The orchestra has been playing in the Pacific Ocean for over a century. Some instruments carry a fast melody, others draw out a slow bass line, and together they create a pattern of warm and cold patches on the water's surface. Until now, climatologists could only hear the overall chord: the method for El Niño and the Pacific Decadal Oscillation (PDO) produced static maps on which it was impossible to tell who was playing at any given moment.
The new method, called extreme-point symmetric mode decomposition (ESMD), works like a sound engineer who separates a concert recording into individual tracks. For each ocean point, it extracts oscillations with different periods, from the seasonal cycle to half-century waves. A test at the equator showed that the sum of the extracted modes almost exactly reproduces the well-known Niño-3.4 index, with a correlation of 0.84.
Once the tracks were separated, it became clear that El Niño's main parts are played by two modes: a quasi-four-year mode with a period of 2.6 years and a quasi-five-year mode with a period of 5.5 years. The PDO has its own lineup: a quasi-five-year mode (5.1 years), a decadal mode (11 years), and an interdecadal mode (22 years). The shared quasi-five-year mode is the very instrument that makes the two phenomena sound alike. It is this mode that explains the "ENSO-like" variability of the PDO.
The integrated version of ESMD allowed researchers to trace how anomalies move across the ocean. In the 1997–2000 episode, the warm El Niño anomaly off the coast of South America gave way to cold La Niña, and simultaneously in the North Pacific there was a shift from a warm to a cold regime. The orchestra is not just playing; it is changing soloists.
The method can be applied to any climate time series, from temperature to precipitation. This will help improve forecasts of El Niño and the PDO, and thus allow better preparation for droughts, floods, and changes in fisheries. The orchestra can finally be heard in its individual parts.
The ESMD method was developed earlier by the author and patented in China in 2019 as the "Integrated ESMD method for spatiotemporal data analysis."