Jing et al. (2025) Improve the accuracy of SAR-based soil moisture retrieval by coupling Bayesian inference and water cloud model

Identification

• Journal: Journal of Hydrology
• Year: 2025
• Date: 2025-12-23
• Authors: Haibo Jing, Linna Chai, Shaomin Liu, Diyan Chen, Shaojie Zhao, Zhongli Zhu
• DOI: 10.1016/j.jhydrol.2025.134826

Research Groups

State Key Laboratory of Earth Surface Processes and Disaster Risk Reduction, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China

Short Summary

This study proposes and evaluates a novel scheme, BIWCM, which couples Bayesian inference theory with the Water Cloud Model to improve SAR-based soil moisture retrieval accuracy by addressing the WCM's limitation of neglecting vegetation volume scattering. The BIWCM significantly enhanced retrieval accuracy over maize-covered areas (14.44% RMSE decrease) and marginally in bare soil areas, demonstrating its dynamic compensation ability for retrieval errors.

Objective

• To improve the accuracy of SAR-based soil moisture retrieval by coupling Bayesian inference theory with the Water Cloud Model (BI_WCM) to address the WCM's inherent limitation of neglecting vegetation volume scattering.

Study Configuration

• Spatial Scale: Middle reaches of the Heihe River Basin’s artificial oasis ecological hydrological experimental area.
• Temporal Scale: Not explicitly stated for data acquisition in the provided text.

Methodology and Data

• Models used: Bayesian inference theory, Water Cloud Model (WCM), BI_WCM (Bayesian Inference coupled with WCM), WCMHV (WCM estimates from HV-polarized Radar), WCMHH (WCM estimates from HH-polarized Radar).
• Data sources: Radarsat-2 images (Synthetic Aperture Radar), WATERNET observational dataset, airborne flight ground synchronous dataset.

Main Results

• The BI_WCM achieved overall R-values of 0.817 for bare soil and 0.816 for maize-covered areas.
• The BI_WCM yielded RMSEs of 0.033 m³/m³ for bare soil and 0.029 m³/m³ for maize-covered areas.
• Compared to WCMHV, the BI_WCM reduced the overall RMSE by 0.001 m³/m³ (p = 0.425, 95% CI = [0, 0.001]) in bare soil areas.
• For maize-covered areas, the BI_WCM reduced the overall RMSE by 0.005 m³/m³ (a 14.44% decrease, p = 0.003 < 0.05, 95% CI = [0.002, 0.008]) compared to WCMHV.
• The Bayesian inference error term (ε) was observed to gradually increase with NDVI, confirming its dynamic compensation ability for retrieval errors in soil moisture, particularly over maize-covered areas.

Contributions

• Proposes and implements a novel BI_WCM scheme that integrates Bayesian inference with the Water Cloud Model to overcome the WCM's limitation of neglecting vegetation volume scattering.
• Demonstrates a significant improvement in SAR-based soil moisture retrieval accuracy, especially over vegetated (maize-covered) areas, through quantitative validation.
• Introduces a dynamic compensation mechanism for retrieval errors via the Bayesian inference error term, which shows a positive correlation with NDVI.

Funding

Not explicitly stated in the provided text.

Citation

@article{Jing2025Improve,
  author = {Jing, Haibo and Chai, Linna and Liu, Shaomin and Chen, Diyan and Zhao, Shaojie and Zhu, Zhongli},
  title = {Improve the accuracy of SAR-based soil moisture retrieval by coupling Bayesian inference and water cloud model},
  journal = {Journal of Hydrology},
  year = {2025},
  doi = {10.1016/j.jhydrol.2025.134826},
  url = {https://doi.org/10.1016/j.jhydrol.2025.134826}
}