From Sunspot-Climate Correlations to Operational Decadal Climate Forecasts with Earth System Models: A History of Decadal Climate Variability and Predictability Research in the Modern Era

Decadal climate variability (DCV) and predictability research has a multi-century history, but it was always associated with 11- and 22-year Sunspot cycles and went out of mainstream climate science because of transient correlations between Sunspot numbers and climate, a lack of understanding of the ocean-atmosphere system, and a lack of long-term measurements of solar radiative and particle fluxes. The term DCV was coined when synergistic developments in climate science instigated an explosive growth in the mid-1990s which led to DCV’s Modern Era (ME). In this ME, multidecades to century long time series of in situ, satellite-based, and model-assimilated data led to the discovery of new or little-known DCV phenomena and their associations with worldwide climate variability. Among these discoveries were the Interdecadal Pacific Oscillation (IPO)/Pacific Decadal Oscillation (PDO) and the tropical Atlantic sea surface temperature (SST) gradient (TAG), and the Atlantic Multidecadal Oscillation (AMO). Concurrently, the development of global climate models facilitated simulations of these DCV phenomena which began to resemble the observed phenomena.

These observations and simulations became the foci of numerous international workshops on DCV from mid-1990s for the next 15 years. Every aspect of DCV and its predictability was discussed in these workshops, with spirited discussions about, for example, quality of SST data in much of the world’s oceans before the 1950s, statistical significance of decadal spectral peaks in relatively short time series, and the speed of oceanic Rossby waves in the-then low-resolution global climate models. New research was discussed threadbare, many a times before the papers were published in peer-reviewed journals. It was a very exciting time to be working on DCV research at the beginning of the ME!

Then, in the mid-2000s, experimental multiyear to decadal predictions with global climate models began to inspire hope that some aspects of the DCV phenomena and possibly their worldwide impacts may be predictable with usable skill, spurring serious efforts for decadal climate prediction. Two other developments simultaneously increased this new interest in decadal climate prediction with global models – discovery of substantial societal impacts of DCV phenomena, and DCV’s confounding of detection and attribution of anthropogenic climate change. So, now, understanding and predicting DCV and its impacts changed from a “merely” intellectual, emerging scientific topic on the horizon to a matter of great practical importance. The impetus provided by these twin motivations has sustained the interest and commitments by the World Climate Research Program to include decadal prediction in the 5^th and 6^th editions of the Coupled Model Intercomparison Project (CMIP). Even though the multiyear to decadal prediction skill is still low and not practically usable in many parts of the world, the World Meteorological Organization’s Global Annual to Decadal Climate Updates now operationally publish five-year climate forecasts of surface air temperature, precipitation, and integrated quantities such as global-average surface air temperature, and indices of phenomena such as the IPO and the Atlantic Multidecadal Variability. Many climate modeling centers participate in these endeavors and continue to improve data assimilation and ensemble sizes, along with post-production processing, to increase the skill, reliability, and applicability of multiyear to decadal climate forecasts. This is truly an epochal achievement.

Some of the most important markers on the road to the present are in situ, satellite-based, and model-assimilated Earth System observations; contributions of curiosity-driven DCV research by dozens, if not hundreds, of researchers worldwide; CMIPs; DCV impacts research; and computing power and facilities unthinkable in the beginning of the ME. Also, the Sun has risen again! Satellite-based and in situ observations of solar visible and ultraviolet radiation, and charged particles; and galactic cosmic rays in the last 20-30 years began to build a record of solar variability beyond just Sunspot numbers. These observations and consequent improved understanding of stratospheric dynamics, stratosphere-troposphere-ocean interactions, the use of these radiation and particle data in empirical studies, and better understanding of DCV generated by coupled ocean-atmosphere dynamics have led to the recognition that the untold number of Sun-climate researchers who toiled over centuries were right after all. Their many frustrations about Sunspot-climate correlations inexplicably changing signs and their steadfast belief in the face of ridicule are now being rewarded. Practically all DCV phenomena – the IPO, the TAG, the NAM, the SAM, and others – are now being conceptualized and modeled as combinations of externally-forced and internally-generated variabilities. This is the second epochal achievement of the ME.

Although the progress has been nothing short of astounding, there are also problems which I will mention and discuss after reviewing the progress outlined above. I will also discuss many early premonitions about DCV that have come true. Having been a front-row participant and observer in this endeavor, I will end this talk with some personal recollections of DCV research researchers in the ME.