GNGTS 2016 - Atti del 35° Convegno Nazionale

524 GNGTS 2016 S essione 3.2 contamination source that causes widespread salinization (Sciarra, 2015; Bonzi et al. , 2016). The dominant ions of deep wells are rich in sodium chlorides and bicarbonates of magnesium and calcium (Mg/Ca ratio is about 1.5) and significant are the amounts of Potassium. Very low is the concentration of sulphates. In fact, a new discipline of research projects has been developed in this direction in the last few years whose aim is to Manage Artificial Recharge activities (MAR projects: Nieto et al., 2012; Pezzi et al. , 2014). This means that their use do not focus only on the geometrical definition of aquifer boundaries but goes a step further towards monitoring the dynamic behaviour of the salinization/de-salinization process in strict relation to the modality being used for AR. Within this context, the main focus of the present work is to present and discuss the use of these techniques as a cost-effective tool for the monitoring of AR activities which constitutes one of the fundamental pillars of any MAR project. The geoelectrical survey. At Copparo’s test site, the subsurface around the artificial lake, once was a clay quarry, was investigated utilizing 10 2D ERT and IP profiles whose locations are shown in Fig. 1. Repeated measurements, although irregular, were executed on ERT profile No. 5 (Fig. 1) at certain periods in coincidence and following surface AR operations. The experimental data were collected using the ABEM SAS1000/ES464 geo-resistivity- meter (Sweden, http://www.abem.se/products ). This equipment acquires, simultaneously the resistance data and the IP information in the “time-domain” mode, i.e. by measuring the apparent chargeability. In all sites, the Wenner-Schlumberger electrode array was used which represents an acceptable compromise between lateral and vertical resolution. Electrode spacing ranged between 2.5 m to 3.0 m. Resistance data was acquired with two cycles of energization, while the IP were acquired using one time window of 100 ms length starting after a 10 ms delay, following current switch-off. The apparent resistivity and chargeability data were inverted using the RES2DINV (2012) commercial software that implements an inversion algorithm based on the smoothness-constrained least-squares method and uses the quasi-Newton approximation for optimization (Loke and Barker, 1996). The inversion strategy is essentially of the Occam’s type (deGroot-Hedlin and Constable, 1990). The inverted resistivity and chargeability 2D images showed RMS relative errors, describing the discrepancy between field and predicted apparent resistivity data, generally less than 2% and 1% (absolute RMS) for resistivity and chargeability inversion models respectively. This low fitting error was achieved thanks to the good quality of field data. Fig. 1 – Location map of Copparo, NE Italy, AR test site. Coordinates are in meters with respect to the European Datum 1950 (ED50). ERT-1: No. of the ERT profile, P9: piezometer, dot and circle: borehole (lithology), blue close curve: artificial lake approximate limits (ex-quarry for clay).