Survey sinking migration using the time-space localized dreamlet one-way propagator
Wu Bang-Yu,Wu Ru-Shan,Gao Jing-Huai,Xu Zong-Ben
DOI: https://doi.org/10.6038/cjg20170919
2017-01-01
Abstract:The dreamlet seismic imaging method is seeking to develop the theory and algorithm for the decomposition and compression of seismic wavefields and the one-way wave equation operator, that is, wave propagation and imaging in the compressed domain. Due to the temporal and spatial non-stationary property of the seismic signal, the decomposition must be localized both in time and space in order to achieve sparse representation. The dreamlet is generated by the tensor product of the time and space localized atoms, and can be taken as impulse-beamlet wave atoms. During migration, the time-space localized dreamlet one-way propagator downward continues the wavefield in depth and the seismic signal propagates in one direction along the time axis. With depth increasing, part of data used to image the shallow structures is abandoned after reaching its depth location, therefore, the valid time length of the seismogram is shorter and the computation cost is decreased for the wavefield depth stepping. In order to fully take advantage of this property, in this paper, we introduce the theory and method of combining the dreamlet propagation with survey sinking migration. Survey sinking migration downward continues the entire surface received data simultaneously. The sunk wavefield is equivalent to the data acquired by deploying sources and receivers at that depth. In dreamlet survey sinking migration, it only keeps the data for imaging the structures beneath the sinking survey system and gets rid of the data already used to image structures above it. Numerical tests on the two dimensional SEG/EAGE post-stack and Marmousi prestack data demonstrate properties of the dreamlet survey sinking. The numerical results show that the combination of dreamlet migration and survey sinking imaging can drastically reduce the data to be propagated without giving for the image quality. It has the potential to be developed as a highly efficient migration method.