Downscaling

GCMs caot provide information at scales finer than their computational grid (typically of the order of 250 km), and processes at these unresolved scales are important. Thus, the usefulness of the raw output from a GCM for climate change assessment in specific regions is limited. To bridge the spatial resolution gaps for GCMs to produce realistic local climate projections, downscaling techniques are usually applied to the GCM output; therefore downscaling addresses the disparity between the coarse spatial scales of GCMs and observations from local meteorological stations.  Downscaling techniques divide in two groups: (a) dynamic climate modeling and (b) empirical statistical downscaling. This technique involves nesting a higher resolution Regional Climate Model (RCM) within a coarser resolution GCM. RCMs use the GCM to define time-varying atmospheric boundary conditions around a finite domain from which the physical dynamics of the atmosphere are modeled using horizontal grid spacing of about 20 50 km or less. RCMs are attractive to those seeking process understanding and causative simulation, but most downscaling is currently empirical. Statistical downscaling uses a statistically-based model to determine a relationship between observed local climate variables such as precipitation and temperature (known as predictands) and large-scale climate variables, GCM outputs, (referred to as predictors). The derived relationships between the predictors and predictands are applied on similar predictors from GCM simulations in the statistical model to estimate the corresponding local or regional climate characteristics.