100-year flood?

As illustrated by the above graphics, there was yet another round of heavy rain and flash flooding a few days ago in the central U.S., this time in southern Michigan and in the Kansas City metro area.

Fortunately, the deluges were localized and short-lived enough to not raise river levels significantly higher; although the weather pattern is such that there could be additional clusters of heavy thunderstorms at times in the Midwest, hopefully the worst of the large-scale flooding, which started really getting out of hand exactly a month ago (June 6-8), is mercifully over.

Just how epic was it? A 100-year flood? A 500-year flood? A gazillion-year flood? And how can, for example, "100-year" floods happen again and again within a much shorter time frame?

Here are the opening sentences from this recent Associated Press article on the Midwest flooding:

"Fifteen years ago, after the Midwest was swamped with what was pronounced a "100-year" or even a "500-year" flood, some folks figured they would never again see such a disaster in their lifetime. Some even dropped their flood insurance. Big mistake."

Why is such terminology so misunderstood? There are all kinds of problems and complications with it.

First of all, there's the overall total magnitude of a flood in a region (worst flood in X years) vs. the "recurrence interval" in a particular location, such as at one's home or business.

Misconceptions are furthered by the way things are portrayed in the media. "Unprecedented Midwest Flooding" blared one headline. "Flood of 2008 to be worse than Flood of '93" announced another. Which flood was actually worse? It depends. In some places the water was much higher in 2008; others such as the St. Louis metro area on the Mississippi River fared much better than 15 years ago, and overall the Great Flood of '93 was larger in scope than the 2008 flood. Locations such as Iowa City got hit hard both times.

Also, individual major weather events can occur very irregularly. For example, the area around Stuart and Port St. Lucie, Florida, had gone a long time since getting hit by the core of a strong hurricane making landfall from the Atlantic Ocean, yet within three weeks in 2004 the center of two such hurricanes (Frances and Jeanne) came in at nearly the exact same spot.

Then there's the length of those sorts of time periods (100 years, 500 years) vs. the length of the historical record for which good information is available. To confidently assess how frequently on average a very rare event occurs you'd need thousands of years of accurate data, which in this case doesn't exist. Absent that long period of record, values can be extrapolated from recurrence intervals of less-rare events, but that's a dicey proposition.

The amount by which the previous record crest of 20' in 1851 (and tied in 1929) was smashed in 2008 in Cedar Rapids would give the impression that 31+ feet must have an average recurrence interval of at least 150 years and probably much more. But we don't know when the last time that level was reached, nor do we know when it'll be reached again.

In Iowa City, given that the water level during the flood of 1993 substantially exceeded that of 1851, could a valid conclusion be drawn that it'd be at least 142 years before another flood of that magnitude? Look at the height of the crest on June 15, 2008 for the answer. But again, we don't know when that'll be equaled or topped! Will it be in 500 years or 100 years or 10 years or 1 year? (Let's hope not again in 2008!)

Add on top of all that: changes in climate, by whatever cause(s), can in turn change the frequency of weather extremes. (Maybe one reason why the extrapolations don't seem to be working very well lately?)

Plenty of material providing information about 100-year etc. floods has been readily available on the web, such as this and this from USGS, this from FEMA, and even this Wikipedia entry.

The upshot: a "100-year" flood means that based on available data (which might flawed), there's a 1-in-100 (1%) chance of a given location/elevation in a particular watershed being flooded in any given year -- not that a flood will necessarily occur exactly once every 100 years or on any other regular timetable.