They are generally agreed to be unknown.  Many hydrology methods accept an extrapolated estimation (sometimes as simple as a linear multiple) of the 100-year.  I don't know if this is valid w/ TR-20 and TR-55, though.

Thanks Iha

I ended up using a USGS formula that estimates stream discharge based on location and area.  I then altered the rainfall depth to approximate the discharge for the basin in question.

Why do you need a 500-year depth?  If you are dealing with regulations that cite a 500-year event, they should also give you a source for the rainfall data.  Frankly, there isn't enough observed rainfall data to determine the 500-year event.

psmart,

Are you sure there is not enough observed data to determine the 500-year storm event.

Remember, an X-year event does NOT mean the event only happens every X years.  It means there is a 1/X chance in ANY given year to be equalled or exceeded.

I am not a statistics and probability wiz, but I would imagine one does not have to have 500 years of records to estimate a probability.

True, there is only a 1/500 chance of a 500-year event occuring in any given year, but 50+ years of good rainfall data seems unlikely to predict such a rare event, if only because of changes in weather patterns.  In any case, the rainfall amount is primarily a matter of public policy, which was the reason for my original reply: Why do you need a 500-year storm?

psmart (Civil) Aug 15, 2004
Why do you need a 500-year depth?  If you are dealing with regulations that cite a 500-year event, they should also give you a source for the rainfall data.

I thought the same thing but that is what my boss wanted.

Your should (respectfully) tell your boss that there is no accepted data for the 500-year event.  Although it might be possible to estimate a 500-year depth, that may not really address your needs.

For example, if you are trying to study an extreme rainfall event, such as rainfall on saturated or frozen ground, this can be accomodated by adjusting the CN values or Antecedent Moisture Condition, along with the 100-year event.  With certain models, it is also possible to model back-to-back storms.  There are several techniques that may be appropriate to your needs.  It all depends on what you're trying to model.

I ran across this problem last year.  The issue at hand was disproving that the Zone B* as depicted was an area of 100-year flooding subject to shallow flooding, and the reason it had to be disproven was that a local bylaw prevented the inclusion of land subject to 100-year flooding from the unit density calculations for multi-family housing.  You can predict the probable maximum event by plotting the other know rainfall amounts for the area on log paper: Rainfall in inches on the Y axis, Return Frequency in years on the X axis 0 log. You can look up the 6-month, 1-, 2-, 5-, 10-, 25-, 50-, and 100-year rainfalls in TP-40.  Then connect the dots with a best-fit line and extend out to 500-year.

* FEMA defines Zone B as EITHER an area subject to 500-year flooding OR and area subject to 100-year shallow flooding (one foot or less).

In the City of Tucson methodology for computing a 100 peak event, they include basin factors. slope, imperviouness, the 100 year rainfall depth and other factors to calculate the 100 year peak. Then they multiply that value by .1 (2 year), .37(10 year) and .77 (50 year) to get the intermediate peak magnitudes.
It is assumed that since you know the rainfall, you will be able to estimate the peaks. The fact ids you do not know the rainfall for any recurrence inteval at any location at any time.
Data does not exist anywhere in Arizona to validate that data. Worse than that, you do not know the runoff amount generated from any amount of rainfall at any intensity. The entire procedure and data is an estimate. No one knows the magnitude of error, plus or minus. This translates to over or under sizing of storm drains, detention dams etc.
This is where we are in Hydrology and where we have been the last 50 years. Until we instrument the watersheds with extensive data collection programs, we will never know the rainfall nor rainfall runoff relations.

I apologize for my incorrect response previously.  Not only does the NOAA have this data, but it is free, readily available AND goes up to the 1,000-year event!  This is news to me.

According to page 9-535 of the Pennsylvania Stormwater
Best Management Practices
Manual
DRAFT - JANUARY 2005

9.5 Precipitation Data for Stormwater Calculations
In 2004 the National Weather Service’s Hydrometeorological Design Studies Center published updated
precipitation estimates for much of the United States, including Pennsylvania. NOAA Atlas 14
supercedes previous precipitation estimates such as Technical Memorandum NWS Hydro 35 and
Technical Papers 40 and 49 (TP-40 and TP-49) because the updates are based on more recent and
expanded data, current statistical techniques, and enhanced spatial interpolation and mapping
procedures. (Bonnin et al., 2003 and NWS, 2004) The “Precipitation-Frequency Atlas of the United
States,” NOAA Atlas 14, provides estimates of 2-year through 1000-year storm events for durations
ranging from 5 minutes to 60 days as shown for Harrisburg in Table 9-2 (available online at http://
hdsc.nws.noaa.gov/hdsc/pfds/). Users can select precipitation estimates for Pennsylvania from over
300 observation sites, by entering latitude/longitude coordinates, or by clicking on an interactive map
on the Precipitation Frequency Data Server. These new rainfall estimates should be utilized for all
applicable stormwater calculations.

Remember: The Chinese ideogram for “crisis” is comprised of the characters for “danger” and “opportunity.”
-Steve