Modeling of larger (1700ac) watershed 2

[murdock]

I have been tasked with creating a model of a 1700 acre watershed of a brook. This model will be used to find at what rain event a diversion structure (basically a long weir along a portion of the streambank) will start to come online.

A previous consultant performed a model, but did not use any of the prefab CNs. All of them seem to be too low for the land use (e.g. 58 for woods in C soils, instead of 70). Also, the Tcs he used did not have any sheet flow, and the shallow concentrated flow lengths seem a bit long for what would could be plotted on a USGS map (my primary map to check against).

However, his numbers track fairly closely to the (limited) amount of data I have on streamflow (I have about a year's worth of rainfall, stage and discharge data slightly upstream of my study point).

Any attempt for me to create a model of this site results in very, very large flows down the stream (200-300 cfs for a 1-year event vs. 20-30 for the other model and data).

Does anyone have advice for a good way to model these larger watersheds? Here are the questions I'm batting around in my head:

Should I be breaking down the site into smaller subcats, or just use a single subcat?

Should I play with the other consultant's model, or just create a new one from scratch?

Are the generic landform CNs (1/2 acre lots, etc.) good enough at this scale, or do I really need to trace out all the impervious surfaces (possible with the aerial photos I have)?

Most importantly: How do I model a stream channel? Do you need to have the baseflow in it, even though storm flows will be significantly higher than the baseflow?

There are probably a few missing pieces of info I forgot that would help, let me know what I'm missing. Hope you fine folks can help...

 

[RWF7437]

A few suggestions:

You probably don't want to use the previous study. It appears you already have doubts about it.

Whatever method you choose to use, if at all possible, CALIBRATE your hydrology model. All of the available models will have errors of estimate of about 50 % OR MORE. Even a calibrated model will still only get you within plus or minus
30%.

Your choices will be limited by the time, money and importance of your project. As you know. Among those models available look at:
HydroCad
Hydraflow Hydrographs
National Flood Frequency (NFF)
HEC-RAS
HEC-HMS
WINTR55
EPA SWMM5
Local Regression Equations
Local streamflow data

and some others. All of these are either free or reasonably priced. Stay away from the high priced, over hyped models which can give you no better answers than these, given the probabalistic nature of hydrology.

Finally, ask yourself what are the consequences if my answers are no better than plus or minus 50% ? Calculate the probability that your calculated flow will be equaled or exceeded over the "life" of your project.

good luck

 

[queque]

Contact the Surface Water Specialist at the US Geological Survey office in your state Capital.
He will provide you with data of actual 100 year flood estimates and drainage areas.
Select the area within your study area, take the 100 year peak flow data provided and respective area,
divide the peak flow data by the given area and that is a estimate of the flow/unit area (ratio). Take your area, multiply it by the ratio, and you will a crude estimate of the 100-year peak flow.
Ask him for the regression equation developed for that area. Use that equation to get an estimate of the 100-year peak flow. That will give you a estimate of the error inherent on the use of limited data to calculate peak flows.

 

[DuckEngineer]

As far as the generic CN's for 1/2 acre lots, I had an engineer that buggied off all of the impervious areas in a 1/4 acre lot subdivision. The number was right on target with the NRCS assumed percentages of impervious.

 

[murdock]

Thanks for all your replies. Fortunately, I believe I found my answer without tackling this modeling question (I have found streamflow/stage data).

The target storm event I'm looking for is under 1 inch, so I know that TR55 stuff will not give me a reasonable number (double-checked against the data I have). RWF, do you have any links to information that talks about the error in the storm models? I didn't realize that there was so much variability with these things. I feel a bit better knowing that my data, while not perfect, will probably give me a better answer than +/- 50%

 

[RWF7437]

murdock,
Please see:

http//smig.usgs.gov/SMIG/features_0398/runoff.html

and,

http://fl.water.usgs/Abstracts/wri96_4158_trommer.html

and thread162-198581 in this website.

good luck

 

[RWF7437]

also

http://www.forester.net/sw_0607_voodoo.html