Estimation of Soil Arsenic Content with Hyperspectral Remote Sensing

Area of Science:

• Environmental Science
• Remote Sensing
• Soil Science

Background:

• Arsenic pollution in soil poses a significant global environmental challenge.
• Accurate detection of soil arsenic is crucial for environmental protection and remediation efforts.
• Hyperspectral remote sensing offers a promising tool for monitoring heavy metal contamination in soils.

Purpose of the Study:

• To develop an efficient and accurate estimation model for soil arsenic (As) content using hyperspectral data.
• To address the challenges of complex nonlinear relationships and data redundancy in hyperspectral analysis for soil arsenic.
• To evaluate the performance of different modeling techniques for soil arsenic estimation.

Main Methods:

• Collection of soil samples and spectral reflectance data from Daye and Honghu, Hubei Province.
• Application of continuum removal (CR) and a coupled stable competitive adaptive reweighting sampling algorithm with successive projections algorithm (sCARS-SPA) for characteristic band selection.
• Development and comparison of Partial Least Squares Regression (PLSR), Radial Basis Function Neural Network (RBFNN), and Shuffled Frog Leaping Algorithm optimized RBFNN (SFLA-RBFNN) models.

Main Results:

• The sCARS-SPA method effectively reduced data redundancy and collinearity.
• The SFLA-RBFNN model demonstrated superior performance compared to PLSR and standard RBFNN.
• The sCARS-SPA-SFLA-RBFNN model achieved high prediction accuracy for soil As content across different land-use types.

Conclusions:

• The developed sCARS-SPA-SFLA-RBFNN model provides a robust and accurate method for estimating soil arsenic content.
• Hyperspectral remote sensing, coupled with advanced data processing and modeling, is a viable approach for monitoring soil arsenic pollution.
• This study offers a scientific and effective solution for assessing and managing arsenic contamination in soils.