Paper No. 166-12
Presentation Time: 4:30 PM
CENOZOIC MODEL SYSTEMS FOR A NO-ANALOGUE FUTURE
Due to rising atmospheric greenhouse gas concentrations, Earth’s climate system is moving towards a state with no close precedent in the instrumental record or societal experience, which challenges both adaptation planning and model testing and predictive accuracy. The large and well-documented climate variations in the Cenozoic record thus provide an essential series of geological model systems for studying the effects of climate change on ecological, cryospheric, anthropogenic, and other Earth system components. No past period is a perfect analogue for the 21st century, but many carry instructive parallels to different aspects of current and future global change (e.g. carbon cycle disruptions, abrupt changes, past warm periods). Here we identify the closest geological and historical analogues for future climates via climate-analog analyses for two future scenarios (Representative Concentration Pathways [RCPs] 4.5 and 8.5) and six geohistorical time periods: Early Eocene, Mid-Pliocene, Last Interglacial, Mid-Holocene, pre-Industrial climates, and a 20th-century snapshot (1940-1970 CE). All climate-analog analyses employ the Mahalanobis distance metric, four variables (summer and winter temperature and precipitation), and a three-model ensemble (HadCM3, GISS, CCSM). Under RCP4.5, climate stabilizes at Pliocene-like conditions by 2040 CE. Under RCP8.5, future climates most closely resemble Mid-Pliocene climates by 2030 CE and resemble Eocene climates by 2150 CE. A small fraction of future climate gridcells had no close geological analogue, reaching 1% for RCP4.5 and 8.7% for RCP8.5. Both species distribution models and community-level models show a steady decline in skill as climate novelty increases, based on hindcasts for the late Quaternary. More work is needed on climate and biospheric dynamics of the Pliocene and other warm periods of the Cenozoic, given their close similarity to the expected climates of the coming decades.