About the results
Each of you that downloaded and ran the experiment on your computer contributed to the graph below. But how?
Dr Nick Faull, a climate scientist from the Oxford University team running the experiment, explains.
UK temperature change
Click the numbered circles on the graph to find out what he has to say.
- What is plotted on the graph?
- Nick: “Each line on the graph belongs to a person who took part in the experiment. The graph shows all these lines on top of each other. The lines run from left to right showing the predicted increase in average world temperature from 1960 to 2080. Each person’s computer ran a slightly different model, that’s why each of the lines is slightly different.”
- Why does the graph start in 1960?
- Nick: “We start in the past so that we can compare each person’s model run to what happened in the real world in the past. That gives us a measure of how good that particular model is. The better the match with the real world, the more we trust that model, and therefore the more we trust the forecast which that model produces.”
- Why are the lines all bunched together between 1960 and 1980?
- Nick: “Model runs which didn’t closely match the recorded temperature data for 1960 to 2005 were downweighted. Only model runs that closely predicted past temperature data are used for future climate prediction. Actual recorded temperature data is shown by the thick white lines.”
- Why is the line that represents the actual recorded temperature so thick?
- Nick: “This represents the idea that we don’t have precise measurements of what the actual temperature was in the past. The thickness of the line represents the uncertainty in the real world.”
- Why don’t we have precise measurements of temperature for the past?
- Nick: “Land-based thermometers measure the temperature in specific locations in the UK. These specific temperatures must be generalised to represent a much larger area. This introduces small errors and uncertainty. Measurements from satellites also have some errors associated with the technology.”
Open University article on measuring temperature
- What do the different colours mean?
- Nick: “Model runs that match well with what really happened between 1960 and 2005 are ranked on a scale from ‘more likely’ to ‘less likely’. The more likely models are coloured pink and the less likely models are blue/black. By running each person’s model to simulate the past, we can compare the model to the real world data. The better the fit with the past, the better the model is.”
- What were the differences between the different model runs?
- Nick: “Each model is an approximation of the real world. No one model is perfect, that’s why we create slightly different representations of what the real world is like. We change the values of different parameters in the models. For instance, rain drops are a good example. We don’t know exactly at what point rain drops form in a cloud, but we have a parameter in the model that represents this. We can then change the value of the parameter in different models to see how it will affect the climate.”
Detailed information about the parameters used in the model
- Why does the temperature dip between 1960 and 1970?
- Nick: “The dip is attributed to a volcano in Bali called Mount Agung that erupted in 1963. Volcanic eruptions can create a cooling effect across the planet known as ‘global dimming’. This is due to volcanic ash, pumped out into the atmosphere during eruption, that reflects sunlight.”
- Some lines seem very erratic – what’s going on?
- Nick: “Each person’s model is different. Some of these models are not stable, and they behave unrealistically. We keep these on the graph so that we can compare all the results. It can also be interesting to look more closely at these models and work out what it is that causes them to become unstable.”
- Do the models take into account the Gulf Stream?
- Nick: “Yes, but the models don’t include scenarios where the Gulf Stream totally shuts down. We are less confident in making specific predictions about ocean changes. That’s because the scale of motion (of currents and circulations) occurs on a much smaller scale than those in the atmosphere. There are large scale motions in the oceans as well, but a lot takes place on a finer scale than the experiment models. We also have far fewer measurements of the ocean compared to the atmosphere, which makes it harder to verify the models against real world behaviour.”
- Can we be more certain about some of the results than others?
- Nick : “We know that the experiment models some areas less well than others, but that’s exactly the reason why we run it with so many variations. The parts of the model that are less certain are also the parts where we introduce most variety between different models. For instance, clouds and oceans are harder to model. In clouds, changes occur on a much smaller scale than the blocks that the globe is split into for the purposes of the experiment. A block is about 300km across, whereas clouds are roughly 1–10km across. So the model can’t deal with individual clouds, which means we must take an average. This approximation leads to uncertainties. With oceans, although we are using a full dynamic experiment.”