GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 3.1 GNGTS 2024 compared against the traditonal Occam’s inversion (Fig. 2 – middle panel) obtained by using EMagPy (McLachlan et al., 2021). In the case of deterministc inversion, special care was devoted to the selecton of the Tikhonov’s parameter to reach the chi-squared values in the range of the optmal value 1 (Fig. 2 – top panel). Despite the similar data ftng, the deterministc inversion seems to favor vertcally elongated structures (opposite to our expectaton about the presence of a peat-clay interface and the informaton from the boreholes). In contrast, the probabilistc approach, whose spatal consistency is atributed to complex prior informaton, seems to ofer a more accurate portrayal of the peat unit's morphology. Fig. 2 – Vertcal profle presentng the chi-squared value of the deterministc (blue) and maximum likelihood (red) models in the top panel. The deterministc model obtained with the standard Occam’s inversion is shown in the middle panel; the maximum likelihood model obtained with the probabilistc inversion in the botom panel. The horizontal blue and red lines in the fst panel represent the average chi-squared values associated with the two models. Figure 3 depicts the probability of having peat and clay in the examined secton. Clearly, the porton between the boreholes B26* and B11 is well correlated with the peat-clay interface. The big diference between the peat probability and the Occam’s result highlights one more tme the infuence of diferent prior informaton in the two approaches (probabilistc and deterministc): the