A study on fluid prediction methods using ocean bottom node (OBN) seismic data: a case study of ultra-deep sub-salt biogenic limestone reservoirs in S Oilfield, Santos Basin

Accurate fluid prediction in ultra-deep sub-salt biogenic limestone reservoirs is of great significance for hydrocarbon resource development. The eastern uplift zone of the Santos Basin hosts two sets of such ultra-deep biogenic limestone reservoirs. Due to their great burial depth (averaging >5 000 m) and the strong seismic signal shielding effect of the overlying thick salt layer, conventional marine towed-streamer seismic data cannot achieve a detailed prediction of internal fluid distribution. An innovative rock physics modeling method based on a Bi-LSTM network was proposed. By utilizing the deep learning mechanism of neural networks to replace the calculation processes for rock matrix and rock frame, this method enabled efficient and accurate S-wave data prediction. Building upon the S-wave data predicted by this innovative method, ocean bottom node (OBN) seismic data, combined with well log data from the S oilfield development area, were used for fluid prediction in the biogenic limestone reservoirs. A comparative analysis of its fluid identification results with those of conventional towed-streamer seismic data was conducted. The results indicate that OBN seismic data, owing to its multi-azimuth angle information and higher signal-to-noise ratio of seismic data, demonstrates a clear advantage over conventional towed-streamer seismic data in reservoir fluid identification. Furthermore, statistical validation using multiple wells for the prediction results from both seismic data shows that the coincidence rate of the OBN seismic data predictions is 20% higher than that of the towed-streamer seismic data predictions. This confirms the important application value of OBN seismic data in reservoir fluid identification studies.