Unsupervised Seismic Facies Classification based on Multiattribute Analysis in the Asmari Reservoir Ramshir Oilfield

The unsupervised seismic facies classification has been increasing used in reservoir characterization over the past two decades, and their popularity and application in the process of geophysical interpretation as a means of estimating hydrocarbon resources continue to grow. In this study, in order to identify seismic facies, based on their seismic attributes, we used simultaneously 3D seismic data (seismic attributes) and electrofacies (petrophysical rock type) in the studied wells, then variations of rock types in Asmari reservoir of Ramshir field have been determined. In this study, neural network approach and k-means clustering method were used to define unsupervised seismic facies based on seismic attributes. Then we used principal component analysis which is a good solution to reduce the number of inputs and thereby the complexity of the model. Finally, the seismic attributes that best show seismic facies distribution were determined. The attributes used include the dominant frequency, envelope derivative, acoustic impedance, thin-bed indicator and spectral decomposition 50 Hz. Ultimately, using this method in the Asmari reservoir of Ramshir oil field, seismic facies related to sandstone, limestone and dolomite were defined. Also, the distribution map of reservoir facies in this field was extracted and interpreted.