Anne Stoner, Graduate Student

Atmosphere-Ocean General Circulation Models (AOGCMs) are capable tools to generate projections of future changes in large-scale climatology in response to human emissions of greenhouse gases and other radiatively-active substances; however, most AOGCMs have a resolution far too coarse to provide useful information at the scale of individual cities or weather stations. To compensate for this inadequacy, downscaling is used to generate sub-grid-scale information from the coarser-resolution AOGCM output fields. The objective of this study is to build on the statistical asynchronous regression method first introduced by O’Brien et al. (2001) to develop a simple statistical technique that directly downscales AOGCM temperature or precipitation outputs to local station scale temperatures or precipitation using quantile-based piecewise regression. This technique determines a relationship between two quantities that are not measured simultaneously, such as an observed and a model-simulated time series of daily maximum temperature. It takes advantage of the hypothesis that although the two time series are independent they describe the same variable, at approximately the same location (here: observed data are at a single point, whereas the model output is for one or several grid cells covering a larger area), and have similar probability density functions. The method is enhanced by several key steps, the most important being pre-processing of the AOGCM input fields using principal component analysis and the addition of monthly piecewise regression. This way of downscaling is highly efficient compared to regional modeling, since it does not involve retention of large-scale dynamical flow patterns, and thus does not require a large amount of computational time.

Bio

I’m from Denmark. I got my bachelor’s degree in 2003 in Geology from Aarhus University in Denmark. In 2006 I got my Master’s degree from Dept. of Atmospheric Sciences at Urbana-Champaign.