Machine Learning on Phase-Field Models of Sea Ice Fracture
H. Dinh, D. Giannakis, G. Stadler, and J. Slawinska
New York University
Abstract
Fractures of individual sea ice floes influence large scale parameters such as floe size distribution and basin scale rheology. The complexity of fracture patterns sharply changes as material shape, stiffness and external forcings vary. We have implemented a phase-field model of brittle fractures, on ice floes, that simulates complex fracture behavior in response to physical parameters. Additionally, we are developing a machine learn ing framework to detect and predict changes in the fracture patterns generated by any model of fracture.