How reliable are the computer models of the Earth’s climate?

Climate models are mathematical representations of the interactions and events between the atmosphere, oceans, land surface, ice – and the sun. As such modelling is clearly a very complex task, models are built to estimate trends rather than events. For example, a climate model can tell you it will be cold in winter, but it can’t tell you what the temperature will be on a specific day – that’s weather forecasting. Climate models look at longer time trends – usually 30 years and longer.

Climate models have to be tested to find out if they work. We can’t wait for 30 years to see if a model is any good or not. So all models are first tested in a process called Hindcasting – in which the models are tested against the past, against what we know happened. The models used to predict future global warming should be able to accurately map past climate changes. The thinking is: If they get the past right, there is every reason to think their predictions of future could be right too.

The IPCC gives three reasons for its confidence in large-scale climate modelling: the fact that the fundamentals of the models are based on well-established physical laws; the success of models at predicting or reproducing observed patterns and variability in our current and recent climate; and the success of models at reproducing past changes in our climate, including global temperature changes.

Where models have been running for sufficient time, they have also been proved to make accurate predictions – be they about eruption of volcanoes, unexpected warming of certain regions of the world, and more. During the last few years, for example, climate scientists have made dramatic advances in their ability to track two key climate indicators—ocean heat content and Arctic sea ice seasonal cycles—and are confident that current models can reproduce surface temperature increases since 1870, including the rapid warming in the second half of the 20th century.

Comparing models developed independently by different centres around the world provides additional confidence where those models agree on the response.

Advances in modelling also have enhanced scientists’ ability to separate natural causes from human activities’ influence on the climate system.

There remain some uncertainties with climate model performance. Scientists are still trying to nail down cloud processes, aerosol distribution, ocean models, and sea ice changes.

In addition, when models are used to provide information about more localised parts of the climate – for example, over a particular country or region – the results become more uncertain. However, the quality of regional models is also improving.