Numerical modeling for groundwater protection in the Venetian plain between the Brenta and Piave Rivers

The Ph.D. project tackled the scientific challenges that a water utility company in the northeast of Italy, Alto Trevigiano Servizi, must face in the elaboration of the Water Safety Plan (WSP), which is the most effective preventive tool to ensure good quality water and consumers health protection. The WSPs guidelines were defined by the World Health Organization and were subsequently implemented in a European Directive and Italian law. The thesis, after an introduction on the scientifical issues, started with the description of the work done to reproduce in CATHY the model that the PhD student Tommaso Trentin built using the software FeFlow. The study area has an extension of around 900 km2 and is delimited to the north-east by the Piave river, to the west side by a flow line parallel to the Brenta river, while the southern boundary is closed by the Risorgive area, and the North boundary by the Montello and colli Asolani. The north part is characterized by an undifferentiated aquifer, while the southern part hosts a multilayer system with 8 confined aquifers. Some modifications, e.g., the mesh refining, the sensitivity analysis, were implemented in the model to try to improve its performance. Also, the soil conductivity of the shallowest soil layer (1 m) was changed following the indications of Carta della permeabilità dei suoli from ARPAV site and the boundary conditions of the norther part of the domain were better defined. Before the calibration step, the initial mesh that hosts the multilayers systems of 8 aquitards and 8 aquifers was cut at the bottom of the first unconfined aquifer. This allowed to speed up the calibration and focus on the aquifer directly influenced by the atmospheric boundary conditions and subject to recharge variability. The calibration was performed alternating FePESt and CATHY. FePEST, having already implemented the PEST algorithm, allowed to easily implement the pilot points method that in CATHY would have require too much time. Both the bottom of the unconfined aquifer and the hydraulic conductivity field were calibrated. The improvement in terms of RMSE was relevant, the errors being reduced to 1/3. Once the calibrated model was obtained, also a validation step was performed. The resulting model allowed us to investigate an irrigation variation scenario, planned in compliance with the European directive indication, to save water: currently a large area of the domain is interested by flood irrigation considered no more sustainable, since it requires a large amount of water. The scenario considered a switch to sprinkler irrigation only. The results show a slight groundwater head decrease in the wells located in the area affected by the irrigation technique conversion. This result was confirmed by the difference of the total cumulative recharge over the domain in case of sprinkler and flood irrigation and sprinkler irrigation only. The model seems to be not particularly affected by the irrigation modification but more sensitive to the hydraulic conductivity values: a map of the mean distribution of the recharge shows that the larger fraction of the recharge occurs where hydraulic conductivity is larger. Parallelly to the continuation of this project, also a study on the analysis of numerical dispersion affecting CATHY model was carry out. This study will be useful for future simulations on vulnerability to contaminations that require an accurate solute transport modeling. Due to lack of time it was not possible to investigate the contaminants transport phenomenon in the area of study to accurately define the wells’ head protection areas, important part of the WSPs, but the preliminary results obtained from the model we built can be considered a good starting point for future transport studies.