Ensembling geophysical models with Bayesian Neural Networks

TL;DR

This paper introduces a Bayesian Neural Network-based ensembling method for geophysical models that enhances prediction accuracy and uncertainty quantification, demonstrated on ozone prediction with significant improvements over traditional methods.

Contribution

Develops a novel data-driven ensembling strategy using Bayesian Neural Networks that models spatiotemporal variations and heteroscedastic uncertainties in geophysical projections.

Findings

49.4% reduction in RMSE for temporal extrapolation

67.4% reduction in RMSE for polar data voids

90.6% of data points within 2 standard deviations

Abstract

Ensembles of geophysical models improve projection accuracy and express uncertainties. We develop a novel data-driven ensembling strategy for combining geophysical models using Bayesian Neural Networks, which infers spatiotemporally varying model weights and bias while accounting for heteroscedastic uncertainties in the observations. This produces more accurate and uncertainty-aware projections without sacrificing interpretability. Applied to the prediction of total column ozone from an ensemble of 15 chemistry-climate models, we find that the Bayesian neural network ensemble (BayNNE) outperforms existing ensembling methods, achieving a 49.4% reduction in RMSE for temporal extrapolation, and a 67.4% reduction in RMSE for polar data voids, compared to a weighted mean. Uncertainty is also well-characterized, with 90.6% of the data points in our extrapolation validation dataset lying within 2 standard deviations and 98.5% within 3 standard deviations.