It's Worse Than We Thought: 'Gulf Stream Collapse Debunked' Becomes 'AMOC Ate My Global Warming' Which Becomes 'AMOC IS My Global Warming!'

The Daily Sceptic has the following article on the perennial climate alarmist favourite ‘Day After Tomorrow’ scenario, i.e. the collapse of the Gulf Stream due to global warming resulting in . . . . . northern Europe freezing.

Chris Morrison quotes the conclusion of a Royal Society study which says:

“If it is not possible to reconcile climate models and observations of the AMOC in the historical period, then we believe the statement about future confidence about AMOC evolution should be revised. Low confidence in the past should mean lower confidence in the future.”

But it’s worse than Chris thinks; in fact it’s worse than the authors of the study think! Not only does the study cast serious doubt upon the latest “highly likely” forecast of AMOC (Atlantic Meridional Overturning Circulation) slowdown in IPCC AR6, it actually inadvertently casts doubt upon the principal cause of global warming post 1950.

Before we examine why this is so, a quick word about the distinction between the Gulf Stream and AMOC. As Chris points out:

The Gulf Stream is part of a wider system of currents known as the Atlantic Meridional Overturning Circulation (AMOC). By bringing warmer waters from the south, it is estimated to increase coastal area temperatures in parts of the northern hemisphere by up to 5°C.

AMOC is a complex system of surface and deep water currents driven by the Thermohaline Circulation stretching from pole to pole, northwards and southwards across the equator, in the Atlantic Ocean. The Gulf Stream (oceanic component) is just one small part of that overturning circulation. The Gulf Stream is also partly wind driven, caused by the rotation of the earth which, along with the differential solar heating between equatorial regions and higher latitudes, drives the global atmospheric circulation. But the upshot is, it’s much warmer in western Atlantic coastal regions than it is in western Pacific coastal regions at the same latitude because of the heat transported northwards by the Gulf Stream.

Climate crisis zealots just love regaling us with grim fairy tales of imminent catastrophic AMOC collapse due to global warming supposedly melting the Greenland icecap, thus diluting the cold north Atlantic oceans with trillions of gallons of fresh water, effectively preventing the dense salty water sinking to depths and so shutting down the ‘conveyor belt’. If telling us that the sinful burning of fossil fuels will alternately or simultaneously drown, drench, parch, starve, boil, or incinerate us is their day job, then warning that AMOC collapse will freeze the nuts off our brass monkeys is definitely the full time hobby carried out during holidays and lunch breaks.

It came to a head a few years ago when two papers were published in 2018 warning us of a potential ‘catastrophic’ slow down in AMOC which began after 1950 and was predicted to get worse over the 21st century, supposedly because of human emissions of greenhouse gases. I covered those two studies and the inevitable MSM scaremongering hysteria here.

What I said then stands now, probably even more so:

OK. Reality check. One study (Rahmstorf’s) says that AMOC has weakened by about 15% since the 1950’s and that this is ‘unprecedented’ in the last 1600 years and therefore is mainly attributable to the sharp increase of atmospheric GHGs (which have melted the Arctic) since then. The other study suggests that the 15% decline in the strength of AMOC (not the Gulf Stream note, which is only a small part of the North Atlantic ocean circulation) “occurred either as a predominantly abrupt shift towards the end of the LIA, or as a more gradual, continued decline over the past 150 years”. They were unable to distinguish one possible scenario from the other. The important point is that it started as a result of the termination of the Little Ice Age which, despite the attempts by climate alarmists to attribute to the beginning of the Industrial Revolution and changing land use, was almost certainly an example of past natural global warming.

It is entirely possible that Rahmstorf et al’s identification of an ‘unprecedented’ decline in AMOC since 1950 is also due to a combination of multidecadal natural variability (which RAPID measurements confirm can be an order of magnitude greater than the decline predicted by climate models) and a millennial decline in the strength of AMOC instigated by not only the ‘recovery’ from the LIA, but by the most intense solar activity in 8000 years.

