In this work, we have proposed impedance-based texture attributes approach to accentuate the reservoir hetrogenity in mapping facies distrubuation and lateral variation. Impedance-based texture attributes account for rock  property and seismic signature, making them efficient in demarcating zones with different  rock and seismic properties such as channel deposits, hydrocarbon-filled sand, salt dome, etc. The presented case study indicates that impedance-based attributes have important implications for visualizing and demarcating zones with different facies distribution. Further, this study shows that the hydrocarbon-bearing zone mapped using our approach is more accurate than the conventional approach

In this work, we applied a hybrid global optimizer, genetic evolutionary ADAM (GADAM) to address the issue of convergence at a local optimum in a semisupervised deep sequential convolution network-based learning framework to solve the nonconvex seismic impedance inversion problem. GADAM combines the advantages of adaptive learning of ADAM and genetic evolution of genetic algorithm, which facilitates faster convergence, and avoids sinking into the local minima

In this study, we propose an application of deep sequential convolution network (DSCNet) to retrieve low frequency information thereby to achieve better acoustic impedance inversion results. In addition, we have implemented a hybrid global optimizer i.e., GADAM (genetic-evolutionary adaptive moment estimation), to ensure convergence at optimal minima. Thus, we are proposing a workflow for improving the resolution of the impedance model by incorporating the response of low frequencies. We have validated the performance of our model using an open-source 3D post-stack seismic data of F3, block, Offshore North Sea.

In this study, a novel optimization algorithm, i.e., G-ADAM was applied, which is capable to learn the variable parameters for deep learning models. Being a hybrid approach G-ADAM integrates the advantages of Adam and genetic algorithm, which enables it to achieve a globally optimal solution by avoiding the sinking into the local optima, and can be deployed on distributed computing platforms. 

In this study, an attempt has been made to delineate the detailed configuration of such a system from full stacked 3D seismic data using multi-attribute analyses together with state-of-the-art artificial neural networks. The study concluded that the deep reservoir of Jurassic age (i.e., Plover formation) is acting as the source of gas that is migrating through the available fault network of this area.