Acoustic full-waveform inversion with acoustic and elastic data using a hybrid optimization method

Abstract

Seismic full waveform inversion (FWI) has been widely applied in different geophysical applications. By fitting both the phase and the amplitude of the observed seismic data using computational modeling engine, FWI can render high resolution subsurface model estimations compared to travel-time based inversion methods. Although FWI applications typically require a modeling engine that accurately approximate the physics that produces the observed data, acoustic FWI in the elastic world might still be able to provide valuable information in many circumstances, especially in marine cases, if proper methodology is adapted. Here, we study acoustic FWI for acoustic and elastic data with a recently introduced hybrid optimization method. The hybrid optimization based FWI is much less sensitive to the starting model, and is able to explore the parameter space to find possible solutions. We use a synthetic model that simulates the geology of deep-water salt province to demonstrate the behaviour of acoustic FWI based on hybrid optimization method when inverting acoustic and elastic data. Even though, the numerical tests illustrate some limitations of such implementation, the inversion results are very promising, the acoustic FWI based on hybrid optimization using the elastic pressure data can generate a reasonably good background Vp model without an accurate starting model, and the result does not degrade too much compared to the FWI result generated using the acoustic data.