Climatological analysis of the real-time NSSL 4km WRF-ARW

Tuesday, 16 December 2014
David Charles Goines and Aaron D Kennedy, University of North Dakota, Grand Forks, ND, United States
In recent years, the Weather Research and Forecasting (WRF) model has been used for dynamic downscaling of
Global Climate Models (GCMs) to forecast smaller scale phenomena that GCMs cannot resolve at coarse
resolutions. High resolution convection-allowing (CA) WRF simulations have gained popularity in recent years
due to their ability to resolve the structure of high impact phenomena such as topographically induced
precipitation, mesoscale convective systems, and supercell thunderstorms. An accurate representation of these
extreme events is important for climate mitigation and adaptation strategies by policy makers. With the usage of
downscaling and fine resolutions of WRF simulations becoming more recurrent, the question still remains: do
high resolution CA WRF simulations correctly represent climatological precipitation? This study examines the
climatology of precipitation over the U.S. Central Plains produced for 7 years (2007-2013) by the National
Severe Storms Laboratory (NSSL) high resolution (4km) CA WRF model. Precipitation forecasts for various
forecast hours are analyzed to determine whether the model climatology is similar to observations. The
Meteorological Evaluation Tool (MET) Method for Object-Based Diagnostic Evaluation (MODE) is utilized to
compare the precipitation forecasts to observations. The National Centers for Environmental Prediction (NCEP)
Stage IV multi-sensor precipitation analysis is used as the truth for model assessment. Model performance is
investigated for a variety of synoptic regimes using self-organizing maps (SOMs).