The Environmental Modeling Center's Model Evaluation Group

The Environmental Modeling Center (EMC) established a Model Evaluation Group (MEG) in the spring of 2012. Inspired by model evaluation efforts at other numerical weather prediction centers, MEG's goal was to evaluate EMC model performance in a more comprehensive and timely way than before, emphasizing weekly synoptic discussion of model performance over the continental United States. MEG has been a major success in its first 3 years; several model problems have been noted, brought to the attention of EMC modeling teams and users of EMC models, and thoroughly investigated by model developers. Many of the problems, such as a near surface cold/wet bias in the Global Forecast System over the eastern United States, the initialization of the Short Range Ensemble Forecast System (SREF) and initial model snow cover, have been corrected or at least reduced in operations. MEG has also conducted thorough post-mortems of high-impact and sometimes poorly forecast events such as Superstorm Sandy, the June 2012 Ohio Valley-Mid-Atlantic derecho, and significant winter storms. Lines of communication have been opened between EMC and the National Centers, the NWS regional and local offices, and private customers to alert them of model biases and issues and provide a forum for model users to report problems they have seen in EMC forecast systems. The purpose of MEG is to help EMC developers improve the models by evaluating EMC models and soliciting information from users of EMC forecast systems and to provide information to users about the performance of the models and proposed model changes. MEG has been requested to play a larger role in working more closely with other NCEP centers and NWS regional and field offices to evaluate and understand proposed model upgrades earlier in the development and implementation process as well as daily issues in operational models. To meet this role and expand EMC's understanding of model performance, MEG needs to expand. How MEG can expand and better incorporate the experience and expertise of model users is under consideration. This talk will review the successes of MEG in its first three years and current EMC forecast system problems and discuss how MEG can most effectively be expanded to provide effective constant communication between EMC model developers and model users throughout the National Weather Service and the meteorological community.