You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

Hurricanes And Power Outages: Better Robocalls Through Science

After breathless earthquake coverage earlier this week, today in Washington it’s all hurricane, all the time. And if constant news coverage weren’t enough to spur endless chatter among locals about the impending storm, a pessimistic robocall going out to area residents should do the trick: Across the region today, scores of people were treated to a greeting from the hated utility company Pepco, warning them that their power will probably go out this weekend. (The surly reaction to this announcement is not due to the likely power outage, an event over which the company has little control—it’s in anticipation, given the company’s widely-criticized record, of the long, long waits for it to come back.) How can utility companies improve their record of preparing for, and responding to, power outages? Is there any way to avoid blanket robocalls to area customers, or any way to anticipate where repairs might be needed most?

A 2009 study on hurricanes and power outages called “Improving the Predictive Accuracy of Hurricane Power Outage Forecasts Using Generalized Additive Models” reports that it’s an awfully tall order. “Managing power outage risk and properly preparing for poststorm recovery efforts,” the authors write, “requires rigorous methods for estimating the number and location of power outages before a storm makes landfall. These estimates must be geographically detailed and accurate while also accounting for the complicated relationships between a number of possible explanatory variables and power outages.”

The authors employ a new mathematical model which, unlike a previously-used model tested in the study, doesn’t assume simple linearity for one crucial variable—and so can incorporate a more complex data set, which is important in the complicated circumstances leading to a power outage. The authors report that their new approach “allows the form of the relationship between the explanatory variables and the measure of interest, here power outages during hurricanes, to be estimated directly from the data.” They record promising results and say their work (which was funded by an unnamed utility company) will improve our ability to predict both the number and the distribution of power outages. And who knows? Someday, it could lead to a more efficient robocall.