A Methodology for Improving Tornado Damage-Based Intensity Ratings

Fujita introduced the concept of rating tornadoes based on observed damage in 1971.  Fujita’s method of damage-based intensity classification used photos and brief word descriptions to describe typical damage for each F scale intensity level (F0-F5).  His method of rating tornadoes based on the maximum observed damage within the tornado path continues today with the Enhanced Fujita (EF) scale rating system. The EF scale was adopted by the NWS in 2007 and implemented into the NWS’s EF Scale Toolkit. The EF scale includes 28 damage indicators (DIs), each with various degrees of damage (DoD), and subjectively-estimated windspeed ranges for each DoD.  

The process of transforming tornado damage intensity ratings into windspeeds is a critical step in the development of tornado windspeeds for engineering design. In order to achieve acceptable performance for tornado-resistant designs,  the engineering framework for the analysis and design of structures (forward process) for specified tornado windspeeds and associated load effects should be consistent with the methods used for the estimation of windspeeds from damage (reverse process).

This paper illustrates how the results of engineering modeling/analysis of tornado simulated damage to single family residential houses can be used to estimate windspeed from damage observed in the field. We use Bayesian methods to develop probability distributions of engineering-based windspeeds given a field-observed DoD.  The proposed  method addresses many issues associated with the current deterministic mapping of EF DoDs to tight windspeed ranges: it is engineering based with a design-consistent analysis framework; it considers inherent variations in tornado wind loads and building position within the path; it considers variation in building practices/codes, and inherent variations in material/structural strengths; it includes significant differences in field-observable structural characteristics; it reflects uncertainties in non-observable structural details; and the method naturally considers overlapping DoD windspeeds. The method produces a progressive refinement in the tornado windspeed estimation process as more DIs are evaluated in the area of maximum windspeeds. Each additional DI adds more information to the process and may either sharpen or broaden the estimated windspeed distribution, depending on the level of consistency of additional observations/ratings. Single family homes were selected for analysis since they are one of the most frequently used DI’s to rate tornadoes in the EF scale and, also, there is a significant amount of field damage data and model validation on the performance of these structures in other wind storms.  

The paper concludes with suggestions for improving the number of DIs and the DoDs used for single family homes in the EF scale. Comparisons are made of the windspeed distribution estimates with the current EF DoDs for single family homes vs. a refined set of DIs/DoDs developed in this research.