RESERVOIR POROSITY DETERMINATION FROM 3D SEISMIC DATA - APPLICATION OF TWO MACHINE LEARNING TECHNIQUES

摘要

This paper proposes a method for solving 3D seismic data inversion problems for prediction of porosity in hydrocarbon reservoirs. An actual carbonate oil field in the south-western part of Iran was selected for this study. Taking real geological conditions into account, different synthetic models of reservoir were constructed for a range of viable porosity values. Seismic surveying was performed next on these models. From seismic response of the synthetic models, a large number of seismic attributes were identified as candidates for porosity estimation. Classes of attributes such as energy, instantaneous, and frequency attributes were included amongst others. Applying sensitivity analysis, the two most significant attributes were determined as Envelope Weighted Phase and Envelope Weighted Frequency, which were subsequently used in our machine learning algorithms. In particular, we used feed-forward artificial neural networks (FNN) and support vector regression machines (SVR) to develop relationships between the known synthetic attributes and synthetic porosity values in a given setting. The FNN consists of six neurons in a single hidden layer and the SVR method uses a Gaussian radial basis function. Compared with real values from the well data, we observed that SVM outperforms FNN due to its better handling of noise and model complexity.