Deep learning multiphysics network for imaging CO₂ saturation and estimating uncertainty in geological carbon storage

Abstract

Multiphysics inversion exploits different types of geophysical data that often complement each other and aims to improve overall imaging resolution and reduce uncertainties in geophysical interpretation.Despite the advantages, traditional multiphysics inversion is challenging because it requires a large amount of computational time and intensive human interactions for preprocessing data and finding tradeoff parameters. These issues make it nearly impossible for traditional multiphysics inversion to be applied as a real-time monitoring tool for geological carbon storage. In this paper, we present a deep-learning (DL) multiphysics network for imaging CO 2 saturation in real time. The multiphysics network consists of three encoders for analyzing seismic, electromagnetic, and gravity data, and shares one decoder for combining imaging capabilities of the different geophysical data for better predicting CO 2 saturation. The network is trained on pairs of CO 2 label models and multiphysics data so that it can directly image CO 2 saturation. We use the bootstrap aggregating method to enhance the imaging accuracy and estimate uncertainties associated with CO 2 saturation images. Using realistic CO 2 label models and multiphysics data derived from the Kimberlina CO 2 storage model, we evaluate the performance of the DL multiphysics network and compare their imaging results to those from the DL single-physics networks.Our modeling experiments show that the DL multiphysics network for seismic, electromagnetic, and gravity data not only improves the imaging accuracy but also reduces uncertainties associated with CO 2 saturation images. Our results also suggest that the DL multiphysics network for the non-seismic data (i.e., electromagnetic and gravity) can be used as an effective low-cost monitoring tool in between regular seismic monitoring.

Electromagnetic imaging and deep learning for transition to renewable energies: a technology review

“…From a computer science perspective, the multi-physics inversion process is understood as complex algorithms trying to find accurate solutions to highly challenging parameterdependent problems. Thus, inverse modeling in realistic applications requires high computational effort and intensive human interaction (Um et al, 2022). In this context, the use of DL technologies to solve coupling relationships between different geophysical data types is undoubtedly a highlight.…”

“…DL methods can learn both prior parameter relationships and structural similarity of the multiphysics data sets. Therefore, the accuracy of EM imaging may be greatly improved (Chen et al, 2022;Guo et al, 2022), and the uncertainty of imaging could be reduced (Um et al, 2022). These aspects are crucial for DL decision-making inversion and need further study (Oh and Byun, 2021).…”

Automating hyperparameter optimization in geophysics with Optuna: A comparative study

“…Alongside this, the network architecture and its layers will be shown. However, hyperparameter optimization (HPO), a crucial part of this analysis, is often overlooked (Kaur, Pham, et al., 2020; Lähivaara et al., 2019; Kaur, Fomel, et al., 2020; Colombo et al., 2021; Um et al., 2022) or at best, only briefly discussed (Côrte et al., 2020; Tian et al., 2020; Aleardi et al., 2022). This pertains to the justification of why a certain network architecture, depth, learning rate or other hyperparameters were chosen.…”