GNGTS 2016 - Atti del 35° Convegno Nazionale

GNGTS 2016 S essione 3.3 597 Then, this straight line is gradually modified by applying Snell’s law to subsequent triplets of points of intersection with the grid. The scheme continues until the last computed ray path dif- fers from the previous one by less than a certain threshold. The inversion scheme used assumes an initial homogenous medium with a p-wave velocity of 3500 m/s. Then, the difference between the travel times computed from the ray paths and the picked ones is minimized with a Simultaneous Reconstruction Technique (SIRT) algorithm for 200 iterations. Since we want to resolve velocity anomalies which are generally horizontal, we select only those rays which have an angle lower than 30° from the horizontal line. Despite this, ray cove- rage is sufficient, with a total of 6775 rays. Results of the inversion are shown in Fig. 3c. As one can see, the main velocity anomalies of the domain are well resolved. As for the residuals, mean RMS is 1.1 %, with residuals well distributed around 0, as can be seen in Fig. 3d. Fig. 3 – a) Synthetic seismogram for shot n. 80, with the source located at a depth of 950 m; b) the Ricker source time history; c) inverted dataset and d) residual distribution. Conclusions. We presented modeling of a synthetic, yet realistic, geological sequestration of CO 2 experiment. We setup a 2D geological model consisting of a deep (>800 m) sandstone aquifer with some lens-shaped shale intrusions, surrounded by shales. We simulated the two- phase flow of the supercritical CO 2 and the brine using STOMP commercial software. Output of this simulation are saturation, pressure, density profiles for both gas and liquid phases. From here, using White’s model, we computed the seismic velocities, which are the input of the viscoelastic code used for the seismic modeling. The direct arrivals were then picked in these synthetic seismograms, which were inverted using our in-house tomographic code CAT3D. The results are encouraging, confirming the validity of the modeling and that a tomographic inversion of direct arrivals is able to detect the velocity anomalies caused by the presence of the CO 2 .