What is the logic behind using the 24-Hour Hydrograph? 5

[CivilMan1980]

What is the logic behind using the 24-Hour Hydrograph in Watershed modelling and Design? Why not a 12, 36, 48, or 72 Hour Hydrograph? Anyone got any comments? All are welcomed.

Thanks

 

[DMcGrath]

I guess the quick answer would be "because that's what we have data/models/etc. in place for." However, if you think about it from region to region (in the US), 24 hours may just be a convenient compromise.

From my college days in Seattle, I saw it rain for seemingly days on end, with low accumulation (< 1"). Conversely in the Southwest and Southeast US, brief, intense thunderstorms can drop inches of rain in less than an hour and then be done.

I do remember reading in the "preamble" of TR-55 that the peak-hour intensity for the xx-th year storm is contained in the 24-hour event that the rainfall tables cover. We had to explain this once to a local government reviewer that the 24-hour, 25-year storm we used for design exceeded the 10-year, 2-hour event he was looking for. Lots of charts and explaining later I think we got our point across.

Dan

 

[francesca]

I work with someone who was City Engineer in the town of 100,000+ people for around two decades. He was involved in setting the standard of storm volume calculations being conducted using a 24-hour storm. He said that, while several factors were used in the decision, ultimately, the selection was arbitrary: You have to use something, why not the 24-hour storm?

Really, on the scale at which we work - i.e. developments within a town, and even on the streams in the town which all, bar the Cumberland River, have a time of concentration much shorter than 24 hours. Except in the case of zero-outflow stormwater features (e.g. sink holes) or stormwater controls that attenuate flow for more than 24 hours (e.g. NPDES Phase II wet ponds*), there isn't much point in analyzing longer storm events as the (flow) peak will happen in the first 24 hours and will be lower than a 24-hour storm event.

We only use the 100-year, 24-hour storm to size volume-sized stormwater controls, such as detention/retention basins. For flow-rate sized stormwater controls (culverts), we use the 10-year, 5-minute (or time-of-concentration length**) storm event.

* Note that NPDES Ph II wet ponds are not intended to detain the 100-year storm, but instead to have it pass over the spillway, so even this example might be moot.

** If you use a storm event shorter than the time of concentration, you're missing the effect that some flows will have passed through the system before the peak flow arrives from the far reach of the basin. In order to have peak flows arriving from all sections of the basin, it must continue to rain for the entire time of concentration.

 

[psmart]

The SCS/NRCS developed a number of synthetic rainfall distributions, such as the Type I, IA, II, III, etc. The distributions usually have an overall duration of 24-hours, but have the benefit of including intensity information for all events up to 24-hours, all in a single curve. So the 24-hour distribution INCLUDES the 1-hour, 2-hour, 12-hour, etc. rainfalls, all in a single table.

Roughly speaking, a 24-hour rainfall is suitable for any watershed that will fully "respond" within a 24-hour time period. This typically means that the overall time-of-concentration and travel times are somewhat less then 24-hours. A longer rainfall distribution (such as 48-hours) is required only for larger watersheds (but larger areas may require a different runoff methodology anyway.)

Although many sites could use a shorter version of the SCS/NRCS rainfalls (such as 12-hours) this complicates the analysis without any real benefit. Of course, there are exceptions, such as a shorter duration "water quality" event, but these are normally used in addition to a traditional 24-hour study, which is necessary for volume-sensitive calculations, such as detention ponds.

Finally, by standardizing on a 24-hour duration we are able to use a consistent measure of rainfall depth for any situation.

For more background on the SCS/NRCS method, and the effect of different rainfalls, see the "Hydrology" section of the hydrology training slides at

http://www.hydrocad.net/training/

 

[RWF7437]

My understanding is that the 24 hour storm was originally selected for the simple reason that more 24 hour rainfall data was available than for any other duration. There is nothing magic about it. It is arbitray, although it may be reasonable, just as a red light meaning "stop" is arbitrary but reasonable.

