One-Way Waveform Inversion (Owi)

Abstract

Summary Reflection Waveform Inversion (RWI) has become an efficient method to generate low-wavenumber velocity models. It has shown a better performance for updating the deep model structures compared to the standard Full Waveform Inversion (FWI). However, RWI is well known to suffer from the slow convergence problem making it high computational demanding. Time consistent Wave equation Inversion (TWIN) has recently been proposed to mitigate this problem by considering the consistency between velocity-depth positioning. Nevertheless, both RWI/TWIN rely on the assumption of the scale separation which requires a sufficiently smooth background velocity model. In case of lack of smoothness, the background velocity model will generate reflections that cannot be matched to the observed data. This issue may lead to a noisy sensitivity kernel that adds more non-linearity to the inversion scheme. In addition, the forward modeling which is an essential component of RWI/TWIN still represents an overburden step because solving the wave equation using the two-way propagator is highly expensive. In this proposal, we introduce an alternative to the two-way wave equation by using a one-way approach for the reflection waveform inversion. After a brief theoretical description of the One-way Waveform Inversion (OWI), we validate its accuracy on Chevron SEG 2014 benchmark.