About the SA Blog Network



Opinion, arguments & analyses from the editors of Scientific American
Observations HomeAboutContact

Damage from Extreme Weather Increasing

The views expressed are those of the author and are not necessarily those of Scientific American.

Email   PrintPrint

Hurricane Irene is part of a worsening trend. Weather disasters have grown more frequent and more costly over the past 30 years in the U.S., according to data that was released today by the National Oceanic and Atmospheric Administration (NOAA).

On Thursday afternoon, NOAA posted a map of the 99 weather disasters that caused at least $1 billion in damages in the United States between 1980 and 2010.

Given that Hurricane Irene is the 10th billion-dollar weather disaster in the U.S. so far this year—breaking a record set in 2008—I wondered what sort of trend might show up over those three decades. By doing simple linear regression on the underlying data, I saw pretty clearly that we can expect to see a lot more frequent and a lot more costly billion-dollar weather events in the coming years.

In the following chart, based on the NOAA data, I focused just on the number of billion-dollar weather events:

line graph of billion dollar weather events

And in the chart below, I focused on the total annual costs from these extreme weather events. Notice the spike in 2005, due in large part to Hurricane Katrina. The total damage estimate from Hurricane Irene has not yet been determined. But so far this year, the US has already experienced $35 billion in damage from billion-dollar weather events.

line charts showing costs of billion dollar weather events in US

Oddly enough, the NOAA map does not make the trend data clear. Fortunately, it was easy enough to do on my own, using basic Excel and Acrobat tools.

I first learned about the NOAA map on Thursday afternoon when Justin Kenney of NOAA tweeted its existence. I followed the link and saw that all the information was in the form of .pdfs—not terribly useful for analyzing the data on your own. So I pinged Justin to find out how to get the underlying data and, as he suggested, contacted the National Climatic Data Center in Asheville, N.C.

They took my request and while I waited for them to get back to me, I manually entered the information from a NOAA table in pdf form into Excel to create a simple spreadsheet that totaled number of events and costs. Then I used Excel’s chart wizard to generate a simple linear regression line on both the number of events and costs and used Acrobat to add source information.


About the Author: Christine Gorman is the editor in charge of health and medicine features for SCIENTIFIC AMERICAN. Follow on Twitter @cgorman.

The views expressed are those of the author and are not necessarily those of Scientific American.

Rights & Permissions

Comments 18 Comments

Add Comment
  1. 1. mtobis 7:54 pm 09/1/2011

    I don’t see anything about “constant dollars” and so this is partly a graph of inflation. Also, dollar damage goes up with wealth and real estate value. So it is partly a graph of that. Finally, the hurricane component of the graph partly indicates an increasing population density at the coasts.

    This matter is hotly debated and the graph makes it seem misleadingly obvious. While I understand that there is something to the increase in damaging weather events, especially floods, these graphs greatly overstate it.

    Link to this
  2. 2. Christine Gorman 11:10 pm 09/1/2011

    Nope, inflation is not the issue. The data are in constant dollars. As the y-axis of the second graph makes clear, the figures are based on the value of the dollar in 2007.

    Obviously, growing coastal population and infrastructure play roles, especially with respect to hurricanes. But a lot of the economic damage in these graphs reflect flooding in the Midwest, drought and fires in the West–which can be verified by following the links to the underlying data.

    Link to this
  3. 3. cfeagans 11:57 pm 09/1/2011

    I like that you took publicly available data and used it to literally paint a picture (the graphs) that show the effect extreme weather events are playing on our economy.

    I’ll be very surprised if this doesn’t get picked up by other media. While there’s certainly something to be said about regressing other variables, I think you did right by keeping it simple and straightforward. You’re pointing out the obvious and your work gives rise to questions that, hopefully, others will begin to address and answer.

    Link to this
  4. 4. Cogitari 10:17 am 09/2/2011

    Interesting, but the data set is too limited. According to the linearization in the first chart, the number of events goes negative prior to 1978, which obviously is physically impossible. I think that showing the data for a much longer period is necessary to really see if there is anything going on. Certainly it is possible that we have reached a threshold where highly damaging events are suddenly increasing, but this data set is not conclusive in that respect.

    Link to this
  5. 5. mike_midwest 10:22 am 09/2/2011

    If you toss out 2005 do you still get much of an upward trend?

    Link to this
  6. 6. geojellyroll 10:56 am 09/2/2011

    The author must have flunked statistics. His projection arises from negative costs. His trend implies that ‘money was made’ from weather disasters in the late 70′s and earlier.

    One could have a hundred tornadoes with no damage or 1 tornado with a billion dollars worth of damage. Trying to extrapolate ‘meaning’ from destruction is not a scientific undertaking.

    Link to this
  7. 7. cajeffo 12:57 pm 09/2/2011

    I also had the same concern about the meaning of this study – as population density increases, there is simply more dollar value in a given area.

    I took the data in chart 2 and removed 2005 (Katrina). There is still an increase in the trend, but the increase is reduced by a factor of 4, only increasing from 16 Billion in 1980 to 20 Billion in 2010, rather than increasing from 12 Billion to 33 Billion as shown in Figure 2.

