--> Applying PCA in Seismic Attribute Analysis for Interpretation of Evaporite Facies: Lower Triassic Jialingjiang Formation, Sichuan Basin, China

AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

Applying PCA in Seismic Attribute Analysis for Interpretation of Evaporite Facies: Lower Triassic Jialingjiang Formation, Sichuan Basin, China

Abstract

The lower interval of the second member of the Lower Triassic Jialingjiang Formation is dominated by anhydrite, limestone, dolostone and calcareous mudstone, and contains important hydrocarbon reservoirs in Central Sichuan Basin, southwest China. We used principle component analysis method to optimize seismic attributes and use the PCA results to calculate the distribution of anhydrite, dolostone and mudstone. Meanwhile, the Read-Green-Blue merge technology was applied to estimate the content of anhydrite, dolostone and mudstone, and to reconstruct evaporite facies. Density data were used to separate anhydrite and tight dolostone from limestone and mudstone. GR value was used to distinguish mudstone from other lithologies. By using density and GR, the total thicknesses of anhydrite, dolostone and mudstone for each well were calculated by summing up every single layer of the three lithologies. Then the total thickness of the three lithologies was divided by the gross thickness to obtain contents of the three lithologies for each individual wells, which were used to test seismic attributes and the later PCA results. Characteristics of anhydrite on seismic reflection vary dramatically depending on sediment environment and lithology combination. It is not easy to differentiate anhydrite from dolostone by ordinary seismic attributes because of their similar acoustic impedances with each other. Therefore, over fifty seismic attributes were extracted from a 2,500 km2 three-dimensional seismic data, which was rotated into 90-phase. There are some relationships between seismic attributes and lithology content at well locations. But none of the single attributes has correlation coefficient higher than 0.6. Simple combination of these seismic attributes failed to improve the correlation coefficient. However, it was improved dramatically when the first two principle components were used as new factors. The PCA method preserves common useful information related to lithology in the target layer, while removing disturbing signals. The predicted contents of anhydrite, dolostone and mudstone are shown in blended colors. And then detailed lithofacies of T1j21 in the study area are reconstructed and analyzed accordingly. In this study, the PCA method has been proved to be a powerful tool in optimizing seismic attributes. The knowledge gained from this study may be helpful for the seismic analysis of lithology and sedimentary facies in similar carbonate-evaporite successions.