On Climate, and Causes in Complex Systems

Why do so many travellers, such as those marooned by the Eyjafjallalokull ash cloud, invoke a panic response on finding they can’t leave a foreign land? I remember that when I was on a memorable trip to Albania in, if I recollect correctly, 1996, our group was playing leapfrog, as it were, with another minibus full of tourists, along a road to the coast. When we stopped – often – for a passenger to relieve the symptoms of one or other of the local stomach-bugs, the other bus passed us, only for us to see them stopped by the roadside a few minutes later. Eventually we pulled up beside them to chat. It turned out that Albania’s borders were shut, in the hope of trapping whoever had blown up the Tirana police-chief. The other group were cutting their trip short. So illogical. We simply carried on with our holiday. [I drafted this a few days ago, but, since then, I’ve heard, Radio 4’s Today programme is discussing the psychology – and even genetics – of the have-to-get-home phenomenon, right now!].

It beats me why so many people spent thousands of euros hiring cars to drive across Europe. Surely staying until the ash alert blew over would have been both cheaper and less stressful.

Whatever they did, though, even those who have spent the last week hitch-hiking from Athens to Calais abroad cannot fail to have heard that, apparently, global warming will lead to more volcanic eruptions.

Is this something we should worry about?

In short, no.

Volcanoes and ice ages

The scare seems to be based on a study of the end of the last ice age:

“Huybers and Langmuir spliced two databases of volcanic eruptions worldwide over the last 40,000 years.

Eruption levels stayed low until around 12,000 years ago, then suddenly they suddenly shot up. The melting ice released so much pressure that the newly liberated volcanoes erupted at up to six times their normal rate, the researchers estimated.

The inferno lasted for 5,000 years and could have pumped enough CO2 into the atmosphere to raise concentrations between 40 and 50 parts per million, the researchers estimate. Changes in ocean chemistry probably released the rest.”

I love the eruption levels “suddenly” shooting up “suddenly”!

Now, a couple of kilometres of ice over a volcano is one thing. It’s reasonable to suppose that would prevent the pressure in a magma chamber that would otherwise have caused an eruption from doing so.

But melting a kilometre or so of ice takes quite some time. And besides, since the last ice age, there aren’t so many ice-sheets left. Worst case, in a few centuries, perhaps, we could feel the effects of some pent-up volcanic activity.

In the meantime, the worst that could happen is that some eruptions are brought forward by a few years.

The hype around the volcano scare exploits our innate difficulty in conceiving long periods of time. It also resonates with research a while back which noted more eruptions at certain times of year. The suggestion was that particular weather conditions – changes in pressure – around volcanoes, could set them off.

This triggering is an entirely different kettle of fish.

Volcanoes and the weather or short-term climate change

Consider a simple model of volcanic eruptions as the sudden release of something we might call “pressure” that builds up over time. Let’s suppose that the main cause of the build-up of “pressure” is geological. Let’s also assume that the weather can cause seasonal variations in pressure. In this model, eruptions will occur when the total pressure crosses some threshold, as in the following diagram:

Because I’m lazy, and Powerpoint is a step back from pen and paper (and reverting to that and scanning is a hassle right now), I’ve shown the total pressure (dotted line) as the sum of the geological pressure and seasonal variations for the first eruption only, but hopefully you get the idea.

It’s not very usual for volcanoes to erupt every 3-5 years, of course – 50 years or so might be more usual – and in real life every eruption is different.

Hopefully, though, it’s fairly easy to see that eruptions are much more likely in this system during the period when the seasonal effect tends to increase pressure. Over this period the total pressure (dotted line) increases much faster than when the seasonal effect is to decrease pressure.

In fact – and hold this thought – if the rate of increase in pressure due to short-term variability is faster than the long slow build-up of pressure, then eruptions, according to this simple model, will always occur during the short-term upswing in pressure.

My proposition is that it is very easy to exaggerate the effect of the seasonal cycle as a “cause” of eruptions. It is merely a trigger.

You can also see that, if, say, the seasonal pressure changes – a gradual trend on top of the annual fluctuations, perhaps, or an increase in amplitude of the cycle – it will not have a large effect on the frequency of eruptions over a long period. The periodicity of the system will still be driven by geological processes. The weather is a secondary driver in this system.

