Abstract: Reef reservoir is important for high-yield carbonate oil and gas field, and it has large development efficiency and economic value, but there are many uncertain factors influencing reef reservoir study. Firstly, it's difficult to process seismic data of mountain area, and the problems include static correction, prestack noise removing, and migration imaging, etc. Secondly, it is difficult to identify and interpret potential reef reservoir. In the light of these problems, this paper studies reef reservoir of Changxing Formation in the west of Longgang area, Sichuan Basin. Equivalent near-surface model tomographic static correction and global optimization of residual static correction are respectively employed to resolve the long-wavelength and shortwavelength static correction problems. Then frequency-dependent noise attenuation method and conical shaped filter method in the F-Kx-Ky domain are selected to be used to improve the S/N ratio and solve the problem of prestack noise removing. High-resolution wide-band constrained deconvolution, surface consistent deconvolution and single-channel predictive deconvolution techniques are used to improve vertical resolution. Finally, anisotropic ray-bending prestack time migration is used to improve the imaging accuracy. Besides, many techniques for reef reservoir prediction and fluid detection are tested, and techniques suitable for this area are selected, achieving better results. The techniques suitable for reef reservoir prediction and fluid detection of the area are summarized and refined, and they could be taken as guidance for reef reservoir study and oil/gas detection in similar area.

Key words: reef, seismic data processing of mountain area, petrophysics forward, prestack inversion, reservoir prediction, fluid detection, low-frequency shadow, Sichuan Basin