Dynamical downscaling for Alaska

Alaska, Canada, and Russia
Aug 2013 to Aug 2016

Dynamical downscaling uses regional models to downscale the coarse output of global climate models and reanalyses to finer spatial and temporal resolution so that such data can be better used by stakeholders for research and planning for future climate change. In this project we use the Weather Research and Forecasting (WRF) model to downscale the ERA-Interim reanalysis and CMIP5 NCAR-CCSM4 and GFDL-CM3 (historical and RCP8.5 simulations) to a 20km grid encompassing all of Alaska for 1970-2100. These data will provide hourly temperature, precipitation, wind speed and direction and numerous other variables on a consistent grid. The ERA-Interim downscaling has been completed for 1979-2013 and downscaling of the two CMIP5 models is currently in preparation.
Comparison of the downscaled with both seasonal PRISM-based gridded and station temperature and precipitation observations shows that the downscaling adds detail to the coarse reanalysis especially in areas of significant topography. For example, in Interior Alaska temperatures are warmer over the mountains relative to the surroundings in the downscaling and gridded observations while unrealistically smooth in the coarse reanalysis. Similar improvements in precipitation related to topography are also seen. However, some biases are present and many of the biases that existed in the original reanalysis were carried into the downscaling. In some instances biases were improved, such as temperatures over higher terrain in winter for example, or worsened like in an enhanced wet bias over higher terrain in the Interior summer. Once the CMIP5 historical downscaling is completed for 1970-2005 it will be compared with the ERA-Interim downscaling to help prepare bias corrections.