Optimizing Volcanic Hazard Modeling with Physics-Informed Neural Networks (PINNs)

Abstract

Volcanic hazards represent a constant challenge for the communities that coexist around these dynamic geological phenomena. The ability to accurately forecast volcanic activity is crucial to mitigating associated risks and protecting life and resources. In this context, Physics Informed Neural Networks (PINNs) emerge as a promising tool to improve volcanic hazard modeling, fusing the power of neural networks with deep knowledge of the physical laws that govern these complex natural phenomena. The main objective of this work is to provide an initial overview of the application and development of methodologies based on PINNs to improve accuracy and prediction in volcanic hazard modeling. By integrating physical principles through PDEs, we have noticed that PINNs overcome the limitations associated with the scarcity of data and allow us to describe the behavior of a system in an optimal way considering the scarcity of information.