GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 3.1 GNGTS 2024 Probabilistc Petrophysical Inversion of Ground- Based FDEM Data for Alpine Peatland Characterizaton: A Case Study in the Italian Dolomites N. Zaru 1 , S. Silvestri 2 , M. Assiri 3 , P. Bai 4 , T.M. Hansen 5 , G. Vignoli 1,6 1 DICAAR - University of Cagliari, Italy 2 University of Bologna, Italy 3 University of Padua, Italy 4 SINOPEC, China 5 Aarhus University, Denmark 6 Near Surface Land and Marine Geology Department, GEUS, Denmark Introducton Peatlands, vital carbon reservoirs, ofen pose challenges in efciently and reliably characterizaton of their geometry and volumes, partcularly in alpine setngs (Silvestri et al., 2019). Here, we investgate a probabilistc petrophysical method applied to ground-based frequency-domain electromagnetc inducton (FDEM) data to quantfy peat thickness and extension in an alpine bog situated in the Italian Dolomites, where peat overlays an electrically conductve clay substrate. While traditonal FDEM inversion retrieves electrical resistvity, a direct estmaton of the probability of peat or clay occurrence is preferable as that is the ultmate goal of the geophysical investgaton (Grana et al., 2022). Additonally, the proposed probabilistc strategy addresses the limitatons of oversimplifed constraints and has practcal applicatons for real-tme feldwork optmizaton. Generally, the ill-posedness of FDEM data inversion necessitates prior informaton. Traditonal deterministc approaches utlize regularizaton terms, such as the L2-norm of the model gradient, which may oversimplify complex geological structures (Klose et al., 2022; Klose et al., 2023). The proposed probabilistc approach is capable to accommodate more complex (and realistc) prior informaton. Additonally, it naturally provides uncertainty estmates, crucial for the assessment of the reliability of inferred geological features. In this study, we apply probabilistc petrophysical inversion to an FDEM dataset collected on an alpine peatland, comparing results against traditonal deterministc Occam's inversions and borehole logs. Methodology The research is focused on characterizing Alpine peatland structures in the Italian Dolomites through ground-based Frequency Domain Electromagnetcs (FDEM). The study area is situated near Danta di Cadore in the North of Italy, encompassing a peatland of partcular interest. A systematc survey was conducted across this targeted peatland area using the GEM-2 instrument