Operational Use of Cell Trends as a Predictor of Non-Mesocyclone Tornadogenesis

This research paper presents new methodology for the early detection of non-mesocyclone tornadogenesis using WSR-88D Cell Trends display capabilities in conjunction with standard WSR-88D products. Statisitcal verification of the technique yields results of a Probability of Detection of 1.00, a False Alarm Rate of 0.33, a Critical Success Index of 0.67, and a Heidke Skill Score (HSS) of 0.78 (HSS indicates technique skill relative to random forecasting, where 1.0=a perfect score and 0=random forecasting). Statistical scores provided are for dependent data sets, as constraints have not yet allowed for an independent case study analysis.

This research focuses on warning techniques for convection producing landspout-like tornadic activity where the tornadic storm cells may display only weak rotation at low radar beam elevation angles on WSR-88D velocity products. Such minimal tornadoes are a difficult warning challenge using standard warning methods (i.e., mesocyclone tornadic activity).

Initial procedures included construction of severe weather and synoptic weather databases, constructed using NOAA Storm Reports and NOAA Synoptic Charts for 188 separate wet-season cases of tornadic activity that had occurred across Central Florida during the last decade. Upon completion of these databases WSR-88D Archive Level 4 data was collected via the NWSFO Melbourne radar for 405 separate storm cells over five storm event dates.

Synoptic chart information was analyzed for frontal system peninsular-orientation, pressure center orientation and intensities, and 500mb contour velocity data. Further synoptic information included surface station analyses (dew point, wind information, temperatures). WSR-88D information included: Composite and Base Reflectivity data, Base and Storm Relative Velocity (SRM) data, and Cell Trends display information (Vertically Integrated Liquid (VIL), Storm Top, Peak Reflectivity, and Peak Reflectivity Height). All reflectivity products were used for analysis of boundary-layer interactions (storm cell and surface boundary collisions), while all Velocity products were utilized for assessment of rotational couplets possessing storm rotational velocity values >20 knots. Cell Trends display functions and raw data was utilized for assessment of any tornado predictive signatures.

All above data sets were utilized for the development of severe weather warning signatures for operational use. From the data sets, a proposed method for early identification of wet-season, non-mesocyclone tornadic activity was developed. This method requires that a convective cell's VIL, Storm Top, Peak Reflectivity, and Peak Reflectivity Height values all have rapid (i.e., one radar scan) increase with value increase of VIL>10 kg/m**2, Storm Top>5000 feet, Peak Reflectivity>2dBZ, and Peak Reflectivity Altitude increase>6000 feet. Furthermore, the storm cell should first encounter a low level boundary during or prior to these increases. If these conditions are met and the storm develops or possesses an SRV couplet with velocity values>20 knots, then tornadogenesis is probable.

Of the 405 storm cells analyzed, six produced signatures meeting the above criteria. Of the six storm cells meeting the criteria, four produced tornadic activity.

Additional research is recommended for additional data sets yielding a larger sample size, and to determine if the signature is viable for Florida cool-season convection, non-mesocyclone tornadic activity elsewhere, or for more classical mesocyclone tornadic activity.