Improving the anticipation of potential tornado intensity across different time scales
Previous studies of tornadoes have focused on understanding and predicting tornadogenesis or diagnosing the intensity of an ongoing tornado. Given that the majority of damage and fatalities are caused by strong to violent tornadoes, there is a need for robust operational tools that focus on anticipating tornado intensity rather than simply on tornadogenesis or ongoing tornadoes. Thus motivated, this study will improve the understanding of and provide new tools for the anticipation of tornado intensity before tornadoes form from the tornado watch time scale to the pre-tornadic stages of ongoing thunderstorms.
The first objective of this research uses an observational pre-tornadic radar and near-storm environmental dataset to confirm and further explore relationships with tornado intensity. Analyses of Doppler radar data and environmental parameters are used to propose an alternative framework for tornado intensity prediction during pre-tornadic stages of ongoing storms, conditional on tornadogenesis. A robust linear relationship (R² = 0.69) is found between pretornadic mesocyclone width and the EF rating of the subsequent tornado. Relationships between environmental parameters and tornado intensity depend in part on how the tornado-intensity categories are distributed [i.e., nonsignificant (EF0–1) versus significant (EF2–5), or weak (EF0–1) versus strong (EF2–3) versus violent (EF4–5)]. Low-level shear parameters discriminate the environments of significant tornadoes from nonsignificant tornadoes, but not the environments of violent tornadoes from strong tornadoes. The converse is true for thermodynamic parameters. The need for real-time, automated quantification of mesocyclone width in addition to intensity and other attributes for operational implementation of this framework for the purposes of impact-based warnings is described. The information gained from this pre-tornadic analysis would allow an operational forecaster to be aware of and communicate information about potential tornado intensity to the public before a tornado forms to better protect life and property.
The second objective of this research uses the relationships described in the first objective to explore tornado intensity prediction for ongoing thunderstorms using pre-tornadic radar data and the near-storm environment using binary classification machine learning. Several classification machine learning algorithms were implemented and used to examine their skill in predicting significant or non-significant tornado intensity for a given storm. The most-skilled classifiers were logistic regression, random forests, and gradient boosting as measured by cross-validated accuracy (~89%), precision (~94%), and recall (~73%), and other binary-classification metrics. The predictors of radar-derived pre-tornadic mesocyclone width and differential velocity were the most important, followed by environmental vertical wind shear and composite parameters. The results demonstrate a skilled, binary prediction of tornado intensity, conditioned upon tornadogenesis, and the potential for these machine learning applications to become a helpful resource in an operational setting through the fusing of the near-storm environment with real-time radar observations.
The third objective of this research uses High-Resolution Rapid Refresh (HRRR) model forecasts of storm-scale diagnostics such as updraft helicity (UH) and vertical vorticity as well as other environmental parameters such as the significant tornado parameter (STP) prior to convection initiation to examine their skill in predicting whether a severe weather event will be associated with no tornadoes, non-significant tornadoes (EF0-1), or significant tornadoes (EF2+). Each event and the associated HRRR forecasts are sampled within Storm Prediction Center tornado outlook regions using different thresholds of storm-scale diagnostics in combination with the large-scale environment.