Natural disasters may not always hit hardest where you’d expect

A post by Chris RowanThe week before last, the southeastern US was pummelled by a swarm of tornadoes that killed more than 300 people in 6 states, including Alabama (which appears to have borne the brunt of the damage) and Mississippi. Although tornadoes are hardly an uncommon occurrence in these states, if you were simply assessing the hazard based on the frequency of tornadoes, you might expect to be in more danger in states like Kansas and Oklahoma to the west. However, a recent study has shown that this is not the case: the most fatalities caused by tornadoes do not occur in the regions with the highest frequency of tornados, but in the states further to the east – like Alabama and Mississippi.

A comparison of the frequency of tornados in the USA (top) with the frequency of tornadoes that have caused fatalities (bottom). Note that the two peaks do not coincide.


(For another way of looking at the dataset plotted above, the New York Times has a nice interactive map/timeline of deadly tornados since 1950)

There are several reasons for this mismatch, the most important of which seem to be the higher density of mobile homes, which are much more susceptible to tornado damage, in the more easterly states; and less appreciation of tornado risks because of the lack of a focussed tornado season, and the perception of safety due to ‘Tornado Alley’ being further to the west. Sadly, the reduced risk of tornadoes occurring is more than matched by the lower resilience and preparedness of communities, which magnifies tornadoes’ impact when they do occur.

I’ve been wondering if similar factors are also an issue when considering the risk from earthquakes. Not unreasonably, a lot of attention is focussed on the seismic hazard in places which are located on plate boundaries, and where damaging earthquakes strike fairly regularly: places like California, or New Zealand, or Japan. However, while most large earthquakes do occur at the edges of plates, earthquakes large enough to do significant damage can and do occur in the interior of plates, the most famous example in the mainland US being the three magnitude 7.5-7.7 earthquakes that struck the New Madrid region between December 1811 and February 1812. In general, the risks from ‘intraplate’ earthquakes within plates are much more poorly known than the risks from ‘interplate’ earthquakes between plates: they follow less regular patterns in time and space, and because strain build up is much, much slower away from plate boundaries, there are much longer periods between one big earthquake and the next in any one area. If an active fault capable of producing large earthquakes only ruptures every few centuries or millennia rather than every few decades, they may not have left their mark in our historical records, and we may be almost completely ignorant of the danger they pose.

In the absence of this knowledge, similar factors to those that increase fatalities from tornadoes in less active areas may be important. The damage and potential casualties from even a moderate intraplate earthquake will be amplified because it will be striking an area with many, many more buildings that are not designed to withstand seismic shaking. Second, away from the regular, low level seismic activity near plate boundaries, people will not only not realise that large earthquakes are a possibility, they will we far less aware of how to respond if one does strike. Events like the recent Central US Shakeout, timed to coincide with the 200-year anniversary of the New Madrid earthquakes, are attempting to at least partially address the public awareness problem. But the fact remains that for many natural hazards, the actual risk is not purely a function of frequency and magnitude: politics, regulation and psychology are also a large influence on the potential human impact.

Categories: earthquakes, geohazards, society
Tags: , , ,

Comments (5)

  1. Lab Lemming says:

    Don’t the states east of the Mississippi have higher overall population densities as well?
    In general, cratonic interiors tend to have lower population densities to active margins, as active margins have lots of water and good soil.

    I think most of the nasty recent quakes have been on unknown faults in tectonically active areas, or known faults but larger that historical records.

  2. Jason says:

    Did you control for tornado size? What do the data look like for that? Is there a chance more, but smaller and therefor less destructive tornadoes, inflate the numbers for your first map. Also, I came here to say what Lab Lemming said.

  3. stuff says:

    Did you correlate for the time of day that the tornado struck. Tornadoes that occur after sundown are classically more deadly because fewer people are aware of them. In tornado alley (classic alley) the tornadoes are most often before sundown. In the Ozarks, they occur after sundown.

  4. k0ny says:

    Are you going to analyze the flooding on the Mississippi River and tributaries?

  5. Chris Rowan says:

    I should point out that this is not research of mine, so I can’t control or correlate for anything. The story I link to above does cite more tornados occurring at night as another possible factor in the elevated death rates in the south-east. I don’t know about population density, but that could also have an effect. Nonetheless, these issues are somewhat tangential to the main point I was trying to get across: that the popular perception of where the most dangerous areas are may not match up to reality, and this can have dangerous consequences.

    Lemming – taking the long(ish) view, there’s no such thing as a tectonically inactive area in continental crust – there’s just more and less active bits. My views on the population density of cratonic interiors may have been skewed by having lived in two rather large cities smack in the middle of cratons (Jo’burg, Chicago); all I know is that I find the idea of a magnitude 5-6 near Chicago – well within the bounds of possibility – a little worrying…

    Links (1)
  1. Pingback: Vulnerability to natural hazards, and other linky stuff « Geodoctor