Tornadic storms are random and short-lived. Even though there are close to 1000 tornadoes recorded annually, it has not been possible to obtain direct measurements of ground level wind speeds in tornadoes. Indirect assessment of wind speed through photogrammetric analysis (PA) of movement of debris and engineering analysis (EA) of damage to buildings has been used since the early 1970's. Both of these methods have limitations. Some of the limitations of photogrammetric analysis are that 1) PA is restricted to debris moving across the line of sight, 2) PA is feasible only at a few meters above the ground because of lack of visibility close to the ground, and 3) PA has uncertainty because the focal length of the lens and distance to the moving debris are unknown. The engineering analysis technique's limitations are that 1) in most cases it provides only a lower bound wind speed, 2) damage varies with proximity to the center of the tornado and with the amount of debris available, 3) structural and material properties are not accurately known, and 4) a weak structural link can collapse a building at a low wind speed.

In spite of its limitations, the EA technique provides a good measure of wind speeds when it is viewed in aggregate rather than by individual cases. EA of damaged buildings has been performed in major windstorms since 1970 (even earlier on occasion), when a severe tornado caused a large amount of damage in the city of Lubbock, TX, with a population of 150,000. During the last three decades, numerous engineering analyses of damaged buildings have been conducted in Lubbock (1970), Xenia, OH (1974), Monticello, IN (1974), Mobile, AL (1979 Hurricane Frederic), Iowa tornadoes (1979), Kalamazoo, MI (1980), Jefferson County, AL (1998), Oklahoma City, OK (1999), Fort Worth, TX, and others. These analyses in the aggregate provide a good understanding of wind speeds in tornadoes and the variability in the analysis procedure.

The paper will assemble the engineering analyses of buildings in tornadic storms. In addition, it will review credence levels of the resulting wind speed to assess uncertainty. The goal is to put engineering analysis in proper perspective and to assess its contribution in establishing wind speeds in tornadoes.