3D CNN-PCA: A Deep-Learning-Based Parameterization for Complex Geomodels

TL;DR

This paper introduces CNN-PCA, a deep learning method for efficient 3D geological model parameterization that produces realistic models consistent with reference data, aiding in uncertainty quantification and history matching.

Contribution

The paper develops a novel 3D CNN-PCA algorithm that combines PCA with convolutional neural networks, including a new supervised reconstruction loss and style loss from pretrained video classification CNNs.

Findings

CNN-PCA generates realistic 3D geomodels matching reference features.

Flow statistics from CNN-PCA models align with those from reference models.

CNN-PCA successfully applied to history matching with ESMDA.

Abstract

Geological parameterization enables the representation of geomodels in terms of a relatively small set of variables. Parameterization is therefore very useful in the context of data assimilation and uncertainty quantification. In this study, a deep-learning-based geological parameterization algorithm, CNN-PCA, is developed for complex 3D geomodels. CNN-PCA entails the use of convolutional neural networks as a post-processor for the low-dimensional principal component analysis representation of a geomodel. The 3D treatments presented here differ somewhat from those used in the 2D CNN-PCA procedure. Specifically, we introduce a new supervised-learning-based reconstruction loss, which is used in combination with style loss and hard data loss. The style loss uses features extracted from a 3D CNN pretrained for video classification. The 3D CNN-PCA algorithm is applied for the generation of conditional 3D realizations, defined on $60\times60\times40$ grids, for three geological scenarios (binary and bimodal channelized systems, and a three-facies channel-levee-mud system). CNN-PCA realizations are shown to exhibit geological features that are visually consistent with reference models generated using object-based methods. Statistics of flow responses ($\text{P}_{10}$, $\text{P}_{50}$, $\text{P}_{90}$ percentile results) for test sets of 3D CNN-PCA models are shown to be in consistent agreement with those from reference geomodels. Lastly, CNN-PCA is successfully applied for history matching with ESMDA for the bimodal channelized system.