Reliability of climate models
Climate models are the best tool available for making climate change projections. Modern climate models are extremely sophisticated. Models replicate the complicated factors and feedback loops that affect the climate, including atmosphere, ocean, sea ice, land surface, aerosol (tiny particles in the air), and carbon cycle components.
The reliability of climate models has improved over time, due to advances in computing power combined with an increasing understanding of the processes at play within our climate system. On-going research and improvements will result increasing reliability of climate models over time.
Of course, given climate models are predicting future climate, it is difficult to truly assess how reliable a climate model really is. In order to help assess the reliability of climate models, they are also run to simulate the observed climate to see how accurately they can recreate the observed climate (see figure below).
Generally speaking, the better a model recreates past climate, the more confident we can be that it will predict future climate. To further improve the reliability of climate modelling, scientists carry out an ensemble approach, which takes the average results from a number of different individual climate models. GCM ensembles are now widely accepted as the most valuable method of modelling the future climate.
As an example of complex climatic features that can be generated by models, this figure shows simulations from the UK Met Office Hadley Centre global climate model (Hadgem1) of 30-year average precipitation (1970-1999 from 20c3m run) for the months of December through to February, together with the corresponding satellite derived observed precipitation pattern from the Tropical Rainfall Measuring Mission (TRMM).