How Fantasy Becomes Fact

UPDATE 9/21: Kudos to Bryan Walsh of Time who has updated and corrected his article.  That is the good news. The bad news is that a Google search of "boulder 1000-year flood" gives 3,190,000 results. 

UPDATE 9/20: Aslak Grinsted has a nice post up which runs rainfall and flood return periods for Boulder, showing similar information to that reported in this post.

The flooding in Colorado has wreaked tremendous devastation, by one estimate the costs will total about $2 billion. Some people remain unaccounted for and it will be many months if not longer before the Front Range recovers. If you'd like to contribute to the recovery, please see this page with resources.

As is often the case in the aftermath of extreme events and disasters, people look for some way to put them into a bigger perspective. With respect to floods, a common way of establishing this perspective is through the N-year flood, which is defined as a flood with 1/N probability of occurring in any given year. So the 100-year flood, used in floodplain regulations, is a flood with an expected 1% chance of occurring in any given year.

Earlier this week, I presented some of my objections to the utility and meaningfulness of the concept of the N-year flood. In this post I show how the concept of the N-year flood can be used to turn fantasy into fact.

In an article titled "The Science Behind Colorado' Thousand-Year Flood" Time magazine explains:

Parts of Boulder are experiencing a 1-in-1,000 year flood. That doesn’t literally mean that the kind of rainfall seen over the past week only occurs once in a millennium. Rather, it means that a flood of this magnitude only has a 0.1% chance of happening in a given year.

Time is a fixture of the mainstream media and what is written there is widely read and repeated.

A big problem with Time's article is that Boulder did not actually experience a "1,000-year flood." In fact, according to an analysis presented by fellow CU faculty member John Pitlick yesterday, using standard hydrological methods, Boulder experienced between a 25- and 50-year flood. (I am focusing here on Boulder, I have not seen similar analyses for other Colorado streamflows, though they are sure to come.) Pitlick further noted that the flood waters did not reach the 50-year flood marker on the Gilbert White memorial (seen at the top of this post.)

Pitlick walks us through maps of estimated inundation for 25-yr, 50 & 100 flood. Observations over recent days look more like 25-yr maps.
— Laura Snider (@lauracsnider) September 18, 2013

How is it that the "1000-year flood" has come to characterize the flood in Boulder? Let's take a quick look.

The Time article points us to an article on the floods at Climate Central, a non-profit group focused on reporting all things climate change. That article made the following claim:

The Boulder, Colo., area is reeling after being inundated by record rainfall, with more than half a year’s worth of rain falling over the past three days. During those three days, 24-hour rainfall totals of between 8 and 10 inches across much of the Boulder area were enough to qualify this storm as a 1 in 1,000 year event, meaning that it has a 0.1 percent chance of occurring in a given year.

So right away we see an error. Climate Central was discussing rainfall which Time mistakenly converted into floods. They are not the same thing. As John Pitlick explained yesterday, return periods for rainfall and flooding for the same event can vary by several orders of magnitude.

There is a further problem. The Climate Central article points us to NOAA's point precipitation frequency analyses, where the 1000-year storm claim originated.

What readers are not told is that in 2007 NOAA was considering discontinuing its presentation of 500-year and 1000-year precipitation return periods, because of their massive uncertainties. At the time NOAA solicited feedback from the expert community and received 122 responses, which you can see here in PDF (ultimately deciding to keep them, despite their problems).

One commenter to NOAA explained:

I pretty much feel 1000 year estimates are in the realm of fantasy.

Fantasy or not, anyone can look up online flood frequency analyses of questionable value, plug them into a story about climate change with no discussion of uncertainties. Those numbers can then be transformed by a larger and more widely read media outlet from referring to rainfall to referring to a flood. That information can then be shared, tweeted, repeated around widely as a "fact" by the mainstream media and presto ... the 1000-year flood is created.