-
Bayesian weighted time-lapse full-waveform inversion using a receiver-extension strategy
Authors:
Sergio Luiz E. F. da Silva,
Ammir Karsou,
Roger M. Moreira,
Marco Cetale
Abstract:
Time-lapse full-waveform inversion (FWI) has become a powerful tool for characterizing and monitoring subsurface changes in various geophysical applications. However, non-repeatability (NR) issues caused, for instance, by GPS inaccuracies, often make it difficult to obtain unbiased time-lapse models. In this work we explore the portability of combining a receiver-extension FWI approach and Bayesia…
▽ More
Time-lapse full-waveform inversion (FWI) has become a powerful tool for characterizing and monitoring subsurface changes in various geophysical applications. However, non-repeatability (NR) issues caused, for instance, by GPS inaccuracies, often make it difficult to obtain unbiased time-lapse models. In this work we explore the portability of combining a receiver-extension FWI approach and Bayesian analysis to mitigate time-lapse noises arising from NR issues. The receiver-extension scheme introduces an artificial degree of freedom in positioning receivers, intending to minimize kinematic mismatches between modeled and observed data. Bayesian analysis systematically explores several potential solutions to mitigate time-lapse changes not associated with reservoir responses, assigning probabilities to each scenario based on prior information and available evidence. We consider two different subsurface models to demonstrate the potential of proposed approaches. First, using the Marmousi model, we investigate two NR scenarios associated with background noise in seismic data. Second, using a challenging deep-water Brazilian pre-salt setting, we investigate several NR scenarios to simulate real-world challenges. Our results demonstrate that combining Bayesian analysis with the receiver-extension FWI strategy can mitigate adverse NR effects successfully, producing cleaner and more reliable time-lapse models than conventional approaches. The results also reveal that the proposed Bayesian weighted procedure is a valuable tool for determining time-lapse estimates through statistical analysis of pre-existing models, allowing its application in ongoing time-lapse (4D) projects.
△ Less
Submitted 10 July, 2024;
originally announced July 2024.
-
A graph-space optimal transport FWI approach based on κ-generalized Gaussian distribution
Authors:
Sérgio Luiz E. F. da Silva,
G. Kaniadakis
Abstract:
The statistical basis for conventional full-waveform inversion (FWI) approaches is commonly associated with Gaussian statistics. However, errors are rarely Gaussian in non-linear problems like FWI. In this work, we investigate the portability of a new objective function for FWI applications based on the graph-space optimal transport and $κ$-generalized Gaussian probability distribution. In particu…
▽ More
The statistical basis for conventional full-waveform inversion (FWI) approaches is commonly associated with Gaussian statistics. However, errors are rarely Gaussian in non-linear problems like FWI. In this work, we investigate the portability of a new objective function for FWI applications based on the graph-space optimal transport and $κ$-generalized Gaussian probability distribution. In particular, we demonstrate that the proposed objective function is robust in mitigating two critical problems in FWI, which are associated with cycle skipping issues and non-Gaussian errors. The results reveal that our proposal can mitigate the negative influence of cycle-skipping ambiguity and non-Gaussian noises and reduce the computational runtime for computing the transport plan associated with the optimal transport theory.
△ Less
Submitted 24 May, 2024;
originally announced May 2024.
-
Generalized statistics: applications to data inverse problems with outlier-resistance
Authors:
João V. T. de Lima,
Sérgio Luiz E. F. da Silva,
João M. de Araújo,
Gilberto Corso,
Gustavo Z. dos Santos Lima
Abstract:
The conventional approach to data-driven inversion framework is based on Gaussian statistics that presents serious difficulties, especially in the presence of outliers in the measurements. In this work, we present maximum likelihood estimators associated with generalized Gaussian distributions in the context of Rényi, Tsallis and Kaniadakis statistics. In this regard, we analytically analyse the o…
▽ More
The conventional approach to data-driven inversion framework is based on Gaussian statistics that presents serious difficulties, especially in the presence of outliers in the measurements. In this work, we present maximum likelihood estimators associated with generalized Gaussian distributions in the context of Rényi, Tsallis and Kaniadakis statistics. In this regard, we analytically analyse the outlier-resistance of each proposal through the so-called influence function. In this way, we formulate inverse problems by constructing objective functions linked to the maximum likelihood estimators. To demonstrate the robustness of the generalized methodologies, we consider an important geophysical inverse problem with high noisy data with spikes. The results reveal that the best data inversion performance occurs when the entropic index from each generalized statistic is associated with objective functions proportional to the inverse of the error amplitude. We argue that in such a limit the three approaches are resistant to outliers and are also equivalent, which suggests a lower computational cost for the inversion process due to the reduction of numerical simulations to be performed and the fast convergence of the optimization process.
△ Less
Submitted 28 January, 2022;
originally announced January 2022.
