AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

Development of an Early Miocene Carbonate Formation From Integrated Well Data, Modern Analogs and Geophysical Information: Case Study From Baturaja Formation, South Sumatera Basin, Indonesia


Baturaja formation is an early Miocene carbonate formation located in Pagardewa Field, South Sumatera Basin, Indonesia. The highest amplitude on conventional seismic sections of often misinterpreted as top of carbonate rocks, when in fact it is a thick sandstone body underlying a thin carbonate layer. The author has used the acoustic impedance derived from application of a Butterworth filter which gives an accurate boundary between overlying shale and the top of carbonate rock. Unlike clastic rocks, the spatial distribution of carbonate rocks across the field is difficult to predict. The distribution of carbonate rock, based on well data and seismic data, is strongly heterogeneous and requires the solution of a non-linear data fitting problem. Multi-layer neural network and genetic algorithm are proposed herein as a useful method to address the problem. A neural network with multi-layer structure is highly recommended to achieve lithology classification since higher order statistical information is generated. A genetic algorithm, which has a well-known analogy to biological evolution processes, leads to an efficient exchange of information between models evaluated by the neural network, permitting rapid assimilation and exploitation of the neural-network results to find a better-fitting model. The main limitation of this hybrid method is the huge amount of computational resources needed for higher-level classification. The combined multi-layer neural network and genetic algorithm inversion recovers accurate acoustic impedance information from conventional seismic sections. The obtained acoustic impedance cube was then used to characterize the development of carbonate formations during the early Miocene. The facies and the carbonate depositional environment were interpreted by integrating information from well data (e.g petrography and lithofacies) with the acoustic impedance values obtained from the inversion. Study of modern carbonate facies in Great Bahama Bank also lends support to the interpretation of facies and carbonate depositional environment of Baturaja Formation. This study shows how Miocene facies and carbonate depositional environment can be interpreted using integrated well data, a geophysical approach (i.e. acoustic impedance obtained from multi-layer neural network and genetic inversion) and comparisons with modern occurrences of carbonate systems.