Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Well Log-Guided Synthesis of Subsurface Images from Sparse Petrography Data Using cGANs

Ali Sadeghkhani, A. Assadi, B. Bennett, A. Rabbani

Abstract

Pore-scale imaging of subsurface formations is costly and limited to discrete depths, creating significant gaps in reservoir characterization. To address this, we present a conditional Generative Adversarial Network (cGAN) framework for synthesizing realistic thin section images of carbonate rock formations, conditioned on porosity values derived from well logs. The model is trained on 5,000 sub-images extracted from 15 petrography samples over a depth interval of 1992-2000m, the model generates geologically consistent images across a wide porosity range (0.004-0.745), achieving 81% accuracy within a 10\% margin of target porosity values. The successful integration of well log data with the trained generator enables continuous pore-scale visualization along the wellbore, bridging gaps between discrete core sampling points and providing valuable insights for reservoir characterization and energy transition applications such as carbon capture and underground hydrogen storage.

Well Log-Guided Synthesis of Subsurface Images from Sparse Petrography Data Using cGANs

Abstract

Pore-scale imaging of subsurface formations is costly and limited to discrete depths, creating significant gaps in reservoir characterization. To address this, we present a conditional Generative Adversarial Network (cGAN) framework for synthesizing realistic thin section images of carbonate rock formations, conditioned on porosity values derived from well logs. The model is trained on 5,000 sub-images extracted from 15 petrography samples over a depth interval of 1992-2000m, the model generates geologically consistent images across a wide porosity range (0.004-0.745), achieving 81% accuracy within a 10\% margin of target porosity values. The successful integration of well log data with the trained generator enables continuous pore-scale visualization along the wellbore, bridging gaps between discrete core sampling points and providing valuable insights for reservoir characterization and energy transition applications such as carbon capture and underground hydrogen storage.
Paper Structure (4 sections, 3 figures)

This paper contains 4 sections, 3 figures.

Table of Contents

  1. Introduction
  2. Method and Theory
  3. Results
  4. Conclusions

Figures (3)

  • Figure 1: Comparison between training images (top two rows) and synthetic images (bottom two rows) generated by the cGAN model. Each image, with dimensions of 256 × 256 pixels, is labeled with its corresponding porosity value, demonstrating the model's ability to generate geologically consistent images across various porosity ranges.
  • Figure 2: Quantitative validation of generated images showing calculated porosity versus porosity labels (with corresponding porosity ranges shown on x-axis). Green points indicate samples within the acceptable 10% margin of their target porosity range, while red points show outliers. The model achieves 81% accuracy within the specified margin.
  • Figure 3: Well log-guided image synthesis showing the porosity log (blue curve) and corresponding synthetic thin section images generated at selected depths. The generated images demonstrate the model's ability to produce geologically consistent representations based on well-log porosity values, enabling continuous visualization of pore-scale features along the wellbore.