Monday, 30 January 2017

When will the Gulf Stream ‘shut down’ and how will it affect our part of the World?

Quick answer: It probably won’t shut down completely in the next century or two. It will almost certainly weaken progressively during the 21st century, and will have an effect on both UK climate and US sea level. 

 This means that the UK may experience an amount of cooling relative to the global average and the long-term local average, and the US Eastern seaboard may experience an amount of sea level rise greater than the global average and the local long-term average, both within this century.

In simple terms, even if the Gulf Stream does not shut down completely, the impact of a weakening of the system will be felt this century, and conceivably within the next 20-30 years. It has already been felt at least once (2009-2011), and the current CO2 concentration pathway suggests that this kind of thing can easily be repeated, is likely to be repeated.

Back in 2006 I spent some time trying to understand how the Gulf Stream worked and what the implications of climate changes would be for me, living in the UK. This included exchanging some emails with a few of the people who were active at the time in analysing the data and models, and a lot of reading.

In 2009, the UK Climate Impacts Programme (UKCIP09) published it conclusions about projected changes to the UK in the 21st century, which also leaned heavily on similar sources to infer likely scenarios for change to our climate.

Since that time, the issue has stayed in the scientific literature, and occasionally pops up as a matter of concern on the internet or in the media.

Most recently, the technically reliable RealClimate has drawn my attention again to the matter in a post: The Underestimated danger of a breakdown of the Gulf Stream System. 

This summarises a paper which draws attention to some of the limitations of the modelling of the Gulf Stream, or AMOC.

Whilst it is a very useful reference, including links to several recent developments in both analysis and modelling, it does not address a question which I raised in the comments section but have not yet had answered: When is this going to happen?

Even more recently, The open access Journal of Ocean Science has published a paper which, though it focuses on a different aspect (sea level rise), has a strong section on the AMOC and some interesting material on modelling: Changes in extreme regional sea level under global warming.

The Scripps paper (Wei Liu et. al.) applies a flux correction to better model the freshwater exchange in the system on which so much depends. It suggest that the Gulf Stream could lose about a third of its strength by 2100.

The Brunnabend et al  paper uses a reasonably new high resolution ocean model, POP, which is able to capture the mesoscale eddy formations in the north Atlantic and the impact this has on regional sea level rise. It also, as you might expect, provides a greater resolution for modelled AMOC flow under the scenarios used (RCP8.5). The model runs out to the end of this century, from an original baseline of 1950, and a comparison between the first and last twenty years of the model.

In that model: The maximum AMOC at 26◦ N decreases from about 20 Sv to about 5 Sv (red curve in Fig. 4a). The spatial pattern of the AMOC does not change, but the North Atlantic Deep Water shallows by about 1000m (Fig. 4b–c). The maximum strength of the AMOC at 35◦ S decreases (blue curve in Fig. 4a) by more than 60 %. The decline in the AMOC causes a rise in mean DSL of up to 0.4 m near the North American continent, mostly because of a redistribution of ocean mass towards these regions (see Fig. 1a).

*Note: Sv refers to Sverdrups: "...used almost exclusively in oceanography to measure the volumetric rate of transport of ocean currents..." This is derived from a model which assumes a high CO2 concentration pathway and associated global temperature change, so might be considered a ‘worst case’ result.

Where does this get us?

So far, I am at the stage where I’m reasonably confident that the Gulf Stream will demonstrate a severe case of Hiccups over the next 80 years, with has already begun. As in 2009-2010, there will be years when the current slows by 30% or more, resulting in seasons which are noticeably cooler than the long-term trend would otherwise suggest. 

This is the UKMO comment on the Winter of 2010, which followed the most recent hiccup:

The following represents an assessment of the weather experienced across the UK during winter 2009/10 (December 2009 to February 2010) and how it compares with the 1971 to 2000 average (the period used for the seasonal forecast). 
Mean temperatures over the UK were 2.0 °C below the 1971-2000 average during December, 2.4 °C below average during January and 1.6 °C below average during February. The UK mean temperature for the winter was 1.6 °C, which is 2.0 °C below average, making it the coldest winter since 1978/79 (1.2 °C). Over England and Wales it was also the coldest since 1978/79. Over Northern Ireland and Scotland, winter 2009/10 was comparable with 1978/79 and 1946/47, with only winter 1962/63 significantly colder in series from 1910. For northern Scotland, it was the coldest winter on record, with the highest number of frosts. A generally mild first 10 days in December was followed by a colder period. This cold spell persisted for the first half of January, with some severe frosts. After mid-month, temperatures rose to around normal before a return to colder conditions. These persisted for most of February, with only a few brief milder interludes mostly in the west and south.

For the sake of simplicity, I’ll assume that a slowdown of 30% roughly equals a drop in mean winter temperature of around 2 degrees. I am not aware of a similar ‘spike’ in the US Eastern Seaboard sea level and am interested to know if this has been researched.

This occasional and regional cooling has to be taken into the context of expected rises in global mean temperature (which will go on irrespective). As things stand, depending on the mitigation scenario, this is likely to mean a temperature rise of between 2 and 4 degrees out to 2100. Again, for simplicity, I’m going to use a median figure of 3c by 2100.

So, by about 2070, the cooling effect of the AMOC slowdown will be offset by the warming effect of the global mean, which at that stage would be between 1.5 and 2.5 degrees. This suggests that the UK will experience a number of cooler winters, comparable to 2010, for the next forty years. After that, its more likely that the cooler winters will be statistically more like recent seasonal averages, i.e., mild.

Meanwhile, back on the US Eastern seaboard, I am speculating that there is a chance that AMOC weakening events could enhance dynamic sea level rise and add 20-30 cm of extra level by the end of the century. This means that a rise of a metre in some areas is plausible by 2060-2070. Add the effect of storm surges and increased offshore cyclonic activity, and New York could experience ‘Sandy-type’ surges on several occasions, with the probability and frequency rising as time goes on.

This is a complex area of study and I have not really done it justice here; my hope is that the average lay person can get a bead on what a slowdown of the Gulf Stream actually means for us and our children.