Quantifying the Impact of Radar Data Assimilation in the Community Leveraged Unified Ensemble

Location: FL2-1001
Speaker:
Patrick Skinner, Cooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma and NOAA/OAR/National Severe Storms Laboratory
Description:

The proliferation of convection-allowing numerical weather prediction models (CAMs) motivated developers to conduct controlled experiments to determine best practices for CAM ensemble design during the 2016 Spring Forecasting Experiment of the Hazardous Weather Testbed.   The resulting Community Leveraged Unified Ensemble (CLUE) consisted of 10 sub-ensembles used to conduct 8 experiments.  This study focuses on one of these experiments, the impact of Doppler radar assimilation on resulting ensemble forecasts.  Forecasts of severe thunderstorm hazards from two 10-member CAM ensembles with identical configurations, except that one ensemble includes 3DVar assimilation of radar reflectivity, are considered for 24 days during May and June of 2016.  Forecasts of simulated composite reflectivity and 1-hour updraft helicity tracks are verified against Multi-Radar Multi-Sensor (MRMS) gridded radar observations. 

Initial results will focus on the development of methodologies for verification of simulated thunderstorm hazards against radar-derived proxies.  Specifically, improved quality control techniques for creating rotation tracks and development of model and observation-based reflectivity and rotation climatologies will be presented.  These results are then used to identify appropriate thresholds for neighborhood and object-based verification of forecasts using the Model Evaluation Toolkit (MET) and Method for Object-Based Diagnostic Evaluation-Time Domain (MODE-TD) software.  Verification results will focus on how long the radar data assimilation influences resulting forecasts, including differences in the climatology and accuracy of forecasts in the 18-30 hour range.