Uncovering Bifurcations in Modern Weather Data

Dates: January 2023 - June 2024

We preformed several experiments to see if a Long Short Term Memory Model (LSTM) trained with the early warning signals of normal form bifurcations could be used to predict the occurrence of heatwaves and cold snaps in weather station data. This project is currently on hiatus do to the start of the Software Solutions for Physical Risk Assessments project.

Top: Cooking up some time-series in the Narval Cluster

Bottom: (L) Detected fold bifurcation, (R) Increasing probability of Bifurcation