What can the recent Kentucky tornado disaster teach us about earthquake early warning?

The recent tornado disaster in the Midwest was preceded by copious warnings, giving time for protective actions, yet the destruction and suffering caused was huge. At PNSN we are in the midst of building and implementing the US Geological Survey's ShakeAlert Earthquake Early Warning (EEW) system on the US West Coast, so we are compelled to examine whether the storm warning experience contained lessons for seismological warnings. While data and experiences are still being collected, analyzed, and interpreted, an article in the December 14 New York Times and another from the Washington Post (printed in the Dec 16 Seattle Times) have provided enough analysis that I can speculate a bit about the major similarities and differences between tornado and earthquake warnings.

 

Damage from the Joplin, Missouri EF-5 tornado in May 2011. (Source.)

 

1. The storm/tornado warning system worked as designed, performed well, and no doubt saved many lives. A similar storm in the same region almost 100 years ago killed nearly 10 times as many (un-warned) people. Like the case of earthquakes, the storm warnings mitigated a hazard that is very uncommon at any given location.  There is no reason to think that EEW couldn’t have similar successes, when the ShakeAlert system is fully operational. 

 

2. Relative to EEW, storm warnings were leisurely—providing minutes of warning; EEWs will give only a few seconds. The additional time afforded for storms is a bit of a two-edged sword. People have more time to review emergency procedures, make decisions, and carry out mitigating actions.  So effective procedures and decisions need to be ingrained and almost automatic. There is no substitute for training exercises. For earthquakes the shorter warning times make actions pretty simple: Drop, Cover, and Hold on. That’s what you will have time to do—and what will reduce injuries.  There is no time to reflect or second guess a warning. For an infrequent hazard action must be almost a reflex. Again, there is no substitute for training. 

 

3. Early warning systems will probably cause some “false alarms”… and that is OK. Of course false alarms due to system errors must be eliminated if possible.  Nevertheless even a well-functioning alarm system will alert many more people to take action than explicitly need to be in order to escape a particular risky situation. That may be interpreted as a false alert. For EEW, as noted above, all one has time to do is Drop, Cover, and Hold on, an activity with almost no cost or risk in and of itself. The benefit/cost ratio for EEW actions is high.

 

4. Early warnings do not do much, if anything, to mitigate damage to most types of built infrastructure. Structures must be well-designed, well-built, and code-compliant in order to best protect occupants. EEWs can provide protection for equipment or procedures that shaking may affect by initiating automated protective actions, even without total failure and collapse of the structure (think fabrication or surgery).  This type of damage isn’t as dramatic as building collapse, although the economic consequences can be immense. Nevertheless, much more effort must be made to ensure our infrastructure is more resilient to all sorts of natural hazards.  

 

5. The damage distributions of earthquakes may differ from the storm images we saw from communities in Kentucky. Air photos of the tornadoes’ paths show broad swaths of structures with damage that is difficult to put into words: …flattened…razed…utterly demolished.  Dramatic structural failures caused by earthquakes seem to me to be generally more patchy or spotty. (Exceptions to spotty earthquake damage are places with extremely and uniformly poor construction, and the inundation zones of strong tsunami.) I realize this is an arguable hypothesis; it is true that impacts of smaller tornadoes can also be very mercurial and scattershot. But for both earthquakes and storms, warnings provide time to prepare for potential impacts.  For earthquakes, there are specific sites where the geological conditions enhance shaking (or liquefaction impacts, say) or where structural types are more vulnerable.  Both hazards, for different reasons, are difficult to “tune” exactly to expected effects, although EEW may eventually be more site-specific.  But with our current generation of ShakeAlert EEW much of the injuries caused by earthquakes that can be reduced result from indoor hazards like heavy falling objects and tripping hazards, and this isn’t necessarily obvious in aftermath air photos.

 

6. Tornadoes, as far as I know, have no equivalent to the aftershocks experienced in earthquakes. The storm passes, the destruction happens, and then the heartbreaking images reveal the extent of damage, and the shocked population must start digging out.  And the rest of us must contribute what we can to help.  After most large earthquakes, we can anticipate a sequence of aftershocks that endanger rescue and recovery operations.  Another, often-unrecognized impact of aftershocks on people is to engender feelings of fear and anticipation that can deepen their depression and shock. EEW provides hope to mitigate these impacts of aftershocks by providing time for emergency workers to clear or stabilize a site, and giving people a few moments to prepare mentally  for shaking.  An EEW system like ShakeAlert must maintain operations, and system users/clients must be trained for this warning venue.

 

So while it is early in the aftermath of the midwest storms, let’s keep the discussions going about the implications for short-term warnings we provide for other natural hazards like earthquakes (and volcanoes). I hope this will stimulate discussions about what constitutes a successful early warning and how to nurture, assess, and advance EEW implementation in the Pacific Northwest.

 

---

Grade 8 students at the Notre Dame Regional Secondary School practice how to react to an earthquake early warning alarm. (Photo from the CBC.)