GNGTS 2018 - 37° Convegno Nazionale

666 GNGTS 2018 S essione 3.2 GEOLOGICAL SETTINGS CHARACTERIZATION OF A CONTAMINATED AREA IN AN URBAN CONTEXT: ASSIMILATION OF ARCHIVES, GEOTECHNICAL AND GEOPHYSICAL DATA B. Mary 1 , J. Boaga 1 , L. Busato 2 , G. Cassiani 1 Dipartimento di Geoscienze, Università degli Studi di Padova, Padova, Italy Centro Interdipartimentale di Ricerca “Ambiente” (CIRAM), Università degli Studi di Napoli Federico II, Napoli (NA), Italy Introduction. The field site is a metallurgic industry located in the region of Friuli-Venezia Giulia (north East Italy), in an urban context which takes a surface of approximately 34.500 m2. Before turning off the main plating activity, the manufacture rejected contaminants into the soil via the open drainage, nowadays still allowed in that area. Soil contamination rate and diffusion depends on the geological features and hydrology of the site. Information about the hydro-geological context of the region is essential but limited. Presence of paleo channels passing through the factory was assumed (this has been further investigate using geophysical interpretation of the cross section). In 2002, a first geophysical (18 profiles inside the factory) and geotechnical survey (more than 50 wells in the south of the factory) had been done to characterize the geological settings of the site. First results highlighted that the site is characterized by thick gravel deposits interrupted by the presence of an impermeable thin clay layer at 7 m depth, which prevents the contaminants to reach the deeper main aquifer. Nevertheless, this clay layer resulted to be discontinuous with several interruptions that threaten the local seal. A short-term remediation plan has been set up using pumping stations to collect and treat the waste water infiltrated in the south part of the manufacture. There is a need for non-destructive imaging of the clay layer discontinuity in order to propose solution such as the implementation of cost remediation devices such as new trench drainage. Objectives. Our study overcomes the limits identified after the 2002 prospection with the following concrete goals: - Firstly, there is a need to better define the clay layer discontinuities. A combination of modelling and accurate constrained inversions give new insights to a better understanding of the electrical resistivity variations observed. Calibration of the clay layer electrical behaviour, was also investigate though the implementation of two cross hole ERT devices monitoring under changing condition of water table. - Secondly, there is a need to compute a 3 dimensional model of the entire manufacture area and of its boundaries (in particular to build the initial model of hydrological simulation). Due to the non-regular spatial location of the geophysical and geotechnical data, new ERT data were collected. In order to use all the available source of inversion possible we aimed to fusion and interpolate the data using a geostatistical analysis. The originality of the study relies on the use of transversal concepts to geophysical imaging such as data fusion, geostatistical interpolation, cluster analysis and data assimilation via hydrogeophysical modelling. Results. Documentation from archives allowed us to define a plan for the new acquisition. The wells showed that the soil is horizontally layered with a first layer often composed of gravel. When existing, the clay layered of about 1m thick is located around 7m of depth. Electrical resistivity tomography data from the 2002 acquisition showed that the position of the clay layer correlated with a high electrical conductivity anomaly is difficult to identify. Its depth is rather well defined but its tick is overestimated. Identical feature has been observed for the cross sections acquired during the new campaigns in 2018 using a simple smooth inversion (Binley, 2013). Only the cross-hole prospection using boreholes electrodes was able to derive the same information than the well data. Monitoring the cross-hole sections under the influence of a changing water level table did not show any change in the clay layer identification (although the water table did not change significantly). From these results and additional simulations, it appeared essential to process ERT sections using a dedicated constrained inversion dictated