Downscaling Seasonal Climate Predictions for Local Applications

Different sectors of society and the economy are increasingly utilizing seasonal climate forecasts for decision-making. However, there are several issues inherent to the applicability of climate outlooks for many users related to forecast spatial scale and extent, and their probabilistic nature. Climate predictions, such as those disseminated by the Climate Prediction Center, generally represent the forecast information across large regions often with ambiguous boundaries. Yet, small-scale climate variations occur over space as the major climate patterns interact with local topography, land cover, and storm track variability. Consequently, large-scale climate forecasts are not necessarily as effective as they might be for use at the local-scale. Downscaling methods can be applied to these forecasts to enhance their functionality at the local-scale, by expressing their probabilistic content in terms of local station data probabilities. By increasing the specificity of a more general forecast, the added utility and accessibility can increase potential use by local decision-makers. In this paper, we downscale probabilistic climate forecasts for a range of locations and evaluate the approach itself as well as the issues arising from scale translations, including relationships between local station data and the climate ‘mega-divisions' used in climate forecasts.