GNGTS 2017 - 36° Convegno Nazionale

532 GNGTS 2017 S essione 3.1 geostatistical information and to define the hard data constraints. The information pertaining to the blind well is used to validate the final estimates. Fig. 3a shows the posterior distributions provided by the LUBI approach compared with the actual petrophysical properties derived from the blind well. We note that the porosity is the best resolved parameter while the shaliness, but particularly the water saturation, are less resolved due to their minor influences in determining the AVA response. As a result, the estimated maximum a posteriori (MAP) solution for the porosity, perfectly matches the actual value, whereas the estimated MAP solutions for the shaliness and water saturation overestimate the true values. Fig. 3b displays the results obtained by the proposed method. By comparing Figs. 3a and 3b we note that the additional geostatistical information and hard data constraints, yield a higher resolution for the porosity and the shaliness estimates and move the MAP solution for the shaliness toward the true value. Differently, the water saturation still remains poorly resolved. Conclusions. We presented a seismic-petrophysical Bayesian inversion that combines sequential geostatistical simulations with the analytical solution of a linear AVA inversion. The implemented approach incorporates geostatistical and hard data constraints and assumes that the petrophysical properties are distributed according to a linear combination of Gaussian distributions. Compared with the more conventional laterally unconstrained Bayesian approach, the implemented method suppresses the noise and the scattering in the estimated 2D maps of petrophysical properties and achieves better resolution and more reliable estimations of shaliness and porosity. On the contrary, in the investigated case the water saturation remains poorly resolved due to its minor influence in determining the P-P wave reflection coefficients. Acknowledgments The author wishes to thank Edison for the permission to publish this work and Dr. Timur Gukov for his continuous support to this project. Fig. 3 - Blind test results obtained by the LUBI approach (a) and by the proposed approach (b). Note that the additional geostatistical information and hard data constraints, yield higher resolutions for the porosity and shaliness estimates and move the MAP solution for the shaliness toward the true value. The water saturation remains poorly resolved in a) and b).