El Niño – riding the climate roller coaster
Box 2 | Modelling climate
Scientists run simulations to help them understand the response of the real world to changes in the atmosphere. These are mathematical models and they involve the use of sophisticated supercomputers. These models are called general circulation models (GCMs). They take account of many processes that together determine the behaviour of the atmosphere.
Conditions on the surface of the Earth – such as temperature, humidity, wind speed and atmospheric pressure – as well as those at various heights in the atmosphere are entered into the computer. Through decades of observing and measuring the behaviour of the atmosphere, scientists have constructed mathematical equations that describe the known movements of heat into and out of the Earth and the atmosphere; how changes in atmospheric composition should cause these heat transfers to change; the rates of evaporation and precipitation of water; the major currents in the air and oceans and their distribution of heat; and the frequency of tropical cyclones.
The model of this land-ocean-atmosphere system is represented by a number of points, making up a grid that acts like a bare outline of a picture.
Just as the picture on a television screen is made of many changing dots organised in 625 lines in order to represent the reality that was filmed, so the changing parameters at each grid-point of a GCM represent the reality of Earth's climate system.
Current GCMs may contain one point every 500 kilometres of surface – whether on east-west or north-south transects – and between two and twenty points above each one on the surface to represent the height of the atmosphere. So, for example, they could have a total of 2000 points on the Earth's surface and 20,000 in the atmosphere. The solutions to a number of equations must be calculated for every one of these 22,000 points in order to move the model forward in time.
Each step of model-time could be an hour, or even less, and the model calculates how the conditions have changed during that time-step. Each point in the grid is given its starting conditions, using the chosen parameters of temperature and so on. Then a supercomputer solves the equations for each point. This creates a change in conditions, and the new information is used to run the equations all over again with the new numbers inserted.
Powerful supercomputers are capable of carrying out more than 500 million calculations per second. However, to solve all the equations necessary to run an Earth-atmosphere system for a model time of 1 year takes about 2 hours of computer time. As more computer power becomes available, the model can become more detailed, with points at closer intervals. The tighter the spacing of dots, and the smaller each time-step of the model, the more accurate can be the representation of reality, exactly as a finer-grained photograph or a greater number of frames per second in a moving film gives greater clarity.
The test for a computer model is to see how well it simulates known real events. Scientists can then see how well the model predicts the already known outcome. For example, you could feed into a GCM some of the temperatures, atmospheric pressures, wind and ocean movements that existed in the world at the end of 1982, which was the start of the strong El Niño event that happened during 1982-82 – and then run the model for 2 years of model time. Would the outcomes given by the model resemble what we know really happened to temperature and rainfall around the world?
When compared with reality the best models have performed well; scientists therefore regard them as good representations of some aspects of reality, but of course they are not complete and cannot predict accurate outcomes for local regions, mainly because they still space the points of the grid too far apart.
Box
Box 1. The Walker circulation and weather forecasting
Related sites
Modelling climate (CSIRO Atmospheric Research, Australia)
Climate modelling: General circulation models (Global Climate Change Student Guide, Manchester Metropolitan University, UK)
Models 'key to climate forecasts' (British Broadcasting Corporation, UK)
Page updated January 2010.