Atmosphere Ocean Science Student Seminar

Constraining estimates for sea level extremes using uncertainty-permitting machine learning

Speaker: Andrew Brettin, CAOS

Location: Warren Weaver Hall 1314

Date: Friday, October 21, 2022, 4 p.m.


In this preliminary work, we use a neural network trained on CM2.6 data to constrain estimates for coastal sea level extremes using only information about the local bathymetry and atmospheric forcing variability. The neural network is trained using a maximum-likelihood loss function, which quantifies an uncertainty range for the extremes given this minimal information. We compare our predictions to a linear model baseline and assess the assumption of Gaussianity in parameterizing uncertainties. Finally, we try to identify which drivers contribute to higher uncertainties by looking at the probability distribution of the features conditioned on the predicted uncertainty.