Calibrating the model in HEC-HMS

[DPAJR]

I have a question about ‘calibrating the model’ in HEC-HMS.

I have pretty strong evidence that a storm of 1.8 inches produced no runoff in a desert gully in Eastern Washington. (State and county highway departments plus personal observation.)

The 100 year storm is 3.2 inches.

I believe the way to calibrate is to calculate the 1.8 inch storm and the 3.2 inch storm and subtract the runoff values for the 1.8 from the runoff values for the 3.2.

The alternative would be to calculate a 1.4 inch storm (3.2 inch-1.8 inch) I believe this approach ignores the fact that the soil would have been thoroughly soaked and runoff would be fully developed before the last 1.4 inches falls.

Is there any other way to calibrate a model?

Just trying to head off a potential disagreement.

Thanks

 

[beej67]

Initial abstraction?

Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com

 

[SMIAH]

Ia.

Calibrate the model parameters (whatever you use) to calculate a runoff of 0 when imputing a hyetograph of 1.8 inches.

Then use the same model to estimate the runoff volume for a 3.2 inches storm.

Shame that you don't have more data to calibrate it though (e.g. a 2.0 inches storm).

 

[DPAJR]

Hi SMAIH,

Your explanation of calibration looks a lot more valid that what I had proposed. I'll do that next.

I actually have over 50 years of storm data. This is just the largest storm during that entire period and it produced no runoff except for the uppermost of five culverts which got some sand and gravel in the upper end of a small culvert.

I believe the NOAA 100 year storm number is not valid but FEMA is not buying that argument because the 52 years of data is daily instead of hourly or at 15 minute intervals. The feel that the 1.8 inches (and other maximum storms for other years) could have been part of a larger storm that spanned midnight. Valid point.

However, another rain gage which only has 17 years of data available in 15 minute or 60 minute intervals did show that the entire rainfall for that day was over by 22:00. If only the second rain gage had more data.

Thanks for your help.

beej67,

Do you mean simply boost the intial abstraction as a means of calibration rather than adjusting curve numbers etc? That would be a lot simpler alright. I'll mess with both methods.

Thanks

 

[cvg]

I have to agree with FEMA on the validity of the 100-year storm rainfall. 3.2 inches is quite low. From my experience in Mohave and Sonorran Desert, I have rarely seen lower. I would have expected a bit more in Washington.

for instance 24 hour rainfall in some of the dryest deserts I have seen are:
Coyote Wells, California - 3.52 inches
Death Vallley - 2.39
Gila Bend Arizona - 3.19
Twentynine Palms, California - 4.47
Needles, California - 3.47

You might check to see if there is more storage in your basin you are not accounting for. If so, you can route through the storage basin and reduce your flow that way.

 

[beej67]

Yes, that's what I meant. Use CNs out of the book, and fiddle with Ia until your model matches your historical data like SMIAH suggests.

Whatever calibration procedure you use, be sure to document it, and it wouldn't hurt to get written buy-in from your reviewing agency. Always a good idea whenever you start turning the "DO NOT TURN" knobs in the model.



Hydrology, Drainage Analysis, Flood Studies, and Complex Stormwater Litigation for Atlanta and the South East -

http://www.campbellcivil.com