Towards real-time monitoring: data assimilated time-lapse full waveform inversion for seismic velocity and uncertainty estimation

Abstract

Summary Rapid development of time-lapse seismic monitoring instrumentations has made it possible to collect dense time-lapse data for tomographically retrieving time-lapse (even continuous) images of subsurface changes. While traditional time-lapse full waveform inversion (TLFWI) algorithms are designed for sparse time-lapse surveys, they lack of effective temporal constraint on time-lapse data, and, more importantly, lack of the uncertainty estimation of the TLFWI results that is critical for further interpretation. Here we propose a new data assimilation TLFWI method, using hierarchical matrix powered extended Kalman filter (HiEKF) to quantify the image uncertainty. Compared to existing Kalman filter algorithms, HiEKF allows to store and update a data-sparse representation of the cross-covariance matrices and propagate model errors without expensive operations involving covariance matrices. Hence HiEKF is computationally efficient and applicable to 3D TLFWI problems. Then we reformulate TLFWI in the framework of HiEKF (termed hereafter as TLFWI-HiEKF) to predict time-lapse images of subsurface spatiotemporal velocity changes and simultaneously quantify the uncertainty of the inverted velocity changes over time. We demonstrate the validity and applicability of TLFWI-HiEKF with two realistic CO2 monitoring models derived from Frio-II and Cranfield CO2 injection sites, respectively. In both 2D and 3D examples, the inverted high-resolution time-lapse velocity results clearly reveal a continuous velocity reduction due to the injection of CO2. Moreover, the accuracy of the model is increasing over time by assimilating more time-lapse data while the standard deviation is decreasing over lapsed time. We expect TLFWI-HiEKF to be equipped with real-time seismic monitoring systems for continuously imaging the distribution of subsurface gas and fluids in the future large-scale CO2 sequestration experiments and reservoir management.

Critical Zone Structure by Elastic Full Waveform Inversion of Seismic Refractions in a Sandstone Catchment, Central Pennsylvania, USA

“…In addition, due to data noise, not fully accounted physics for wave propagations in the application of field data, the uncertainties of elastic velocity inversion (Huang & Zhu, 2020; Liu & Peter, 2020), etc., there may be updates of Vp and Vs with unreasonable Vp/Vs values in applications with field data. Here we design an automated procedure to reject the possible unreasonable velocity updates by introducing additional constraint Vp/Vs > 1 at each iteration of multiscale frequency procedure (see details in Text S1 in Supporting Information S1).…”

“…In addition, due to data noise, not fully accounted physics for wave propagations in the application of field data, the uncertainties of elastic velocity inversion (Huang & Zhu, 2020;Liu & Peter, 2020), etc., there may be updates of Vp and Vs with unreasonable Vp/Vs values in applications with field data.…”

Toward Automated Early Detection of Risks for a CO2 Plume Containment From Permanent Seismic Monitoring Data

“…A neural network would provide a safe scenario of the CO 2 migration, then a sufficiently large deviation from the predicted plumes would indicate a potential leakage and raise an alarm. Another application of the neural network is prediction of the states of subsurface in Kalman filtering of the continuous monitoring data in a way similar to the previously mentioned work by Huang and Zhu (2020). In this paper, we examine the feasibility of a neural network‐based algorithm for rapid plume forecasting from permanent seismic monitoring data.…”

“…Instead of solving the inverse problem, a practical solution uses state‐space approach as implemented in Kalman filtering (Evensen, 2009; Huang & Zhu, 2020; Ma et al., 2019). This approach still requires a robust transition model that can predict future states of the subsurface, such as, pore pressure and saturation.…”

Learned coupled inversion for carbon sequestration monitoring and forecasting with Fourier neural operators

“…Over the years, several attempts have been made to mitigate this challenge by improving the repeatability of time-lapse seismic, including the forward and backward bootstrapping method [Asnaashari et al, 2015], the double-difference method [Watanabe et al, 2004, Denli and Huang, 2009, Zhang and Huang, 2013, Yang et al, 2015, the central-difference method [Zhou and Lumley, 2021a], data assimilation via Kalman filtering [Li et al, 2014, Eikrem et al, 2019, Huang and Zhu, 2020 and the joint recovery model , Oghenekohwo and Herrmann, 2017a, Yin et al, 2021. While these methods resulted in improvements in repeatability of time-lapse seismic, they ignore the fact that the dynamics of CO 2 plumes, to the leading order, adhere to two-phase flow equations.…”