The Australian Academy of Science’s new report, The risks to Australia of a 3°C warmer world, has grim predictions for the nation’s future under current carbon emissions policy and action. All of these predictions are based on climate models.
Among the now-familiar predictions of ecological and economic damage, the report points out that Alice Springs could see a 213% increase in energy demand by the end of the century (page 57), Hobart could see a 45% increase in Ross River virus cases by 2079 (p 64), and one in every 19 property owners could be facing unaffordable insurance premiums by 2030 (p 59).
How do climate researchers come up with numbers like this? What’s involved in climate modelling? How are climate models applied?
Unsurprisingly, this varies hugely with what you’re trying to predict. But there are a few key things that researchers need to predict the effects of climate change.
Expertise from different areas
Climate predictions demand expertise from a range of different scientific and economic fields.
“The more connections you have, and the greater the range of perspectives that are participating, the more robust your models are,” says Ove Hoegh-Guldberg, chair on the report and a professor of marine studies at the University of Queensland.
One researcher may have a close understanding of how warmer temperatures affect certain crops, but it takes a different field of study – and thus a different researcher – to understand how their yields might differ at a large scale.
Hoegh-Guldberg’s work on coral reefs in the 1980s suggested that they were vulnerable to warmer ocean temperatures, but it required work from global systems scientists and modellers to reveal the extent of the risk to coral reefs.
“When it comes to global models, I’m a user, not a builder,” he says.
Predicting the climate
Before we understand how we’ll be affected, we need to understand how the climate will change. And that can be difficult to do precisely.
“The ability to predict the future changes in climate is a fine art,” says Hoegh-Guldberg.
“There are uncertainties in terms of emissions and that’s a function of population and technology and policies and consumption patterns,” says Mark Howden, a member of the report’s expert panel and director of the Australian National University’s Institute for Climate, Energy & Disaster Solutions.
“For any given sort of level of greenhouse gas emissions there’s scientific uncertainty about how that will translate to temperature increases. And then there’s uncertainties about how well we will adapt to those increases, say, in health outcomes.”
This is why predictions vary so much, and why they’ll often be reported with large error bars.
That said, climate models are tested against past events, and can prove to be more accurate.
“In many cases, it can be useful to see how well models can hindcast,” says Hoegh-Guldberg. “Essentially going back in time to see how well your model explained what actually happened.”
If the models are able to accurately guess how the Earth responded to volcanic eruptions, for instance, then they’re probably going to be effective at predicting the temperature rise in the next century.
There are myriad climate models and simulations that can be used to predict variations in temperature. The more that are used, the better: multiple simulations can produce a more accurate result between them.
One popular tool is General Circulation Models, or GCMs. The website CoastAdapt, which can predict sea level and temperature rise for individual council areas under different emissions scenarios, uses these models. It stresses that they’re not neat predictions, and a number of models should be used to explore plausible futures for your suburb.
Human activity also needs to be taken into account. A simple example is energy use – for instance, if Australian cities are experiencing more heatwaves, they’ll also be consuming more energy and producing more emissions to mitigate these heatwaves. That’s just one among many more complicated feedback loops.
Predicting human outcomes with models
It can be even more complicated to use these models to predict economic and health effects on people – but there are also a lot of resources poured into this field. Insurance companies, governments and commercial enterprises have a vested interest in knowing what’s going to happen in their area over the next few decades.
“Essentially what we’re doing is creating devices which allow us to make decisions and operate complex systems,” says Hoegh-Guldberg.
Economic costs are assessed via complex computer models, using a variety of software packages. A couple of common models used for predicting climate are computable general equilibrium models, or CGEs, and integrated assessment models, or IAMs. CGEs provide detailed pictures of the economies of countries and regions, while IAMs are more focussed on the interactions between the economy and the environment, and how they both affect one another.
There are a few other methods to assign a price tag to losses from climate change. One estimate might assess the value of all of the infrastructure in a certain area, while another might examine the cost of an insurance payout if the area was affected by a natural disaster.
Similarly, health outcomes can be judged in a few different ways – from predicting incidence of a particular disease, to changes to population-wide life expectancy. Again, it needs medical specialists, public health experts and climate modellers working together to make predictions.
A thoughtful way of reporting predictions
It’s one thing to make predictions about climate futures, and quite another to report them. Uncertainty, in particular, can be a very difficult thing to communicate.
In scientific reports, predictions will be listed with error bars. For instance, one study (described on page 53 of the report) suggests that given 3°C of warming, by 2090 Darwin could be experiencing 180–322 days each year with temperatures over 35°C, with the mean estimate at 265 days.
So how do you tell people what you found? Do you lead with the (relatively) optimistic 180 days, the worst-case 322 days, or the most likely value of 265?
“I think all of the above,” says Hoegh-Guldberg.
“Understanding the range in the number of days affected provides important insights, as does the mean.”
These grim predictions are very carefully developed, but there is always going to be some inaccuracy. It’s perhaps better to focus on events that have already happened to gauge the seriousness of the climate crisis.
Understanding current losses and disasters, like the Black Summer bushfires, is enough to realise how swiftly emissions need to fall, says Howden.
“The urgency of the situation, I think, is very apparent to anyone who wants to look at what’s already going on,” he says. “Then we can draw the dots to the future.”