Shan et al. (2025) Refined Leaf Area Index Retrieval in Yellow River Delta Coastal Wetlands: UAV-Borne Hyperspectral and LiDAR Data Fusion and SHAP–Correlation-Integrated Machine Learning

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

Identification

• Journal: Remote Sensing
• Year: 2025
• Date: 2025-12-23
• Authors: Chenqiang Shan, Taiyi Cai, J Wang, Yufeng Ma, Jun Du, Xiang Jia, Xu Yang, Fangming Guo, Huayu Li, Shike Qiu
• DOI: 10.3390/rs18010040

Research Groups

Not explicitly provided in the text.

Short Summary

This study developed and evaluated machine learning models using UAV-borne hyperspectral and LiDAR data fusion for accurate leaf area index (LAI) retrieval in coastal wetlands, demonstrating significant accuracy improvements over single-source methods and identifying key contributing features.

Objective

• To improve the accuracy and overcome spatial constraints of leaf area index (LAI) retrieval in coastal wetlands by utilizing UAV-borne multi-source remote sensing data (hyperspectral and LiDAR) combined with machine learning algorithms.

Study Configuration

• Spatial Scale: Coastal wetlands in the Yellow River Delta, China.
• Temporal Scale: Not explicitly provided in the text.

Methodology and Data

• Models used: Random Forest (RF), Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost).
• Data sources: UAV-borne hyperspectral sensors, UAV-borne LiDAR sensors. Features extracted: 38 vegetation indices (VIs) from hyperspectral data, 12 point cloud features (PCFs) from LiDAR data.

Main Results

• Multi-source feature fusion significantly improved LAI retrieval accuracy, with the Random Forest (RF) model achieving the highest performance (coefficient of determination (R²) = 0.968, root mean square error (RMSE) = 0.125).
• LiDAR-derived structural metrics and hyperspectral-derived vegetation indices were identified as critical factors for accurate LAI retrieval.
• The feature selection method integrating mean absolute Shapley Additive Explanations (|SHAP| values) with Pearson correlation analysis enhanced model robustness.
• The intertidal zone exhibited pronounced spatial heterogeneity in the vegetation LAI distribution.

Contributions

• Demonstrated the superior performance of UAV-borne multi-source remote sensing data fusion (hyperspectral and LiDAR) for high-accuracy LAI retrieval in complex coastal wetland environments, addressing limitations of traditional and single-source methods.
• Identified specific critical features (LiDAR structural metrics and hyperspectral VIs) contributing to accurate LAI estimation.
• Introduced an enhanced feature selection methodology combining Pearson correlation and SHAP values, improving model robustness.

Funding

Not explicitly provided in the text.

Citation

@article{Shan2025Refined,
  author = {Shan, Chenqiang and Cai, Taiyi and Wang, J and Ma, Yufeng and Du, Jun and Jia, Xiang and Yang, Xu and Guo, Fangming and Li, Huayu and Qiu, Shike},
  title = {Refined Leaf Area Index Retrieval in Yellow River Delta Coastal Wetlands: UAV-Borne Hyperspectral and LiDAR Data Fusion and SHAP–Correlation-Integrated Machine Learning},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs18010040},
  url = {https://doi.org/10.3390/rs18010040}
}