Dynamical and statistical downscaling of multidecadal global climate models provides finer spatial resolution information for climate impact assessments [Wilby and Fowler, 2010]. Increasingly, some scientists are using the language of “prediction” with respect to future regional climate change and impacts [e.g., Hurrell et al., 2009; Shapiro et al., 2010], yet others note serious reservations about the capability of downscaling to provide detailed, accurate predictions [see Kerr, 2011]. Dynamic downscaling is based on regional climate models (usually just the atmospheric part) that have finer horizontal grid resolution of surface features such as terrain [Castro et al., 2005]. Statistical downscaling uses transfer functions (e.g., regression relationships) representing observed relationships between larger-scale atmospheric variables and local quantities such as daily precipitation and/or temperature [Wilby and Fowler, 2010]. These approaches have been successful in improving the skill of numerical weather prediction. Statistical downscaling can also be used as the benchmark (the control) against which dynamic downscaling skill is judged [Landsea and Knaff, 2000]