Abstract: Fractured reservoirs are important exploration and development targets for increasing oil and gas reserves and production. Conventional post-stack seismic data is not effective for fractured reservoir identification. This paper proposes a relative separation technique of reflected wave field from scattered wave field based on diffusion filtering. This method predicts fractured reservoirs by taking optimal diffusion coefficient and iterations to separate post-stack seismic data into reflected wave field and scattered wave field, and then extracts seismic attributes such as coherence, curvature, and ant body from the scattered wave field. Application for predicting the fractured reservoirs of bedrock weathered crust in the Songliao Basin and the fractured reservoirs in the Maokou formation in the Sichuan Basin has obtained good results, demonstrating the method has higher resolution in predicting fractured reservoirs. The coincidence rate between the predicted result and the actual drilling data from two wells is 67% on the bedrock weathered-crust fractured reservoirs in a Daqing exploration block. It provides a reference for exploration, development, and well deployment in similar reservoirs.

Key words: scattered wave field, reflected wave field, fractured reservoir, seismic attribute, coherence, curvature