Retrieval of chromium and mercury concentrations in agricultural soils: Using spectral information, environmental covariates, or a fusion of both?

Abstract

The universal contamination of arable land with potentially toxic elements (PTEs) poses a severe threat to food security and jeopardizes worldwide efforts to meet the United Nations Sustainable Development Goals (SDGs). How to obtain more reliable concentrations of PTEs in regional agricultural soils is a priority problem to be solved. Multispectral satellite remote sensing, with its advantages of high spatial and temporal resolution, broad coverage, and low cost, offers the potential to acquire spatial distribution of PTEs in agricultural soils over large areas. However, owing to the complexity of soil environments and the insufficiency of spectral information, the mechanism for retrieving concentrations of PTEs in agricultural soils via multispectral satellites is not yet clear, and the accuracy needs to be improved. In this study, we aimed to assess whether employing a fusion of spectral information and environmental covariates, results in more accurate retrievals of PTEs, specifically chromium (Cr) and mercury (Hg), in croplands than does employing spectral information alone. Three machine learning algorithms—kernel-based support vector machine (SVM), neural network-based back propagation neural network (BPNN), and tree-based extreme gradient boosting (XGBoost)—were developed to retrieve Cr and Hg concentrations in agricultural soils. The results showed that the fusion of spectral information and environmental covariates combined with the XGBoost model performed best in retrieving both Cr and Hg concentrations in agricultural soils with coefficient of determination (R2) values of 0.73 and 0.74, respectively. Environmental covariates were important variables for determining Cr and Hg concentrations in agricultural soils, but the ability to retrieve these element concentrations by utilizing spectral information alone was limited. High Cr concentrations occurred in central towns and southern hilly mountains. High Hg concentrations were detected in the northwestern region and southern hilly mountains. The potential of fusing spectral information and environmental covariates to precisely retrieve PTE concentrations in agricultural soils can serve as a reference for agricultural soil health information monitoring worldwide.

Keywords

• Potentially toxic elements (PTEs)
• Sentinel-2
• Environmental covariates
• Machine learning
• Farmland