3D numerical modeling of the Pisa coastal plain (North-Western Tuscany)

Since prehistoric times, the Pisa coastal plain, lowest portion of the Arno and Serchio Rivers basins, was the preferred site for human settlements. It was therefore involved in important drainage and hydraulic reclamation works. They began before Roman Empire and produced an almost entirely artificial hydrographic network, including significant modifications to large part of the main river courses. In the last two centuries, population quick rise, urban centers, tourist and industrial activities growth raised both water and dry land demand. Irrational water pumping near the coast produced severe seawater intrusion in the coastal aquifers. 3D finite differences numerical simulations (Visual Modflow 4.2) were applied to the phreatic aquifer of the Pisa coastal plain (Wide Area Model - WAM), with dedicated zooms on the densely inhabited areas of Marina di Pisa and Tirrenia. In particular, the WAM has his boundary at the Arno River to the North, the Navicelli Canal to the East, the Scolmatore Canal to the South and the seashore to the West. The WAM output was used as boundary condition to implement the zoomed models of Marina di Pisa and Tirrenia (Visual Modflow Seawat code). The work required a first step of acquisition, systematization and processing of bibliographic and field data, in order to build the geological and hydrogeological conceptual model; the conceptual model was then translated into numerical input. The implementation of the numerical model based on the hydrostratigraphic structure, rainfall pattern, thermometric and piezometric data collected from October 2007 to December 2009 (calibration stage), as well as on the unconfined and first confined aquifer salinity data logs, and on the main hydraulic parameters (hydraulic conductivity, porosity, storage coefficient) of the studied aquifer. The unconfined aquifer, 10-15 m thick, is hosted in sandy coastal dunes, characterized by medium-high hydraulic conductivity (K ≈ 10-2 cm/s) and overlies an aquiclude (K ≈ 10-7 cm/s) generally constituted by clay and silty clay. Besides numerical modeling the correlation between rainfall and piezometric level data was analyzed in order to identify the phreatic aquifer recharge time. For almost all the monitored piezometers, a high correlation coefficient (0.75<R2<0.85) between rainfall and water table data was detected for a span of 75 days. The identified correlation helped the model simulations interpretation. The WAM was implemented and calibrated in order to analyze the water flow system, while the zoomed models of Tirrenia and Marina di Pisa were carried out and calibrated to calculate the extent and saline content of the seawater intrusion. The results reached a significant statistical correspondence to observed data sets. Regarding Marina di Pisa and Tirrenia models, simulation was also performed to forecast possible future scenarios relating the water table trends and seawater intrusion. Both models show a modest degree of seawater intrusion near the shoreline. In particular, in the Marina di Pisa area, close to the dune system, the seawater intrusion is just a thin level below a thick freshwater body. On the other hand, in the Tirrenia area the salt water completely saturated the unconfined aquifer. In the internal portion of the study area, between the dune system and the Navicelli Canal, in ancient times characterized by lagoons, marshes and swamps, the salt content raises, from high to very high for the whole thickness of the aquifer. In conclusion, the simulation models show the important role of the natural dune system as freshwater pseudo-hydrostatic barrier, effectively contrasting the seawater intrusion in the unconfined costal aquifer. Moreover, simulations identified the main cause of the seawater intrusion in massive pump driven dewatering by the artificial drainage network, which causes a significant lowering of the water table, from 1.70 to 2.50 below the sea level.