Identifying faults and structural features from seismic wavefields is a pivotal approach in hydrocarbon reservoir prediction. Although seismic attributes and edge detection techniques are widely used in reservoir prediction, they are often inadequate for effectively delineating complex structures, particularly deep carbonate reservoirs exhibiting karst cave characteristics. Based on the stochastic process theory, we have introduced a new algorithm that applies spectral moments and statistical invariants to analyze the amplitude variations in the seismic wavefield. Seismic spectral moment elements represent the slopes variance and the covariance of seismic data within a calculation window in different directions. Statistical invariants are derived from second-order spectral moments, including the scratch coefficient designed to capture finer details within lower-amplitude anomalies. We develop a new scratch coefficient, termed the seismic scratch, which accurately captures the structural geometry of karst cave bodies. A comparative evaluation is conducted among the root mean square (rms) attribute, coherence, curvature, edge detection technology, and the seismic scratch method. The study reveals that (1) the seismic scratch method surpassed the rms amplitude attribute in detecting reservoirs with weak amplitudes; (2) coherence and curvature can only predict faults, whereas the seismic scratch method more accurately characterized the geometry and structure of karst cave reservoirs; and (3) the seismic scratch method outperformed edge detection technology in depicting carbonate reservoir features and preserving amplitude variations. The application of this method in the Tazhong region demonstrates its efficacy in accurately mapping fracture-cave units and enhancing predictive accuracy for oil and gas well locations.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.