Yu et al. (2026) Soil Salinity Mapping from UAV-Borne Hyperspectral Imagery with Soil Moisture Correction

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

Identification

• Journal: Agronomy
• Year: 2026
• Date: 2026-09-15
• Authors: Haiye Yu, Muyan Yu, Ranzhe Jiang, Xin Zhang, Zhu Guo, Yaohui Fu, Xingbang Liu, Xingyu Sun, Bingze Li, Yuanyuan Sui
• DOI: 10.3390/agronomy16181812

Research Groups

• Department of Agricultural Engineering, University of California, Davis
• Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences
• School of Environmental Science and Engineering, Shanghai Jiao Tong University

Short Summary

This study evaluates the performance of six spectral transformation methods combined with machine learning algorithms for soil salinity inversion using drone-based hyperspectral remote sensing data. The results show that a Stacking ensemble model integrating these base learners achieves the highest accuracy and stability.

Objective

• Investigate the effectiveness of combining spectral correction techniques with machine learning algorithms for improving soil salinity estimation from UAV hyperspectral imagery.

Study Configuration

• Spatial Scale: Local to regional scale (field to 100 km2)
• Temporal Scale: Short-term to seasonal scale (days to months)

Methodology and Data

• Models used: Stacking ensemble model, K-Nearest Neighbors (KNN), Support Vector Regression (SVR), Multi-Layer Perceptron (MLP), XGBoost, Random Forest (RF)
• Data sources: UAV hyperspectral imagery, ground truth measurements of soil salinity

Main Results

• The Stacking ensemble model achieved the highest accuracy and stability among the evaluated models.
• The FDR + OSC–Stacking combination provided the best validation performance, with Rp2 = 0.87, RMSEP = 0.67 mS·cm−1, and RPD = 2.93.
• Compared to the Ref–Stacking model, Rp2 increased by 0.32 (from 0.55 to 0.87), while RMSEP decreased by 0.58 mS·cm−1 (from 1.25 to 0.67 mS·cm−1).

Contributions

• This study provides a novel approach for improving soil salinity estimation from UAV hyperspectral imagery by combining spectral correction techniques with machine learning algorithms.
• The results demonstrate the potential of using Stacking ensemble models for site-specific salinity management in precision agriculture.

Funding

• National Key Research and Development Program of China (2018YFC1503700)
• National Natural Science Foundation of China (41971301)

Citation

@article{Yu2026Soil,
  author = {Yu, Haiye and Yu, Muyan and Jiang, Ranzhe and Zhang, Xin and Guo, Zhu and Fu, Yaohui and Liu, Xingbang and Sun, Xingyu and Li, Bingze and Sui, Yuanyuan},
  title = {Soil Salinity Mapping from UAV-Borne Hyperspectral Imagery with Soil Moisture Correction},
  journal = {Agronomy},
  year = {2026},
  doi = {10.3390/agronomy16181812},
  url = {https://doi.org/10.3390/agronomy16181812}
}