Subsurface Model

Subsurface modeling aims to create accurate and efficient representations of geological formations and fluid flow within them, primarily to optimize resource extraction and environmental management. Current research heavily utilizes deep learning, employing architectures like Fourier neural operators, DeepONets, and convolutional neural networks, often combined with graph neural networks for handling complex geometries, to build surrogate models that significantly accelerate simulations compared to traditional methods. These advancements are crucial for applications ranging from geological carbon sequestration and enhanced oil recovery to groundwater resource management, enabling faster, more cost-effective decision-making in high-stakes scenarios. The focus is on improving model accuracy, particularly for extrapolation and handling noisy or limited data, while maintaining computational efficiency.