Abstract. Groundwater recharge (GWR) is recognized to be a strategic hydrologic variable, necessary to estimate when implementing sustainable groundwater management, especially within a global change context. However, its simulation at the regional scale and for long-term conditions is challenging, especially due to the limited availability of spatially-distributed calibration data and to the rather short observed time series. The use of a superficial water budget model to estimate recharge is appropriate for this task. A reliable regional-scale estimate of GWR that can be updated relatively easily using widely-available data is essential for the implementation of long-term water use policies and is clearly lacking in southern Quebec (Canada; 36 000 km2). This study aims to test the ability of a spatially-distributed water budget model, automatically calibrated with river flow rates and baseflow estimates, to simulate GWR at a regional-scale from 1961 to 2017 in southern Quebec (monthly time step, 500 m × 500 m spatial resolution). The novelty of this work lies in the simulation of the first regional-scale GWR estimate for southern Quebec and in the development of a robust approach to implement a superficial water budget model at the regional-scale and for a long period. The HydroBudget model was specifically developed by a team at Université du Québec à Montréal for regional-scale simulation and cold climate conditions, and uses parsimonious input data (distributed precipitation, temperature, and runoff curve numbers). The model was regionally calibrated with river flows and baseflows (recursive filter on river flow data), and the automatic calibration procedure of the R package caRamel allowed a satisfying calibration quality (KGE = 0.72) to be reached. Across the study area and based on the exceptionally long spatialized time series, the simulated water budget was divided into 41 % runoff (444 mm/yr), 47 % actual evapotranspiration (501 mm/yr), and 12 % potential groundwater recharge (139 mm/yr). This partitioning was influenced by precipitation, temperature, soil texture, land cover, and topography. Groundwater recharge peaked during spring (44 % of annual recharge) and winter (32 % of annual recharge). A novel and particularly useful result from this work was to show that the seasonality of recharge was driven by the regional temperature gradient, with decreasing temperatures from west to east, and that winter GWR presented a statistically significant increasing trend since 1961 due to increased precipitation and warming temperatures. Another original contribution of this work was to show that at the regional scale, water budget models, such as HydroBudget, can be easily calibrated with river flow measurements and baseflows, and therefore represent a good option with which to acquire knowledge about regional hydrological dynamics. Being accessible, they are a useful approach for scientists, modellers, and stakeholders alike to understand regional-scale groundwater renewal rates, especially if they can be easily adapted to specific study needs and environments.
Precipitation isotope time series predictions from machine learning applied in Europe