Are tornadoes becoming more common because of climate change?
While the jury is still out on just how exactly tornado events will be impacted by climate change, there have been several studies performed by other scientists that give us clues about how things might play out in the future.
Before we discuss the science, there are a few very important points that need to be made about tornado climatology.
In general, the state of the climate investigates long term trends of atmospheric conditions over space and time. At the minimum, a climatology must include at least 30 years’ worth of data. In the context of global climate change, scientists investigate trends over much longer time frames – generally on the order of thousands of years. In historical terms, climatologists tend to focus on a time frame of approximately 25,000 years before current era. This time frame coincides with what is referred to as the interglacial Pleistocene, the timeframe after the last ice age. (Although, it should be noted that climate data exists for much longer timeframes too!) However, the interglacial Pleistocene is representative of a period of moderately constant conditions over space and time, when the physical construction of the Earth’s surface such as continental distribution, is what we are familiar with today. Scientists can obtain information about global and even regional temperature and moisture conditions from proxy measurements such as tree rings and through direct measurements of certain chemical constituents known to be related to temperature conditions, the presence of certain species of plant and animals, the presence of dust and sediment deposits, etc.
THE US HISTORIC TORNADO RECORD: In contrast to what we can obtain for general understanding of Earth’s past temperature and moisture, there is no way to retrieve information about historic tornadoes except through direct human observations. This makes constructing a historic database of tornadoes incredibly problematic because it introduces a non-physical requirement to the observations. Simply, someone somewhere must have seen the tornado and reported it. In fact, after a brief period of interest in the late 1880’s, there was actually a ban on forecasting or reporting tornadoes from 1905-1938 based upon the belief that the panic incited by the mere threat of a tornado was more harmful than the effects of an actual tornado event itself (Bradford, 1999). Note that it was during this timeframe that the historic “Tri-State Tornado” occurred (18 March, 1925), lasting 3.5 hours, traversing 219 miles through Missouri, Illinois and Indiana, and killing 695 people (https://www.weather.gov/pah/1925Tornado_ss). Severe weather forecasting was re-introduced to the U.S. Weather Bureau, coincident with the development of the Severe Local Storms Forecasting Unit (SELS) in 1952, but it wasn’t until the development of the Fujita Scale in 1971 that tornadoes were officially recorded and assigned an F-scale intensity value. The F-scale used damage incurred to single-residence, well-built homes to assign an approximate wind speed value for the tornado winds. Modern US tornado records only date back to 1950 and the tornadoes that appear in the database between 1950 and 1971 were assigned an F-scale rating retroactively. Scientists and students carefully combed tens of thousands of newspaper articles and old scientific papers to find reports of tornadoes and then added them to the database. Because the event had to be newsworthy in the first place, many weak tornadoes are surely missing. Furthermore, damage ratings were assigned from pictures that may or may not have represented the true intensity of the tornadoes. After 1972, the National Weather Service began recording tornado reports in a manner similar to what is currently practiced today. Formal damage surveys were introduced, many performed by Fujita himself, but there were still likely many events that occurred in rural areas or were weak events to begin with that were never reported. Then, with the introduction of a National Doppler Weather Radar network in the early 1990’s, meteorologists were better able to identify areas of rotation without human reports. This created a non-physical spike in tornado reports that began this decade. Even more recently, with the advent of cellular based internet, video streaming, and a boom in interest in storm chasing, tornadoes are much more likely to be seen and observed. An update to the Fujita Scale, coined the Enhanced Fujita (EF) Scale was implemented in 2006. This update changed the wind speed values assigned to the 0-5 intensity values, and added a wide range of damage indicators to the list of objects that could be used to provide an EF-scale rating beyond a well-built home.
Putting all this together, the US historic tornado database is fraught with inaccuracies, inconsistencies, and non-physical biases, making it unwise to use without substantial investigation and correction for these biases. Taken at face value, tornado reports appear to be increasing with time almost consistently from 1950 onward. However, this is not truly representative of the actual numbers of all tornadoes that occurred during this era. Unfortunately, we will just never know how many tornadoes occurred in the past. As we move forward past the year 2000, we are obtaining more consistent numbers, but it is inappropriate to construct a picture of tornadoes in the climate system with such a relatively short duration of reports.
