By Dan Hodson
After 140 years of observations, we now know that the temperature of the surface of the Atlantic ocean slowly varied over time, cooling and warming over periods of decades (Figure 1). These slow variations in temperature sit atop the background global warming trend (A), the contrast with other regions of the globe can clearly be seen in spatial maps of SST difference (B). The term Atlantic Multidecadal Oscillation (AMO) was initially coined to describe these variations around the global mean trend, but recently the more general term Atlantic Multidecadal Variability (AMV) has been adopted by the community.
Figure 1. A) Black: Atlantic multidecadal Variability (AMV) (mean over black box). Green: mean over region outside black box
B) annual mean Sea Surface Temperatures: (1965–75) minus (1951–61).
The origin and mechanisms by which the AMV arises are still a matter of debate. It is ultimately impossible to deduce the origins from using observations alone (although we can hazard some educated guesses), so we have to turn to model studies. Some argue that the AMV arises due to internal ocean variability – involving variations in the heat transported by ocean dynamical processes, such as the Atlantic Meridional Overturning Circulation (perhaps responding to stochastic forcing from the atmosphere). Many coupled climate models do display AMV that arises due to this. 1 2. Others argue that models show that the historical AMV arose due to changes in external forcings, or question the role of ocean dynamics altogether.
Jon Robson has recently written about ongoing efforts to predict the evolution of the AMV by using ocean observations to initialize ocean models. These studies suggest an ocean-origin for the AMV is more likely. Whatever the origin of the AMV, and independent of our ability to predict it, we can still ask – what are the climatic impacts of the AMV? Again, we have to turn to models to start to answer this question. Multiple attempts have been made over the the past two decades to examine the possible impacts of the AMV on climate. Ten years ago we examined the idealized impact of a fix AMV pattern on climate in an Atmosphere-only model 1 2. We discovered significant, and potentially important, impacts on surface temperatures, rainfall and atmospheric circulation (Figure 2) – notably, these were consistent with the observational record in a number of regions.
Figure 2. (A to C) Observed JJA (Warm-Cold AMV periods). (A) Sea-level pressure.(B) Land precipitation (mm/day). (C) Land surface air temperature (°C). (F to H) As in (A) and (B) but Model response to AMV (warm – cold). D and E are AMV Warm – Cold composites from a model run with historical SSTs.
Motivated by this, during the DYNAMITE project, we repeated these experiments in a range of other atmosphere-only models, we discovered a range of similar responses, but a number of key uncertainties – e.g. the magnitude of the impact on rainfall.
Experiments such as these are the first step in elucidating the climatic impact of the AMV. However, since these experiments used atmosphere-only models with fixed sea surface temperature (SST), it wasn’t possible to investigate dynamical feedbacks – for example, how the atmospheric response to the AMV in turn affects the ocean, such feedbacks may ultimately modify the final atmospheric response. Modelling studies to date suggest that such feedbacks could be significant.
In order to address this, a new international multi-model experiment is underway to resolve these questions. It will run as part of CMIP6:DCPP – the Decadal Climate Prediction Project component of the fifth Coupled Model Intercomparision Project (there was no CMIP4) – the Decadal Climate Prediction Project . The experiments within DCPP will examine the impact of the AMV in coupled climate models. Each of the models in the experiment ensemble will allow SSTs in the models to evolve with the underlying ocean model, but will periodically nudge those in the Atlantic towards a warm AMV pattern. The idea behind this is to drive the models with a warm AMV, but without restricting the ocean coupling or responses. Reading are talking part in this international effort by using the MetUM-GOML2 coupled mixed layer ocean model developed by Nick Klingaman and Linda Hirons here in Reading. First results are just beginning to arrive, and it looks like we may have some interesting differences from the old AGCM results – most notably, the AMV appears to have a significant impact on the Pacific ocean across the globe. If these results are born in other models, it may point to a greater role of the Atlantic in modulating global climate than has hitherto been expected. Watch this space!