GNGTS 2018 - 37° Convegno Nazionale

GNGTS 2018 S essione 3.2 667 spatially by the well logs (when available) and qualitatively by the cross-hole (estimation of the clay conductivity) using the open source python code pyGIMLI (Rücker et al., 2017). Simple simulations using a dipole-dipole configuration on a two-layers models were done. Different geological settings were tested using a conductive thin layer (clay) surrounding within a resistive soil (gravel). The results highlighted that the conductivity above the conductive layer are affected. Using a non-constrained inversion did not allow to retrieve the real clay feature while significant improvements on the estimation of the clay layer tick (associate with lower error on the estimated model) were observed using a constrained inversion. The ultimate objective of the study was to build the 3 dimensional model. To this end, several steps were needed each providing partial results: - Both geophysical and stratigraphy data from an old report made in 2002 were digitized with a particular care to make the data reusable (georeferenced) and comparable (scaled) with the new data acquired in 2018. Three dimensional projection of wells and geophysical Fig. 1 - (left) Forward modeling of a clay discontinuity (20Ohm.m) for a two layered soil sand (200Ohm.m) over gravel (500Ohm.m). (right) result of the inversion using pyGimli showing the overestimation of the clay layer thickness without constraints during the inversion. Fig. 2 - 2D ERT inverted data showing different sources of errors (presence of paleochannels, bad inversion) on the estimation of the clay layer depth and thickness . Integration of geotechnical wells and geostatiscal analysis using kriging interpolation aims to refine the model.