Appraisal of reservoir porosity using a machine learning approach: A study from the Eocene‐Miocene interval of Upper Assam Basin, NE, India

Abstract

Porosity estimation is essential for the exploration of hydrocarbons in sedimentary basins, which are affected by several factors such as the burial depth, lithology changes, sedimentary environment and diagenetic degree. Hence, detailed understanding of reservoir porosity is essential for estimating potential economic reserves and developing an explored hydrocarbon field. The present study uses high‐quality three‐dimensional (3D) seismic reflection data to decipher the porous zones within the Eocene‐Oligocene‐Miocene intervals of the Upper Assam Basin. The coherency seismic attribute has brought out several fault‐bounded structures that control the reservoir geometry. Random forest‐based machine learning technique has been utilized for the porosity estimation within the formations of interest. It is observed that the north‐east (NE) parts of the targeted intervals are porous with varying porosity of 0.28–0.42. The Miocene period witnessed a high amount of sedimentation within the basin. Different structural highs and thicknesses are pronounced in the NE part of the study region. Thus, the Miocene intervals are favourable leads for the exploitation of hydrocarbons.