When Will We Observe a Significant Trend in the AMOC?

Abstract:

We consider an observed trend of the Atlantic Meridional Overturning Circulation (AMOC) at the RAPID 26°N array (2004-2018) and use pre-industrial control runs of CMIP6 models as a benchmark to assess whether the trend is consistent with natural variability. We also consider the sensitivity of the trend to statistical methods used to calculate it from the raw data.

The observed trend is situated relative to a PDF of trends inferred from short time series of the model data and is found to be insignificant at the 95% level. Further analysis reveals that the observed trend must persist for 4-6 additional years to achieve significance at the 95% level, assuming natural variability in the models is comparable to that in nature.

The sensitivity of the results to how the annual cycle is removed is also considered. Specifically, we compare removing a constant annual cycle with one whose year-to-year amplitude can vary. The latter is achieved by following a multi-step algorithm described in Pezzulli et al. (Journal of Climate (2005) 18(1) 71-88). We find that permitting a variable-amplitude annual cycle allows for trends of a given amplitude to be detected from a shorter timeseries.

Finally, we address the question of whether the results are unique to 26°N or whether they are robust across latitudes and attempt to identify the presence of an ‘optimal latitude’, one at which a significant AMOC trend would emerge first.