SIParCS 2015- Richardson

Lee Richardson, Carnegie Mellon University

Statistical Downscaling- R Package and Application to Snow

(Slides)  (Recorded Talk)

Statistical downscaling is a commonly used method to relate large scale climate variables with local variables. It is used to resolve the different spatial scales produced by Global Climate models and those needed for local impacts assessments, such as determining an individual cities risk of flooding. We have created a R package, called sdsmR, which replicates the functionality of a widely used tool, Statistical Downscaling Model (SDSM), in the R language. This package gives users the same modeling options they had in SDSM, but also provides access the powerful statistical and graphics capabilities built into the R language. Next, we explore different models to statistically downscale Snow Water Equivalent (SWE) at the Niwot site in the Rocky Mountains. We exploit the consistent, annual structure of the SWE to build a realistic statistical model which does an adequate job of predicting snow trajectories in future years.

Mentors: Rachel McCrary and Doug Nychka, CISL IMAGe