LAUSR

Abstract The seismic imaging has three difficulties in the Keshen area of the Kuqa depression in Tarim Basin: difficult static correction, poor original data, difficult velocity modeling and pre-stack depth migration. A dual-scale tomography inversion approach based on BP neural network and LSQR was developed to obtain the accurate near-surface velocity of the complex near-surface structure, to address the static correction of subsalt high steep structure imaging. On the basis of applying high-precision static correction and root-mean-square (RMS) velocity to the seismic data, three dimensional cone filtering and spherical spreading amplitude compensation were used to enhance the signal to noise ratio and restore the deep effective signals to cope with the poor quality of original seismic data. Under the constraints of geologic, well logging and drilling data, the dual-scale velocity modeling technology based on model-based velocity updating and grid-based tomography was adopted to obtain the precise velocity model of the complex substructure, and then the pre-stack depth migration was taken to improve the imaging effect of structure with complex surface conditions, to solve the problem of subsalt high steep structure velocity modeling and pre-stack depth migration. By applying these three techniques, the high-quality imaging achievements of subsalt high steep structure were obtained. The results of seismic imaging prediction are in good agreement with drilling results and three ultra-deep wells have been drilled successfully.