However, AMOC variability appears to be much more complicated than what the simple ‘freshwater hosing’ climate model experiments suggest. Climate scientists keep telling us that a melting Arctic inhibits the formation of deep salty water in the north Atlantic by adding trillions of tons of freshwater at the surface, thus slowing down the subsurface transport of deep cold water to the south, which in turn affects the ‘return’ currents of warmer surface waters from the south. In truth, nobody really understands the intricate mechanisms which drive the Atlantic thermohaline circulation and predictions of abrupt slowdowns of AMOC initiated by melting glaciers at high latitude are probably a gross over-simplification. As such, the theory of an anthropogenically enhanced or caused rapid slowdown in AMOC is very suspect, to say the least.

The fact is, AMOC variability can be driven by factors other than so called ‘freshwater hosing’. In demonstration, I give you this study: Hosed vs. unhosed: interruptions of the Atlantic Meridional Overturning Circulation in a global coupled model, with and without freshwater forcing.

However, other model simulations have shown that spontaneous changes in the AMOC can occur in the absence of freshwater inputs (Winton, 1993; Sakai and Peltier, 1997; Hall and Stouffer, 2001; Ganopolski and Rahmstorf, 2002; Schulz, 2002; Loving and Vallis, 2005; Wang and Mysak, 2006; Colin de Verdière, 2007; Friedrich et al., 2010; Arzel et al., 2011; Kim et al., 2012; Drijfhout et al., 2013; Peltier and Vettoretti, 2014; Vettoretti and Peltier, 2015). Although uncommon, these “unhosed” oscillations show that the AMOC can vary as a result of processes internal to the ocean–atmosphere system, which have been linked to oscillations in the strength of the vertical density gradient in the North Atlantic (Winton, 1993; Arzel et al., 2011; Peltier and Vettoretti, 2014) as well as to the existence of unstable states of sea-ice extent in the North Atlantic (Li, 2005; Li et al., 2010; Siddall et al., 2010; Petersen et al., 2013).

So, on the evidence we have so far, The Day After Tomorrow will never come however hard climate alarmists rub their magic lanterns in the hope of conjuring up wished for catastrophes. AMOC may however decline further and we may pass a ‘tipping point’ where the global climate (especially that of northern latitudes) shifts to a rather abrupt cold state. The wheels were probably put in motion early in the 20th century and thoroughly oiled from 1950 onwards, so such a shift may be inevitable, or it may not. I have an idea we might know for sure within the next decade or two. Meanwhile, install a multi-fuel burner before they’re banned by Western governments and stock up on plenty of wood!

Strangely, wood-burning stoves are gradually being outlawed and ever more strictly regulated in urban and suburban areas (and even on canal boats) - because of ‘air pollution’ - but the government is keen to bankrupt the country by subsidising ever more ridiculously inefficient, environmentally harmful and costly wind turbines.

But it’s natural internal variability which I’d like to focus on. In contrast to externally forced variations in AMOC (supposedly due to GHG emissions). What - if any - effect does natural internal variability have on AMOC? Quite a significant one according to that latest Royal Society study, maybe even a dominant one.

The Royal Society study mentions those two AMOC papers released in 2018 in the context of changing IPCC assessments:

Whether the AMOC will weaken in the future, or perhaps has already begun to do so, is an important question. However, this question cannot be answered with certainty, which is reflected in the different assessments of the state of the AMOC as given in the IPCC reports over the last decade. While the IPCC AR5 report concluded that there was no observational evidence of a long-term AMOC decline, in 2019, the IPCC SROCC report stated with medium confidence that the AMOC had weakened relative to 1850–1900. However, by 2021, the IPCC AR6 report went back to stating there was low confidence in a twentieth century AMOC decline. How did this happen? In 2018, two independent studies were published in Nature that concluded that the AMOC had weakened by about 15% over the course of the twentieth century [13,14]. While the temporal evolution suggested by these proxies (Caesar and Thornalley indices in figure 2) did not fully agree with the latest climate models at that time (CMIP5) that suggested only a weak downward trend [23,24], they at least agreed on the sign of the change. Yet this changed as the latest generation of climate models (CMIP6) was released that showed, in the multi-model ensemble mean, a slight strengthening of the AMOC over the historical period [25].

