Convolution-based multi-scale envelope inversion

• DOI: 10.1007/s12182-019-00419-8
• Publication Date: 2020-03-11T11:02:58Z
• Authors: Ding-Jin Liu; Jian-Ping Huang; Zi-Ying Wang
• Keywords: Maxima and minima; Wavenumber; Envelope (radar); Component (thermodynamics); Convolution (computer science)

Abstract Envelope inversion (EI) is an efficient tool to mitigate the nonlinearity of conventional full waveform (FWI) by utilizing ultralow-frequency component in seismic data. However, performance envelope depends on frequency and initial model some extent. To improve convergence ability avoid local minima issue, we propose a convolution-based method update low-wavenumber velocity model. Besides, multi-scale strategy (MCEI) also incorporated accuracy while guaranteeing global convergence. The success this relies modifying original data expand overlap region between observed modeled data, which turn expands minimum basin misfit function. accurate provided MCEI can be used as migration or for FWI. numerical tests simple layer complex BP 2004 verify that proposed more robust than EI even when coarse high.