Sun et al. (2025) Season-Specific CNN and TVDI Approach for Soil Moisture and Irrigation Monitoring in the Hetao Irrigation District, China

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

Identification

• Journal: Agriculture
• Year: 2025
• Date: 2025-09-14
• Authors: Yinghao Sun, Dongliang Zhang, Ze Miao, Shaodong Yang, Quanming Liu, Zhongyi Qu
• DOI: 10.3390/agriculture15181946

Research Groups

Not explicitly mentioned in the provided text.

Short Summary

This study develops a year-round, field-scale framework for soil moisture retrieval and irrigation mapping in arid regions by introducing a season-stratified TVDI scheme and a multi-source inversion. The framework, leveraging Sentinel-1 SAR and Landsat data with a CNN regressor, successfully maps complementary seasonal irrigation patterns and provides operational evidence for water management.

Objective

• To develop a year-round, field-scale framework for retrieving soil moisture and mapping irrigation in arid irrigation districts, addressing limitations of static, single-season approaches and TVDI application during non-growing seasons.

Study Configuration

• Spatial Scale: Field-scale within the Yichang sub-district of the Hetao Irrigation District, China.
• Temporal Scale: Year-round, utilizing multi-sensor image time series from 2023–2024.

Methodology and Data

• Models used: Season-stratified Temperature-Vegetation Dryness Index (TVDI) scheme (based on LST–EVI feature space), Partial Least Squares Regression (PLSR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Convolutional Neural Network (CNN).
• Data sources: Sentinel-1 Synthetic Aperture Radar (SAR) imagery, Landsat imagery (features including Enhanced Vegetation Index - EVI), in situ topsoil moisture measurements, meteorological records, local cropping calendar, district statistics.

Main Results

• The CNN regressor achieved the highest accuracy for non-growing-season soil moisture retrieval (test R² ≈ 0.56–0.61), outperforming PLSR, RF, and XGBoost by approximately 12–38%.
• Integrated mapping revealed distinct seasonal irrigation patterns: spring irrigates 40–45% of farmland (e.g., 43.39% on 20 May 2024), summer peaks around 70% (e.g., 71.42% on 16 August 2024), and autumn stabilizes near 20–25% (e.g., 24.55% on 23 November 2024).
• Significant spatial contrasts in irrigation were observed, with intensively irrigated southwest blocks and drier northeastern zones.
• The combination of season-stratified edges and multi-source inversions enables reproducible, year-round irrigation detection at the field scale.

Contributions

• Developed a novel year-round, field-scale framework for soil moisture retrieval and irrigation mapping in arid irrigation districts.
• Introduced a season-stratified TVDI scheme based on LST–EVI feature space with phenology-specific dry/wet edges, extending TVDI applicability to non-growing seasons.
• Pioneered a non-growing-season inversion fusing Sentinel-1 SAR and Landsat features with advanced regressors (CNN showing superior performance).
• Provided operational evidence for refining irrigation scheduling, optimizing water allocation, and supporting drought-risk management and precision water governance.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{Sun2025SeasonSpecific,
  author = {Sun, Yinghao and Zhang, Dongliang and Miao, Ze and Yang, Shaodong and Liu, Quanming and Qu, Zhongyi},
  title = {Season-Specific CNN and TVDI Approach for Soil Moisture and Irrigation Monitoring in the Hetao Irrigation District, China},
  journal = {Agriculture},
  year = {2025},
  doi = {10.3390/agriculture15181946},
  url = {https://doi.org/10.3390/agriculture15181946}
}