Climate modelling is the simulation and prediction of climate trends. Researchers combine data on the atmosphere, oceans, land surface and ice to forecast how the planet's climate might change in the future. 

The models do this by first dividing the earth into cells. They then look at factors such as temperature, surface pressure, humidity and rainfall and calculate how they will interact in each new cell. The models repeat these calculations over and over again at set time intervals, or 'time steps'. This means that the more precise the prediction, the more time and computing power the model needs.

The vast and complicated nature of the interactions limits how precise the models are. While weather models recalculate every few minute to predict weather three days away, even supercomputers can't calculate at that level of accuracy for decades into the future. Scientists can, however, test their models' accuracy through a process called ‘hindcasting’. By using past data they can see if their models can predict current climate trends. If their model can predict weather that’s already happened, then it can be trusted to make accurate predictions of the future.

In order to prepare for climate change we need ever more precise, localised climate models that can be translated into policy.