The spring of 2011 will likely be remembered as one of the most extreme tornado seasons ever to occur in the United States. As of June 16th there had been a total of 1482 tornado reports since the start of the year, with 537 deaths from 56 killer tornadoes, and an estimated $20 billion worth of damage caused (SPC). The vast majority of the death and destruction occurred during the ‘Super Outbreak’ of April 25–28 and the Joplin tornado of May 22. The former was the largest tornado outbreak in recorded history, with nearly 500 tornado reports and 322 deaths over the 4 day period. Most of these occurred in Mississippi and Alabama on the 27th which was the deadliest tornado day since the 1925 “Tri-State” tornado outbreak. The EF5-rated Joplin tornado meanwhile killed 131 people and caused an estimated $3 billion worth of damage in Jasper County, Missouri, making it one of the most deadly and costly tornadoes in US history. Other notable events included the April 14–16 outbreak, which caused 38 deaths across the states of Oklahoma, Arkansas, Alabama, Mississippi, North Carolina and Virginia, and the unusual New England tornado outbreak on June 1st, which killed 3 people in Springfield, Massachusetts. The extreme nature of this tornado season can be seen in the tornado trend graph below – currently 2011 is on track to have the highest annual count in the record.
Of course the question which immediately comes to mind is why? Why has this year seen so many tornadoes across the United States? And why has the death toll from many of these been quite so high? The second of these questions is much easier to answer than the first. The majority of US tornadoes occur in the vast open countryside of Great Plains, where major population centres are few and far between. Furthermore, most tornadoes are rather small phenomena, typically only a few hundred metres in diameter, so their damage paths tend to be similarly narrow. Unfortunately, this year a number of large and violent tornadoes passed through major towns and cities, including Tuscaloosa AL, Birmingham AL, El Reno OK, Philadelphia MS, Rayleigh NC and, of course, Joplin MO. There is unlikely to be any scientific explanation for this – it is simply bad luck. Of course increasing population (and population density) and the associated expansion of towns and cities has increased the probability of tornado fatalities over time. However, this has likely been offset by improved warnings and methods of communicating these to the general public (see graph below). The severe weather outbreaks this year were all forecast several days in advance and for most of the high-impact tornadoes, warnings were issued with significant lead time – for example, 24 minutes in the case of the Joplin tornado (NOAA). A huge amount of media coverage was also provided during the events (as is typical these days) increasing the likelihood that individuals would hear about an approaching storm. Nevertheless, it is clear that many people still failed to get the message, while others were simply unable to get out of harms way in time. Apathy could be part of the problem – Alabama meteorologist James Spann suggests that high false alarm ratios (FARs) have made many people complacent over the years. He argues that drastic changes are required in the way warnings are issued, with better targeting of NOAA severe weather radio broadcasts are greater use of social media by forecasters in communicating with the general public. This year’s tragic loss of life (see below; NOAA / Washington Post) certainly illustrates that much work must still be done to reduce the vulnerability of the US population to tornadoes.
Returning to our first question – why this Spring saw so many tornadoes – we note that the real meteorological question we are asking is ‘why were conditions so favourable to the development of tornadoes this Spring?’ To answer this we must first consider what conditions in general favour the development of tornadoes, or, more specifically, mesocyclone tornadoes – those that form from thunderstorms with a rotating updraught (supercells) – which are responsible for the vast majority of tornado damage and fatalities in the US. Like all forms of deep moist convection, supercells require instability to provide energy for air parcels to ascend, moisture for condensation of cloud drops and the associated release of latent heat, and a lifting mechanism to bring air parcels to their level of free convection (LFC). However, they additionally require strong vertical wind shear (changes in the speed and direction of the horizontal wind with height) to generate horizontal vorticity which can be tilted into the vertical and stretched by the storm updraught to provide rotation. A cap on convection, provided by a temperature inversion above the boundary layer, is also desirable as this allows convective available potential energy (CAPE) to build up to high values before it is released. These conditions frequently exist during Spring in the area east of the Rocky Mountains through a combination of favourable geography and synoptic weather patterns. Midlatitude storm systems, directed by the jet stream, provide vertical shear (through their baroclinicity), with southerly winds near the surface veering to westerly aloft. This results in the advection of warm, moist air from the Gulf of Mexico under dry air with steep lapse rates from the elevated desserts of the southwest US and Mexico, giving high values of CAPE. Lifting of air parcels to the LFC meanwhile may be provided by low-level convergence (along drylines, fronts or outflow boundaries) and ageostrophic motions associated with the jet stream. Of course, not all supercells produce tornadoes, and the exact mechanisms by which the latter form remains an area of active research. However, certain factors have been identified as favourable for tornadogenesis, including high boundary layer humidity and strong vertical wind shear in the lowest 1 km of the atmosphere (Craven and Brooks, 2004).
For much of this spring, and particularly during April, the conditions described above were frequently observed thanks to a persistent weather pattern which saw numerous storm systems pass through the eastern half of the US. A strong jet stream associated with upper-level troughs provided dynamic forcing while at the surface, strong southerly winds advected very warm, moist air from an abnormally warm Gulf of Mexico (see figures below; NCDC). According to Harold Brooks of the National Severe Storms Laboratory (NSSL), such a sustained pattern happens only two or three times a century (New Scientist).
The next question we might ask is whether the observed anomalies in wind patterns and sea-surface temperatures are related to larger-scale patterns of climate variability, such as El Niño / Southern Oscillation (ENSO), or long-term climate change. The occurrence of extreme weather often compels members of the general public and the media to ask whether these events are being ‘caused’ by things like ENSO and climate change. This can cause frustration for severe weather experts like Chuck Doswell of the University of Oklahoma, particularly when links are confidently drawn without much regard for the science (see, for example, the Guardian). However, the curiosity which drives these questions is not a crime, and the relationships between large-scale climate variability/change and tornadoes have generated significant interest in the scientific community.
Links between ENSO and US tornadoes have been drawn by Cook and Schaefer (2008) and Bove (1998), among others. Both of these studies suggest that La Niña conditions favour more frequent and severe tornado outbreaks, which is consistent with the events of this year, for which lingering cold anomalies in the west Pacific were present. However, a recent review by the Physical Sciences Division of the Earth System Research Laboratory (part of NOAA) argues that the relationship between ENSO and major destructive tornado events appears to vary over time and, in general, is difficult to infer due the small sample size. It suggests that future research should examine the sensitivity to the evolutionary history of the ENSO cycle and teleconnections which may pre-condition the atmosphere for destructive tornado outbreaks.
Numerous recent studies have also considered how anthropogenic climate change may affect the frequency and severity of tornadoes. Some have tried to directly infer changes using historical records of tornadoes while others have used climate models to simulate future changes under various emission scenarios. Of course, the resolution of these models is too low to represent even convective storms, let alone tornadoes themselves. Changes in tornadic activity must therefore be inferred from related environmental parameters, such as static stability, column moisture and vertical wind shear. Several studies using this methodology (e.g. Trapp et al., 2007), have found an increase in CAPE but a decrease in vertical wind shear, making it difficult to know how the occurrence of severe storms will be altered. It is possible that the former change will outweigh the latter, but in any case the effect on tornadoes remains a mystery.
Meanwhile, the approach of using records of tornado activity to infer changes is fraught with difficulties of its own. Observations of tornadoes (particularly weak ones) have increased significantly over time (see the example for Illinois below) but this appears to be almost entirely due to more people living in tornado-prone areas, and improved observation and reporting systems. Even if a trend does exist, there is so much interannual variability that it will be many years before we can detect it. The IPCC echo this point for localised weather phenomena in general in their 4th Assessment Report: “There is insufficient evidence to determine whether trends exist in…small-scale phenomena such as tornadoes, hail, lightning and dust-storms”.
The bottom line for this year’s tornado season is that, even if it does set a record we cannot take it as a harbinger of climate change. Dr Anthony Del Genio (NASA GISS) provided a good analogy for this in a quote to New Scientist: “It’s sort of like judging a baseball player who comes up from the minors and hits a home run at his first bat. Is that a fluke, or is this guy going to be the next great baseball player?”