Parametrization and generation of geological models with generative adversarial networks

TL;DR

This paper explores the use of Wasserstein GANs to parametrically generate complex geological models, effectively capturing geological structures and flow statistics for uncertainty analysis in subsurface modeling.

Contribution

It demonstrates the application of Wasserstein GANs for geological model parametrization, preserving multipoint statistics and flow features, advancing data-driven subsurface modeling techniques.

Findings

GANs preserve geological structures visually and quantitatively

Generated samples accurately reproduce flow statistics

Method effective across different permeability patterns

Abstract

One of the main challenges in the parametrization of geological models is the ability to capture complex geological structures often observed in the subsurface. In recent years, generative adversarial networks (GAN) were proposed as an efficient method for the generation and parametrization of complex data, showing state-of-the-art performances in challenging computer vision tasks such as reproducing natural images (handwritten digits, human faces, etc.). In this work, we study the application of Wasserstein GAN for the parametrization of geological models. The effectiveness of the method is assessed for uncertainty propagation tasks using several test cases involving different permeability patterns and subsurface flow problems. Results show that GANs are able to generate samples that preserve the multipoint statistical features of the geological models both visually and quantitatively. The generated samples reproduce both the geological structures and the flow statistics of the reference geology.