While most people now understand that the enhanced greenhouse effect means a much warmer planet, communicating regional shifts in weather remains a significant challenge.

As with most complex science, nuance is everything. But how do you communicate complexity and nuance in a world increasingly geared to a 140-character limit? This is the fourth and final part of a series looking at the relationship between climate change and rainfall. Part three is here.

If you’ve been hanging round your friendly neighbourhood climatologist of late, you might have heard them talking about the IPO. No, not the Facebook initial public offering, the real IPO – the “Interdecadal Pacific Oscillation”. Yeah you nod and smile and wink a bit, but do you really know what this thing is?

No?

Well let’s be honest – neither do the boffins.

Well, not completely.

Yet.

Let’s first step back a notch. Many people would probably think that being a climate scientist is just one long party, yakking about hockey sticks and rising sea levels and disappearing ice. In reality most climatologists spend their days pondering the natural modes of climate variability – the swings and roundabouts that have been driving climate for millennia and that regularly push our mean weather from one state to another; wet to dry, hot to cold and all things in between.

Climate scientists want to know how these modes work now, and how they might behave in the future, just as James Risbey explained in part 3 of this series.

For instance, one of the natural modes we’ve all heard of is El Niño/La Niña, otherwise known by climatey types as ENSO (the El Niño – Southern Oscillation). ENSO describes how changes in the year-to-year location of heat in the tropical Pacific Ocean changes the dominant weather patterns over large parts of the globe. ENSO clearly has a strong influence over Australian climate, as we’ve all seen of late: Australia’s wettest two-year period on record corresponds to one of the stronger La Niña periods we’ve seen in the past century.

So why is this IPO thing interesting?

Just like ENSO, the IPO is also thought to be a natural climate mode that affects global weather patterns. For those that need to predict these tremendous swings in our climate, it’s obviously of interest that some decades have more wet or dry years than others.

The IPO moniker was coined in the late 1990s by a Bureau research team led by Scott Power. It describes decadal swings in climate. During “negative phases” of the IPO the eastern Pacific Ocean tends to be cooler than average — much like La Niña, but operating over years rather than months. During positive phases the same regions tend to be warm — much like El Niño, but again on these long timescales.

For Australia, these extended periods of cool and warm corresponded to persistent periods of wet and dry. For instance during the 1970s, the 1950s and the Teens, Australia had several very wet years. It was true gumboot weather. And during the 1920s, the 1940s and the nou/aughties, things tended to be dry; thong-wearingly (or flip-flop-wearingly) dry.

Early work by Scott and his colleagues showed that during IPO positive (warm) phases, ENSO had limited impact upon total Australian rainfall. This means that El Niño and La Niña events were still occurring, but their influence on local rainfall was diminished. During IPO negative (cool) phases it flipped – ENSO impacts hit Australia hard.

There has since been a lot of research around the world addressing the whats and whens and wheres of the IPO, so we have a good idea of what the IPO looks like in both space — the IPO pattern in the Pacific — and time — these long and semi-regular cycles.

But science hasn’t been able to fully nail the why and how. Particularly, why is there a change in the frequency and influence of El Niño and La Niña events from decade to decade?

And that’s where the hard yakka is being done.

Some suggest that the IPO may be related to the long, slow, deep circulation of the oceans.

Which kind of makes sense – waters in the deep ocean certainly do move on long timescales, shuffling heat and salt about at a snail’s pace compared to the rapid changes in the day-to-day weather we’re all standing in. And we now know such long-timescale ocean changes have a degree of predictability.

But in the data we have, there is no clear evidence that the deep oceans are the “driver” of the IPO. We are limited by lack of historical deep ocean information, and hence such an explanation remains as a very interesting idea worth exploring.

Indeed people continue to study the physical whys and hows and it is hoped that we’ll eventually have a better idea of what degree of predictability we might have for decadal swings in Australian rainfall – a lot or a little. There’s a stack more to be done here folks: it’s the joy of science.

What many scientists believe may be happening, though, is ENSO red noise.

“Red Nose?”

No.

Noise.

Noise, in science, is a statistical concept which essentially means random variations. The static on your grandad’s radio is classic noise, where the sound from one moment to the next is purely random rather than anything the ears can make sense of.

However with red noise, changes in a system from one moment to the next are still randomly determined, but the system has some “memory” of where it was a moment ago. This memory typically means that the next random change carries some part of the previous state with it. Climatologists call this memory “auto-correlation”.

So, back to the IPO. All this means is that if it is red noise, then we don’t have a clear physical driver for the decadal changes in our climate (at least not that we know of yet). Rather, the long period changes are an artefact of what’s going on at the year-to-year level.

Try this for an analogy.

The drunken walk.

Imagine you’ve just come back from a good night on the tiles and have had a few too many sherbets. You jump out of the taxi, look at the front door, extend your arm out with key in hand aiming for the lock, and make a go at it.

