Using Network Theory and Machine Learning to predict El Ni\~no

TL;DR

This paper introduces a hybrid model combining classical time series analysis and machine learning, utilizing climate network properties to improve El Niño predictions up to one year ahead, outperforming existing models at shorter lead times.

Contribution

The paper presents a novel hybrid approach that integrates ARIMA and neural networks with climate network attributes to enhance long-term El Niño prediction accuracy.

Findings

Hybrid model outperforms CFSv2 ensemble for 6-month lead predictions.

Predicts 12-month lead time with similar skill as shorter predictions.

Uses climate network topology as input features for neural network.

Abstract

The skill of current predictions of the warm phase of the El Ni\~no Southern Oscillation (ENSO) reduces significantly beyond a lag of six months. In this paper, we aim to increase this prediction skill at lags up to one year. The new method to do so combines a classical Autoregressive Integrated Moving Average technique with a modern machine learning approach (through an Artificial Neural Network). The attributes in such a neural network are derived from topological properties of Climate Networks and are tested on both a Zebiak-Cane-type model and observations. For predictions up to six months ahead, the results of the hybrid model give a better skill than the CFSv2 ensemble prediction by the National Centers for Environmental Prediction (NCEP). Moreover, results for a twelve month lead time prediction have a similar skill as the shorter lead time predictions.