Data assimilation for subsurface flow using latent diffusion model parameterization: performance of ensemble-Kalman and Monte Carlo techniques

Problem
This work addresses the gap in data assimilation (DA) methodologies for subsurface flow modeling, particularly the challenge of calibrating model parameters to observed data while maintaining geological realism. The authors note that existing literature lacks a systematic comparison of DA algorithms in the context of high-dimensional geological models, especially when employing latent diffusion models (LDMs). This paper is a preprint and has not undergone peer review.

Method
The authors propose a systematic comparison of DA algorithms applied to large-scale 3D channelized geomodels characterized by hierarchical geological uncertainty. They utilize the ensemble smoother with multiple data assimilation (ESMDA) for both model-space and latent-space updates. The LDM serves as a parameterization technique that maps high-dimensional geological model space to a low-dimensional latent space, facilitating the inversion process. To address the computational demands of Markov chain Monte Carlo (MCMC) and Sequential Monte Carlo (SMC) methods, the authors develop a fast surrogate flow model that approximates well-rate responses. The evaluation is conducted across three synthetic test cases, focusing on the performance of MCMC and SMC in the LDM latent space compared to ESMDA.

Results
The results indicate that MCMC and SMC methods yield lower data mismatch and greater uncertainty reduction compared to latent-space ESMDA. Specifically, the MCMC and SMC approaches maintain geological realism while achieving significant improvements in uncertainty quantification. The authors report that model-space updates, while effective in reducing uncertainty, often lead to geologically unrealistic posterior models. The paper highlights that the ensemble Kalman methods may overestimate posterior uncertainty when applied to highly nonlinear parameterizations, underscoring the advantages of rigorous Monte Carlo sampling techniques.

Limitations
The authors acknowledge that while MCMC and SMC provide improved performance, they are computationally intensive, necessitating the development of surrogate models to enhance efficiency. Additionally, the study is limited to synthetic test cases, which may not fully capture the complexities of real-world geological scenarios. The authors do not address potential limitations related to the scalability of their methods to larger or more complex geological models.

Why it matters
This research has significant implications for the field of subsurface flow modeling and data assimilation, particularly in enhancing the reliability of parameter estimation under geological uncertainty. The findings suggest that rigorous Monte Carlo methods, when combined with fast surrogate models, can offer a more robust alternative to traditional ensemble Kalman approaches. This work contributes to the ongoing discourse on improving DA techniques in geosciences, as published in arXiv.