Acoustic Wave Reverse Time Migration And Low Wavenumber Noise Suppression

Deconvolution、superposition and migration are the main three techniques in seismic prospecting data processing, prestack depth migration is the hot point in geophysical prospecting worldwide in recent years. Migration is applied to converge the diffracted waves and return reflected waves to the right position underground, ultimately we could get the seismic section that represents the reflecting layers underground. The methods of seismic migration can be classified to prestack and poststack by the way of gathers superposition, and to time and depth migration by the adaptability to transverse velocity changes. With the objectives of seismic prospective becoming more and more flexible, reverse time migration(RTM) has been paid more and more attention by the advantage of high precision imaging. RTM belongs to prestack depth migration, though it’s presented in 1980 s, the huge demand of computation and storage restricts its wide application. Now because of the great development of computer technology, RTM is becoming more and more popular.RTM is based on two-way wave equation, there is no limit in layer dip angle underground, it’s capable of imaging steep dip angle structures precisely, and has a good imaging result for strongly transverse velocity change. According to the amplitude-preserved imaging, we can conduct lithology analysis, it’s helpful to predict hydrocarbon reservoir. This thesis conducts RTM by acoustic wave equation. In the process of RTM, the reflected wave on the border will be eliminated by perfectly matched layers(PML), the damping factors in PML will be analyzed. Because the conventional cross-correlation imaging condition needs to store source wave field every time step, it demands huge storage. In the thesis we make use of the source wave field reconstruction technique, this technique only needs to store border wave field at every time step and full wave field of the last two time steps, which will be used as boundary condition and initial condition respectively. With this technique we needn’t to store the source wave field at every time step. With regard to the excitation time imaging condition, it needn’t to store source wave field but the initial wave arrival time at each node as the imaging time, so its demand for storage is smaller. When receiver wave field is extrapolated in a reverse time way, the value of every node at its imaging time will be extracted as imaging result. Under the condition of conventional cross correlation imaging condition, this thesis introduces normalized cross correlation imaging condition, which can make up energy spread dissipation and have a better deep imaging result.Conventional reverse time migration will produce strong low wave number noise in shallow layers as a result of applying the cross correlation imaging condition. When the angle between source wave field vector and receiver wave field vector is considerable or in the vicinity of 180, low wavenumber noise will arise. In order to suppress the noise and improve the imaging resolution, this thesis introduces three methods: modifying the imaging condition、filtering the imaging result and modifying wave equation. The laplacian filter and wave field decomposition imaging condition will be paid much attention in this thesis. Wave field decomposition refers to divide source and receiver wave fields into upgoing and downgoing wave fields, then apply cross correlation imaging condition to right wave fields that are selected out, as a result, the low wavenumber noises will be suppressed considerably. Because the normalized cross correlation imaging condition and wave field decomposition imaging condition are helpful to eliminate the noise and improve the imaging resolution, this thesis will combine the two conditions, and apply the laplacian filter to the imaging result.Lastly, this thesis will conduct reverse time migration with the combined imaging condition on the horizontal layered model and Marmousi model. From the imaging result we can conclude that the normalized cross correlation imaging condition based on wave field decomposition can eliminate low wavenumber noises effectively. Then the imaging result is filtered by the laplacian in order to improve the resolution and make the imaging result more clear.