CarbonNet: How Computer Vision Plays a Role in Climate Change? Application: Learning Geomechanics from Subsurface Geometry of CCS to Mitigate Global Warming

TL;DR

A new approach using computer vision to predict the land surface displacement from subsurface geometry images for Carbon Capture and Seques- tration (CCS) and shows ResNetUNet outperforms the others thanks to its architecture in static mechanics problem, and LSTM shows comparable performance to transformer in transient problem.

Abstract

We introduce a new approach using computer vision to predict the land surface displacement from subsurface geometry images for Carbon Capture and Sequestration (CCS). CCS has been proved to be a key component for a carbon neutral society. However, scientists see there are challenges along the way including the high computational cost due to the large model scale and limitations to generalize a pre-trained model with complex physics. We tackle those challenges by training models directly from the subsurface geometry images. The goal is to understand the respons of land surface displacement due to carbon injection and utilize our trained models to inform decision making in CCS projects. We implement multiple models (CNN, ResNet, and ResNetUNet) for static mechanics problem, which is a image prediction problem. Next, we use the LSTM and transformer for transient mechanics scenario, which is a video prediction problem. It shows ResNetUNet outperforms the others thanks to its architecture in static mechanics problem, and LSTM shows comparable performance to transformer in transient problem. This report proceeds by outlining our dataset in detail followed by model descriptions in method section. Result and discussion state the key learning, observations, and conclusion with future work rounds out the paper.

Figures (11)

• Figure 1: A summary of workflow of this project, and the goal is to leverage computer vision intelligence for carbon storage surveillance in order to mitigate climate change. ResNetUnet assists the image prediction in terms of land surface displacement, and we use LSTM and transformer to predict displacement change over time, that is a video problem in computer vision domain.
• Figure 2: The layout of the computational problem. The end of the computational program is to capture the displacement given the configuration of the loading, material distribution and boundary conditionsbathe2007finite.
• Figure 3: The architecture of ResetNetUnet, designed by ourselves for the displacement prediction of heterogeneous computational mechanics problem. The detail of the architecture is provided in the supplement section.
• Figure 4: The implementation of the ResNet block in the ResNetUNet architecture.
• Figure 5: The architecture of the LSTM model to predict the evolution of vertical displacement.