HEC-RAS 2 year storm looks same as 100 year storm 2

[peterkugl]

Dear Forum,

I have to put together a report using someone else's HEC-RAS analysis. The stream in question flooded during a hundred year event last year and the Village engaged us to model the 2/10/25/50 and 100 year events. While the 100 year storm matches anecdotal evidence of flooding levels, all the higher frequency storms modeled show flooding levels far higher than anything experienced in the village. Put another way, the 2 year storm (3.5 in/day) as modeled shows the stream overtopping its banks and flooding 90% of the 100 year storm (7.5 in/day) flood area to a depth 90% of the 100 year storm.

Is there any well known systematic error that could be the culprit here.

My best notion concerned the output of the HEC_HMS analysis used. There were only 4 nodes for subbasins modelled so that along the 2 mile stream only four different discharges were used. I've seen HydroCAD models of watersheds that were half the size that had 15 nodes; so that in any basin there was a steady increase in discharge from profile to profile. As a first crude cut I thought of interpolating the discharges I have and applying them to the profiles. This would keep the furthest upstream stations in any subbasin from using discharge values from 1/2 a mile downstream, rather they would be marginally larger than the discharge of the immediate upstream basin.

This is one of my first encounters with HEC-RAS so I'm no champ with it but I dislike the notion of going to the village meeting and putting up nearly the same map 5 times and having people tell me, there's no way they flood that often.

I hope I've explained my problem clearly. Any answers you have would be appreciated greatly.

 

[cvg]

it appears that you have missed a minor, but critical step in your analysis. You need to calibrate both your hydrology model as well as your hydraulic model. This is to make sure it correctly reflects reality as you know it. Apparently, these models don't. I would question the hydrology at this point rather than the hydraulic model. A complex hydrology model is not necessarily any better than a simple one. Also, look at your starting water surface elevation conditions in the hecras model.

 

[Highway27]

I agree with cvg, you need to check your hydrology model, and see if the discharges that are being modeled in your HEC-RAS analysis are realistic for the more frequent storms. I would worry less about having more discharge points along the stream, as long as you're correct with the discharges at the points being modeled, you should have a representative model. If there's a big jump in discharges between the 4 points you've got, maybe you should get some intermediate locations, but if they aren't too different you should be fine.

Like cvg, what I think is happening is that your discharges or starting water surface elevations are just too high for those more frequent storms, and you're modeling too much water in the stream.

 

[francesca]

You need to factor in your times of concentration. In a river, as opposed to a small draw, most of the "node" flows have long-since passed their peak when the river peak comes down. Think in terms of the flooding on the Mississippi right now. The river did not crest at St Louis until this week, but the storms were well over a week ago. Rain on the tributaries and adjacent lands in St Louis peaked and left and might be in the Gulf of Mexico already.

A good rule of thumb I once heard over coffee and subdivision plans rather than a peer-reviewed journal was that "bank high" is about a 1.5 year storm. Of course that does not take into account soil types, antecedent moisture conditions, ground water levels, basin geometries, storm characteristics, etc. But high-recurrence interval floods are often higher than you'd expect.

You can't think of conveyance only in terms of depth. Once you're on the overbanks, a slight increase in depth correlates to a large increase in flow area. As your overbank depth increases slightly, so does the velocity and the conveyance. Make sure you have the right Manning's coefficients for your overbanks, as that's where the majority of the flow will be and so will play a big role in HEC-RAS's headloss calculation.

It's a good point about the downstream water elevation. The HEC-RAS manual says to use the normal depth, but it also says to include extra downstream cross-sections to allow the calculation to settle. If you've got enough downstream cross-sections, you can set a ridiculously low WSL and the calculation should pop back up to something reflecting the conditions you've modeled.

The thing about anecdotal evidence of flooding is that when it starts to rain, most people stay inside instead of running down to the creek to see if the peak breaks the banks.

 

[cvg]

check your hecras cross sections to make sure you include all of the overbank areas. Otherwise hecras assumes a vertical wall. if the surveyor cut the cross sections short, then the analysis could predict higher than normal depths.