GNGTS 2021 - Atti del 39° Convegno Nazionale

393 GNGTS 2021 S essione 3.1 parameters until the response fitted at best the observed seismic data. It was found that AVA can detect – at least qualitatively – the presence of kerogen, predicting a class IV response (fig. 1, right). However, the calculated anisotropy parameters matching observed seismic AVA response do exceed Thomsen’s weak anisotropy hypothesis for large TOC values, and this limits the application of current AVA inversion. Seismic attributes were then considered and analyzed as a method to identify the presence of kerogen. Since kerogen is known to have both low compressional impedance and small anelastic attenuation (Q factor), it was expected that frequency-dependent attributes could show – at least in principle – some degree of sensitivity to the presence of kerogen. The main issue in using seismic frequency attributes based on an instantaneous frequency computation (Taner et al. 1979) is their relatively high noisy response. A Tikhonov- regularized frequency estimator (De Tomasi 2016) was developed, to improve the frequency estimation with noisy signals. Then the instantaneous “sweetness” S(t) was calculated, defined as The drawback of impedance inversion is that it requires at least one well to calibrate inversi as an exploration tool could be limited. We then considered the AVA response for ker sensitivities. Trial AVA models were computed, changing anisotropy parameters until the re best the observed seismic data. It was found that AVA can detect – at least qualitatively – t kerogen, predicting a class IV response (fig. 1, right). However, the calculated anisotro matching obs rved seismic AVA response do xceed Thomsen's weak anisotropy hypothesi valu s, and this limits the application of current AVA inversion. Seismic attributes were then considered and analyzed as a method to identify the prese Since kerogen is kn w to have both low c mpressional impedance and small anelastic factor), it was expected th t frequency-d pendent attribut s could show – at least in pr degree of sensitivity to the presence of kerogen. The main issue in using s ismic frequency a on an instantaneous frequency computation (Taner et al. 1979) is their relatively high no Tikhonov-regularized frequency estimator (De Tomasi 2016) was developed, to improve estimation with noisy signals. Then the instantaneous "sweetness" S(t) was calculat S ( t ) = A ( t )/ √ F ( t ) , where A(t) is the instantaneous amplitude and F(t) the instantane Results for impedance inversion and sweetness are shown in figure 2 together with the pl inverted impedance versus the estimated sweetness for the source rock layer. As it ca sweetness attribute and the compressional impedance are well inversely correlated. The o , here (t) i i li F(t) the instantaneous frequency. R sults for impedance inversion and sweetness are shown in figure 2 together with the plot showing the inverted i pedance versus the estimated sweetness for the source rock layer. As it can be seen, the sweetness attribute and the compressional impedance are well inversely correlated. The only drawback in using sweetness is lacking of a robust quantitative rock physics calibration to infer the absolute TOC value, so its application is limited to a qualitative estimate. Finally, since the inverted compressional impedance is – up to now - the only geophysical parameter available to estimate quantitatively the TOC content, a method for inverting seismic to acoustic impedance without well control was developed, thus allowing the use of few impedance constraints, carefully predicted from rock physics models according to the expected litho-fluid setting, in place of data directly measured at wells. Results obtained using this new inversion methodology are still preliminary (fig. 3) but show globally a reasonable agreement when compared to the results of well-constrained inversion. The potential application of this method to TOC investigation will be further investigated in the near future. Fig. 2 - Comparison between inverted seismic impedance (top left) and sweetness attribute (bottom left). The plot shows acoustic impedance vs. sweetness.