Identification of carbonate cave reservoirs using wavelet reconstruction and WLRPCA based background modeling

Abstract

Carbonate cave reservoirs are controlled by large strike-slip fault, which are home to incredibly rich oil and gas resources and hold immense exploration potential. The effective identification of these reservoirs using geophysical methods is quite important, but often challenging due to low signal-to-noise ratio and strong background reflection shielding of raw seismic data. In order to fully exploit the information of carbonate reservoirs contained in seismic data and improve the interpretation accuracy, we involve the innovative use of wavelet reconstruction and background modeling techniques to enhance the identification accuracy of carbonate cave reservoirs based on post-stack seismic data. The wavelet reconstruction in the first step mitigate the impact of noise and background reflection, allowing us to reveal the bead-like reflections hidden underneath strong interference energy. Then, we propose a background modeling method based on the weighted logarithmic robust principal component analysis (WLRPCA) technique in the second part to further eliminate background and noise signals. Compared with traditional principal component analysis (PCA) and robust principal component analysis (RPCA), WLRPCA exhibits superior recognition accuracy and noise immunity in the background modeling process, which effectively separates the target reflection from interference reflection and provides a favorable data basis for subsequent reservoir identification. Finally, a comprehensive attribute analysis is conducted on the processed seismic data, which confirms that the sweetness is the most effective attribute in detecting cave reservoirs. The proposed method has been successfully applied to both synthetic data and real data from an oil field in northern China, demonstrating their feasibility and effectiveness in identifying carbonate cave reservoirs.