What do we know about how tornado events might change in the future? We use advanced climate models, which represent the full range of atmospheric processes including land-air, water-air, and ice-air interactions to investigate how the large-scale atmospheric variables that are known to be associated with tornado producing storms (called proxy variables) change with time. These variables include various measures of atmospheric instability, placement of the jet stream and surface lows, and vertical wind shear. Climate models are too coarse spatially to be able to resolve an individual tornado, but by identifying changes in these larger-scale atmospheric parameters which are resolvable, we can get a feel for what the future might hold. Studying changes in these environment proxy variables can be done historically as well, by the use of reanalysis datasets. From these studies we can have moderately high confidence in the following conclusions: i. Tornadoes across the SE US are increasing in frequency over the past several decades, and they have decreased in frequency over the traditional “tornado alley” of the central US Plains states. ii. The number of days in the future with tornadoes will likely decrease. But, the days that have tornadoes will likely have a larger geographic extent and will have more tornadoes. iii. There is some evidence to suggest increasing numbers of tornadoes in areas like the Ohio Valley and New England, although the increases are not statistically robust. iv. There will likely be increased inter-annual variability in tornado counts. We are likely to see more frequent years with below-average tornado events juxtaposed against more years with above-average events.
In conclusion, we expect that the net change in the number of tornadoes per year is 0 – the average number of tornadoes per year will ultimately stay the same. There will be fewer days where tornadoes happen, but more tornadoes on those days. However, the geographic distribution of where the tornadoes occur might change with time.
Dr. Houser’s personal work on climatology has specifically focused on creating the database for severe weather events in Ohio found here. Future work is ongoing and mostly specific to Ohio weather.
References and further reading:
Agee, E., Larson, J., Childs, S. & Marmo, A. Spatial redistribution of US Tornado activity between 1954 and 2013. J. Appl. Meteorol. Climatol. 55, 1681–1697 (2016).
Bates, F. C., 1962: Severe Local Storm Forecasting. Bulletin of the American Meteorological Society, 43 (7). 288-291.
Bradford, M., 1999: Historical Roots of Modern Tornado Forecasts and Warnings. Weather and Forecasting, 14 (4), 484-491. DOI: https://doi.org/10.1175/1520-0434(1999)014<0484:HROMTF>2.0.CO;2
Brooks, H. E., Carbin, G. W. & Marsh, P. T. Increased variability of tornado occurrence in the United States. Science 346, 349–352 (2014).
Dixon, P. G., Mercer, A. E., Choi, J. & Allen, J. S. Tornado risk analysis: Is Dixie alley an extension of tornado alley? Bull. Am. Meteorol. Soc. 92, 433–441 (2011).
Doswell and Burgess, 1988: On Some Issues of United States Tornado Climatology. Monthly Weather Review, 116 (2), DOI: https://doi.org/10.1175/1520-0493(1988)116<0495:OSIOUS>2.0.CO;2
Farney, T. J. & Dixon, P. G. Variability of tornado climatology across the continental United States. Int. J. Climatol. 35, 2993–3006 (2015).
Galway, J., 1992: Early severe thunderstorm forecasting and research by the United States Weather Bureau. Wea. Forecasting,7, 564–587. https://journals.ametsoc.org/view/journals/wefo/7/4/1520-0434_1992_007_0564_estfar_2_0_co_2.xml
Gensini and Brooks, 2018: Spatial trends in United States tornado frequency. Nature: Climate and Atmospheric Science. https://www.nature.com/articles/s41612-018-0048-2
Guo, L., Wang, K. & Bluestein, H. B. Variability of tornado occurrence over the continental United States since 1950. J. Geophys. Res. Atmos. 121, 6943–6953 (2016).
Moore, T. W. Annual and seasonal tornado trends in the contiguous United States and its regions. Int. J. Climatol. 38, 1582–1594 (2018).
Tippett, M. K., Lepore, C. & Cohen, J. E. More tornadoes in the most extreme US tornado outbreaks. Science 354, 1419–1423 (2016).
Verbout, S., H. Brooks, L. Leslie and D. Schultz, 2006: Evolution of the U.S. Tornado Database: 1954-2003. Weather and Forecasting 21 (1), 86-93. DOI: https://doi.org/10.1175/WAF910.1