Menu Close

Explainer: climate modelling

Climate models allow us to look at the planet’s future climate.

We know the climate is changing because that’s what climate models tell us. But what exactly is a climate model, and are they cutting-edge science or modelling madness?

What is a climate model?

Climate models are used to describe the behaviour of all, or part, of the earth’s climate.

They can be simple computer programs that describe one or more components of the climate in a specific location (how temperature changes with cloud cover in Hobart, for example). Or they can be immensely complex systems that attempt to describe the behaviour of the earth’s entire climate system.

Models that simulate the climate of the entire earth are called general circulation models, or GCMs.

The two major components of the climate are the atmosphere and the ocean. Other elements include sea ice, land surface, snow cover, lakes and rivers.

The movements of the atmosphere and ocean currents are described using the Navier-Stokes equations.

These mathematical equations have been used for more than 200 years to describe the movement of fluids. As well as being used to model climate systems, these equations are used by engineers to describe water flow in pipes and air flow around objects such as aeroplane wings.

Can we predict the future?

The future state of the climate has been of interest to people for thousands of years. The Babylonians used cloud patterns to forecast the weather as far back as 650 BC, and Chinese and Indian astronomers independently developed weather prediction methods by 300 BC.

The practice of using a mathematical model to predict the weather has been around since the 1920s, but it wasn’t until the advent of computers in the 1950s that numerical weather predictions first produced realistic results. The first GCMs were developed in the late 1960s.

Average annual daily maximum temperature (in °C) in Tasmania, observed (top) vs. computer model (bottom).

Weather vs climate

Numerical weather prediction (weather forecasting) and climate projections are closely related sciences.

Weather forecasting is about predicting the day-to-day variation in the atmosphere, including temperature, rainfall and humidity. Climate is the average weather pattern for a region.

In this way, climate projections seek to describe the condition of the atmosphere (and oceans) over long periods of time, often decades to centuries.

The weather varies on a daily basis, whereas the climate of a particular region is consistent and predictable.

For example, the climate of Hobart in winter sees rain falling on approximately half the days of each month with daily maximum temperatures around 10 – 15°C.

Climate projections (and climate models) do not tell us about the weather at any particular time, but about the average weather over a long period.

A climate model can’t tell us if Christmas Day, 2085, in St Helens (on Tasmania’s east coast) will be fine and sunny with a maximum temperature of 25°C.

But it can tell us the average (or “mean”, more accurately) temperature for December towards the end of the century, in this region, is likely to be 3°C warmer than it is now.

Climate models also tell us that on Tasmania’s east coast the end of the century will see more rain falling in summer, and that this rain is likely to fall in relatively more intense, short bursts.

Why should we trust climate models?

An important difference between numerical weather prediction and climate models is how they use observations.

In order to make a weather forecast as accurate as possible, observations are used to define the current state. As new observations become available these are incorporated into the simulation so that it becomes more accurate with time.

In contrast, climate models are initialised using an approximation of the current (or past) climate, but once running do not correct themselves using observations in any way.

That climate models produce a climate that is so similar to observations is purely due to the physical processes simulated in the model. This close comparison between observations and simulations is a strong argument for the efficacy of climate modelling.

A model state

In collaboration with the Tasmanian Government, the Antarctic Climate and Ecosystems Cooperative Research Centre recently completed its Climate Futures for Tasmania project.

The project featured a suite of climate models running from 1961 through to 2100 to generate highly detailed projections of the current and future climate of Tasmania.

The simulations were configured so that the concentration of CO₂ in the atmosphere matched the observed (increasing) CO₂ levels between 1961 and 2007 and then continued to increase in line with projected CO₂ emissions from 2007 to 2100.

Between 1961 and 1990 the simulations reported a state-wide mean temperature within 0.1°C of the observed mean Tasmanian temperature for the same period (see images above).

Similarly, the simulations reported a mean annual rainfall within five millimetres of the observed value of 1,390 milllimetres – an accuracy of around 99.6% (see images below).

The strong agreement between the simulations and observations shows just how capable climate models can be in simulating current climate.

Mean annual rainfall for Tasmania (in millimetres), observations (top) vs. model (bottom)

While the ability to simulate current climate does not guarantee the ability to simulate future climate, it is an important precursor.

Today’s climate models can accurately reproduce current climate, and have been able to reproduce changes to the climate that have been observed in recent years.

Many of the processes that drive the climate are well understood and the ability to capture these processes in models has been tested and is constantly being improved.

Climate models are not perfect, but they are the best tool we have available for explaining the current behaviour of our climate and predicting likely changes to the planet’s future climate.

Want to write?

Write an article and join a growing community of more than 184,200 academics and researchers from 4,969 institutions.

Register now