Application of full waveform inversion velocity model-building technology for the fractured-vuggy reservoir

Velocity is the key to pre-stack depth migration. Carbonate reservoir in Halahatang region of Tarim basin is buried deeply (deeper than 5500m) and has strong heterogeneity. Multi period igneous rock with uneven thickness and different velocity in Permian had developed overlying target formation. Velocity modeling based on tomography is difficult to obtain accurate velocity field. In this paper, we carry out full waveform inversion (FWI) method to improve imaging accuracy of the fractured-vuggy reservoir. By detailed analysis of FWI theory and its application challenges for land seismic data, taking into account of the actual situation of seismic data in YM area, we propose special strategies for successful application of FWI in this area: (1) Joint denoising technology based on surface wave modeling and curvelet transform is used in preprocessing stage to retain low frequency effective signal as far as possible, the five-dimensional regularization method is also used; (2) we utilize the high frequency component of static correction to eliminate the high frequency of near-surface velocity model and then use first arrival and reflected wave tomography to build a more accurate initial velocity model; (3) With the initial velocity model provided by traveltime tomography, the hierarchical multi-scale FWI method in time domain is applied. We utilize refraction waveform information to invert for shallow velocity while reflection waveform for deep velocity model. The actual data processing result shows that the joint denoising technology based on surface wave modeling and curvelet transform can protect low frequency information more effectively, maintaining the dynamics and kinematics information and meeting the basic requirement of FWI method. The initial velocity model provided by tomography can satisfy the accuracy of FWI. The hierarchical multi-scale FWI in time domain can characterize the velocity of igneous rock with high accuracy. The result of pre-stack depth migration indicates that the imaging of target formation under the igneous rock is improved obviously, eliminating the "fault" phenomenon caused by rough velocity model, and the imaging of fractured-cavernous reservoir is better. Although FWI is a high-precision velocity modeling technology with perfect theory, its application for land seismic data is still a challenge. The accuracy of initial velocity model and useful low frequency seismic information are keys to affect its result. We believe that the productive application of FWI for land seismic data will be developed gradually with the development of acquisition and processing technology.