Theme 1: Downscaling and Probabilistic Forecasting
Global climate models currently resolve physical scales on the order of 100 km or greater, and current regional climate models over Canada resolve features larger than 45 km (although a resolution of 15 km will be achieved shortly). Even these anticipated higher resolution simulations are too coarse to characterize local climate in the presence of strong surface heterogeneity, such as in coastal areas or regions of rough topography. British Columbia is a province with complex mountainous terrain and a long, crenulated coastline: local predictions of future climates require the development of tools to statistically "downscale" climate information from global or regional forecast models. Because of the highly complex nature of the terrain in BC, increases in model resolution are expected to improve predictive skill but will not eliminate the need for downscaling tools. Furthermore, these tools can downscale not only climate predictions but also weather forecasts, allowing improved local forecasts in areas of abundant historical local data (e.g. probcast.washington.edu ).
Not only are these technologies needed but people skilled in their development and implementation must be trained. Research and training of students and postdoctoral fellows focuses on both the development of theoretical and methodological aspects of these tools as well as their application at the scales at which impacts and vulnerabilities are experienced. Along with temperature and precipitation, the prediction of surface winds is an area of increasing interest because of their importance for wind power generation, their role in mediating surface fluxes, and the severe damage that can be caused by extreme winds (as the December 2006 windstorm in Vancouver vividly demonstrated).