A Bayesian Machine Learning Algorithm for Predicting ENSO Using Short Observational Time Series

TL;DR

This paper introduces a Bayesian machine learning algorithm that effectively predicts ENSO events using only a 20-year observational dataset, outperforming traditional models and extending forecast skill to nearly 10 months.

Contribution

The study presents a novel Bayesian ML approach that leverages short observational data and multiscale features to improve ENSO prediction accuracy and robustness.

Findings

Outperforms model-based ensemble and standard ML forecasts

Achieves nearly 10 months of skillful prediction starting from spring

Utilizes multiscale features to extend forecast horizon

Abstract

A simple and efficient Bayesian machine learning (BML) training and forecasting algorithm, which exploits only a 20-year short observational time series and an approximate prior model, is developed to predict the Ni\~no 3 sea surface temperature (SST) index. The BML forecast significantly outperforms model-based ensemble predictions and standard machine learning forecasts. Even with a simple feedforward neural network, the BML forecast is skillful for 9.5 months. Remarkably, the BML forecast overcomes the spring predictability barrier to a large extent: the forecast starting from spring remains skillful for nearly 10 months. The BML algorithm can also effectively utilize multiscale features: the BML forecast of SST using SST, thermocline, and wind burst improves on the BML forecast using just SST by at least 2 months. Finally, the BML algorithm also reduces the forecast uncertainty of neural networks and is robust to input perturbations.