Methods for Generating Hydrograph for Large Drainage Areas

[TRSPEMN]

Hello. First post here. I am working on a large 2D H&H model using Flo-2D, and am looking to import a hydrograph for a large offsite area (295 sq mi). However, I am not sure the best way to generate a hydrograph for import. It seems that most methods I am used to (Rational Method and TR-55 for example), are limited to a small watersheds. Anyone here work in a regional H&H capacity that might be able to shed some light on methods/programs for generating hydrographs for event storms? Thanks!

 

[beej67]

That's a big watershed.

Do you have any stream gauges upstream you can use to rake time series data of prior floods?

Most of the time consultants work with areas that big, we're using USGS regression equations to generate flows, but those don't give you hydrographs, just peak flow rates. I suspect you'll want a hydrograph for Flo-2D to make it do the fun stuff. If you weren't using time series flow data, you'd be just as good dialing back to HEC-RAS. If you want to go back to the root of all evil, you can check NEH 630. It's about as comprehensive a look as you'll get.

Do you like Flo-2D? Every demo I've seen looks really cool, but also really time consuming to set up.

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

http://www.campbellcivil.com

 

[jgailla]

You can use HEC-HMS for this to generate flows, although I think you'll have to use HEC-RAS for the output instead of Flo-2D. Both are available for free.

 

[beej67]

I'm sure Flo-2d has a direct hydrograph input option, where you can import time series data generated by HMS. The question is what method to use.

NEH 630 Chapter 10 (above link) outlines the NRCS method (aka SCS method) which is what we usually use for watersheds smaller than that. Chapter 16 outlines how to develop a runoff hydrograph from convolution of dimensionless unitless hydrographs, which is what TR-20 does. TR-55 then uses tables and such to match the results from TR-20. Any number of softwares, including HMS, our local EngTips favorite HydroCAD, or others, can do this for you. But in a watershed that big, without a gauge to compare the results against, you're likely to get answers that are very different than what you experience on the ground. Especially given how difficult it is to generate quality, reliable times of concentration to use in the NRCS method.

USGS regression equations are purely statistical, and based on stream gauges. In my experience, the results from USGS regression equations for peak flow in streams are often much lower than what you'd get from an uncalibrated NRCS analysis. So if you're just looking for a conservative estimate, I'd say use NRCS, but if you're doing a steady state model for a flood study, definitely use USGS regression equations. That's literally why they were developed, and they're pretty accurate. They don't give you a hydrograph though.

Your 295 square mile watershed probably has a lot of sub watersheds in it, with reaches. In theory, if you model each sub watershed individually, and then do reach routing for each as it progresses down the watershed, you'll get a result that's closer to the USGS regression equation as the reach routing will have ameliorative effects on each hydrograph before they're summed at the bottom where your study point is. Your watershed also probably has a bunch of culverts in it which act as storage zones, so you could go so far as to model each as a detention pond. It's a question of how complicated you want to get, or need to get, for your analysis.

If you really want to get fancy, you might consider developing a NRCS hydrograph per the methodology in NEH 630 Chapter 16 (or HMS, or HydroCAD, or Hydraflow Hydrographs, should be very similar) and then also run a peak analysis with the USGS regression equations, and compare them. Then see if there are numbers within your NRCS model you could tweak (calibrate) to get your hydrograph model to match. Goose the Tcs and CNs a little in each sub watershed, basically. I would feel confident as an engineering professional doing this procedure for a simple flood model for a land development project, because you're typically doing that in a pre/post way, so any error in one model would show up in the other as well so your comparisons are still very likely valid. I'm less sure that FEMA would accept this sort of calibration procedure on a county flood study though. There are probably rules for this buried deep in the FEMA review manuals. And if you're that deep into it, the engineering answer starts to sound a little like "Hire Dewberry."

NEH 630 Chapter 16 should be required reading for anyone posting here, particularly when it comes to the process of convoluting unit hydrographs to make a runoff hydrograph. Their explanation is as good or better than any explanation I've seen in a textbook.


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

http://www.campbellcivil.com

 

[gbam]

Have you done any research for the area? There may be previous studies in the area and there could be storm models already and you would be reinventing the wheel. I presume you are in the USA so there are sources out there just need to do some leg work. Try the local college, university, or local-state-federal agencies. USGS has conducted regional studies in my area and I know that the BLM looks at Watersheds as well. Good luck.