455 150-Year CONUS Quasi-Dynamical Downscaling Product

Tuesday, 24 January 2017
4E (Washington State Convention Center )
Trude Eidhammer, NCAR, Boulder, CO; and E. Gutmann, M. Clark, J. R. Arnold, L. D. Brekke, R. Rasmussen, and K. Nowak

To produce hydrological projections of change due to anthropogenic climate change, downscaling of climate models is necessary to provide information on relevant spatial and temporal scales. Here we present 12 km downscaled projected precipitation over the Contiguous United States (CONUS) from 10 CMIP5 models (Coupled Model Intercomparison Project Phase 5) for RCP4.5 and RCP8.5 from the years 2006-2100. In addition, we downscaled the historical runs (1950-2006) of the same CMIP5 models. We employed a quasi-dynamical downscaling approach using the Intermediate Complexity Atmospheric Research (ICAR) model. This model uses a complex linear model to describe the perturbations to the wind field caused by topography, and includes detailed physical schemes for advection, microphysics, and land surface modeling.  For input to the ICAR downscaling, the CMIP5 runs were first bias corrected against ERA-I reanalysis (1980-2005), then downscaled by WRF to a 50 km grid resolution to provide hourly input to ICAR. The ICAR CONUS simulations (using ERA-I as input) are compared with gridded observational datasets. We evaluate the downscaled CMIP5 results based on the historical runs, and then assess the variability in the projected precipitation between the different downscaled CMIP5 models from the ~100 year future downscaled ICAR results.  Finally, the downscaled ICAR results are compared with other statistically downscaled datasets, such as the Bias Corrected Statistical Disaggregation (BCSD) climate projection, and a hybrid analog-regression statistical downscaling technique based on the WRF-50 km simulations.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner