On September 27, 2018 Version 2.0 of the National Oceanic and Atmospheric Administration’s Atlas 14, Volume 11 was released and a press conference was held. As I’ve described in previous posts, Volume 11 provides us with a more accurate picture of the annual probability of rainfall of various depths and durations across the state of Texas. These data were last updated across the entire state in 1961. Volume 11 was created by evaluating rainfall records as early as the 1870’s up through 2017.
As I written previously, the depth of 1% annual chance, 24-hour rainfall is a key design storm we use for drainage and flood risk reduction infrastructure in the Harris County area. Prior to Volume 11, this key design storm was about 12.4-inches in western Harris County and about 13.5-inches in southeastern Harris County, as shown below.
After Volume 11’s release the new 1% annual chance, 24-hour duration rainfall map for Harris County looks like this:
We used to believe that the 1% annual chance, 24-hour storm was 2.6-inches to 4.5-inches smaller than it actually is. This was because the rainfall records did not span as long a period as they do now. Our uncertainty bands have gotten a bit narrower and the overall magnitude of most storm frequencies and durations have increased. The actual depth of the 100-year, 24-hour duration rainfall ranges from 21% to 33% higher than we thought.
The updated information is presented by NOAA using a web-based map interface, located here: https://hdsc.nws.noaa.gov/hdsc/pfds/pfds_map_cont.html. It allows anyone to double click on the map at any point of interest and instantly get both a tabular and graphical display of the rainfall stats.
Here’s how it works for my neighborhood in west Houston.
This first image is from the map interface. I have zoomed into my part of town and have double-clicked on the map near my house. The red cross-hairs show the click point.
After double-clicking the following information appears below the map:
This shows that my house has a 1% annual chance of receiving 16.5 inches (or more) of rain in 24-hours. If you read the fine print you will see that there is a 90% chance that the actual rainfall depth for this 1% annual chance, 24-hour storm ranges will be somewhere between 11.6 inches and 23.5 inches, because of the uncertainty of our statistical estimate. The website also provides a graphic version of the same data:
The website also provides a graph that illustrates the confidence bands around the estimated rainfall depth. The figure below shows confidence range for all 24-hour storms. The green line shows the upper limit of the 24-hour rainfall depth and the red line shows the lower limit. There is 90% chance that the “true” rainfall depth will fall in between those values (for any given recurrence interval). Please recall that “recurrence interval” is the inverse of annual probability. So the 1,000-year storm has a 1/1,000% chance of occurring in any given year. If you convert the fraction to a decimal, that’s a 0.001% annual chance. Notice how that is not zero?
See how the confidence range grows for the larger, more rare events? The 100-year recurrence interval corresponds to the 1% annual chance event. The math tells us that there is 90% chance that the 1% annual chance, 24-hour event is somewhere between 11.6 inches and 23.5 inches. The corollary is there is 10% chance that the real depth is outside that range.
The expanding confidence interval – the gap between the green line and red line – makes sense because we have statistically evaluated rainfall records that only go back to, at best the 1870s, and, at worst 1970’s. This is a period of record of only between 47 and 148 years. So our certainty about 500-year or 1,000-year recurrence interval events is low.
In my next post I will be providing some thoughts about the cost impacts of this refined view of our rainfall statistics.