AMOC Trends From 1850 to 2100 At Interannual To Multi-Decadal Time Scales Corroborated By Changes In Salinity Budget

Abstract The Atlantic Meridional Overturning Circulation (AMOC) is a key player in the global coupled ocean-atmosphere climate system. To characterise the potential of an AMOC slowdown, a past and future trend probability analysis is applied using 16 models from the Coupled Model Intercomparison Project Phase 5. We determine the probability of AMOC annual to multidecadal trends under the historical period and two future climate scenarios (`business-as-usual’ scenario - RCP8.5 and `stabilisation’ scenario - RCP4.5). We show that the probability of a AMOC decline in model data shifts outside its range of intrinsic variability (determined from the pre-industrial control runs) for sustained 5-year trend or longer. This suggests that interannual AMOC events are not significantly affected by future climate scenario, and so potentially neither by anthropogenic forcing. Furthermore, under the ‘business-as-usual’ scenario the probability of a 20-year decline remains high (87\%) until 2100, however in a ‘stabilisation’ scenario the trend probability recovers its pre-industrial values by 2100. A 20-year unique event is identified from 1995 to 2015, marked by simultaneous unique features in the AMOC and salinity transport that are not replicated over any other 20-year period within the 250 years studied. These features include the maximum probability and magnitude of an `intense’ AMOC decline, and a sustained 20-year decline in subpolar salinity transport caused by internal oceanic processes (as opposed to external atmospheric forcing). This work therefore highlights the potential use of direct salinity transport observations, and ensemble mean numerical models to represent and understand changes in past, present, and future AMOC.