A key way of understanding our climate and making projections about how it may change in the future is to use climate models.
These are essentially simulations of the Earth’s climate system. They are made up of millions of lines of computer code which represent the physical processes which govern our atmosphere and oceans.
Supercomputers then run the code using observations of modern day climate, with the models able to recreate the past (hindcasting) or give projections of the future (forecasting).
Looking at the past is important for understanding historical changes and influences on climate, and it also allows scientists to gauge how accurate the models are (by comparing model output to reality).
Looking at the future enables researchers to see how things might change given various different scenarios – such as changing levels of greenhouse gases.
The Met Office uses models to look at many different timescales and to study different aspects of the Earth’s climate system.
You can find out more about how climate models work in our YouTube video.
Official blog of the Met Office news team 
Climate models have so far proved to be completely useless. What happened to the IPCC AR4 climate model predictions of 0.2C per decade of warming, which they claimed to have considerable confidence in?
http://www.ipcc.ch/publications_and_data/ar4/wg1/en/spmsspm-projections-of.html
Point us to 1 climate model which has consistently managed to match observations please.
The argument here is not whether the globe has warmed but whether the warming rate is alarming and whether it’s due to manmade emissions. The warming trend over the entire temperature record is less than a quarter of the rate that the IPCC climate models have predicted.
Assuming that a 152 year temperature trend is better than a 30 year temperature trend one wonders if alarmists like the Met Office are actually fixated on a 30 year temperature trend purely because 1 of only 2 periods in which the earth has warmed happened to start just over 30 years ago?
http://woodfortrees.org/plot/hadcrut3vgl/to:1932/trend/plot/hadcrut3vgl/plot/hadcrut3vgl/from:1938/to:1981/trend/plot/hadcrut3vgl/from:1998/trend
Fitting computational models of high Kolmogorov complexity (see: “millions of lines of computer code”) to a single run of a unique physical instance (terrestrial climate system) – Is. Not. Science.
For science goes the other way around: it seeks simple models fitting multiple runs of a wide class of instances.
Go for a general theory of quasi steady state non equilibrium closed thermodynamic systems with many internal degrees of freedom, radiatively coupled to their environment. Verify it on a wide variety of members of this class fitting into a lab setup.
As soon as it is done, come back to us, because only then we are ready to apply this knowlege to any specific system which happens to be too large to be studied experimentally.