P26 One-way downscale on a 4-km continental scale climate simulation and test on land surface coupling strength

Zhang, Zhe, Yanping Li, University of Saskatchewan, and Fei Chen, National Center for Atmospheric Research

High-resolution regional simulations are required to provide information for local scale hydrology and agriculture study. Two sets of one-way dynamical downscaling simulations have been performed using the 13-year continental scale 4-km WRF (CONUS) simulation as forcings, focusing on the central U.S. during a heatwave event in 2006 summer. The first simulation (i.e., the control (CTL) run) is conducted with the exact same model configuration as that of the CONUS run, except the spatial resolution changed from 4-km to 1-km. In the second simulation (the Z0H run), the land surface module is modified to mitigate the warm bias in the 4-km WRF simulations, by changing the roughness length for heat and moisture exchange, hence to reduce the coupling strength between land and atmosphere. The 2-m temperature and precipitation data were evaluated for both 1-km simulations and the original 4-km CONUS simulation against gridded observation data PRISM and METAR station data. Compared to PRISM gridded data, the monthly mean temperature of the 4-km CONUS run is overestimated but within 2 degrees C bias from April to June, and the bias is greatly intensified over the entire domain in July. The same bias is found in the 1-km CTL run, but with greater error. However, the 1-km Z0H simulation shows slight cold bias from April to June, but less warm bias compared to the original 4-km CONUS simulation. The two 1-km simulations show more precipitation in April and May while less precipitation in June and July, with the 1-km Z0H simulation produces better precipitation pattern than that of the 4-km CONUS simulation. When compared to METAR station observation, the two 1-km simulations show reasonable daily minimum temperature but overestimate the daily maximum temperature. Two sets of offline HRLDAS simulations are implemented over the same 1-km domain, one using default setting and the other changing Z0H to show the impact of modifying land-atmosphere coupling strength on regional climate. These two sets of 1-km simulations  (offline and coupled) suggest that the land-surface model parameterization is critical for simulating precipitation and near-surface weather variables in regional dynamic downscaling to very high resolution.