Publication: Evaluation of light non-aqueous phase liquid (LNAPL) contamination detection by integrating electrical resistivity and machine learning.
Loading...
Date
2025-10
Authors
Ola Tareq, Abdulsayed
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This study introduces a novel, integrated, and non-invasive approach for detecting and predicting Light Non-Aqueous Phase Liquid (LNAPL) contamination in subsurface environments, addressing critical limitations of conventional monitoring techniques. Four controlled laboratory experiments were conducted using a custom-designed soil box filled with stratified site soil, where LNAPL was injected to simulate fresh spills. Its migration was monitored through electrical resistivity (ER) measurements and digital image analysis. Artificial Neural Networks (ANN) and K-Nearest Neighbors (KNN) were trained on two datasets: ER with LNAPL concentration, and image-derived spread area and velocity. A strong inverse correlation (R² > 0.9) was found between ER and LNAPL concentration using a power regression model. Image analysis enabled precise quantification of spread behavior, while AI models achieved high accuracy in predicting both concentration and spatial distribution. This integrated strategy enhances early-stage LNAPL detection and provides a reliable tool for environmental risk assessment and decision-making.