Integrating Score-Based Diffusion Models with Machine Learning-Enhanced Localization for Advanced Data Assimilation in Geological Carbon Storage

2025-11-07

AI Summary

To address unreliable risk assessment in geological carbon sequestration (GCS) caused by inaccurate characterization of subsurface heterogeneity, this paper proposes a machine learning–enhanced localization method for fractional-order diffusion models. The approach employs FLUVSIM to generate channelized prior permeability fields, leverages a fractional diffusion model to efficiently produce large ensembles (Nₛ = 5000), utilizes a lightweight machine learning surrogate to map state variables, and embeds the scheme within the Ensemble Smoother with Multiple Data Assimilation (ESMDA) framework. Validation on a CO₂ injection scenario using the DARTS reservoir simulator demonstrates that, compared to a no-localization baseline, the method significantly improves covariance estimation quality and ensemble diversity while maintaining high data misfit accuracy—thereby enhancing uncertainty quantification and reliability of storage safety assessment. The core contribution lies in integrating physics-informed fractional diffusion modeling with ML-accelerated localization, achieving a balance between representational fidelity and computational efficiency.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Stochastic Optimization
• Intelligent Robots: Localization, Mapping, and Navigation

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Machine learning and data science for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate characterization of subsurface heterogeneity is important for the safe and effective implementation of geological carbon storage (GCS) projects. This paper explores how machine learning methods can enhance data assimilation for GCS with a framework that integrates score-based diffusion models with machine learning-enhanced localization in channelized reservoirs during CO$_2$ injection. We employ a machine learning-enhanced localization framework that uses large ensembles ($N_s = 5000$) with permeabilities generated by the diffusion model and states computed by simple ML algorithms to improve covariance estimation for the Ensemble Smoother with Multiple Data Assimilation (ESMDA). We apply ML algorithms to a prior ensemble of channelized permeability fields, generated with the geostatistical model FLUVSIM. Our approach is applied on a CO$_2$ injection scenario simulated using the Delft Advanced Research Terra Simulator (DARTS). Our ML-based localization maintains significantly more ensemble variance than when localization is not applied, while achieving comparable data-matching quality. This framework has practical implications for GCS projects, helping improve the reliability of uncertainty quantification for risk assessment.

Problem

• Characterizing subsurface heterogeneity for safe geological carbon storage projects
• Enhancing data assimilation using score-based diffusion models in channelized reservoirs
• Improving uncertainty quantification reliability for CO2 injection risk assessment

Innovation

• Integrates score-based diffusion models with machine learning
• Uses large ensembles for improved covariance estimation
• Applies ML-enhanced localization to maintain ensemble variance