Lithostratigraphic Interpretation of Seismic Data for Reservoir Characterization, Mahesh Chandra, A. K. Srivastava, V. Singh, D. N. Tiwari, P. K. Painuly

Paradigm shift in hydrocarbon exploration and development strategies has increased utilization of seismic data many fold for reservoir characterization. Complicated and nonlinear relationship between seismic attributes and reservoir properties has been addressed recently using artificial neural network techniques for lithofaices classification and prediction of reservoir properties through unsupervised and supervised analysis. In lithofacies classification, analysis presumes no prior knowledge of the object to be classified and neural network looks for pattern itself. This type of analysis is very useful for areas where even no well control is available. In supervised analysis, reservoir properties are predicted away from the boreholes in inter-well regions after establishing the relation between multi-seismic attributes and well log data. In a real data example, a series of thin clastic reservoirs are sandwiched between coal and shale layers and are discrete in nature. These multi-pay sands having thickness from 2m to 8m are the main hydrocarbon producers and exhibited severe lateral lithological variation, which has affected the porosity distribution. Low porosity zones are found devoid of hydrocarbons. For better reservoir characterization, lithofacies classification and effective porosity distribution were carried out using Kohonen self organized maps(K-SOM) and probabilistic neural networks(PNN) respectively, which have demonstrated the effectiveness of these techniques. The lithofacies and effective porosity maps have provided more realistic reservoir model, which has helped in understanding the subsurface image and internal reservoir properties. The analysis has added a significant value to the exploration and development of hydrocarbons in the study area.