial series approach (SSA) for soil moisture (SM) retrieval from a single Synthetic Aperture Radar (SAR) image, effectively addressing the limitations of low temporal resolution inherent in traditional time-series methods.
• Decouples the SM signal from confounding factors like vegetation and roughness by implicitly compensating for their effects using backscattering intensity ratios (BIRs) derived from spatially homogeneous points, thereby avoiding explicit and potentially biased parameter estimation.
• Establishes a new observational dimension for SM monitoring, providing a viable solution for rapid response applications and scenarios where multi-temporal SAR data are unavailable or sparse.
• Demonstrates operational versatility and compatibility with various scattering models (e.g., Dubois, Kim) and sensor configurations (full-polarimetric, dual-polarimetric).
• Lays the groundwork for a scalable spatiotemporal-series inversion framework, promising enhanced accuracy for large-scale SM retrieval by integrating the strengths of spatial density and temporal continuity.

Funding

• National Natural Science Foundation of China (NO. 42501462)
• Hunan Provincial Natural Science Foundation (NO. 2025JJ60236)

Citation

@article{Han2025Spatial,
  author = {Han, Wentao and Wang, M.X. and Cao, Yangyang and Luo, Zhengdong and Zhou, Cui and Zhu, Jianjun and Fu, Haiqiang and Xie, Qinghua},
  title = {Spatial series approach to estimate soil moisture over wheat fields from a single SAR image},
  journal = {Agricultural Water Management},
  year = {2025},
  doi = {10.1016/j.agwat.2025.109883},
  url = {https://doi.org/10.1016/j.agwat.2025.109883}
}