IMAGE PROCESSING IN POLLUTION VULNERABILITY ASSESSMENT OF SICILY AQUIFERS

Abstract

It is generally acknowledged that vulnerability mapping represents an essential tool for territory management. Intrinsic vulnerability can be generally described as “the susceptibility of the aquifer systems, in their different components and in different geometric and hydrodynamics situations, to swallow and diffuse, also mitigating the effects, a fluid or water-transported pollutant, in such a way to produce impact on the groundwater quality, in space and time”. Assessment procedures of vulnerability consider a number of parameters, the estimation of which is usually difficult and very often insufficient to correctly represent aquifer’s features. As aquifers are closely influenced by numerous factors, their behavior is difficult to be predicted. For example, the most common methods to evaluate vulnerability, like SINTACS, assign scores and weights to a number of parameters, mainly depending on granulometric and textural properties, the estimation of which is frequently arduous and sometimes inadequate to perform reliable and accurate models of water systems. This paper shows the first results of a study about the expected correlation among selected maps of SINTACS system. They belong to the original set of seven parameters, utilized to perform a vulnerability cartography for a generic zone. Several SINTACS applications have been realized on sites of Italian Peninsula characterized by meaningful social-economic importance. Advanced modifications to the available standard are producing new protocols more appropriate for the surveyed areas. The aim was to validate the hypothesis to obtain similar information as respect to the original method by using just a selection of the seven maps. In order to assess this potential correlation resulting in superfluous information, a full statistical analysis has been performed on data from some urbanized areas of Sicily - namely Carini, Acquedolci and S. Agata di Militello - and results are reported in the present work. More specifically, Pearson's linear correlation coefficient matrixes have been computed and analyzed for each different couple of variables, maps being shifted both horizontally and vertically. In order to determine boundary values of those parameters indicating a similarity grade between a couple of identical maps, the correlation coefficient trend has been also investigated among the mentioned couple of maps, again shifted both horizontally and vertically. It results that in the analyzed data the investigated correlation actually exists. This strongly suggests to develop adjustments to the available standards, trying to create new protocols more adherent to the particular sites under investigation. In this particular case, for example, a considerable refinement of the SINTACS approach could be achieved using just five parameters instead of seven. In fact, it results that the map giving the vulnerability indexes - obtained after such parameters reduction - is statistically equivalent to that resulting through the original SINTACS approach. Software utilized in the shown procedure has been implemented in Matlab® environment. The numerical approaches have methodically permitted to realize a severe selection and reduction of parameters and to outline - in certain Sicily areas - the real importance of selected SINTACS factors, clearly influencing the transport processes of contaminants towards the aquifer. As a matter of fact, the investigated aquifers can be defined as complex ecological systems, their depiction seeming likely to be closely conditioned by a number of geometrical and hydrogeological parameters, which rarely portray the real hydrogeological behavior of the underground systems. Authors show how the statistical elaboration, applied to vulnerability settings, can contribute to an implementation of the hydrogeological passive scenery offered by complex aquifers as the considered ones.