An Improved Dynamical Downscaling Method with GCM Bias Corrections and Spectral Nudging

The traditional dynamical downscaling (TDD) approach employs a continuous integration of a regional climate model (RCM) where global climate model (GCM) data are used to provide initial conditions and lateral boundary conditions. Differences between TDD simulations and observations can result from biases in GCMs, RCMs, or both. To reduce the biases in the regional climate downscaling simulations, an improved dynamical downscaling method with GCM bias corrections and spectral nudging is developed and assessed over North America. Regional climate simulations are performed with the Weather Research and Forecasting (WRF) model embedded in the National Center for Atmospheric Research (NCAR) Community Atmosphere Model (CAM). To reduce the GCM biases, the GCM climatological means and the variances of interannual variations are adjusted based on the National Centers for Environmental Prediction-NCAR global reanalysis products (NNRP) before using them to drive WRF. Spectral nudging is introduced to further reduce the RCM-based biases. Our study builds on a recent paper. Xu, Z.-F. and Z.-L. Yang, 2012: An improved dynamical downscaling method with GCM bias corrections and its validation with 30 years of climate simulations, J. Climate, in press; see http://journals.ametsoc.org/doi/abs/10.1175/JCLI-D-12-00005.1.

Two sets of WRF downscaling experiments are conducted. In the first set, the GCM-driven RCM simulations without spectral nudging are compared with NNRP-driven RCM simulations to assess the GCM bias correction effects. All WRF experiments are identical except that the initial and lateral boundary conditions are derived from the NNRP, the original GCM output, and the bias corrected GCM output, respectively. In the second set, the GCM-driven RCM simulations with bias corrections and spectral nudging (IDDng) are compared with those without spectral nudging (IDD) and North American Regional Reanalysis (NARR) data to assess the additional reduction in RCM biases relative to the IDD approach. When the spectral nudging is applied, it will introduce the effect of GCM bias correction into the RCM domain, thereby minimizing the climate drift resulting from the RCM biases. Our analysis will focus on assessing the impacts of the new method on modeling extreme climate events.

The new dynamical downscaling method can be applied to regional projection of future climate or downscaling of GCM sensitivity simulations.