A Contribution to the Hydrogeological Knowledge of the Cremona Aquifer System and to the Explotation of new Water Resources/ Contributo alla conoscenza idrogeologica del sottosuolo di Cremona ed al reperimento di nuove risorse idropotabili

This paper reports the results of some hydrogeological and hydrochemical investigations carried out on the aquifers where the new Eastern Cremona well field is to be built. The Cremona aquifer system is a layered one, and has low permeability levels acting as aquicludes which separate the confined aquifer levels – exploited for drinkable purposes - from each other. In the Cremona area, such as in the entire Lombardia Plane, fresh water rests on salt water which was originally stored in the marine sediments which form the deeper levels of the Lombardia Plane. In the studied area the interface between fresh and salt water ranges between a depth of 400 and 500 meters. The new Eastern well field is part of a wider project, consisting in two new well fields, with which the company supplying water to Cremona (AEM) wants to replace the existing wells, in order to meet the new water quality requirements of the Law n° 31 of February 2nd 2001. The eastern well field is supposed to supply about 450 l/s and should consist of ten new wells which will join the Postumia well, located in the same area, and currently in operation. Up to now four piezometers (Pz1, Pz2, Pz3 and Pz4) and two wells (A and B) have already been drilled. The piezometers filters have been located at different depths, so that each one can be fed by different aquifers. This has enabled every single aquifer to be characterized both from the hydrogeological and from the hydrochemical point of view. The A and B wells are also fed by different aquifers, as shown in figure 2. The correlation between the piezometers and wells stratigraphyies has enabled the fundamental stratigraphic sequence of the area to be defined. Multi-step and long-term pumping tests have been carried out on both A and B wells: the piezometric variations have been measured both inside the wells and on the surrounding piezometers. This enabled the transmissivity (T) and the storage coefficient (S) of each aquifer to be estimated (T=10-2 m2/s; S=10-3÷10-5) and the well losses and formation losses of each well to be defined according to the Jacob-Rorabaugh equation (sw = BQ + CQn)(FIGG. 4, 5, 6 and 7). The pumping tests made it evident that, for a discharge of 0.057 m3/s, the drawdown recorded in well A (≈ 9 m) is about three times higher than the drawdown recorded in well B (≈ 3 m), although the non-linear losses are comparable with each other. This indicates the aquifer layers feeding well B to yield more water than the ones feeding well A, as confirmed by the slightly lower transmissivity and storage coefficient detected in well A. The hydrochemical analysis recorded the presence of some “undesiderable” chemicals such as Arsenic, Manganese, Ammonia and Iron in relevant concentrations, close to - or higher than - the maximum allowed concentration (VMA). Since these chemicals are found naturally in the Cremona aquifers at deep levels also, a water treatment plant will be necessary. In order to have an idea of what the aquifer behaviour would be with the full Eastern Well Field in operation, a simplified model of it was built by using MODFLOW. The layered aquifer was represented by means of 14 horizontal strata, on the basis of the stratigraphic data,. The grid was built in order to allow the piezometric variations around each well and each piezometer to be represented correctly and wide enough (10000×20000 m) for the depression cone generated by the wells not to reach the boundaries. The hydrogeological parameters were assigned to each layer on the basis of the pumping tests results, and the boundary conditions were set in order to simulate the natural groundwater flow directed southward, toward the river Po. The B well, which is fed by layers with better hydrogeological characteristics, was placed in the grid centre. The model was calibrated in steady state on the basis of the data from the long-term pumping test carried out on well B, with a discharge of 0.057 m3/s. The data of both the well B and the piezometers 2 and 3 were used to calibrate the model. A good simulation of the field data was obtained by slightly changing the conductivity values of the layers feeding well B. Once the model was calibrated a new steady state simulation was run, with all of the 10 wells in operation with a discharge of 0.057 m3/s each. In this simulation all of the wells had the same characteristics as well B. This new simulation showed that, under these conditions, the drawdown in well B would be about 4 meters higher than the drawdown recorded when the only well B is in operation, which means that the cones of depression of the wells interfere with each other. Both the results of the model simulation and the data collected during the pumping tests, indicate that the problem of the salt/fresh water interface rise is to be carefully evaluated. According to the Ghyben Hertzberg equation, which can be demonstrated to be valid also for horizontal interfaces, the interface rise is about 40 times higher than the well drawdown, if we consider the salt water density to be 1.025 g/cm3 and the fresh water density to be 1.0 g/cm3; in fact, the relationship between interface rise and well drawdown has the same order of magnitude even when considering density values slightly different from these. The Ghyben Hertzberg equation indicates, therefore, that the interface rise would be of about 240 m if the well field pumping scheme will be the same as in the model simulation, and it would be as high as 120 m even with the only well B in exercise. It has to be pointed out that the time required for the interface to rise this much is probably rather long, and it is strongly influenced by the presence of low permeability levels. Nevertheless the information collected up to now indicate that , if the well field is going to be built and managed according to the model scheme, and if the only hydrogeological levels exploited by well B will be used, sooner or later the wells will pump salt water. It has to be pointed out that, according to the Dagan & Bear criterion on horizontal interfaces, if the interface rise Z is as high or higher than 1/3 of the distance between the well bottom and the static interface L, the interface does not come to an equilibrium but it continues to rise indefinitely. In order to avoid or limit these problems a way could be drilling the next wells in such a way that they exploit at the same time all of the four productive aquifers levels which feed well A and well B. This would allow the global transmissivity of the system to be higher so that – for a given discharge - the drawdown would be lower. In order to quantify this fact the model was newly run after having added two filters intervals to each well, at the same depths as in well A. The simulation was run again with all of the wells pumping at 0.057 m3/s, and the results showed that, in these conditions, the maximum drawdown would be about 3.5 rather than 6 meters, as it was in the former simulation. Although this drawdown is still not safe in terms of upcoming, it would limit the problems, which could be definitively solved by planning to pump at intervals, in order to allow at least a partial recovery of the piezometric level. It has to be pointed out that, the upcoming takes generally decades to occur, and this should allow the proper solution to be determined by changing the pumping and/or the well field scheme, once new data will be available.