Extended-range statistical ENSO prediction through operator-theoretic techniques for nonlinear dynamics

Abstract

Forecasting the El Niño-Southern Oscillation (ENSO) has been a subject of vigorous research due to the important role of the phenomenon in climate dynamics and its worldwide socioeconomic impacts. Over the past decades, numerous models for ENSO prediction have been developed, among which statistical models approximating ENSO evolution by linear dynamics have received significant attention owing to their simplicity and comparable forecast skill to first-principles models at short lead times. Yet, due to highly nonlinear and chaotic dynamics (particularly during ENSO initiation), such models have limited skill for longer-term forecasts beyond half a year. To resolve this limitation, here we employ a new nonparametric statistical approach based on analog forecasting 1 , called kernel analog forecasting (KAF) 2, 3 , which avoids assumptions on the underlying dynamics through the use of for nonlinear kernel methods for machine learning and dimension reduction of high-dimensional datasets. Through a rigorous connection with Koopman operator theory for dynamical systems, KAF yields statistically optimal predictions of future ENSO states as conditional expectations, given noisy and potentially incomplete data at forecast initialization. Here, using

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

“…The model error often comes from the incomplete understanding of nature and/or the inadequate spatiotemporal resolutions in these models (Palmer, 2001;Kalnay, 2003;Majda and Chen, 2018). More recently, machine learning techniques have become prevalent in forecasting ENSO and many other climate phenomena (Ding et al, 2018;Ham et al, 2019;LeCun et al, 2015;Wang et al, 2020). These machine learning approaches exploit sophisticated neural networks or other nonparametric methods to recover the complex dynamics in nature.…”

Benchmarking prediction skill in binary El Niño forecasts

“…There is ample evidence for the close relationship between El Niño and climate anomalies across the globe (Wallace et al 1998;White et al 2014;Luo and Lau 2019). As El Niño evolves at a slower rate compared to the weather system and is one of the most predictable climate fluctuation on the earth (on lead times of up to several months) (Goddard et al 2001;Wang et al 2020b), it can be used in statistical prediction models to make seasonal forecasts which are unfeasible for dynamical prediction models. One example is the seasonal forecast of tropical cyclones (Wahiduzzaman et al 2020).…”

Simple El Niño prediction scheme using the signature of climate time series

“…Although previous studies suggests that the predictions with expensive climate models beat purely statistical predictions, statistical prediction still seems to have value because of their simplicity (e.g., Penland & Magorian, 1993). Recently, elaborate and quite skillful predictions have been performed based on machine learning, or deep learning, with the use of past oceanic SST and subsurface information (Wang X, 2020;Ham et al, 2019). Nonetheless, there seem to exist few practical prediction studies that only employ the series of multi-dimensional climate indices as learning dataset.…”