Today a moderately powerful earthquake rattled Washington, D.C. and was felt as far north as Massachusetts. Sitting here, feeling the earthquake shake my desk and water glass in central New York State, hundreds of miles from the epicenter, I was reminded that we are never far from the risk of natural disaster.
The Washington Post’s Jason Samenow wrote today that aftershocks are a significant concern. Although the 5.9 magnitude quake did not appear to cause significant damage, earthquakes are rare in the region and people are ill-prepared for them. According to the Post: “McNutt, director of USGS expressed a concern that the earlier quake will precede something more powerful: ‘What the concern is, of course, is that this is a foreshock. If it’s a foreshock, then the worse is yet to come.’ If not a foreshock, Mike Blanpied, associate coordinator for the USGS earthquakes hazards program cautioned aftershocks are possible: ‘Aftershocks could go on for days, weeks, or even months. They’re most likely to be felt under the next three or four days.’”
It got me thinking to ways that risk and data analysis techniques that we use every day in business applications could be applied in this situation. After all, the use of Monte Carlo simulation and and decision trees in DecisionTools Suite software has been used to cope with natural disasters – from volcanoes to hurricanes.
A few years ago the US Geological Survey asked the same question in an interesting study on the use of Monte Carlo simulation for the prediction of aftershocks in California. The paper, published in 2008, notes the typical absence of data specific to a particular earthquake site and examines the usefulness of Monte Carlo simulation for “assessing recurrence from limited paleoearthquake records.” In the absence of data, Monte Carlo simulation can be quite effective.
In a similar situation, the use of neural networks was examined by researchers in China to get a handle on the risk of aftershocks from the enormous 2009 Sichuan province quake. In their paper, published by the Journal of Sustainable Energy and the Environment in 2009, data from initial aftershocks was provided to a neural network so that it could “learn” any patterns in the aftershocks. These patterns were then used to predict future tremors. The concentration and trend of the aftershocks was predicted “precisely,” according to researchers.
In science as well as business, quantitative risk and decision analysis techniques produce tangible benefits that directly impact many of us.
VP, Palisade Corporation