In Functional Urban Areas, management plans need to take into consideration both point sources (PS, corresponding to areas releasing plumes of high concentrations) and multiple point sources (MPS, consisting in a series of unidentifiable small sources clustered within large areas) that cause diffuse groundwater contamination. For the former category, according to the Decree n°152/2006 in case they are suspects linked to historical analysis or evidence, if the analytical results overtake the limit concentration (1.1 µg/l for Tetrachloroethene in the italian law), the site is “potentially contaminated”.
Tools developed within the AMIIGA Project (WEBGIS, multivariate and cluster analysis) were applied in order to reconstruct the historical activity in a specific spatial and temporal context. In order to reconstruct main groundwater flow direction and advective transport of the contaminant in a pilot area in the NW part of Milano FUA, a numerical multi-layered transient model was implemented and calibrated with Modflow. Withdrawal from 4997 wells has been considered as a major source of uncertainty in contaminant directions (PEST). The deterministic approach is not able to consider the uncertainties due to calibration parameters and targets affected by data entry error.
Adopting in PEST a Nullspace Monte Carlo (NSMC) analysis, several sets of K- fields were generated, all respecting the measured transient head targets. Considering the effect of heterogeneity in K-distribution within the aquifer, using MODPATH and placing a number of particles as starting points in a suspected contamination site, 400 forward MODPATH runs were performed starting by a stochastic set generated by the NSMC and minimizing the objective function (composed by head targets in monitoring wells).
Collecting the particle positions in each cell of the multi-layered aquifer for the most suitable realizations (selection was based on acceptable threshold objective function), the stochastic forward tracking technique was able to obtain: 1) probabilistic map of source location and time of source activation 2) wells likely to be impacted downstream of suspected source and 3) correlation between time frequency of passing particles and observed concentration time-series in the suspected wells.
Identification of the sources of groundwater contamination is crucial to enable remediation of contaminates sites, where groundwater concentrations are higher than the threshold limit. Identification is always very complex and very difficult to be solved in old industrial areas where contamination can be very dated. Following a criterion of the “most probable than not”, a probabilistic methodology was developed to quantify the sources of uncertainty affecting groundwater flow over time (abstraction rate from wells and contamination spill in the area) and the uncertainties linked to hydrogeological model (K and vertical discretization). The obtained maps correlate the historically contaminated wells with the “potential source areas”, thus enabling public authorities to focus their investigations of actual contamination sources.