Geospatial Analysis of Tornado Path Locations Within and Near Storm-Based Tornado Warnings

Keywords: tornado, warning, probability, uncertainty, geospatial, analysis
Abstract Type: Paper Abstract

Abstract

Advances in research generating probabilistic hazard information (PHI) to accompany tornado warning risk perception has led to recent risk perception experiments using visual stimuli based on hypothetical probability plumes not built from observations, which could render findings less valid. This research aims to generate probability plumes from observed spatial patterns of tornado occurrences from historical data, which could be useful in establishing an empirical baseline of where tornadoes occur more or less frequently within warnings. A dataset was compiled of tornado warning polygons, paths, and initial Time, Motion, Location (TML) points/lines derived from warning text data from 2008-2019. Space-time matching was used to create a matched dataset with a unique identifier connecting all three datasets. We reprojected all warnings preserving their respective spatial representation of the path within the bounds of the polygon into a common coordinate system using the initial latitude/longitude locations provided within the TML information in each warning as a common reference point. We were able to stack and observe spatial frequencies of where paths tended to fall with reference to the initial point from each warning from years 2008-2019, producing a visual probability plume of tornado frequency based on historical data. Results showed probabilities near fifty percent in areas upstream of the polygon, and we found certain meteorological and warning factors affected overall path probabilities. These results could be useful for weather communications practices and research regarding the development and use of probabilistic hazard information in warning images.