Estimating the nonaqueous-phase liquid content in saturated sandy soil using amplitude domain reflectometry

Groundwater contamination by nonaqueous-phase liquids (NAPLs) has become a serious environmental issue. Therefore, it is necessary to estimate the NAPL content (θ_NAPL) in saturated soil to detect and monitor the NAPL contamination in groundwater. The objective of this study was to investigate three estimation methods for the θ_NAPL in saturated sandy soil, namely, the noncalibration method (NCM), the permittivity regression method (PRM), and the best-fit α method (BFαM). These methods utilize the permittivity (K) measured by amplitude domain reflectometry (ADR) and two dielectric mixing models: a theoretical Maxell-de Loor (MD) model and an empirical α model. The NCM applies the measured K to the models directly. The PRM applies the measured K to the models indirectly using the regression equation between the measured and estimated K. The RMSEs of the NCM for the α model were 0.038 m³ m^-3 for light NAPL (LNAPL) and 0.015 m³ m^-3 for dense NAPL (DNAPL) when the constant α value was set at 0.5. The RMSEs of the MD model, however, were 0.090 m³ m^-3 for LNAPL and 0.070 m³ m ^-3 for DNAPL. Using the PRM, the RMSEs of the MD model became much better than those of the NCM. The RMSEs for θ^NAPL estimated by the BFαM were 0.007 m³ m^-3 for LNAPL and 0.018 m³ m^-3 for DNAPL. These results demonstrate that θ_NAPL was estimated easily and with sufficient accuracy with the α model and the PRM.