Data-driven reflection imaging based on seismic interferometry

Abstract

In conventional land seismic exploration, shallow seismic reflections with low fold are usually contaminated by noise and most conventional acquisition geometries struggle to produce sufficient near-offset fold to process shallow seismic reflections based on the velocity model. Deep seismic reflections tend to occur as weak events due to seismic wave absorption and attenuation. To enhance the shallow and deep seismic reflections, we have developed a data-driven reflection imaging method based on seismic interferometry of cross-coherence type, which first retrieves coherent reflections by cross-coherence of full wavefields with zero-offset reflections and then achieves imaging results with high quality by summing all of the crosscorrelations of full wavefields with coherent reflections in a common-midpoint gather. The novel purely data-driven method does not require any velocity information and can flatten noisy interfering reflection events with no waveform distortions. The focusing effects, which represent the stack of all the seismic reflection events, increase the fold of weak seismic reflections and further significantly improve the signal-to-noise ratio of imaging results. In comparison with data-driven reflection imaging methods based on seismic interferometry of crosscorrelation type, our results illustrate that the data-driven reflection imaging method based on seismic interferometry of cross-coherence type can automatically compress the band-limited source signature and make full use of background noise to broaden the bandwidth of reflection signals for achieving imaging results with a higher resolution.