GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 3.3 GNGTS 2024 Figure 2 shows the inversion results obtained by adding random noise of 0.1% to the amplitudes and traveltimes of 300 seismic traces, which corresponds to only a few samples. Fig.2 – Amplitude and traveltime inversion when a random noise percentage of 0.1% is added to the synthetic data. The smoothed, scaled estimate (dotted red line) fits well the true model (dashed yellow line). To improve the stability of the inversion, we also introduced a lateral smoothing filter with a window length of 31 samples. The instability in the initial estimates (solid blue line) is completely removed by the smoothing and scaling (dotted red line), so that this curve practically matches that of the true model (dashed yellow line). When the random noise increases to 0.5% (Figure 3), the estimated velocity is definitely unstable, but again the smoothed, scaled version (dotted red line) is not too far from the correct solution. The weakest estimate is that of density, which still correctly identifies a minimum value at the center of the mud volcano. CONCLUSIONS The lack of redundancy of a minimal survey, such as a mono-channel Boomer system, can be partially compensated for by interpreting and inverting the amplitudes and traveltimes of primaries and multiples. However, such an inversion requires a separate but coupled inversion of the dynamic and kinematic data to limit the crosstalk of physically independent variables. The results obtained with different noise levels show that we can obtain an encouraging estimate even for density when the signal-to-noise ratio is very good. This information is important for offshore engineering and marine geology.