Data-Driven approach to groundwater level prediction for improving water resource management: the Brenta aquifer system case study (Veneto, Italy)

In the context of climate change, this work aims to model the piezometric levels of the foothill aquifer located in the middle-high Brenta river plain (Veneto, Italy) to support managing a groundwater system that provides drinking water for most of the Veneto Region. Using a Data-Driven approach, predictive Multiple Linear Regression Models were developed for the piezometric level at different wells, and scenarios of groundwater level evolution were achieved under dry periods. Results highlighted the high sensitivity of the aquifer to climate extremes, as well as the need to plan actions for mitigating the effects on such a strategic water supply system. Groundwater hosted in the foothill aquifer represents an important resource. However, these systems are highly sensitive to the variation of Meteo-climatic regimes. At the same time, the exploitations can lead to excessive groundwater drawdown and consequent threats of water scarcity. The Data-Driven approach adopted using long time series of meteorological, hydrometric and piezometric data can represent a valid example in these terms. The groundwater level evolution has been well-reproduced by these models. The equations describing models show the close dependence of groundwater from the Brenta River and the high sensitivity of the aquifer to meteo-climate regimes. Given this sensitivity, the forecast of groundwater level evolution under a dry period, similar to 2022, was performed. Results point out a progressive drawdown of groundwater level. These predictive models can be useful for local authorities to maintain these levels over specific critical values.