About That "Thousand Year" Storm in Colorado....
Last week I expressed my doubts that the storm in Colorado was really, as described breathlessly at the Weather Underground, a once in a thousand year storm (the logic of the article, and many others, being that one in a thousand is the same as "zero" and thus the storm could not have occurred naturally and therefore Global Warming).
Turns out it is not even close. From the Colorado Climate Center at Colorado State University:
How much rain fell on Colorado this week? And where? Colorado residents can help the weather experts at Colorado State University answer these questions.
In response to the incredible recent rains and flooding in parts of the state, the Colorado Climate Center will be mapping rainfall totals and graphing hourly intensities for the entire state for the period beginning Sunday, Sept. 8 (as storms first developed over southern Colorado) through the end of the storm later this weekend
"As is typical of Colorado storms, some parts of the state were hard hit and others were untouched. Still, this storm is ranking in the top ten extreme flooding events since Colorado statehood," said Nolan Doesken, State Climatologist at CSU. "It isn't yet as extreme or widespread as the June 1965 floods or as dramatic as the 1935 floods but it ranks right up there among some of the worst.”
Among the worst, according to Climate Center data, occurred in May 1904, October 1911, June 1921, May 1935, September 1938, May 1955, June 1965, May 1969, October 1970, July 1976, July 1981, and, of course, the Spring Creek Flood of July 1997 that ravaged Fort Collins and the CSU campus.
."Every flood event in Colorado has its own unique characteristics," said Doesken. "But the topography of the Colorado Front Range makes this area particularly vulnerable when the necessary meteorological conditions come together as they did this week."
So it is perhaps a one in fifteen year flood. Note that (by the math in my previous article linked above) a one in fifteen year flood covering an area half the size of Colorado should occur on overage over 60+ times a year around the world. Our intuition about tail of the distribution event frequency is not very good, which is just another reason they make a poor proxy for drawing conclusions about trends in the mean of some phenomenon.
Matthew Slyfield:
" Our intuition about tail of the distribution event frequency is not
very good, which is just another reason they make a poor proxy for
drawing conclusions about trends in the mean of some phenomenon."
Personally I think the problem here is very specifically judging the frequency of tail of the distribution for events distributed in two dimensions, here being space and time.
September 17, 2013, 9:07 amNehemiah:
If an event, tail or otherwise, fits the template, it is used. If enough people repeat it, it becomes conventional thinking, then consensus and finally fact.
September 17, 2013, 10:22 amrandom_eddie:
"Note that (by the math in my previous article linked above) a one in fifteen year flood covering an area half the size of Colorado should occur on overage over 60+ times a year around the world."
Note that your math is wrong, as these are not independent events.
September 17, 2013, 1:19 pmmarque2:
Yeah, I read that article as well. It might be a top 10 storm for the last 150 years in Colorado, which is a far cry from 1000 year storm. And 5 - 7% was caused by AGW per the NOAA - than goodness for Global Warming or the storm would be nothing to write about at all.
September 17, 2013, 7:56 pmmarque2:
No events are truly independent, just reasonable so. His math may be off a bit, but the further away you are from a weather event by distance and time the less dependence there is on that event.
So the math isn't perfect but it is more reasonable than you are leading us to believe.
September 17, 2013, 7:58 pmFelineCannonball:
A 100 or 1000 year flood has a strict technical definition and refers to a particular point on a stream, not to an entire state. A 1% or 0.1% probability of an event occurring at a gauging station within a given year. There are problems with short records, statistical framework, and climate oscillations or drift. There is no problem with the same stream having such events in back to back years. It's probability. And other drainages no matter how close are irrelevant.
The number of such events in a larger geographic area (the state of Colorado for instance) is a function of the granularity of storms, drainages, and measurement stations. 100 year floods should happen in Colorado much more frequently than 100 years.
September 18, 2013, 7:04 pm