May 9, 2014

Tornadoes and Climate Change: Challenges in Interpreting the Record and Looking Forward

Posted by kcompton

By Harold E. Brooks, Senior Research Scientist, NOAA/National Severe Storms Laboratory

With the release of the new National Climate Assessment, the scientific community has put forward our best understanding of the changes that have occurred and are expected to occur as the planet continues to warm. Noticeably, little is said about tornadoes in this document. There’s good reason for this absence. Despite a wide variety of speculation in the online community, there’s little in the formal literature that addresses the problem.

Number of F1 and stronger tornadoes in 12 consecutive months beginning with the time on the x-axis. Based on data from the National Weather Service’s Storm Prediction Center.

Number of F1 and stronger tornadoes in 12 consecutive months beginning with the time on the x-axis. Based on data from the National Weather Service’s Storm Prediction Center.

The challenges in coming up with highly confident statements are many. Changes in the way reports have been collected and damage assessed over the years and over space make using the official tornado database (available from NOAA’s Storm Prediction Center) without considering those changes dangerous and make it impossible to compare to the rest of the planet. Verbout et al. (2006) indicate that the counts of F1 and stronger tornadoes starting in 1954 are reasonably consistent. (Tornadoes prior to 2007 are rated on the Fujita (F) scale and more recent ones are on the Enhanced Fujita (EF) scale. For this purpose, the scales appear to be similar in practice.) The weakest (F0) tornadoes have increased from fewer than 100 per year 60 years ago to 700-800 currently. For a shorthand, I’ll refer to the F1 and stronger as “damaging tornadoes.” If we take a running total over 12 month periods, we see that there’s little long-term trend in the number, and the average has remained a relatively stable 500 per year, but the records for most and fewest damaging tornadoes have both occurred since 2010. This appearance of increasing variability is also seen if we look at the timing of the early part of the “tornado season” during the year. If we use the date of the 50th damaging tornado (roughly 10% of the annual average) as a measure of the beginning of the season, 7 of the 9 earliest starts have occurred since 1997 and all 5 of the latest starts have occurred during that time, representing 30% of the record.

Average number of days per year with at least 1 F1 or stronger tornado (black line, left vertical axis) and days per year with at least 25 F1 or stronger tornadoes (red line, right vertical axis). Averages computed over a decade.

Average number of days per year with at least 1 F1 or stronger tornado (black line, left vertical axis) and days per year with at least 25 F1 or stronger tornadoes (red line, right vertical axis). Averages computed over a decade.

Underlying the increased variability is a large decrease in the number of days per year with at least one damaging tornado and a large increase in the number of days per year with a large number of damaging tornadoes, as seen in the decadal averages. In short, we have fewer days with damaging tornadoes being reported now than 40 years ago, but more “big” tornado days. The two changes have balanced each other, so that the decadal average number of tornadoes has remained relatively constant, but variability on time scales from days to years has increased. In effect, tornadoes have been concentrated into a fewer number of days. At this point, no physical mechanism has been found that explains the changes, but it very difficult to imagine a scenario in which the reporting changes are responsible since one aspect would come from more aggressive data collection and the other from more conservative data collection.

The reporting differences have led scientists to look at the distribution of environmental conditions in which severe thunderstorms and tornadoes most commonly form as an approximation of the events. This approach has also been applied to climate models (e.g., Trapp et al. 2007). In general, as the planet warms, the energy available to fuel thunderstorms will increase, but other factors that support conditions favorable for tornadoes could decrease with climate change, leading to a question of which influence will dominate. Diffenbaugh et al. (2013) have suggested recently that the environments supportive of tornadoes will increase over the 21st century, although the large interannual variability seen in the observations may make it difficult for such trends to emerge from the noise for a long time.

Fundamentally, even without a change in mean occurrence, the clustering of tornadoes could have significant impacts on society. In such a scenario, emergency response and insurance would find resources taxed significantly by an increase in the number of outbreaks. It is possible that the long-term casualty and damage averages wouldn’t change, but their distributions might have large changes. Preparing for such a future will require a great deal of thought and planning.

Dr. Brooks was a contributing author for the Third and Fifth IPCC Assessment Reports, writing about severe thunderstorms, as well as writing the severe thunderstorm section of the US Climate Change Science Program Synthesis and Assessment Product report on Weather and Climate Extremes. Dr. Brooks will speak at the 2014 AGU Science Policy Conference on June 17th.