Research Projects

Rapid developments are underway in reservoir geophysics with special interest in applying anisotropy in seismic exploration and reservoir monitoring. Conventionally, the 3 Component (3C) seismic data or dipole logs are used to analyse seismic anisotropy, which are often unavailable or limited. This project proposes a Machine Learning (ML) driven method to predict the anisotropic properties such as seismic velocities and anisotropic parameters (fast polarization direction, delay time, fracture density, etc.), leading towards enhanced production. As a second part, the project aims to use 3C data and natural seismic data of the shelf region of the Assam Arakan (AA) basin for a combined analysis. The anisotropic parameters strongly influence the seismic attributes (Amplitude Variations with Offset (AVO), attenuation coefficients, etc.). The seismic attributes' dependence on the anisotropic parameters can be incorporated in machine learning for predicting the anisotropic parameters. The labelled data, which shows the dependence of seismic attributes with anisotropy, are divided into training and testing data for training and validation, respectively. The validation error gives the reliability of the machine learning model. Synthetic seismic gathers are generated using rock physics model, while anisotropic parameters will be computed using a supervised machine learning model with the synthetic seismic as input. Once reliable results are obtained from synthetic data, the method will be applied to real seismic datasets to predict the anisotropic parameters. The results will be validated with available field data like well logs, resistivity images, etc. After successfully validating the 2D models, the project will focus on using 3C seismic data to predict the anisotropic parameters using the proposed methodology. This project further evolves its scope in estimating seismic anisotropy through the Shear Wave Splitting (SWS) technique using the natural seismic source (local earthquake events) to characterize the crack density, fracture orientation and connectivity, which are vital for reservoir production and management. SWS is sensitive to the orientation and strength of anisotropy, whose variations over a period could be monitored to possibly track the fluid movement and the stress changes within the reservoir. The local earthquake data are beneficial to characterize the shallow subsurface anisotropy by modelling the SWS parameters (fast axis, delay time, crack density, anisotropic percentage) from a micro to macro scale in the reservoir and its larger vicinity. This project work will be a first-of-its-kind approach in dealing with both the artificial and natural seismic sources to delineate the complex anisotropic scenario of the reservoir environment by complementing the robustness and validity of synthetic models, leading towards an enhanced reservoir production (ERP), especially in the tight complex reservoirs like AA basin.