Bai et al. (2025) Near Real-Time Reconstruction of 0–200 cm Soil Moisture Profiles in Croplands Using Shallow-Layer Monitoring and Multi-Day Meteorological Accumulations

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Agronomy
• Year: 2025
• Date: 2025-12-12
• Authors: Zhongrui Bai, Shujie Jia, Guofang Wang, Mingjing Huang, Wuping Zhang
• DOI: 10.3390/agronomy15122864

Research Groups

• Not specified in the provided text (typically associated with Agricultural Engineering, Hydrology, or Water Management departments).

Short Summary

This study developed a machine learning-based model to reconstruct deep-layer soil moisture (0–200 cm) using shallow-layer data and meteorological features. The approach achieves high predictive accuracy (R² up to 0.98), providing a low-cost alternative to expensive deep-probe monitoring for precision irrigation.

Objective

• To develop and validate a near-real-time reconstruction model for soil moisture profiles across the 0–200 cm depth using shallow-layer monitoring (0–20 cm, 20–40 cm) and multi-day accumulated meteorological drivers.

Study Configuration

• Spatial Scale: Agricultural fields in semi-arid regions.
• Temporal Scale: 2023–2025 (Field measurement period).

Methodology and Data

• Models used: Random Forest (RF), Gradient Boosting Decision Tree (GBDT), and Support Vector Regression (SVR).
• Data sources: Field measurements of soil moisture (0–200 cm) and multi-day accumulated meteorological features.
• Input Scenarios:
  • Scenario A: Surface moisture only.
  • Scenario B: Surface moisture + multi-day meteorological accumulation.
  • Scenario D: Dual-layer moisture (0–20 cm, 20–40 cm) + meteorological drivers.

Main Results

• Predictive Performance: Scenario D yielded the highest accuracy, with R² values of 0.96–0.98 for shallow layers (RMSE < 7 mm), 0.85–0.90 for mid-layers, and 0.76–0.84 for deep layers (140–200 cm).
• Meteorological Impact: Incorporating multi-day meteorological data improved R² by 0.05–0.08 compared to moisture-only inputs.
• Optimal Time Windows: The study identified depth-dependent lag effects for meteorological drivers: 5–10 days for shallow layers, 10–15 days for mid-layers, and ≥20 days for deep layers.
• Validation: Rolling validation confirmed high consistency in the 0–80 cm range (R² > 0.90, RMSE < 10 mm).

Contributions

• Cost Reduction: Eliminates the technical and financial burden of installing and maintaining deep-layer soil probes.
• Methodological Innovation: Establishes a clear relationship between meteorological accumulation windows and soil depth, enhancing the interpretability of machine learning models in hydrology.
• Practical Application: Provides a deployable, near-real-time pathway for precision irrigation and water resource management in water-scarce agricultural regions.

Funding

• Not specified in the provided text.

Citation

@article{Bai2025Near,
  author = {Bai, Zhongrui and Jia, Shujie and Wang, Guofang and Huang, Mingjing and Zhang, Wuping},
  title = {Near Real-Time Reconstruction of 0–200 cm Soil Moisture Profiles in Croplands Using Shallow-Layer Monitoring and Multi-Day Meteorological Accumulations},
  journal = {Agronomy},
  year = {2025},
  doi = {10.3390/agronomy15122864},
  url = {https://doi.org/10.3390/agronomy15122864}
}

Generated by BiblioAssistant using gemini-3-flash-preview (Google API)