So the new ‘improved’ CMIP6 models suggested that there should have been a slight increase in the strength of AMOC during the recent past, contrary to (imperfect) observations. Direct observations of AMOC have only been possible since the development of the RAPID project in 2004. According to RAPID data:

The trends observed at RAPID consist of a strong weakening from 2004 until about 2010, following a partial recovery in recent years [32].

Prior to this:

Going further back in time, compilations of AMOC reconstructions suggest that the AMOC weakening in the period after 2004 followed an AMOC strengthening from about 1985–2000 [33] which is also in agreement with the results of forced model runs [34]. Prior to the 1980s few instrumental reconstructions of the AMOC exist but those that do indicate the AMOC weakened from the 1960s to the 1980s [22,31], leading to an overall weakening in the latter half of the twentieth century. The problem is that due to the decadal variability seen in the AMOC reconstructions, the sign of the long-term weakening trend depends on whether the AMOC weakened before the 1980s, i.e. from a time with little observational evidence.

This is the first admission by the Royal Society authors that there exists decadal variability in AMOC which may mask the long term trend or at least render it very uncertain if that decadal variability cannot be accurately quantified and accounted for. The kicker is, even though the CMIP5 and CMIP6 models disagree over the historical period, they both fail to match the observations and even more importantly they comprehensively fail to match the direct measurements made by RAPID since 2004.

Comparing the models’ AMOC evolution to the observational data (figure 3, lower panel), we find that neither the CMIP5 nor the CMIP6 ensemble mean are successful at representing the observational AMOC data. Looking at the only available direct continuous AMOC observations (RAPID data) we can see that the ranges of uncertainty of the 2005–2016 RAPID trend and the standard deviation of the CMIP5 and CMIP6 ensemble distributions of the same trend barely overlap (figure 4a).

So what can we conclude from this? That the models are wrong? Or that they cannot be expected to be right given their limitations? If the latter, what are those limitations? The RS authors phrase it as follows:

5. Possible reasons for the model-observation discrepancy

But why do we see such a discrepancy between model and observational data? We consider three possible reasons:

(i) The models are wrong.

(ii) The observations are wrong.

(iii) Models and observations are not expected to agree.

The authors go through all three possibilities but it’s the third possibility which I find most interesting. They say:

It is also possible that climate models should not be expected to fully agree with observations in the historical period: the multi-model ensemble mean aims to represent the overall forced response of a system, not its internal variability (as this is cancelled out when averaging over enough ensemble members). The true internal variability (i.e. that of the real-world climate system) may only be picked up by a few individual simulations or by none at all. Bonnet et al. [46] found that those model runs that fit the evolution of global mean temperature most accurately showed a declining AMOC over the twentieth century. This suggests that the internal variability of the AMOC could have muted global warming over the last decades—a trend that will reverse in future, making it more difficult to avoid crossing the 2°C threshold. The results of Bonnet et al. [46] stem from only one model and the IPCC AR6 report indicates that most climate models fail to correctly model the larger internal variability in the observed AMOC. When considering the multi-model ensemble mean here, we are not only mixing different ensemble members but also different models. This means that different models may respond to different external forcing, i.e. internal variability need not be the only reason for differences in the simulation. This is well shown by the differences between the CMIP5 and CMIP6 model ensemble means. Even when multiple models are considered, the response to external forcing can still be different. Therefore, we cannot say that the difference between the reconstructions and the model ensemble mean can be narrowed down to internal variability but we can consider the situation where the difference might be predominantly driven by internal variability.

