Data-driven and multi-approach sampling scheme optimization: the Alimini Lakes aquifer case

Due to the high wells drilling cost, monitoring sites are usually selected among existing wells; nevertheless, the resulting monitoring network must assure a good assessment of the main characteristics of the considered aquifer. Groundwater managers, need to find a good balance between two conflicting objectives: maximizing monitoring information and minimizing costs. In this paper, a couple of groundwater monitoring optimization methods are presented, related to the local shallow aquifer of the Alimini Lakes, located in Apulia (South-Eastern Italy) where a large number of existing wells have been pinpointed and the need of optimally reducing exists. The proposed methods differ each other for the required amount of prior information. The first proposed method, namely Greedy Deletion, just requires the geographical position of the available sites, while the second, the Simulated Annealing, also requires the knowledge of the spatial law of the considered phenomenon. The managerial need was to halve the number of monitoring sites minimizing the information loss.