The role of unresolved processes in climate and climate change

The best way we have of predicting the climate of the future is to run complicated models on the world's fastest computers. These models divide the surface of the Earth up into boxes measuring around a hundred kilometres. Anything that measures less than a hundred kilometres (e.g. clouds and many other important features) is therefore missed out of the model. Nobody doubts that this makes the models and their predictions less reliable, but what can be done about it? We cannot make the boxes smaller, because even the world's most powerful supercomputers would not be able to do the calculations fast enough! Some scientists have recently suggested that the best solution might be to add some random noise to the models. Random noise is responsible for what you see on your TV screen when there's no signal, and for what you hear when you're tuning your radio. The theory is that clouds and other small features are pretty random anyway, so why not make them random in the model? Strangely, there is a lot of evidence that this can actually make weather forecasts better. This project will investigate the impacts of random noise on climate models. It is important to make these models as accurate as possible, because governments use them to decide the level of greenhouse gases we can safely emit into the atmosphere, without global warming becoming a danger to us.