Tag
This paper presents a workflow using text-to-image foundation models and variational autoencoders to rank competing geologic interpretations based on their consistency with hydraulic-head observations, improving early-stage subsurface decision-making.