Multivariate Self-Organizing Map Approach to Classifying Supercell Tornado Environments Using Near-Storm, Low-Level Wind and Thermodynamic Profiles Academic Article uri icon


  • Abstract Self-organizing maps (SOMs) have been shown to be a useful tool in classifying meteorological data. This paper builds on earlier work employing SOMs to classify model analysis proximity soundings from the near-storm environments of tornadic and nontornadic supercell thunderstorms. A series of multivariate SOMs is produced wherein the input variables, height, dimensions, and number of SOM nodes are varied. SOMs including information regarding the near-storm wind profile are more effective in discriminating between tornadic and nontornadic storms than those limited to thermodynamic information. For the best-performing SOMs, probabilistic forecasts derived from matching near-storm environments to a SOM node may provide modest improvements in forecast skill relative to existing methods for probabilistic forecasts.

published proceedings


altmetric score

  • 1.75

author list (cited authors)

  • Nowotarski, C. J., & Jones, E. A.

citation count

  • 12

complete list of authors

  • Nowotarski, Christopher J||Jones, Erin A

publication date

  • June 2018