Skip to content

Simulated climate and climate change in the GFDL CM2.5 high-resolution climate model

Key Findings

  • A new high-resolution coupled climate model has been developed¬† and used for climate variability and change studies
  • Many aspects of the climate simulation are significantly improved relative to previous lower resolution models, especially regional-scale precipitation. For example, the simulation of precipitation over the western U.S. is significantly improved.
  • The model has an outstanding simulation of tropical climate, including Indian monsoon rainfall and variability in the tropical Pacific.
  • Studies with this model highlight the important role that ocean mesoscale eddies can play in the climate system.

Thomas L. Delworth, et al., Journal of Climate, 2012, DOI:10.1175/JCLI-D-11-00316.1.

In this manuscript we report on recent work to develop and use a new high-resolution coupled climate model, called CM2.5. The move toward high-resolution is motivated in part by the need to simulate aspects of climate variability and change on regional spatial scales, since those are the scales of great interest to the public and to policy makers. It is also motivated by the need to improve the representation of small-scale processes in the climate system, such as regional circulations and ocean mesoscale eddies.

The model has relatively high spatial resolution: in the atmosphere the grid spacing is 50 Km, roughly uniform over the globe. In the ocean the grid size varies from 28 Km at the equator to approximately 8 Km at high latitudes. In the ocean, mesoscale eddies are well resolved at lower latitudes, but only partially resolved at high latitudes.

The model atmospheric physics are similar to the GFDL CM2.1 model, with the exception of a new land model (LM3), along with increased atmospheric vertical resolution (increase from 24 to 32 vertical levels). Model ocean physics are similar to CM2.1, with some important differences. The CM2.5 model does not use a parameterization of the effects of ocean mesoscale eddies. The CM2.5 ocean component also uses very high-order numerics, and very low viscosity, resulting in a very energetic flow, with very little numerical diffusion.

We have performed a multicentury control simulation and an idealized climate change simulation. We also compare results to a protoype model of even finer resolution (GFDL CM2.6), with ocean resolution varying from 10 Km at the equator to 3 Km at high latitudes.


Many aspects of regional climate simulation are notably improved in the CM2.5 model, especially in the Tropics. Regional rainfall simulations are significantly improved, especially over India, South America, western Europe and North America. The ENSO variability is significantly improved relative to the CM2.1 model. Ocean circulations are far more vigorous, especially boundary currents. Regional ocean circulations, such as flow in the Caribbean and Gulf of Mexico, are well resolved.

Many of the aspects of the response to increasing CO2 are similar in CM2.5 and the lower resolution CM2.1 model. The overall warming is somewhat larger in CM2.5. One difference is in the character of the rainfall reductions over the Mediterranean, which are widespread in CM2.1 and many other models. In the higher-resolution CM2.5 model, the rainfall reductions are largest in regions of steep topography. This could have significant implications for regional water resources.