High frequency intrinsic modes in El Ni\~no Southern Oscillation Index

TL;DR

This paper analyzes daily Southern Oscillation Index data, identifying high-frequency oscillations linked to geophysical phenomena, reconstructs signals for forecasting, and investigates the stochastic nature of extreme events.

Contribution

It uncovers specific high-frequency oscillations in the SOI and links them to known geophysical processes, providing a basis for improved forecasting and understanding of ENSO dynamics.

Findings

Identified significant high-frequency oscillations at 24, 27, 37, 76, 100, and 365 days.

Reconstructed signals based on these periods for 9-month forecasts.

Discovered non-random distribution of extreme event return times.

Abstract

Recent $daily$ data of the Southern Oscillation Index have been analyzed. The power spectrum indicates major intrinsic geophysical short periods. We find interesting ``high frequency'' oscillations at 24, 27, 37, 76, 100 and 365 days. In particular the 24 days peaks may correspond to the Branstator-Kushnir wave, the 27 days may be due to the moon effect rotation, the 37 days peaks is most probably related to the Madden and Julian Oscillation. It is not yet clear the explanations for the 76 days which may be associated with interseasonal oscillation in the tropical atmosphere; the 100 days could be resulting from a mere beat between the 37 and 27 periods, or the 76 and 365 days. Next these periods are used to reconstruct the signal and to produce a forecast for the next 9 months, at the time of writing. After cleansing the signal of those periodicities a detrended fluctuation analysis is performed to reveal the nature of the stochastic structures in the signal and whether specific correlation can be found. We study the evolution of the distribution of first return times, in particular between $extreme$ $events$. A markedly significant difference from the expected distribution for uncorrelated events is found.