Integration of Time-Lapse GPR and Hydrus-1D Modelling to predict Soil Hydraulic Parameters in Boreal Podzolic soil

Knowledge of the soil hydraulic parameters (SHPs) is crucial for many agricultural and environmental applications. However, current methods for directly measuring SHPs are time-consuming, labour-intensive, and often involve processes that disturb the soil. In this study, we combined time-lapse ground-penetrating radar (GPR) data collected during a Beerkan infiltration experiment with Hydrus-1D inverse modelling to non-destructively predict Mualem-van Genuchten (M-vG) parameters and simulate water flow dynamics at a Boreal podzolic soil in western Newfoundland, Canada. Calibration was carried out using the inverse modeling of the first infiltration experiments conducted both in the field and the laboratory, while evaluation was done through forward modeling of two additional experiments under both conditions. Resulting M-vG parameters (θr = 0.04037 cm3/cm3, θs = "0.22630" cm3/cm3, α = 0.04252 cm-1, n =1.66660, Ks = 0.00259 cm/sec) obtained from the inverse modeling of the GPR data in the field were comparable with estimates from Rosetta neural network prediction for our soil and other conventional methods. The evaluation results revealed a good agreement between the simulated and measured soil water content with a root mean square error of 0.0199  0.0365 cm3/cm3 and Nash-Sutcliffe modeling efficiency of 0.2955 ‒ 0.8273. The approach presented in this study offers a promising non-invasive method for large-scale characterization of soil water dynamics and SHPs in Boreal podzolic soils, yet further validation across diverse conditions and adoption of more advanced simulation models, such as Hydrus 2D/3D, could improve the accuracy of water flow simulation and provide spatial visualization of flow paths.