There are several things to consider if you are thinking about using climate projections. How well a climate projection (or set of projections) is suited for a particular application will depend on the climate question that is being asked. Each problem will have a different threshold for how much uncertainty can be accommodated in the decision space, and uncertainty can be partly measured by the quality of the projection(s). The quality of a projection depends on how well regional and local weather patterns are represented in the climate model. A snapshot of weather at any location is a combination of large-scale atmospheric conditions and small-scale weather processes. Although we should not expect climate models to reproduce or predict the exact weather for any given location at a specific time, the models should reflect the correct weather processes for the region. For example, the Great Lakes can receive snowfall from large weather systems moving across the region, but areas within the lake-effect zone can receive significantly greater amounts of snowfall due to how the lakes modify the atmosphere over and near them. Small-scale features such as lake-effect and convective-type precipitation typically are too small for GCMs to capture, however, GCMs can do a reasonable job representing the larger-scale atmospheric conditions. One of the main reasons projections have difficulty capturing the correct weather features in the Great Lakes is because the lakes are not well represented in global climate models (if at all). Downscaled projections do not always provide an improvement upon the global models. How well a particular model represents important local features will depend greatly on the spatial resolution of the model (GCMs typically have coarse spatial resolutions compared to downscaled projections) but higher spatial resolution should not be equated to greater confidence in the projection. Confidence in a projection should come from the ability of the projection to capture weather processes that are important to your problem. Since we can not know exactly what the future weather and climate will look like, most of our confidence comes from how well the projection represnts what has already been observed.