In my own practice, I use the 24 hour storm as a starting point for an analysis. For a large, expensive, system it makes sense to me to adopt an eclectic approach, i.e to try as many approaches as I can think of. Hydrology is more art than science. Even with the best of data and most powerful computers it is more like weather forecasting than it is like designing a structure for "known" loads.

Many enlightened storm water agencies are developing design storms of their own to replace the four so-called standard NRCS storm types.

For some basins, there may be a critical storm duration for design. It is possible, with modern computers, to examine many storm types, durations and rainfall patterns. This kind of approach makes sense to some of us, given that most estimates of peak flow, volume and other factors are no more accurate than plus or minus 30 percent, OR MORE!

Finally, any computer model is no better than the effort put into calibrating it with real world data.


good luck

 

[RWF7437]

And by the way:

We often speak of the selected return period for design as "the 50 year flood" or "the 100 year flood." Many who hear this understand it to mean that this is a flood expected to occur once every 50 or 100 years. By speaking this way we imply a level of protection which is seldom achieved. For example, if we design for a 100 year storm event many people would believe that would mean the system should function for 100 years without ever being over capacity.

Unfortunately, this is not true. Because flood probabilities are based on historical records, their accuracy depends on the length and completeness of those records. Many reporting stations have only a few years of record so that the probabilities calculated from them are less reliable than stations with a record of 40, 50 or more years. As this is being written NOAA is updating their precipitation frequency records adding an additional 30 years of record to the data. This should improve the accuracy of the published data and make our estimates more reliable. Even so, they will remain "estimates" only. That is, they are simplified ways of stating the probability of an unpredictable event occurring.

A possibly better way of viewing these events, and speaking about them, is to refer to them by their annual probability of occurrence. The so called 100 Year storm is, by definition, the storm which has a 1 percent probability of occurring in any one year. If we want to know how frequently such a storm might occur over a longer period of time, that probability can be calculated by:

Px = 1 - ( 1-1/N)^x

Where: Px is the probability of occurrence in x number of years

1/N= the Probability of Occurrence in any one year

For example, if we want to calculate the probability of occurrence of the 100 year flood over 100 years the calculations would be:

Px = 1 - ( 1-1/100)100

Px = 1 - ( 1- .01) 100

Px = 1 - ( 0.99)100

Px = 1- 0.366

Px = 0.634

In other words, there is a 63 percent probability that the 1 Percent storm will occur one or more times over the next 100 years.

This kind of calculation can be done for any selected range of frequencies and time periods. This has been done in Figure 7-3.

Figure 7-3 Probability of the N Year Event Occurring in x Years

C. How Much Insurance Can You Afford to Buy ?

You may find it helpful to think of storm water management measures as insurance. The question for the designer then becomes, "how much protection do I need and how much can I afford ?" If your project is to design protection for the regional hospital in your area you’d probably want a very large "insurance policy" to protect this vital community resource. But, if the only thing downstream from you detention basin is a large community park a much lower level of protection can be justified.

 

[cvg]

The practice here is to analyze using both the 6-hour and 24-hour and generally pick the most extreme event. For larger reservoirs (flood control dams), we have also used a 10-day event which might be worse than the 24-hour.

 

[queque]

In a watersehd modeling effort, you began by knowing the amount and time distribtuion of the rainfall. Then you relate that data to the runoff measured at some know point within the outlet of the watershed.
All you need is the rainfall data and runoff data and then the analyses will show that each event is different. There is no such thing as a 24 hour storm with a presrcibed distribution yiielding a prescribed runoff ditributuion. Ask any USGS Hydrologist about data collection, interpretation, correlation and statistical analyses of the data. He will usually reply that you need several more hundred years of data to prove that there is no correlation between any particular rainfall event, resulting runoff and another similar storm event. Computers simply calculate numbers, not reproduce known events.

 

[psmart]

Due to the unpredictability of actual rainfall events, design work will normally use on the standardized synthetic rainfall distributions (Type I, IA, II, etc).

However, if detailed rainfall data (intensity or depth vs time) is collected for an actual event, this can be used instead of the synthetic storm. Comparing the results to the observed flows will tell you how accurate the model really is. Or conversely, you can use the comparison to calibrate the model for more accurate results.