The Challenge of Low-Frequency ENSO Variability (Invited)

The tropical Pacific is connected to patterns of drought and flooding throughout the world. Global climate models give ambiguous projections about changes in this region, yet anticipating these changes will be crucial to adaptation strategies during the coming decades. Here, we use proxy, observational, and climate model data to address two fundamental questions that are not resolved in our view. First, what is the relationship between frequency and variance in the tropical Pacific? And, second, do observations, climate models, and proxy records provide a consistent view of tropical Pacific SST variability across interannual, decadal, and centennial timescales? To explore these questions, we analyze the power spectra of proxy records from the tropical Pacific as well as observational and climate model data from the same region. Observations suggest that variance is concentrated within the canonical 2-7 year window, but results are ambiguous on longer timescales due to the shortness of the records and other limitations of instrumental data. High-resolution coral δ18O records suggest that the window of ENSO variance extends through multidecadal periods, with exceptionally strong variability concentrated in the decadal band during the late 19th century. Lower-resolution proxy records imply a continuum of variance that increases at lower frequencies through the multicentury time scale. These paleoclimatic analyses suggest that the 20th century does not capture the full range of ENSO variability, even in the last millennium. Climate models, in contrast, support ENSO variability at timescales of 2-5 years, with many models exhibiting overly regular oscillations between El Niño and La Niña conditions. Proxies and observations argue for more energetic low-frequency variability than is seen in the variance spectra of simulated tropical Pacific SST, suggesting that GCMs may be missing key low-frequency behaviors in their internal variability. GCMs that do not capture the observed low-frequency ENSO variability will provide an incomplete representation of future climate conditions and of regional vulnerabilities to prolonged climate extremes.