Reanalysis of global temperature variability during the last 24,000 years

Abstract:

Constraining the evolution of surface temperature since the height of the Last Glacial Maximum (LGM; ca. 21 ka) represents a critical paleoclimatic target. Yet, despite broad efforts and an abundance of climate-proxy data, estimates vary widely, especially on a per-region basis. Here, we pair a new, globally dispersed compilation of nearly 500 sea-surface temperature geochemical proxy records with isotope-enabled CESM time-slice simulations to produce the first gridded reanalysis of surface temperature spanning the last 24,000 years using an offline ensemble Kalman Filter approach. Our reanalysis performs well when validated against globally withheld marine proxies, as well as “external” terrestrial isotope records (ice cores and speleothems) that are not directly assimilated, demonstrating our method’s ability to robustly resolve the regional characteristics of climate variability since the LGM. Therein, we refine prior estimates of global and regional temperature evolution from the LGM to present based on comparably simple methods and (or) fewer proxy records. Our reanalysis confirms that glacial surface cooling was likely -6.1±1.1˚C (2σ), corresponding to changes of -8.4±1.5˚C and -3.7±0.7˚C in the Northern and Southern hemispheres, respectively. Critically, our method also illuminates the spatial patterns of deglacial and Holocene climate change: For example, during periods of rapid North Atlantic cooling (such as the Younger Dryas, ~12 ka), we observe large warming anomalies – upwards of 1-4˚C above pre-industrial levels – in the Southern Ocean, consistent with the bipolar seesaw hypothesis. In turn, during most of the Holocene we find global temperatures were slightly cooler than pre-industrial values, voiding northern Europe and Asia where widespread warming occurred. Collectively, our reanalysis is expected to provide valuable new insights into the mechanisms driving Late Quaternary climate change in a spatially complete, dynamically consistent framework.