US Tornado Damage 1950 to 2021

How much damage from tornadoes would occur in 2021 for each of the past 72 years

The figure above shows an estimate of how much damage would result if each tornado season of the past 72 years occurred with 2021 levels of population and development. The loss data come from NOAA and the methodology is from a paper I co-authored, Simmons et al. 2013 (PDF). Note that 2021 is an estimate, based on early damage figures from NOAA and Karen Clark & Company. The red-capped line indicates that the 2021 final damage figures could come in lower or higher.

One important note: The damage figures presented here, as described in Simmons et al. 2013, are scaled to 2011. That is, our methods calculate total normalized losses as a proportion of official NOAA damage estimates for 2011. Different loss estimation methods will results in different estimates (so be careful about those apples-to-apples). Our methods focus on direct property damage, others include indirect damages like business interruption. Please see our paper (PDF) for all the gory details.

Some quick highlights:

• The largest loss year was 1953, with an estimated $35 billion in damage, which exceeds an average hurricane loss year;

• 2021 (which still has two weeks to go) is estimated at about $6.6 billion (~$4 to ~$10 billion), which would ranks it 16th (~6th to ~37th) of the past 72 years;

• It seems that 2021 will be remembered more for the tragic loss of life than for property damage. I recommend this excellent article for the challenges of saving lives when an intense tornado outbreak occurs.

• Average annual losses were about $7.5 billion from 1950 to 1985 (36 years) and less than half 1986 to 2021 (36 years), at $3.2 billion;

• The data show a long-term decline. The decline can be found in the NOAA loss data even before normalization.

Many people have been wondering about tornado trends, with some claiming that intense tornado incidence has increased dramatically over the period of record since 1950. The IPCC comes to a different conclusion:

• “observational trends in tornadoes, hail, and lightning associated with severe convective storms are not robustly detected”

• “attribution of certain classes of extreme weather (e.g., tornadoes) is beyond current modelling and theoretical capabilities”

• “how tornadoes or hail will change is an open question”

What does our research showing a clear decrease in overall and normalized damages contribute to this discussion? Here is what we said in our 2013 paper, which remains my view, further supported by trends in tornado incidence and damage since our paper was first published:

The degree to which this decrease [in damage] is the result of an actual decrease in the incidence of strong tornadoes is difficult to assess due to inconsistencies in reporting practices over time. However, an examination of trends within sub-periods of the dataset is suggestive that some part of the long-term decrease in losses may have a component related to actual changes in tornado behaviour.