Forecast Improvements
Other Agency – Evaluation of Synoptic-Scale Controls on Tornado Outbreaks
Funding Agency: NSF
Objectives
Determine signals in synoptic scale data that can distinguish between tornado
and non-tornado outbreaks; create fields to be used to initialize the
WRF model for the prediction of tornado outbreaks.
Accomplishments
An important tornado forecasting problem is to decide whether or not a
particular synoptic-scale system is going to produce a significant outbreak
of tornadoes. Although much work has been done on individual case studies
over the decades since tornado forecasting began in the 1950s, this issue
remains problematic for forecasters. In our second year of the project,
a set of 50 tornado outbreaks and another set of 49 primarily nontornadic
outbreaks of severe weather have been selected for analysis. All of these
61 cases have been run by research assistant Chad Shafer using the WRF
model starting with coarse global NCEP Reanalysis data, 24 and 48 hours
in advance. Runs beginning 72 hours in advance are in progress, to be
completed by the fall. Diagnosis of the model simulations is in progress
as well. The model runs include nests down to 2 km grid spacing, but
will not resolve tornadoes, of course. Thus, we are seeking “proxy
variables” to discriminate tornadic from primarily nontornadic
simulations. This effort involving research assistant Andrew Mercer is
underway and preliminary results look promising. Mr. Mercer is also developing
composite analyses for use in the WRF model, based on principal component
analysis of our selected outbreak cases. This effort will involve Eigen
analysis of very large matrices and so is pushing the limits of the on-campus
supercomputing system software.
This project is ongoing.
Publications
Mercer, A.E., C. Shafer, C.A. Doswell III, L.M. Leslie, and M.B. Richman,
2006: Tornado outbreak detection using modern numerical simulations.
23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc.,
CD-ROM.
Mercer, A. E., C. M. Shafer, C. A. Doswell, L. M. Leslie, M. B. Richman, 2006: Analysis of tornado outbreaks using principal components. 23rd Conf. Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., CD-ROM.
Shafer, C. M., A. E. Mercer, M. B. Richman, L. M. Leslie, and C. A. Doswell, 2006: Analysis of WRF and MM5 mesoscale model forecasts to distinguish tornado outbreaks from primarily nontornadic severe weather outbreaks. 23rd Conf. Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., CD-ROM.
Shafer, C. M., A. E. Mercer, M. B. Richman, L. M. Leslie, and C. A. Doswell III, 2007: Analysis of WRF and MM5 mesoscale model forecasts to distinguish tornado outbreaks from primarily nontornadic severe weather outbreaks. 87th Annual Meeting, San Antonio, TX, Amer. Meteor. Soc., CD-ROM.
Subjective forecast quality evaluation for the 24-h WRF model forecasts run using the NCEP/NCAR global reanalysis data as input, for the 50 cases of major tornado outbreaks and 50 cases of primarily nontornadic severe weather. The majority of the forecasts are either "excellent" or "very good" for these cases.