Catalogue of strong nonlinear surprises in ocean, sea-ice, and atmospheric variables in CMIP6

Abstract. The Coupled Model Intercomparison Project Phase 6 (CMIP6) archive was analysed for the occurrence of Strong Nonlinear Surprises (SNS) in future climate-change projections. To this end, we built an automated detection algorithm to identify SNS in a reproducible manner. Two different types of SNS were defined: abrupt changes measured over decadal timescales and slower state transitions, too large to be explained by the forcing without invoking strong internal feedbacks in the climate system. Data of 54 models were analysed for five shared socio-economic pathways for ocean, sea ice, and atmospheric variables. The algorithm isolates regions of at least 106 km2 and utilizes stringent criteria to select SNS. In total 73 SNS were found, divided in 11 categories of which 4 apply to abrupt change and 7 to state transitions. Of the identified SNS 45 % relate to sea-ice cover, 19 % to ocean currents, 29 % to mixed layer depth, and 7 % to atmospheric systems like the Intertropical Convergence Zone. For each category, probability density functions for time-windows of maximal change indicate SNS occurring earlier and at lower global temperature rise than assessed in previous reviews, in particular the ones associated with winter Arctic Sea ice disappearance, northern North Atlantic winter mixed layer collapse and subsequent transition of the Atlantic Meridional Overturning Circulation (AMOC) to a weak state in which the cell associated with North Atlantic Deep Water involved has vanished. This catalogue emphasizes the possibility of SNS already below 2 °C of global warming, even more than the previous assessments based on CMIP5 data.