Now, if you didn’t know that volcanic eruptions are caused by a build-up of “pressure” underground, you might hypothesise that they’re caused by weather conditions. You might collect a lot of data and calculate correlation coefficients to “prove” your theory. You might even argue convincingly that, because we know what causes the weather, the weather must cause volcanic eruptions rather than vice versa and furthermore, it is not the case that both the weather and volcanic eruptions are caused by a third factor.

But you’d be wrong.

Could this mistake happen in other circumstances, though?

Solar cycles and the AMO

That old chestnut, solar cycles, surfaced yet again in New scientist a week or two ago. The claim is that there’s a “compelling link between solar activity and winter temperatures in northern Europe.”

Well, maybe there is.

But anyone who’s stayed awake this far will realise that it’s not enough to determine a correlation between solar cycles and weather patterns. Maybe the solar cycle does trigger a change from one state of the AMO (Atlantic Multidecadal Oscillation) to another. But that doesn’t make it the sole or even the main cause of the variability.

To recap, I first explored the idea of the AMO when I became concerned that the emphasis being put on shrinking Arctic ice as an indicator of global warming (GW) could backfire if the shrinkage reverses. My first post on the topic was therefore titled: Spin Snow, Not Sea Ice, the AMO Is Real!. Back then, I noted that the AMO cycle – likely to be variable in length, especially now we have the extra GW complication – tends to be of the order of 60 years or so, with the previous cooling phase lasting from the 1940s to the 1970s. Maybe we’re entering another one.

That first post suggested a mechanism for the AMO, which I discussed a little more in my second post on the topic, Why the AMO Overshoots. So I won’t repeat myself today.

Later, in 1740 And All That I looked at a historical example of a sudden switch from mild to cold winters in NW Europe. The weather pattern that leads to cold winters might be termed an “anti-monsoon”, as I first discussed back in January in Snow Madness and the North-West European Anti-Monsoon.

Two other posts Ice Pie and Ice Sickle explore aspects of the AMO.

The basic argument in all these posts is that the natural cycle – the AMO – is characterised by a set of feedbacks. Positive feedbacks – perhaps including the effect of lying snow, as considered in That Snow Calculation – produce distinct warming (Arctic ice melt) and cooling (Arctic ice recovery) phases. Negative feedbacks cause one phase to flip to the other. But the exact timing of the tipping-point may be caused by external triggers.

My proposition is that by the end of warming phase of the AMO, the seas (especially the Arctic and the North Atlantic) are relatively warm compared to the land. Any sudden cooling event could trigger a flip to the cooling phase, because the land cools quicker than the ocean, so would become relatively even colder.

Possible sudden cooling events are volcanic eruptions or the change to a cooling phase of the solar cycle, as discussed previously for the case of 1740.

A critical point is that the sunspot cycle is much shorter than the AMO (see AMO discussion and graphs in my first post on the subject):

The sunspot cycle indicates the total irradiance from the sun, and the rate of variation is comparable to that of other causes such as GW:

Further Implications

Not too many scientists claim volcanic eruptions are “caused”, as opposed to triggered, by variation in the weather or by climate change. Most understand that only over the sort of long timescale that is needed to melt an ice-sheet would the frequency of eruptions change.

But far more common is the explanation of apparent climate cycles – such as the AMO – by variations in solar output. In cases such as this, it is necessary to do more than just prove a correlation. The causal mechanism needs to be clear and must be shown to be quantitatively sufficient to explain the observed phenomena.

Considerable care is required whenever attempting to explain the “causes” of complex system behaviour.

The need to distinguish between triggers and underlying causes of cyclic behaviour also applies elsewhere in the climate system. Furthermore, the distinction between triggers and underlying causes may become blurred – both may be of similar magnitude, creating a resonant system. In particular, over longer timescales than so far discussed, the Milankovitch cycles are not enough alone to explain the ice age cycle. Perhaps they resonate with another cycle internal to the climate system.

In other domains too, it is not possible to assume that the “cause” in a complex system is just that which is evident on the surface. The lax lending practices and cheap money that are held to have caused the credit crisis may just be one part of a deeper, more complex cycle of optimism, deregulation, increased trade and globalisation on one hand and retrenchment and nationalism on the other.