A New Approach to Construct Representative Future Forcing for Dynamical Downscaling

Abstract:

Dynamical downscaling of future climate change has traditionally used 6-hourly output from individual simulations by global models as lateral boundary foricng. However, individual model runs often contain large realization-dependent internal variaiblity that can overshadow greenhouse gas (GHG)-induced climate change on decadal and regional scales. Thus, the use of such forcing data may produce misleading results. We propose a new approach to construct representative forcing data from the CMIP5 multi-model ensemble of simulations for downscaling GHG-forced future climate change. The idea is to use the multi-model ensemble mean to smooth out internal variaiblity in the mean state of the foricng data, while keeping the daily variability realistic as it is derived from one selected mode run. Using the WRF-based 4km resolution downscaling of future climate change over the contiguous U.S. as an example, we will discuss the various issues involved in this approach.