March 28 2007 Tornado Outbreak Model Comparison Jonathan D. Finch
Purpose The purpose of this presentation is to stress the importance of using global scale models for severe storm prediction in the 24 to 48 hour time frame. Use of the UKMET and eventually the GFS helped in the forecast process leading up to March 28, 2007.
UKMET 48hr fcst valid 00 UTC March 29
Mar 29 2007 0 UTC Mar 29 2007 0Z
UKMET (Green) NAM/WRF(Orange) 00z Mar 28 2007 initializations
March 29 2007 00 UTC Mar 29 2007 02 Z March 29 2007 00 UTC
Impact of Elevated Heating on Theta-e elev. ........ SLP...........T.......TD......pot. T......MR....theta-E ---------------------------------------------------------------------------- Coolidge 3500ft.......1000..........65......56......84.5F......11......335.1 Angelton 0006ft.......1015..........73......68......70.7F.....14.6....336.7
Bangladesh and East India Tornado Prediction Site More on Elevated Heating see my web page: Bangladesh and East India Tornado Prediction Site http://bangladeshtornadoes.org
March 29 2007 0 Z UKMET
Conclusion In the March 28, 2007 case, The UKMET model outperformed the NCEP models, especially the NAM/WRF.
My personal observations The UKMET usually outperforms the NCEP models, especially the NAM/WRF, in the 48 to 72 hr time frame in the cool season. When large scale processes are dominating (i.e. 3-28-07), the global scale models will usually be superior with day 2 to day 3 and even day 1 placement of synoptic features. In warm season events when the small scale is very dominant(i.e. weak flow at mid levels with MCS clusters on going), the mesoscale models become more important to look at.
Implications for Operations Many forecasters like to use the NAM/WRF because of its high resolution output and severe weather parameters. But if the smaller scale models are in error with synoptic features, then can we expect its mesoscale features to be correct? On the other hand, the large scale models may miss subtle mesoscale boundaries such as outflow boundaries. I would argue that this is mostly important in the shorter range(24hrs?) I think we need to make effective use of all the models, understanding the limitations of each model. Of course, the forecaster must be in touch with observational data at all times as well.