You stagger a bit left (watch the pot plant!), slow down, stagger a little further left (garden bed!), realise you’re headed for the fish pond, sway right, head across towards the kiddies swing, lumber a little more right, realise you’re headed for the fence, try going left again, stagger back and … well you get the picture, until you finally get the key in the lock. Overall, you’ve travelled in a straight line (taxi to door: we’d call this your “climate”) and each stagger (your El Niño or La Niña) took you forward, but you did some ruddy big loops along the way – the fishpond to the fence. Those big loops are your IPO.

And here is the rub – you couldn’t robustly predict where you were headed at any one instant in time, because the staggering was mostly random (granted you always had that front door in mind and knew – roughly – where you were in the garden). Likewise, the IPO may well just be a statistical artefact of natural climate variability oscillating (lurching) one way or another but always balancing out to an average, and hence from one year to the next is pretty much unpredictable.

Such an analogy highlights something profound about data analysis. If you had the remnants of a kebab in your pocket, and left a trail of garlic sauce from the taxi to the door, then you will have left an interesting snail trail for your better half to ponder in the morning. Following the line of garlic sauce, and not knowing that you were full as a boot, it might seem that the night before you deliberately took the time to smell the roses in the garden before opening the front door. But it’s a mirage. Viewed in real time there was no method at all, just random lurches constrained by how far your steps could propel you, and an inability to stop yourself once you got a bit of momentum behind you.

The short-term constraint of momentum is another potential source of predictability in the drunken walk. If the person is light on their feet, then their movement from one step to the next is pretty much decided by how far they can stride, with direction still being mostly unpredictable. However, if they are a bit heavier on their feet, then you might gain some short-term predictability based on their momentum alone. But as anyone who has had to carry drinks through a crowded pub will attest, even the big blokes can change direction all too suddenly. The lack of a true organisation in movement, or a physical climate driver, is what defines random processes.

And that inherent unpredictability is what currently makes using the IPO to forecast or attribute climate, and particularly Australian rainfall, problematic. For instance we’re in a certain phase now. But will we leave it today? Next month? Next year? Or have we already left it? At this stage we simply don’t know.

Why?

Well first of all, its very difficult to know exactly what phase we are in at present, simply because being decadal, and hence a 10-year-plus average thingy, we’d really need to know what will happen over the next five or more years to know exactly what phase we are in now. Sure we can use some fancy stats to compare the typical pattern of an IPO to the current pattern of the oceans temperature, and hence estimate where we might be. But that doesn’t mean we know for sure exactly what phase we are in – it’s simply a well educated guess. It certainly doesn’t tell us if we are just about to flip out of the current phase or if the current phase will get stronger (in other words, if we will stagger into the fishpond or be back on the path).

So how does this all relate to climate change and the recent “big dry” or two-year “wet”?

Any time a climatologists talks about unknowns in the climate system, it gets some people excited. Perhaps this unknown can explain 20th century global warming? It’s important to remember that there are lots of things that we do know about the climate system. The IPO is clearly unrelated to the steady increase in global mean temperature, or the pattern of long-term climate change observed across many different climate variables. Those changes fit with increasing greenhouse gases.

The IPO is just one of a multitude of climate modes that, together, cause our short term climate variability, just as James described. And if the IPO is entirely a statistical leftover from of a bunch of random staggers (still an “if” mind you – remember those deep oceans), then it doesn’t really tell us a lot about what’s physically driven the longer (decadal) swings in our climate, or why.

What we do know is that during the big dry we had roughly equal numbers of El Niños and La Niñas (and indeed the IPO index was actually negative/cool from around year 2000) yet, as noted in part 1 of this series, our autumn and early winter periods down south were dry regardless of ENSO; there was simply never any catch-up from the bone dry years. We also know that high pressure systems progressively became bigger and fronts moved further to the south – something which has been going on longer than just the big dry. Such changes are all typical of what may be expected with climate change: Mediterranean climates (like southern Australia) will generally get drier.

And we only need look west for a potential analogue. South-west Western Australia has undergone a virtual step-change in its climate, with a drying trend since the late 1960s or early 1970s. And why? Well as Karl noted in part two of this series, it’s due to a shift in the type and location of synoptic patterns. The rain hasn’t stopped falling, its just decided that it would rather fall out in the Southern Ocean.

But what’s a little scary is that south-west WA is the area of the world where virtually all the climate models agree; the way you get such a shift is to pump 40% more CO₂ into the pre-industrial atmosphere.

As Karl also indicated in his second article, we are still researching these issues.

The IPO is clearly a difficult beast. It’s a factor in our climate, but we simply don’t know how predictable it is, if it is at all. Those that plan for water resources on five- to ten-year timescales would dearly love to get some more predictability in Australian rainfall out of the IPO.

However, viewing recent changes in eastern Australia’s climate as purely due to the IPO ignores many other factors and a wealth of evidence and scientific studies. Additionally, climate modes like the IPO might themselves be changing due to global warming. As James described yesterday, carbon dioxide is like a control lever that kicks the system into different modes, or even changes the way the modes operate. You can think of global warming as changing the incline of the ground under the person returning from the bar – ever tried to walk a straight line up a steep muddy slope, not to mention with a few sherbets under your belt? To ignore the trend, and to simply label long-term changes as decadal swings, may well have us not just staggering about, but falling on our bum in the pond.