Estimating the hydrodynamic parameters of Light Non-Aqueous Phase Liquid (LNAPL) is a fundamental pillar of contaminant transport studies in hydrocarbon-contaminated aquifers and a key factor in the design and management of remediation projects. However, direct measurement of these parameters through field and laboratory methods is typically costly, time-consuming, and often constrained by operational limitations. In this study, the efficacy of the Gene Expression Programming (GEP) model in estimating LNAPL hydrodynamic parameters within the Ray industrial zone was evaluated. Monthly data regarding LNAPL thickness, groundwater levels, and LNAPL levels and transmissivity—derived from the LNAPL Distribution and Recovery Model (LDRM)—were collected for three representative wells over the statistical period of 2006–2020. Modeling was conducted using GeneXproTools 5 software, and performance was assessed using R2 and RMSE metrics. Results indicated that the model demonstrates high precision in estimating LNAPL levels and transmissivity, with the coefficient of determination (R2) exceeding 0.99 in some instances. Conversely, the model’s performance in estimating hydraulic conductivity was suboptimal, which is attributed to the limited range of variations in this parameter and the relative uniformity of the aquifer sediments. The innovation of this research lies in applying the GEP method to derive explicit, practical mathematical relationships for estimating hydrodynamic parameters in the Ray aquifer. While highlighting the high efficacy of the GEP model in estimating dynamic parameters, the findings underscore the necessity of accounting for input data limitations in data-driven modeling

Keywords: Contaminated aquifers, Hydrodynamic parameters, LNAPL, GEP model.