Estimating Soil Arsenic Contamination by Integrating Hyperspectral and Geochemical Data with PCA and Optimizing...

Area of Science:

• Environmental Science
• Geoscience
• Remote Sensing

Background:

• Soil arsenic contamination poses significant environmental and health risks.
• Accurate and efficient monitoring techniques for soil arsenic are crucial.

Purpose of the Study:

• To develop a novel multi-source data fusion approach for enhanced hyperspectral inversion of soil arsenic concentrations.
• To integrate dimensionality-reduced spectral data with correlated soil components.

Main Methods:

• Principal Component Analysis (PCA) was used for hyperspectral data dimensionality reduction.
• Evaluated Partial Least Squares Regression (PLSR), Artificial Neural Networks (ANN), and Random Forest (RF) models.
• Compared four input variable combinations, including original/reduced spectral data and soil components.

Main Results:

• The Random Forest (RF) model, using PCA-reduced spectra and soil components, achieved the highest inversion accuracy (R² = 0.86).
• This multi-source data fusion approach significantly outperformed simpler models like PLSR (R² = 0.75).

Conclusions:

• The study confirms the effectiveness of multi-source data fusion for improving soil arsenic estimation accuracy.
• The Random Forest model demonstrates superior capability in handling complex, high-dimensional data for environmental monitoring.