Intercomparison of Midwest Precipitation Changes from Statistical and Dynamical Downscaling Methods

Session: Poster Session

Kyuhyun Byun, University of Notre Dame, [email protected]
ASHISH SHARMA, University of Notre Dame , [email protected]
Alan Hamlet, University of Notre Dame, [email protected]
Jennifer Tank, University of Notre Dame, [email protected]
Todd Royer, Indiana University, School of Public and Environmental Affairs, [email protected]

Abstract

Statistical and dynamical downscaling of Global Climate Models (GCMs) simulations are both commonly used for regional and local-scale impacts assessment studies. However, statistical downscaling approaches are considered to be less reliable for simulating fine-scale convective storms, because of their dependency on coarse-resolution GCMs. By contrast, high-resolution dynamical downscaling methods which use physically based Regional Climate Models are arguably better able to simulate small-scale storm dynamics, because they better represent physical processes and land surface features over a region. This study first compares simulations of summer historical convective storms with observations for a) statistically-downscaled GCM simulations produced using the Hybrid Delta statistical downscaling approach and b) dynamically-downscaled simulations driven by the Weather Research and Forecast (WRF) model implemented at 12-km resolution and 4-km resolution. Then, we inter-compare 2080s projections of mean and extreme precipitation in winter and summer using the two downscaling approaches.  Key findings are that changes in winter precipitation are broadly similar between statistical and dynamical downscaling, but simulations of summer convective storms, and especially precipitation extremes associated with convective storms, can be fundamentally different between them. Results from WRF simulations show that the summer mean precipitation and changes in extremes are also essentially decoupled.