Faced with the low effective signal of deep carbonate reservoirs in the Tarim Basin, the seismic response characteristics of the small-scale fracture-cavity reservoirs are unclear, and conventional interpretive processing and reservoir prediction technologies fail to meet the required prediction accuracy. To address this problem, based on the differences between the “string-of-beads” reflections of fracture-cavity reservoirs and the layered stratum reflections and on the principle of feature decomposition, an interpretive processing method based on Tucker decomposition for fracture-cavity anomaly extraction is proposed. First, based on principle component analysis, a singular value cumulative ratio is constructed to quantify the key parameters of Tucker decomposition. Its applicability to fracture-cavity bodies of different scales, types, and development positions is evaluated, and the workflow for anomaly extraction is established. Then, a theoretical geological model of the fracture-cavity reservoir was constructed to verify the feasibility and effectiveness of the method, and the parameter selection for practical applications was justified. The results show that when the singular value cumulative ratio ranges from 0.68 to 0.72, the best fracture-cavity anomaly extraction results are obtained. Field seismic data and drilling results demonstrate that the fracture-cavity anomalies extracted by this method exhibit prominent characteristics in both profile and planar attributes. Compared with the conventional interpretive processing and reservoir prediction technologies, the enhancement effect of small-scale fracture-cavity bodies under strong reflection events is more obvious and is also highly correlated with the actual well features. This method provides a practical and effective way to predict small-scale fracture-cavity reservoirs.
Li et al. (Sun,) studied this question.