3D Porosity Estimation Using Multiattribute Analysis Methods in One of the Persian Gulf Oil Fields

Abstract

Abstract This study focuses on one of Persian Gulf oil fields. The major reservoir unit in the interested area is composed of alternation of thin dolomite with anhydrite layers. In the available 3D seismic data, these layers are not resolved and generate a composite detectable seismic response, making reservoir characterization difficult through conventional seismic attribute analysis (using single attribute transforms). In addition to resolving the major reservoir units the main aim of this study is porosity prediction and making 3D porosity cube for detailed reservoir characterization. For this purpose firstly the 3D seismic volume was inverted to obtain an acoustic impedance cube using three different inversion algorithms. Inversion improves the vertical resolution and resolving the interested reservoir layers. Among the different inversion methods applied, according to cross-validation results, acoustic impedance derived from model-based inversion has been selected as the main seismic attribute. In the next step of this study, acoustic impedance attribute beside other attributes extracted from seismic volume were analyzed using single attribute, multiple attribute regression (with stepwise regression for attribute selection) and neural networks (PNN and MLFN). These linear or non-linear combinations of attribute for porosity prediction result in the improved match between the derived porosity and predicted one. To estimate the reliability of the derived multi attribute transforms, cross validation is used; According to results it's found that cross correlation between actual porosity and estimated one has increased from 80% in single attribute prediction to 88% using multiple regression transform. Also neural networks provide higher cross correlation values rather than both. Finally according to cross validation results, multiple regression transform is used for porosity prediction. Using implemented technique slices prepared which have higher consistency with log data and could lower risk in field development.