Climate Models
Why do we need comprehensive numerical models?
While a positive forcing implies a warming of the climate system and a negative forcing a cooling, the climate system is too complex for the size of a radiative forcing to be related to the global mean climate response in a straightforward manner. The response of the climate system to forcing agents is complicated by feedbacks, the nonlinearity of many processes and the different response times of the different components to any given perturbation.
How does a numerical climate model work?
They essentially divide the globe, both horizontally and vertically, into a finite number of ‘blocks’ or gridboxes approximatelly 200km big. Values of climate system variables such as temperature, water, wind speed, etc. are stored, with one value of each for each gridbox. The physical, chemical and biological equations that govern how all the different variables interact with each other and how they evolve over time are solved on this grid and over a finite number of intervals or timesteps; in other words, they are solved in a `discretised' manner. Some variables are prognostic, that is they are stepped forward in time as time evolves in the model. Others are diagnostic their new values are diagnosed from the new situation that results after time has been stepped forward in the model.
Features of a typical global climate model
The diagram schematically illustrates a typical global climate model. The typical gridbox size is illustrated with reference to the British Isles. A typical column is shown, in this case for an atmospheric model coupled to an ocean model. The yellow arrows illustrate how in the atmosphere there is vertical exchange between layers and horizontal exchange between columns of momentum, heat and moisture (the prognostic variables). In the ocean there is vertical and horizontal exchange of momentum, heat and salinity (salts) by diffusion, convection and upwelling. Information about the ‘orography’ (the vertical terrain profile), vegetation and surface characteristics is included at the surface in each gridbox.
Features of a typical coupled atmosphere–ocean model
the main interactions in a typical coupled atmosphere–ocean model. The incoming radiation that drives the system is input at the top of the atmosphere and the outgoing radiation from the climate system is computed according to its state at any time. Within the atmosphere, the evolution of density, motion and water is computed. In the ocean the evolution of density (including salinity) and motion is computed. The coupling between atmosphere and ocean involves the exchange of heat, momentum and water between the two. For example, winds from the bottom level of the atmosphere provide a wind stress that stirs up the upper layers of the ocean. The effects of the properties of sea ice and the land surface are also included.
Models today in total consist of (starting with the earliest)
- Atmospheric
- Land
- Ocean and sea ice
- Sulphate aerosol
- Non-sulphate aerosol
- Carbon cycle
- Dynamic vegetation
- Atmospheric chemistry
Parametrisation, uncertainty and ensemble modelling
Computers do not have infinite computing power so parameterisation of a sub-grid box is used to represent the function of a larger size box. The parameters include all plausible scenarios and sometime ensemble parametrizationis used to represent a chnage of possibilities. (see http://climateprediction.net) Uncertainty is also present we do not know the nature of any change in Climate forcings may take e.g. output of the sun, volcanic eruptions, change in carbon emissions.
Use of climate models to attribute climate change to natural and anthropogenic causes
One example is plotting the time series of annual global mean surface temperatures from 1860 to 2000. The red line represents observed temperatures and clearly shows the warming towards the end of the period. The grey bands show the results from climate model simulations using (a) natural forcings only, (b) Anthropogenic forcings only, and (c) all forcings, both natural and anthropogenic. The best fit was when all forcings prediction was matched again the actual change in temperature.
A similar pattern is seen if you look at this on a continent by continent basis. Generally, land as warmed more than the ocean given the lower thermal inertia
See more information at http://maps.conted.ox.ac.uk/cenet/modelplots.php#UKandIreland
