THE PROBLEM OF As NBL DEFINITION IN THE DRAINAGE BASIN OF VENICE LAGOON (DBVL), NE ITALY

In northern Italy, especially into the alluvial aquifers of the Po and Venetian Plain, high concentration of arsenic is principally found in reducing environments and it often is associated with the presence of organic matter. Our study is focused on the area of the Drainage Basin to the Venice Lagoon (DBVL) that is a densely populated area notoriously affected by the groundwater arsenic pollution extending on approximately 2038 km2 on the provinces of Venice, Padua and Treviso. The UE’s Groundwater Directive (GWD 2006/118/EC) suggests an arsenic Contamination Threshold Values (CTV) equal to 10 µg/l. In addition, the UE BRIDGE project proposes to use the 90th percentile of the concentration data for the estimation of the Natural Background Level (NBL). Nevertheless, this method provides only a NBL value for the whole area. The dataset used in this work comes from the “A.Li.Na” study (founded by the Regional Environmental Agency) that was aimed to define the NBLs of As, Fe, Mn and NH4+ into the DBVL’s groundwater. Hydro-geochemical parameters were collected by 50 piezometers during four seasonal surveys from 2013 to 2014. The aim of this study is to improve the NBL concept using a sensitive case where the NBL is higher than the CTV. A geostatistical prediction of the arsenic spatial distribution appears a good procedure to spatially highlight zones exceeding the CTV. The relations among As, Fe and NH4+ suggested a cokriging method to spatialize arsenic distribution in the DBVL. The results allow highlighting different zones of arsenic concentration: i) areas with values lower than the CTV; ii) areas with values between the CTV and the median, calculated on the values over the CTV; iii) areas exceeding the median. Subsequently, the 90th percentile is applied to calculate a local NBL for each zone following the BRIDGE suggestions. Hence, the spatial predictions could be useful to define local NBLs related to the different portions of the whole predicted area improving the procedure of NBL calculation. In addition, if necessary (n° pz < 30), in these three classed areas the “A.Li.Na” sampling network should be increased to calculate the LNBLs.