    Link to this
  8. 8. jtdwyer 1:16 pm 09/2/2011

    Forecasting derivatives can be tricky – there may be significant ‘hidden’ or embedded contributing factors.

    In this case, it should be considered that the U.S. population has increased by 35% in the past 30 years.

    Also, there has been a significant migration of rural populations to large urban environments.

    If I recall correctly, 50% of the U.S. population now resides in urban environments within 50 miles of a seashore.

    These I think significant contributing factors are embedded within your simple [annual] “U.S. Billion Dollar Weather Events” metric.

    Link to this
  9. 9. dusheck 1:16 pm 09/2/2011

    Was there an overall increase in weather related damages (as documented by NOAA) in the United States between 1980 and 2010? Sure.

    But you haven’t established either a trend or a mechanism. We also don’t know what limitations there are to what kinds of events got included in this data set. Fires in the west were grouped as if they were one event or all caused by one thing, which is not true.

    If you go back to the NOAA data and graph deaths per million against time (that is the total number of deaths each year from weather related disasters divided by the total population), the overall trend is decidedly downward–even using Excel’s “regression analysis.” Does that prove that people are less and less likely to die in disasters or that global warming isn’t real? Of course not! I think statistics and analysis deserve to be taken a bit more seriously.
    Jennie Dusheck

    Link to this
  10. 10. jtdwyer 1:29 pm 09/2/2011

    That you restricted your data to the weather events that caused more than a billion dollars in damages, the increase in the frequency of such events is at least partially due to the increase in the cost of damages per event.

    If you tracked a fundamental indicator of weather events, such as the number of F3+ hurricanes, for example, they could be decreasing over the same period, while the number of billion dollar F3+ hurricane events could appear to be increasing a a result of embedded, independent, contributing socioeconomic factors.

    While some commentators have complained about your invalid use of statistics, I have to commend you on your clever, manipulative methods…

    Link to this
  11. 11. Christine Gorman 2:10 pm 09/2/2011

    Come on, guys. Talk about straw man arguments. Obviously these simple charts cannot confirm mechanisms–nor are they meant to. But whether you just moved to the coast and your house is destroyed or you just moved to the Oakland hills, or you lived there forever, you’re going to feel the pain.

    You can bet insurance companies are paying attention to these data.

    Link to this
  12. 12. Christine Gorman 2:18 pm 09/2/2011

    In reply to Jennie Dusheck. I’d argue that the decreasing death rate is real and a result of better forecasting methods and evacuation planning. As I wrote above, I am asking for the underlying data set. This is just the first round of querying the data.

    Link to this
  13. 13. jrennie 2:25 pm 09/2/2011

    In reply to #6 geojellyroll: Her. “Her projection,” “Her trend,” etc.

    Link to this
  14. 14. sceptoptimist 3:28 pm 09/2/2011

    Why is the NOAA data normalized to 2007? If they normalized to, say, 2005, then all of the pre-2005 events would still meet the $1B threshold, but 11 of the post 2005 events would fall out, would they not?

    Link to this
  15. 15. Christine Gorman 3:49 pm 09/2/2011

    The question of which year to use for normalization and why (statistically speaking) is one I plan to ask the NCDC metereologist, when I speak with him. Will keep you posted.

    Link to this
  16. 16. jtdwyer 4:22 pm 09/2/2011

    Christine Gorman – “Come on, guys…”!

    Thanks for you considered response to very real issues with the presentation – I’m sorry I wasted my time!

    I can reliably predict that the number of billion dollar weather events will continue to rise.

    If you want to produce some really impressive charts, why not plot the number of billion dollar weather events since 1900? That would be a ridiculous misrepresentation of data, wouldn’t it?

    Link to this
  17. 17. R.Blakely 12:29 am 09/8/2011

    Hurricanes are similar to tornadoes. Both use heat that becomes trapped at the surface of the Earth. The energy of rising air and moisture transforms into wind energy, which causes the damage.
    Water vapor rises like helium in air since water molecules have about half the mass of air molecules. Clouds form as moisture rises and concentrates. This incredible effect controls Earth’s temperature. Clouds float in air because they have more water vapor than surrounding air, and so they act like blimps, which reflect sunlight into space.
    Earth’s climate tends to be self-regulating because it reacts to oppose change. For example, cooler weather causes fewer clouds, which reflect less sunlight. Obviously, global cooling causes more hurricanes.

    Link to this
  18. 18. Karlin 6:31 pm 10/13/2011

    This is a great start though, despite the comments being mostly nitpickingly negative [forget pre-1970!!]

    $35 billion in the USA in 2011, so far, tells the tale.

    Globally in 2011 the annual storm damage of $100 Billion could be accurate… and watch it climb until 2020 when things get really bad.

    The point is that whatever economic benefits the oil industry produces will be more than erased by severe weather due to climate change/global warming.

    Link to this

Add a Comment
You must sign in or register as a member to submit a comment.

More from Scientific American

Email this Article