Remote Sensing-Assisted Characterization of Contaminated Soils for Sustainable Geotechnical Site Investigation and Environmental
Abstract
Soil contamination resulting from industrial effluent discharges, mining activities, agrochemical overuse, and improper hazardous waste disposal poses severe geotechnical and environmental challenges worldwide. Traditional geotechnical site investigations rely on localized intrusive sampling, soundings, and laboratory testing, which are labor-intensive, costly, spatially limited, and prone to missing localized contaminant hotspots. This study evaluates an integrated multi-scale remote sensing framework—combining satellite multispectral imagery (Sentinel-2, Landsat-9), Unmanned Aerial Vehicle (UAV)-borne hyperspectral imaging (VNIR/SWIR, 400–2500 nm), and ground-based spectroradiometry—to assist non-destructive characterization of contaminated soils for sustainable site investigation and remediation engineering. By coupling spectral unmixing, partial least squares regression (PLSR), and random forest (RF) machine learning algorithms with geostatistical kriging, the framework delineates spatial contaminant plumes, estimates heavy metal (Pb, Cd, Zn, Cr) and total petroleum hydrocarbon (TPH) concentrations, and maps altered soil geotechnical properties (plasticity, shear strength degradation, and hydraulic conductivity). Field validation across three industrial brownfield sites demonstrated high predictive accuracy for heavy metals (R² = 0.88–0.93, RMSE = 28.4–34.2 mg/kg) and TPH (R² = 0.86, RMSE = 142 mg/kg). Integrating remote sensing screening reduced required physical borehole locations by 42%, decreased field investigation carbon footprint by 38%, and optimized remediation design. The proposed methodology establishes a transformative, sustainable paradigm for brownfield redevelopment, environmental forensics, and geoenvironmental site risk assessment.
KEYWORDS: Remote Sensing; Soil Contamination; Hyperspectral Imaging; Geotechnical Site Investigation; Environmental Remediation; Machine Learning; Heavy Metals; Total Petroleum Hydrocarbons (TPH).
KEYWORDS: Remote Sensing; Soil Contamination; Hyperspectral Imaging; Geotechnical Site Investigation; Environmental Remediation; Machine Learning; Heavy Metals; Total Petroleum Hydrocarbons (TPH).
Full Text:
PDF 137-150Refbacks
- There are currently no refbacks.