Research Of Seismic Full Waveform Inversion Technology In Near-surface Velocity Model

With the deepening of the oilfield exploration and development throughout the country, the targets of the entire oil and gas exploration turn to deeper layers and peripheral areas with more complicated geological conditions now. At the same time, there also puts forward higher demands of accurate velocity modeling, high precision imaging, fine reservoir description and seismic geological interpretation. Therefore, it is badly eager for the progress of geophysical technologies, especially the progress of the precision of inversion imaging technologies.Velocity modeling is the key problem in seismic migration, and high precision velocity model is the foundation of seismic imaging and attribute inversion. At the same time, the near-surface velocity model which is accurate or not plays a key role in static correction problem. The strategy of current near-surface velocity modeling methods in shallow layer is replaced with constant velocity or refraction tomography inversion velocity, which is difficult to describe the change of velocity in complex near-surface structure and further affects the imaging accuracy of deep strata.Full waveform inversion(FWI) method based on wave equation theory and used to describe the near-surface model overcomes the problem that it is unable to be lighted because of the multiples and uneven ray tracing in chromatography. This method uses the kinematics and dynamic information of prestack seismic wave field to reveal the complex geological background details and lithology structure. So FWI is a high precision, multiple parametric modeling technology.In order to achieve accurate velocity model of waveform inversion, in this paper, on the basis of predecessors, wave field forward modeling process is analyzed in detail, and the wave equation finite difference caused for the difference in accuracy, difference equation stability condition and boundary condition have been studied on the theoretical research. In the paper, using the high order staggered grid finite difference algorithm improves the accuracy of numerical simulation. Adopted the effective storage and PML absorbing boundary conditions, both avoid interference due to boundary reflection wave field and saves the computer memory space. On this basis, selecting step length of time and space which satisfy the stability condition to avoid the instability of solving difference. Through the above method achieving high quality of finite difference numerical simulation. At the same time using the least squares method to achieve the fast convergence of residual wave field, moreover through the CPU/GPU collaborative computing method to achieve the fast inversion and forward work of velocity, completed the near-surface velocity model accurately.