A PROBABILISTIC APPROACH TO ASSESSING GROUNDWATER CONTAMINATION

The implementation of the EU Water Framework Directive (WFD) in Italy, produced a number of national guidelines and decrees for a sustainable water resources management and safeguarding. Within these guidelines, the design and realization of reliable groundwater monitoring well networks are prescribed and a methodology for classifying groundwater qualitative state is defined, fundamentally based on the crossed-evaluation of seven physical and chemical parameters. Consequently, the classification is achieved just in the monitored locations. Some problem can yet arise when punctual information needs to be spatialised. Substantially, spatializing a variable consists in estimating its value at an unmonitored location using, firstly, the neighbouring monitored values and, possibly, any other related, available information. This paper proposes a probabilistic approach, based on geostatistical techniques, for the assessment of groundwater contamination level according to quality classification defined by the Italian law. The Sequential Indicator Simulation (SIS) was applied to 35 punctual index values of groundwater quality evaluated in as many wells located in the Tavoliere plain, located in Southern Italy. Categorical, lithological data were also used as an auxiliary variable with the aim of increasing the precision of groundwater quality spatial prediction. In fact, a relationship between the quality index and the lithology characterizing any considered monitoring well was assessed in terms of frequency of occurrences. This relationship was used as a constraint in the simulation process. Finally, the uncertainty associated to quality index estimations was assessed by means of the standardised entropy of the local simulated distribution. Post-processing of one thousand simulations produced two types of maps: the first representing groundwater quality classes and the second showing the uncertainty associated to any estimation. These two maps, jointly used, could effectively support water managers' activities. The results showed a general good reliability of the methodology and allowed to classify water quality in the considered groundwater system.