Algorithmic analysis towards time-domain extended source waveform inversion

Abstract

Full waveform inversion (FWI) updates the subsurface model from an initial model by comparing observed and synthetic seismograms. Due to high nonlinearity, FWI is easy to be trapped into local minima. Extended domain FWI, including wavefield reconstruction inversion (WRI) and extended source waveform inversion (ESI) are attractive options to mitigate this issue. This paper makes an in-depth analysis for FWI in the extended domain, identifying key challenges and searching for potential remedies towards practical applications. WRI and ESI are formulated within the same mathematical framework using Lagrangian-based adjoint-state method with a special focus on time-domain formulation using extended sources, while putting connections between classical FWI, WRI and ESI: both WRI and ESI can be viewed as weighted versions of classic FWI. Due to symmetric positive definite Hessian, the conjugate gradient is explored to efficiently solve the normal equation in a matrix free manner, while both time and frequency domain wave equation solvers are feasible. This study finds that the most significant challenge comes from the huge storage demand to store time-domain wavefields through iterations. To resolve this challenge, two possible workaround strategies can be considered, i.e., by extracting sparse frequencial wavefields or by considering time-domain data instead of wavefields for reducing such challenge. We suggest that these options should be explored more intensively for tractable workflows.

Figures (12)

• Figure 1: (a) True velocity model including two layers of velocities 1500 m/s and 1800 m/s; (b) Initial velocity model of constant velocity 1500 m/s, where the source and receiver positions are marked by $+$ and $\times$.
• Figure 2: (a) Observed data simulated from true model; (b) Synthetic data by classic FWI (direct arrival from source to receiver is reproduced, but reflection from the interface is missing); (c) Synthetic data by WRI; (d) Synthetic data by ESI.
• Figure 3: (a) Classic FWI, (b) WRI and (c) ESI gradients representing the sensitivity of each method to model variations. ESI gradient presents one-sided illumination, significantly different from the two-sided FWI and WRI gradients.
• Figure 4: The 1D layered velocity model used for generating the initial model to study the variation of the misfit functional with respect to velocity errors. The dashed line indicates the location of the receivers, while the source coincides with the receiver in the top left corner.
• Figure 5: The variation of the misfit functional with respect to velocity errors for classic FWI