Using water level and temperature time series to improve hydrogeological parameterization in a complex alluvial system

A new pumping station was designed in the northern high plain of the province of Padua (Veneto region, north-eastern Italy), aiming to reach an overall abstraction rate of about 2 m3/s, in order to relevantly contribute to the regional drinking water supply. Local unconfined aquifer is a highly permeable alluvial system, hydraulically connected to the Brenta river, one of the most important groundwater recharging sources of the entire hydrogeological basin, and the Camazzole lake, a former open-pit mine. This lake deepens below the water table and is directly connected to the surrounding phreatic aquifer and indirectly to the river, forming a 3-element hydraulic equilibrium. In order to evaluate the sustainability of the groundwater exploitation, this case study required an in-depth analysis of the hydrogeological resource, focusing on the estimation of hydraulic conductivity values and distribution. A numerical simulation was needed since the first step of the study, to plan the following field activities and provide a rough representation of the expectable drawdown in the pumped aquifer, even if the initial model had a very high level of uncertainty. Before the pumping tests no experimental data were available, so a homogeneous distribution of hydraulic conductivity was preliminarily assigned to the entire mesh, referring to a single bibliographic value available for the aquifer. After the analytical interpretation of pumping tests, different punctual values of hydraulic conductivity were estimated, but the parameter field was still very difficult to define, due to the complexity of the hydrogeological context and the non-uniqueness of the possible spatial interpolations. The availability of groundwater level observations at a larger scale allowed to calculate a set of hydraulic conductivity fields through the pilot points method, integrating the pumping tests results and extending aquifer characterization to a wider domain. The numerical model was finally calibrated with groundwater temperature monitored trends, reproducing the interaction between the lake and the phreatic aquifer through a heat transport simulation. The resulting hydraulic conductivity distribution has been considerably refined, especially at the interface between the lake and the aquifer, and the parameterization has been further validated using heat as a groundwater tracer.