Current global models project the global mean temperature to rise 2.5°C to 6°C, over the next 100 years with comparable or larger variation on regional scales. Changes in precipitation have large variation among models, with some indicating substantially lower mid-continental precipitation and others providing increases. Narrowing the range of uncertainty in determining impacts of climate change under such disagreement requires use of ensembles of global models to capture the uncertainty and to bracket the range of possible outcomes.
Downscaling changes in the global climate to regional and local scales is needed for assessing impact on crop production. Computing resources available now and in the next decade do not permit global models to run at resolutions sufficiently high to provide detailed climate information for assessing regional impacts of climate change on agriculture. While global climate models will approach 100-km resolution in the foreseeable future, the U.S. Climate Change Science Program (CCSP) noted that it is unlikely that in the near future they routinely will reach the 10-km scale that regional climate models (RCMs) can simulate (U.S. CCSP, 2003).
Alternative methods are needed to downscale global climate model results to spatial scales of importance to decision making in agriculture. Two candidate methods for achieving needed resolution are regional climate modeling (dynamical downscaling) and statistical downscaling. For defining future scenario climates, both methods rely on global climate model information to provide large-scale states of the climate system that are consistent with global physical conservation principles under conditions that include changes of external forcing (e.g., increases of atmospheric greenhouse gases and sulfate aerosols and changes in land use). The U.S. CCSP (2003) emphasizes RCMs as a means of achieving climate information at resolutions higher than are available from global models.
Was this article helpful?