The short answer is that any regulation which requires detention ponds in a flood plain is stupid and probably a violation of FEMA's flood plain regulations.  Next, some agency will be requiring detention ponds in the floodway.

What can a detention pond in a flood plain possibly be protecting ?

I dont think that the regulation requires it - I have only seen it on plans and on the ground.  Someone explained it to me that it is possible for the detention pond to peak and empty out before the peak of the flood plain even gets there.  It sounds shaky to me.

johnhan76 has part of the answer.  Here are more parts -
1) The detention pond provides attenuation for runoff for all rainfall events, especially lessor than 100 year floods, that do not cause the flood plain to be occupied.
For example, most local roads with curbs and gutters (in my area) , the storm sewer is designed to handle the 5 year rainfall event.  Without detention ponds, roadways in the flood plain would flood in virtually every rainfall, not just excessive rainfalls.
2) The attenuation volume is separate from the flood mitigation volume.  Increasing the effective volume of the flood plain. (Atleast in my region of southwest Florida)
3) Why bother designing any drainage systems, since there will be some flood event (1 year, 100 year, 10,000 year or 1,000,000 year) that will overwhelm your system and flood everything except for the some mountain tops (remember Noah!)?  Face it, drainage systems are designed to regularly fail because we can not afford to make drainage systems failure proof.  We arbitrarily pick a level of failure that we can afford and design around that. Of couse, the wonks claim a detailed cost benefit analysis justifies these levels, but ...
4) Face it, much civil engineering is performed to meet certain regulatory rules.  If the powers that be agree that a rule is "stupid" then it is changed.  Otherwise, just grin and bear it.  
5) The more "stupid" rules, the more I get to charge because more services and time are required.  It is just an additional development cost.  
6) Why don't you research what Civil Site Plans looked 60 years ago, especially the drainage plans.  It was usually "blow and go", ie no attenuation, no water quality treatment, no FEMA rules, no NPDES, no Clean Water act, etc.   

I hope that this help a bit.

Clifford H Laubstein
FL Certified PE #58662

The local Flood Control District in my area allows
detention basins within the 100-year floodplain.
They tend to be favored there. For one, the developers
tends not to fill the floodplain or rectify the
channel too much if they can use the land. Fill is
expensive and so is mitigating the floodplain and
water surface impacts.

The design aspects of putting a basin in the floodplain
does not countradict the purpose of the detention basin.
The basin job is to attentuate the runoff of a development.
Most basins that are built in the floodplain will have
flapgates of some type to stop the streams flows from
backing up into the basin for more frequent events.
Which are the events that occur regularly.
For events like the 50 & 100-year, if the basin is overtopped,
it is still performing its job. Flows into basin are
reduced to the lesser head differential and the development's
storm sewer system and roadways (depending on the elevation
differences) start to become part of the detention system.

In areas like the Gulf Coast Region where you have flat slopes,
a lot of rain, and wide floodplains, basins are often built
in the floodplains. Once they are overtopped, the flows into
the stream from the development are actually less than if
the basin wasn't overtopped (head differentials). The stream
just uses the area as floodplain storage.

No agency requires detention ponds in the floodplain. No
FEMA has no regulations against putting basins in floodplains.
Why should they? If the basin doesn't have a levee around it,
it doesn't cause conveyance or storage impacts.
However putting a basin in a floodway is tricking because of
the difference in the conveyance values used for vegetation
versus the impervious factor of the basin.

Do any of the respondents to this have a real world example of a detention pond in a flood plain ?  Has the pond been modeled in say HEC-RAS ?  Does the model demonstrate that the assertions made are indeed true ?

If so, I'd be very interested to see the evidence.

rwf7437@attbi.com

Bayou - several questions regarding your post - I am a little confused.  If the detention pond doesnt have a levee, then how are you going to detain water.  If it is just a basin dug into the ground, won't the groundwater fill it up?  If you do have a levee with a flap gate, arent you taking up land that was previously used for floodplain storage.  Thanks a lot for your insight - your post brought up some points that I have yet to consider.  I am also in a flat coastal area with lots of rain and run into this situation all of the time.

A basin can be constructed both with or without a levee.

With a levee you are removing floodplain storage. This lends to a hydraulic impact of the conveyance of the receiving stream as well as require floodplain fill mitigation.

Without a levee does not impact the conveyance or require fill mitigation. This will require flap gates or a similar outlet structure to prevent backflow from the stream.

Seepage from groundwater is based on the areas water table and the soils.  If it is a problem, it would probably be a problem if the basin was outside of the floodplain boundary.

Just to make matters even worse than already described by others............ how often does it happen that the flood plain area where you would like to build, or dig, a detention pond is also a jurisditional wetland ???

Does anyone know of any agency that has design criteria for such detention ponds in flood plains ?  Sure would like to see an example of these things !

Bayou,
When you are discharging your detention pond to a stream, how do you estimate the time/stage (ie boundary conditions) of the stream when you are designing your detention pond outfall structure.  The tailwater is variable so it would seem difficult to model.

For typical designs (the floodplain location doesn't matter), the local flood control districts have specific regulations stating the tailwater elevation. Some districts require a frequency TW as a constant. The district I deal with most (Harris County FCD) requires a constant TW 2 feet below the basin's design elevation. This elevation is either 2 feet below the basin berm or 2 feet below the 100-year wsel. A constant tailwater isn't the best design method; however, for simple basin designs, it is quick, it's what is required by the agencies, and it does tend to over-design the basin.

For a detail design of complex systems, a variable tailwater maybe more appropriate. The old method required UNET; however, with the newer releases of HEC-RAS, the process is a lot easier. The DSS integration is easier and quicker and the hydraulic model no longer has to be remodeled.

Typically the overbanks near the streams of situations like this are not wetland areas. The overbanks areas with the streams here still have a slope towards the stream(in this area 0.1% is considered as a good slope).

Bayou you mentioned determining a variable tailwater. I think that Ill be forced to do this on my latest project.  Is there some way that I can estimate this without going out and getting a lot of cross sectional data?  There is a FEMA FIS on this stream.  IS there anything in that report that will be helpful to me?  THanks for all of your help?

Johnhan 76,
Email me at kekillian@civiltecheng.com. It'll be easier to discuss.