Abstract:

Lithofacies, porosity and permeability as well as other petrophysical properties of the reservoir, in recent year, with the robust development of modern technology applied in seismic data processing and interpretation, shall be predicted directly from seismic data using the empirically built relationship, either linear or nonlinear between one or more seismic attributes generated from 2D, 3D seismic data and one or more petrophysical properties calculated from wireline logs ... This paper presents the preliminary outcomes of porosity prediction in Miocene sediments of block 103 in Northern Red River basin. The relationship between porosity and seismic attributes is revealed based on multiple regression and neural network methods, namely MLFN and PNN. For individual approach, one predictive model is derived. In order to reduce the uncertainty of individual predictive model, in this paper, the authors proposed the average committee model that combined all outputs from individual regression and neural network models into a final output of porosity.