Global and Regional Climate Models

Climate models are numerical computer models which calculate the past, present and potential future climate for a specified time period. Generally, climate models are based on meteorological numerical models used for weather forcasts, extended to correctly simulate as many climate relevant processes as possible. Usually, different submodels are coupled, e.g. an atmosphere model, an ocean model, a cryosphere model (covering snow and ice), and a land surface - vegetation model.

Climate modelling on different spatial scales

Global Glimate Models (GCMs) describe and simulate climate relevant processes on the global scale, which means that the spatial resolution of these models can only be rather coarse, owing to the complex and time consuming calculations, even on the largest and most modern supercomputers. In contrast to the GCMs, regional climate models (RCMs) simulate only a small area, but with a much higher spatial resolution. This allows the consideration of much more details, e.g. the simulation of a dense river routing network, as in the BALTEX Integrated Model System BALTIMOS, or other climate relevant topograpic features. GCMs have horizontal resolutions of 300-600 km, with about 10-30 leves into the atmosphere in the vertical. This means that for a grid box of roughly 300 x 300 km x vertical height (which is variable), the model calculates one value for temperature, precipitation, etc. Regional models have horizontal grid sizes of just 20-50 km.

Typical atmospheric grid resolutions used in Global Climate Models (low resolution, left) and Regional Climate Models (high resolution, right). In the regional models (right), landscape features which may have an impact on the regional climate, are much better represented (e.g. the Alps). Figure from BACC, 2008.

Climate models need good "food"

In order to be able to properly calculate "what could be", models need a good formulation as to which environmental properties and processes lead to which outcome. For instance, the appearence of clouds (the assemblage of tiny drops of water in the atmosphere) depends on a number of variables like temperature, humidity, and the concentration and type of aerosols. The numerical description of this is called "parametrization". These parametrizations are the "food" for the models, and the better, i.e. the closer to reality they are, the better the model is able to simulate the processes. The models usually simulate the development of chosen variables (e.g. air temperature, precipitation etc.) over a specified time period, which is divided into time steps, which ranges from minutes to hours.

Climate models can only say what could be, not what will be

As climate models will never be able to "predict" the future climate to a high degree of certainty, modelers speak of "scenarios" rather that predictions. As many combinations of sub models and parametrizations are possible and often equally reasonable, a set of these combinations are usually presented together ("ensembles"). If a set of ensembes gives a certain trend with little variation between the individual model results, the models show a good credibility. The regional climate models can then be used to drive impact models like for hydroligical, biogeochemical, agricultural or social changes.

Climate model results need to be checked against reality

Models must of course be "validated" against reality, e.g. they must be able to simulate the past and current climate correctly. For the past and present, real data and observations exist which are fed into the models. Only if the models can reliably simulate the past and present climate, they can be trusted to model the future climate, under certain preconditions like the development of CO₂ concentrations in the atmosphere.

Future greenhouse gas emissions drive the models

The development of greenhouse gas emissions (like CO₂) essentailly determines the future climate. So their concentrations are a fundamental factor in climate models. The future development of these gases depends on many different factors. To supply the climate models with a number of plausible greenhouse gas concentrations, emission scenarios have been developed, each describing different political, social and technological conditions in the future, leading to different CO₂ emissions. These emission scenarios are summarized in the "Special Report on Emission Scenarios (SRES)" of the Intergovenmental Panel on Climate Change (IPCC) and are therefore called SRES Scenarios.