Effectiveness and computational efficiency of absorbing boundary conditions for full-waveform inversion

Abstract. Full-waveform inversion (FWI) is a high-resolution numerical technique for seismic waves used to estimate the physical characteristics of subsurface region. The continuous problem involves solving an inverse on infinite domain, which impractical from computational perspective. In limited area models, absorbing boundary conditions (ABCs) are usually imposed avoid wave reflections. Several relevant ABCs have been proposed, with extensive literature their effectiveness direct problem. Here, we investigate and compare theoretical several in full After brief review most widely ABCs, derive formulations respective adjoint problems. different implemented highly optimized domain-specific language (DSL) framework, Devito, primarily modelling We evaluate effectiveness, efficiency, memory requirements ABC methods, considering simple models realistic ones. Our findings reveal that, even though popular perfectly matching layers (PMLs) effective at avoiding reflections boundaries, they can be computationally more demanding than less hybrid ABCs. show here that proposed formulation, nested Higdon's conditions, cost-effective method among methods considered here, being as or PML other schemes but also efficient.