Stochastic groundwater recharge modelling in the Harricana river watershed, Québec, Canada

Quantifying groundwater recharge is a critical aspect of water resources management, yet it remains highly challenging due to the inherent complexity of physical processes and the uncertainties associated with hydrogeological conditions. The objective of this study is to estimate groundwater recharge and quantify its associated uncertainties in the Harricana River watershed. The spatial distribution of the recharge was initially estimated using the deterministic HELP model, implemented through the PyHELP interface, integrating a three-dimensional Quaternary geology model with soil, meteorological, and physiographic data. Saturated hydraulic conductivity (Ksat) was identified as one of the most sensitive parameters and was adjusted to calibrate PyHELP-simulated groundwater recharge against observed baseflow hydrographs. The hydrostratigraphic units were assigned probabilistic Ksat distributions using geostatistical simulation, generating multiple realizations of groundwater recharge. The preliminary analysis highlights the sensitivity of recharge to subsurface heterogeneity. Notably, the results suggest that certain hydrostratigraphic contexts are associated with greater recharge variability and, consequently, greater sensitivity to parameter uncertainty. These preliminary results highlight the importance of quantifying recharge uncertainty. Future work will extend this framework to probabilistic groundwater flow and contaminant transport modelling, incorporating multivariate uncertainty characterization and pumping-stress impacts.