If internal variability dominates AMOC evolution in the historical period, is there any hope for its simulation? We do not expect climate models forced by historical emissions to reproduce the weather or, for example, individual storms over the historical period—these simulations are not considered as weather forecasts. If the AMOC is more like a storm, then it should be forecast like a storm using initialized models. This is the approach of decadal climate prediction where climate models are initialized from observations. This approach has also proved successful in simulating decadal Atlantic climate variability with the Atlantic warming in the 1990s and cooling in the 1960s successfully reproduced using this approach [50,51]. Indeed, AMOC hindcasts from decadal prediction systems show similar decadal evolution to the converging lines of evidence from the observations: weakening in the 1960s and strengthening in the 1990s [52]. While decadal predictions may offer a successful way of modelling AMOC evolution, their success could be evidence of large internal variability, and consequently pointing to the unlikelihood of historical climate models reproducing AMOC variability.

These are the two paragraphs which keep giving and I’m sure that the authors were not themselves aware of just how revealing their words were. There’s quite a lot to unpack but basically it comes down to the following:

CMIP5 and 6 models do not individually simulate internal variability at all well or accurately. It is assumed that by running a large number (ensemble) of them that the error in simulating internal variability (positive and negative) will cancel itself out. Maybe. Maybe not. The point is, if internal variability is large relative to the externally forced response (man made greenhouse gases supposedly) over the reference period and the reference period is relatively short (it is, we’re talking post 1950) then it is entirely possible that internal variability will completely dominate the observed AMOC variability over that period, including the trend.

But that’s not all: decadal and multidecadal AMOC variability strongly influences northern hemisphere temperature, in the oceans and on land. 1960s cooling is far more likely due to the coincident negative phase of the Atlantic multidecadal oscillation which is linked to AMOC and that other major mode of Atlantic variability, the North Atlantic Oscillation (NAO) by a complex series of (not very well understood) teleconnections. For this very reason, a 2022 Nature study suggested that indeed, AMOC has been dominated by natural internal variability since 1900. The RS authors confirm that the initialized weather like climate models which treat AMOC as being like a storm successfully reproduce Atlantic warming in the 1990s and cooling in the 1960s. But how could they, if anthropogenic GHGs are supposedly dominating warming trends, even over decades? The authors suggest that AMOC variability has muted global warming over the past decades. But if that’s the case, it might equally have reinforced global warming over certain periods, notably the very rapid 1976-1998 warming which coincided with an upturn in AMO after the downturn during the 1960s when the northern hemisphere especially cooled significantly (hence the 1970s Ice Age scares!).

If AMOC/AMO/NAO internal variability is large compared to the alleged dominant external forcing (GHGs), then it stands to reason that over a relatively short period (since 1950), not only has internal variability dominated the ups and downs of global temperature, but it may also have dominated or at the very least obscured the alleged ‘catastrophic’ man-made global warming trend. The IPCC says no, using a very dubious statistical summation of probabilities and by largely ignoring internal variability, alleging that it ‘cancels itself out’ over the period. But what if it doesn’t? This latest analysis of AMOC suggests it doesn’t.

The Royal Society authors are even more explicit in their Discussion section on the possible climate impacts of large AMOC internal variability, but they only point to the possibility of it masking global warming, not enhancing it, which is equally possible, both in the historic period and in future projections.

Potentially, we do not expect the historical climate models and the observations to match. Internal variability may dominate observations. The multi-model ensemble spread only barely encompasses the range of AMOC observations as shown in figure 3. This suggests that natural internal variability is very large and, while it does not exclude the possibility that certain model ensemble members capture the correct size of internal variability, suggests that it may be the limit of the models’ ability to capture it. In this case, we do not expect the climate models to match observations, but can we explain the magnitude of internal variability and what are the implications for the future? If internal AMOC variability is that big, then future climate could be incorrectly projected because forced models are being tuned to internal variability. AMOC variability may have masked the global warming trend [46]. Likewise, if AMOC variability is all (or at least mainly) internal and multidecadal, the AMOC change could mask other changes in future climate.

So the models are not ‘wrong’, they’re just not fit for the purpose of simulating past and future climate change and thus informing climate policy, precisely because they do not accurately simulate natural internal variability which is large over decadal and multidecadal reference periods compared to external forcing. We’ve been conned by bad science allied with political activism and corporate vested interests.