Real-Time Flood Simulation using Physics-Informed Operator Learning: A Novel Approach for Efficient and Accurate Predictions.

Abstract Submission: Flood simulation is an essential tool for assessing and mitigating the severe impacts of floods on human populations and infrastructure. With the increasing need for real-time, high-resolution flood predictions, particularly during rapidly evolving flood events, there is a pressing demand for more efficient computational approaches. However, traditional methods, such as finite difference or finite element simulations, often require extensive computational resources due to the complexity and non-linearity of flood dynamics in natural environments, making real-time applications challenging. This research introduces a novel approach to real-time, high-resolution flood simulation using physics-informed operator learning. By integrating machine learning techniques with domain-specific physics, the proposed framework significantly reduces computational costs without compromising simulation accuracy. This study advances the use of physics-informed machine learning for addressing complex environmental challenges. The synergy of data-driven models and physical principles enables real-time, high-resolution flood simulations that can improve flood preparedness, response, and the protection of vulnerable communities.

Learning Objectives/Expected Outcome (Optional) :