linked in logo | IS-ENES | Contact

Exploring climate model data

Global models

Global Climate Models or General Circulation Models (GCMs) are based on the general physical principles of fluid dynamics and thermodynamics. They have their origin in numerical weather prediction and they describe the interactions between the components of the global climate system: the atmosphere, the oceans and a basic description of the land surface (i.e. aspects of the biosphere and the lithosphere that are relevant for the surface energy balance). Sometimes they are referred to as Coupled Atmosphere-Ocean GCMs (AOGCM).

As the scientific understanding has increased, as well as available computational resources, the GCMs successively become more and more complex. More components of the global climate system are included in the model and the process descriptions become more detailed. The scientific knowledge has now progressed to the level where global climate models are being replaced by Earth System Models signifying that the models now embrace more components and processes than the physical atmosphere-ocean components traditionally used to describe the climate. For a detailed inventory and/or comparison of the various Earth System components in any of the current generation of GCMs please refer to ES-DOC Comparator.

A schematic image of an atmosphere-only GCM. In a coupled atmosphere-ocean GCM a three-dimensional grid is included to represent the oceans. Image source: NOAA 200th Celebration, accessed 29 June 2012.

Model resolution

The equations used in climate models are complicated differential equations. The numerical integration is accomplished by discretization of space and time. This discretization determines the spatial and temporal resolution of the GCM. Typical values for the spatial resolution ranges from 100–300 km, and the corresponding temporal resolution – often called the timestep – is typically in the range 30–60 minutes. While output data from the model is usually generated at full spatial resolution it is not feasible to save data for each timestep. Instead output data is often generated at 6, 12, or 24 hour intervals. There are two main discretization techniques; finite difference method and spectral method. The use of spectral methods in general circulation simulations has expanded substantially with published results from several major research centres. This external link provides a technical summary of the advantages of spectral methods, and further down that page a translation table of spherical resolutions (T42, T63, T106,...) and the corresponding gridcell resolution.

Many GCMs divide each gridcell into several land cover categories. If data is available for such a ‘tiled’ land surface subdivision this can in a way be regarded as a further increase of the spatial resolution within a gridcell.

Control simulation — Historic simulation — Climate change simulation

To carry out a climate change simulation the GCM has to be ‘started’. This is a complex procedure involving several steps:

Data from suitable climatological datasets are entered into the model and a long run of constant climatic conditions is started. The external forcings (greenhouse gas concentration, aerosols, solar input, etc.) are held constant at levels representative of a certain time period, typically year 1850 or 1860. The purpose of this control run, or spinup run, which spans 500–1000 years or even longer, is to spin up the model into an equilibrium state.

When the control run has achieved equilibrium, one or several instantaneous states of the model are extracted. These are selected so that they are independent  representations of reasonable, but not necessarily observed weather conditions for 1850 or 1860. These snapshots are then used as initial conditions for the next step.

From each of the initial conditions the model is again started. In these runs time-varying external forcings are applied. That is, greenhouse gas contrentrations, aerosols, solar input, etc. are set to change over time according to historic records from 1850/1860 up to the present. In these historic runs the GCM simulates the evolution of the climate over this period.

From the historic run again a time snapshot is extracted representing realistic, but not necessarily observed weather conditions for a desired year, say 1990 or 2000. This snapshot is again used to start a new model run. In this climate change simulation — or scenario run — the external forcings are set to vary according to one of the future scenarios.

Schematic of the relationship between control (spinup) run, initial condition, historic run and scenario run of a GCM. Image source: SMHI Rossby Centre.

To understand and correctly make use of climate scenarios and climate model data it is essential to understand how the initial conditions for initiating a historic run influences how the model simulates climate variability at the decadal timescale. Some relevant links are listed below.

Related links



The IS-ENES project has received funding from the European Union's Seventh Framework Programme for research, technological development and demonstration.