科研之友
微信
新浪微博
Facebook
分享链接
摘要
A significant challenge with dynamical downscaling of climate simulations is the ability to accurately represent convection and precipitation. The use of convection-permitting resolutions avoids cumulus parameterization, which is known to be a large source of uncertainty. A regional climate model ( RCM) based on the Weather Research and Forecasting model is configured with a 4 km grid spacing and applied to the U.S. Great Plains, a region characterized by many forms of weather and climate extremes. The 4 km RCM is evaluated by running it in a hindcast mode over the central U.S. region for a 10 year period, forced at the boundary by the 32 km North America Regional Reanalysis. The model is also run at a 25 kmgrid spacing, but with cumulus parameterization turned on for comparison. The 4 km run more successfully reproduces certain observed features of the Great Plains May-through-August precipitation. In particular, the magnitude of extreme precipitation and the diurnal cycle of precipitation over the Great Plains are better simulated. The 4 km run more realistically simulates the low-level jet and related atmospheric circulations that transport and redistribute moisture from Gulf of Mexico. The convection-permitting RCM may therefore produce better dynamical downscaling of future climate when nested within global model climate projections, especially for extreme precipitation magnitudes. The 4 km and 25 km simulations do share similar precipitation biases, including low biases over the central Great Plains and high biases over the Rockies. These biases appear linked to circulation biases in the simulations, but determining of the exact causes will require extensive, separate studies.
- 出版日期2016-12-16
- 单位南京大学
