Modelling, understanding and predicting the regional climates

Last update: February 2020


Traditional climatic studies are done using numerical models called GCMs (General Circulation Model) that cover the whole earth, represent the atmospheric dynamics and its physical laws. The current tendency when the global climate of the planet is studied is to couple these atmospheric models with models representing the other parts of the Earth system: the ocean, vegetation, rivers, the marine biogeochemistry, atmospheric chemistry, the polar ice-sheets, the carbon cycle.

For reasons of limitation of the available computer power, these models generally have a resolution of about a few hundred kilometres. This resolution does not make it possible to solve certain regional physical processes and phenomena which influence the climate of an individual region of the planet (islands, complex relief, regional winds, precipitation spatial pattern, small scale ocean circulation).

Similarly, with this resolution, these models do not provide spatially accurate data for all climate data users. For example: potential impacts of climate change on tourism, on water resources, on tropical diseases, on marine ecosystems, on renewable energy production ... These fields need simulations with higher spatial resolution. The last disadvantage of the global models is their low capacity to simulate the extreme events (extreme wind speed, heavy precipitation events) which are often related to processes or forcings at small scales. For these reasons, the scientific climate community has developed the regional climate modelling concept in order to better understand the regional climate and their future evolution.

 Regional climate modelling techniques

There are 3 different methods to simulate regional climate. We develop and use the three at CNRM.

- The first method consists in using a GCM with very high resolution and covering the whole earth. Numerical experiments were carried out, for example, with ARPEGE-Climate with a uniform resolution of 50 km. The length of the runs remains limited by their numerical cost.
Références for ARPEGE-Climate at uniform high-resolution : Elguindi et al. 2011.

- The second solution is the use of global model with a zoom. This is the case of ARPEGE-Climate for which the grid can be tilted and stretched to strengthen the horizontal resolution in the area that one wants to study. The stretched version of ARPEGE-Climate allows for longer simulations (few hundred of years). The ARPEGE-Climate stretched model was first used in its "Medias" version with a zoom over Europe and the Mediterranean. However, there now is a tropical Atlantic version and a North America version centred on Winnipeg (Canada).
Références for stretched ARPEGE-Climate : Déqué and Piedelievre 1995 (Climate Dynamics), Déqué et al. 1998 (Climate Dynamics), Gibelin and Déqué, 2003 (Climate Dynamics), Déqué 2007 (Global and Planetary Change).

- The third solution is the use of a limited area model (LAM) as in numerical weather prediction. These models cover only a part of the sphere, Europe for example and are forced at the lateral boundaries by models of weaker resolution (GCM, reanalyses, ...). CNRM-ALADIN is commonly used at CNRM for these applications since the beginning of the years 2000s. This model is quicker than the stretched ARPEGE-Climate at the same resolution, it allows to reach finer spatial resolution (up to 10 km) and to test various types of boundary conditions.
Over the last years, CNRM-AROME is in use at CNRM and allow to carry out regional climate modelling experiments at kilometer scales.

 Examples of specific studies carried out at the CNRM:

  • Study using CNRM-ALADIN: here
  • Study using CNRM-AROME: here
  • Study using CNRM-ALADIN with interactive aerosols: here and here
  • Study using the coupled regional climate system model CNRM-RCSM: here

 Staff dealing with this topic:

At CNRM, this activity is mostly conducted by the members of the MOSCA team, plus Fabrice Chauvin (AMACS) and Lola Corre (DCSC). A big thank to Michel Déqué (retired in 2019), Clotilde Dubois (now in MERCATOR) and Alain Braun (retired in 2012) as well as to PhD students and postdocs.

 To go further:

  • Guidance for the use of regional climate model outputs: here
  • Other links in French here