Downscaling climate data is a strategy for generating locally
relevant data from Global Circulation Models (GCMs). The overarching strategy is to connect global
scale predictions and regional dynamics to generate regionally specific
forecasts. Downscaling can be done in several ways.
Nesting a regional climate model into an existing GCM is one way to downscale data. To do this, a specific location is defined and certain driving factors from the GCM are applied to the regional climate model. A regional climate model is a dynamic model, like a GCM, but it can be thought of as being composed of three layers. One layer is largely driven by the GCM, another layer builds on some locally specific data, and the third layer uses its own physics based equations to resolve the model based on data from the other two. The results are comparatively local predictions that are informed by both local specifics and global models. This process requires significant computational resources because it is dependent on the use of complex models. Currently Canada has just one Regional Climate Model (CRCM).
A second way of downscaling climate data is through the use of statistical regressions. There are a variety of such methods ranging from multiple regressions that link local variables to particular drivers in GCMs, to more complex methods using statistics designed for neural networks. The general strategy of these methods is to establish the relationship between large scale variables, such as the driving factors derived from GCMs, to local level climate conditions. Once these relationships have been developed for existing conditions, they can be used to predict what might happen under the different conditions indicated by GCMs.
A third strategy for downscaling data is also statistically driven (and thus not dynamic like a regional climate model). This strategy uses stochastic weather generators. The weather generator develops a series of statistical linkages among variables to predict weather at that particular location by using long term weather data for a particular area. These empirically based models can be used to downscale data by using data, such as wind speed or other variables, generated from GCMs to predict the local result of driving variables.
All of these techniques are estimations, but they can generate useful local data. In the Canadian context, a range of climate data, both directly generated from GCMs and downscaled, is available for decision makers.
For more information about climate models, please see the materials under "Downscaling Climate Data" in the Recommended Links section. To access data from GCMs and the CRCM visit http://www.cccma.ec.gc.ca/models/crcm.shtml.