Monday, June 16, 2008

About that "500-year flood" thing

As the waters rise and recede in the midwest, the blogosphere, the media, and the mass mind are once again running away with the idea that a "500 year flood" is Just Not Normal, and so this must be yet another sign of the coming apocalypse. Okay, let's look at the plain statistical facts:

A "500-year flood" is a flood level that, on average, hydrologists and climatologists have estimated has a 0.2% chance of occurring at one single point on one single river in one year. Given that none of these rivers actually have anything remotely close to 500 years of historical data, these 500-year flood levels are obviously based on extrapolations and geology, not observed flood recurrence intervals. But, let's skip past that and assume that the 500-year flood level estimates are generally accurate.

The 500-year flood should only happen on average every 500 years at any single point on any single river. But, the continental US has hundreds of large river basins and tens of thousands of smaller drainages. Given this, it is indeed "normal" to see a 500-year flood on a major river, somewhere in the US, just about every year on average. Given that week-to-week and seasonal weather patterns are not randomly dispersed in space and time, but tend to settle into regional patterns over those time scales, it is in fact expected to see an outbreak of "500-year floods" on several large rivers within one region, somewhere in the continent, every few years. With the thousands upon thousands of small stream drainages, localized 500-year floods are expected to happen somewhere damn near all the time! It's just your one single house, if you live in a 500-year floodplain, that should only expect to get hit once every 500 years. In the nation as a whole, somebody's house will be getting whacked by one just about every week.

Case in point:

We lived most of the 90's in Fort Collins, Colorado. In 1997, a "freak" upslope thunderstorm dropped as much as 14" of rain on the foothills west of town. The result? A 100-year flood that killed 5 people and did massive damage to Colorado State University and thousands of houses and businesses. Interestingly enough, in 1976, a virtually identical "freak" flood had happened just 20 miles away, on the Big Thompson Canyon, killing 145 people and prompting the installation of the now-iconic "IN CASE OF FLASH FLOOD ABANDON CAR AND CLIMB TO SAFETY" road signs in foothill canyons throughout Colorado. So what's up with that? Two 100-year floods in the same county in just 21 years? Isn't that really a "freak of nature?" Of course not. Larimer County has dozens of foothill canyons where the Rockies rise abruptly from the Great Plains. So in fact having a localized 100-year flood strike one or two of these canyons somewhere in the county every decade or two is expected and normal, not freakish. Fortunately the emergency planners understand this, even if the collective mass mind doesn't quite get it.

Same thing applies to Katrina, by the way. Based on historical patterns (no climate change needed), several storms of that magnitude are fully expected to strike somewhere on the western and central Gulf Coast in any given Century. There is nothing "freak" about it.

Something I find interesting is that in general, the 500-year flood is not planned or prepared for either at the level of the individual household or the community as a whole. The lines are generally drawn at the 100-year floodplain. In contrast, in much of "tornado country," the recurrence interval for a strong, potentially lethal tornado at any single point (i.e. your house) is also on the order of 500 years. Yet, people do plan, prepare, and engineer for tornados. I'm not sure what the reasoning is behind this difference in approach.

Just remember the adage, truer now than ever before :

"There isn't more weather, just more weathermen."

Site Meter