Update: Since my original posting, there has now been another 1-in-1000 year rainfall event (the 4th in 2016), this time in Louisiana.
This year we've witnessed a number of record-breaking floods. So far, three floods in the US have been called 1-in-1000 year events. At first glance, it can seem astoundingly improbable that such extreme events could all occur in a single year. Yet, tens of thousands of people were affected by these floods and some lost everything they had. In the face of such tragedy, it's good to remind ourselves what a 1-in-1000 year event means in the context of the flood peril. What exactly defines a 1-in-1000 year flood? Is it the amount of rainfall? River gauge height? Flow rate? To help provide some context, let's take a look at the 1-in-1000 year events that have occurred this year.
A few months ago in mid-April, massive amounts of rain fell in the Houston area – northwest Harris County, as well as neighboring Fort Bend and Waller counties. Most of the rainfall occurred within a day, resulting in flash flooding that caused eight deaths and over 7000 damaged buildings.
In June, West Virginia had record amounts of rainfall, causing 23 deaths and destroying 1200 buildings. Total recovery may take years, and even now, residents are still displaced and businesses are still closed, leaving many people unemployed.
Most recently, on the evening of July 30, parts of Howard County, Maryland experienced extreme rainfall within a short period of time. The flooding damaged over 250 buildings, killed two people, and over 120 people had to be rescued. At the heart of the flash flood was Ellicott City where the most rainfall occurred and much of its downtown has been destroyed. In the aftermath of the storm, devastating pictures show Main Street covered with debris, cars piled up, and storefronts cleared out by the floodwaters. Efforts to clean up and rebuild have just started, and businesses are closed indefinitely.
Calling the event a "1-in-1000 year occurrence" helps quantify the significance and rarity of the event. Let's clarify a few points. First, a 1-in-1000 year event suggests a probability: such an event is expected to occur with 0.1% likelihood. Over a thousand years, you would expect to have one event, but there could be more than one, or even none. Second, a 1-in-1000 year event is usually specific to a given region. If we had a model for each of the (roughly) 3000 counties in the US, statistically we might expect to see three "1-in-1000" events each year. Finally, in these three cases, the 1-in-1000 return period refers to the amount of rainfall over a given duration, as opposed to a measure of damages which is usually how underwriters and insurers look at these events.
NOAA's Atlas 14 allows you to point to a specific location in the US and track down the precipitation frequency by duration. In the case of Ellicott City, the measured rainfall on July 30 was six inches within a two hour time period. The 1-in-1000 year rainfall for a two hour duration is 5.45 inches. Since the actual amount of rain surpassed this mark, the event could be called a 1-in-1000 year or greater rainfall event. The significance of this flood was the intensity of the rainfall. Because the six inches of rain in Maryland fell within such a short time period, the water was not able to drain quickly enough. Combined with the fact that the city is at the bottom of a hill, the resulting property damage was amplified.
The same method can be followed to determine the return period for the West Virginia flooding. In Lewisburg, Greenbrier County, over nine inches of rain fell within 24 hours. Atlas 14 suggests that for a 24-hour duration, the 1-in-1000 year rainfall is 7.2 inches. Rainelle, another city in Greenbrier that incurred a lot of damage, had 7.5 inches of rain in 24 hours, and the 1-in-1000-year rainfall for that location is 7.17 inches. The data for both these cities suggests the West Virginia rainfall was a 1-in-1000 year occurrence.
Unfortunately, NOAA is still preparing the Atlas 14 precipitation map for Texas, which is expected to be released in 2018. The flood in April has no single return period associated with it, because the flooding affected a vast area in differing degrees, and there were multiple components that contributed to the flood. First, there were record amounts of rainfall, over a foot in a day in some areas. The return period can vary significantly by location. According to Harris County's flood report, as it was measured over a 12-hour period, the rainfall in the eastern parts of the flooded regions were considered to occur with 1 in 50 year probability, while western regions experienced rainfall with a probability of 1 in 1000+ years. As an extreme example, the city of Pattison, TX had over 23 inches of rain that fell within 14.5 hours and can be considered a 1 in 10,000 year occurrence for that location. Second, the rainfall raised the elevation of several streams which overflowed and magnified the flooding. The flood report shows that certain river gauge elevations reached levels considered to happen with 1 in 500 year frequency. While the overarching 1-in-1000 year label was applied to the whole event, the reality is more nuanced.
In the beginning we entertained the likelihood of three significant floods each associated with the term "1 in 1000 year event" happening in one year. The chance of all three happening is not as rare as we might have thought – it is definitely not one in a billion (or 1 / 1000 cubed). The actual answer is not so simple, but we now know that a flood return period is tied to a specific occurrence (e.g. the rainfall intensity of 6 inches in 2 hours) in a given location (e.g. Ellicott City, MD) in a given year. The chance that you'll see some event of similar magnitude in any city or county in the US is a lot bigger, because there are numerous combinations of regions and possible rainfall intensities.
So now that we know these events are not as rare as one might think, ask yourself this: are you prepared?
Category: Climate/natural disasters: Floods/storms