GNGTS 2017 - 36° Convegno Nazionale

634 GNGTS 2017 S essione 3.2 number of DCs and large distance between DCs near the sand body border that create low lateral resolution and consequent artifact due to interpolation. This could be resolved by retrieving more DCs during processing stage. The comparison of the velocities with those obtained by traditional DC inversion and P-wave tomography show variations mostly under 10%. Conclusion. We have shown that time-average VS and VP can be estimated, even in presence of sharp lateral variations, by directly transforming DCs using Socco et al. (2017) and Socco and Comina (2017). The results of the time-average VS and VP in Fig. 2 shows that the methods can outline, with great level of confidence, the near surface complexity and they could be a powerful alternative in near surface characterizations. The developed hierarchical clustering method grouped the DCs into two main groups outside and inside the sand body. Even though the DCs from CNR are not particularly challenging in terms of clustering, since two clear groups emerges by eye, still the developed clustering algorithm resulted to give very good information with respect to outliers. This could be potentially more useful over wider datasets and more complicated subsurface areas. Acknowledgments The authors would like to thank TOTAL E&P-RECHERCHE DEVELOPMENT for supporting the “Novel geophysical methods for retrieving near surface P-and S-wave velocity models” research project. References Nazarian, S., 1984, In situ determination of elastic moduli of soil deposits and pavement systems by spectral-analysis- of-surface waves method: Ph.D. dissertation, The University of Texas at Austin. Oded Maimon and Lior Rokach. 2010. Data Mining and Knowledge Discovery Handbook  (2nd ed.). Springer Publishing Company, Incorporated. doi: 10.1007/978-0-387-09823-4 Socco, L. V., C. Comina, and F. Khosro Anjom, 2017, Time-average velocity estimation through surface-wave analysis: Part 1 — S-wave velocity: Geophysics, 82(3), U49-U59. Fig. 2 - a) The estimated time-average VS model from the CNR data. b) The estimated time-average VP model from the CNR data; cross section of the sand body is displayed by the black line and locations of the different DCs are shown by crosses whose color indicates the cluster.