Collaborative Research: Detection and Estimation of Multi-Scale Complex Spatiotemporal Processes in Tornadic Supercells from High Resolution Simulations and Multiparameter Radar

The project is to understand thunderstorm conditions that trigger tornados. Each year across broad regions of the United States, atmospheric conditions become favorable for the formation of supercell thunderstorms, the most prolific source of violent tornadoes. Tornadoes ranked EF4 and EF5, the top strength categories of the Enhanced Fujita scale, are responsible for the bulk of fatalities, even though they are the least common, comprising less than 1% of observed tornadoes.

The death and destruction wrought by supercell tornadoes has motivated much observational, theoretical, and numerical modeling research designed to understand and predict these powerful storms. However, despite the many advances that have resulted from these studies, there is currently poor understanding of what determines whether a supercell will produce a tornado or not, and whether that tornado, should it form at all, will be weak or strong, short-lived or long-lived.

This complex question is not only one of the great mysteries of nature but is of critical importance to assuring public safety. The project will investigate these issues by combining observational, numerical, and analytical methods. The project will develop educational exhibits on tornadoes at the Fleet Science Center at Balboa Park, San Diego, CA and the National Weather Museum at Norman, OK.

The project will also provide unique research and education opportunities for undergraduate and graduate students in understanding tornado evolution through high-resolution numerical simulations as well as data analysis and visualization.

The central challenge for understanding the generation and maintenance of violent, long-track tornadoes in supercells is being able to quantify the storm-wide processes that determine whether strong, long-lived tornadoes form. This proposal will use a novel method called the Entropy Field Decomposition (EFD) as a unifying framework to integrate and quantify the complex dynamics of tornadic supercells produced in high resolution physics-based simulations, predicted radar signatures derived from these simulations, and actual observational data of supercells collected in the field.

EFD is a data-agnostic approach to four-dimensional space-time entangled data mining that leverages techniques from Bayesian analysis and the physics theory of fields to identify statistically significant storm “modes" within huge volumes of complex, often noisy, data. In contrast with machine learning approaches, no training datasets are required.

Rather, prior information within individual data derived from space-time correlations, codified in the theory of Entropy Spectrum Pathways (ESP), provides sufficient prior information to extract distinct space-time modes of complex systems. This method will be used to study a first-of-its-kind data set comprised of ensembles of high-resolution simulations that yield a rich variety of tornadic and non-tornadic storms to understand fundamental controls of tornadogenesis, tornadogenesis failure, and tornado maintenance.

This ensemble will also enable some of the first detailed intercomparisons between mobile radar observations and tornado-resolving, idealized simulations.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.