PMF using frequency analysis 2

[SMIAH]

The PMF is often calculated as the flood generated by the most severe precipitation possible at a site at a particular time of year, referred to as the PMP.

I would like to know if someone has ever considered using a standard frequency analysis to calculate the PMF even it the periods of record available are too short to extrapolate for such a low probability.

 

[Drew08]

I've found that the 500-year rainfall depth is generally about equivalent to the 1/4-PMP (as shown in HMR-51) + or - an inch. The accuracy of the estimate is of course only as accurate as the data to generate the estimate...and as accurate as HMR-51 which is missing 33 years of data since its publication (1978).

I haven't used this in practice though or tried to pass it by a reviewer in place of HMR-51. It is just an observation, and by no means universal.

 

[SMIAH]

Thanks,interesting.

I've found a publication (

http://pubs.cwra.org/doi/pdf/10.4296/cwrj2301001)

:

"Since PMFs are presumed to fall in the 10-4 to 10-6 per year
frequency range, stream flow records are of little value for this purpose."

 

[Drew08]

Interesting concept, but I disagree with the author that the full PMP and PMF have a finite probability of occurrence (however small). The author didn’t just “assume” this, but rather “presumed” a probability of occurrence. That doesn’t strike a lot of confidence in me.

If the PMP/PMF has a probability of occurrence, say 1:10^-5, then there is some larger event that can occur with an even smaller probability of occurrence, 1:10^-7. No larger event is physically possible (in the current climate, global warming aside) because the PMP and PMF are physically limited events; they are the occurrence of the maximum amount of water that can saturate the atmosphere and soil released all at once. An accurate chart showing rainfall vs. return period would be asymptotic to the PMP, and stream flow would be asymptotic to the PMF.

I think the probability “presumed” by the author of 1:10^-4 to 1:10^-6 is not the probability of occurrence, but rather the probability of the estimate being right. The one point the author made that I do agree with is, because of this uncertainly, the PMP or PMF are not good indicators of risk.

The term probable maximum flood is a little miss leading, because no matter how it’s defined, it is very improbable.

 

[SMIAH]

The project I am referring to in this thread has a Watershed area of more than 400 000 sq2...

The PMF has to be calculated for an auxiliary dam located on a site where the water level are recorded since over a period of 100 year.

You probably see where I'd like to go with this... Which is basically to avoid the determinist, lumped-parameter model.

Wondering if it's possible with the concept of the PMF.

 

[cvg]

site specific PMP/PMF analysis would be highly recommended over the ordinary HMR49 or HMR51 procedure.

 

[Drew08]

For a first guess, find the peak flow on record for several gages in the climatic region and with an extensive period of record, plot the flows on a Log(flow) vs. Log(area) graph, and choose a representative straight line that just “sits” atop the two most extreme flow rates for their watershed areas. The real PMF for your site would plot somewhere above that line.

A similar method was performed by NOAA for rainfall that aligns somewhat to PMP rainfall depths in HMR-51. Surprisingly, the greatest rainfall depths on record are scattered across the eastern US and Hawaii, not centralized in a coastal hurricane zone as you would expect.

http://www.nws.noaa.gov/oh/hdsc/record_precip/record_precip.html

Side note: Dam on a 400,000 sq.mi. watershed!? (Texas x 2).

 

[SMIAH]

Interesting as NOAA performed it.

Actually it is a (small) auxiliary-dam located on a lake and most of the flow is not passing through this dam. It has however no effect on the water level of the lake and there is a maximum discharge that can physically pass through this dam (i.e. downstream canal).

I think that routing the PMF through the lake, taking account of both the natural discharge at the lake outlet and at the auxiliary dam, would be a "very long process".

A "maximum probable water level" with the 100 years water levels records... ?

Anything to avoid the PMP/PMF on a 400,000 sq.mi watershed.

 

[thatdamengineer]

My experience in Oklahoma indicated that 25% PMF was about equal to the 100yr flood by TP40. This ranged from 20% in the eastern part of the state to 30% in the west. Therefore we used 25% as the 100 yr flood. Other floods were described as % of the PMF

The PMF versus frequency is going to be very site specific. As an earlier responder stated the HPR51 is over 30 years old. It was printed the year we started the inventory of dams in the US

I think I still have my draft copy somewhere in the files.

WWW.thatdamengineer.com