simple RUNOFF question

Suppose your watershed has 80% impervious area but runoff calibration suggests 20% impervious hence hinting that only ¼ of the impervious area is directly connected. Now if you set %Impervious in line H1 of RUNOFF to 20, what happens to the rest of the actually impervious area? Will it be treated by SWMM as pervious area and so the same infiltration, detention storage, and manning n values as impervious area will be used for it? What about the quality (build-up wash-off coefficient of this "undefined" portion of the watershed? The puzzling thing is that I know that in SWMM/RUNOFF the runoff from one surface does not flow over another surface.

Thanks for considering this simple question,
Kaveh Zomorodi, Ph.D.
kzomorodi@nier.org

For the quantity side, we usually account for the non-directly connected impervious area by multiplying the Horton infiltration rates by the actual pervious area divided by the modeled pervious area (pervious plus NDCIA).

Hi,

Though a beginner I will try to answer part of this question. 80% of watershed will be treated as pervious area. Infiltration will be processed only for pervious area. Storage can be defined separately for pervious and impervious areas.

More, 25% (default B4 PCTZER parameter) of impervious area has no storage capabilty.

Regards.
Janek Milewski
milewski@libertysurf.fr

I get two different impressions from the answers: 1) The NDCIA is treated exactly like pervious areas, even though the SWMM manual clearly states that runoff from one area will not go over another area and all runoff from one land cover directly drains to gutters. 2) Statement 1 is not true and nobody knows how NDCIA is treated in SWMM!

I ran a sensitivity analysis for my watershed that has an apparent impervious area about 75% of the watershed (from maps and GIS), and the model likes a DCIA of only about 15%. While the runoff peaks and volumes are very sensitive to DCIA, they are not at all sensitive to infiltration parameters or any other parameters of the pervious area. To me that indicates that the NDCIA is not treated as pervious area. This bring me back to my original question: What is happening to 85%x75% or over 60% of my watershed? Maybe this is not such a simple question after all?!

Kaveh Zomorodi, Ph.D.
kzomorodi@nier.org

Watch out for those simple questions that generate complicated answers.

I get the impression that NDCIA wasn't really envisioned in the genesis of SWMM RUNOFF. Is there enough of a consensus about how NDCIA affects overland flows to warrant a new subroutine? I'm envisioning subcatchment inputs for % DCIA (presently % impervious), % NDCIA, and % truly pervious. Infiltration gets tricky, but should be handled internally. Are there any research papers to guide us here?

Here's a situation and a question that I believe is related: Through a regional agency, I maintain a SWMM RUNOFF/EXTRAN model of a 20 square mile watershed that gets updated annually with land use and channel modifications. The goals of the annual modeling exercise are to provide accurate projections of design storm flows in an Army Corps flood control channel and to monitor how well or poorly the four localities that share the watershed are managing their flow contributions. Under the program, detention storage is required for most developments within the watershed, and localities expect the model to give them credit for this detention storage. While a few large detention storage systems are represented directly, there are hundreds of small detention storage systems that provide peak hydrograph shaving benefits which are impractical to model directly. I have up to 30 of these per subcatchment. And the question would be...

...Does anyone know of a good way to handle this detention storage in the aggregate at the subcatchment level, or do I need to keep breaking up my subcatchments into ever smaller slivers?

Don Waye
NVPDC

You won't getting any infiltration effects if your storm isn't big enough to surpass the initial infiltration rate...That's why we calibrate impervious areas (DCIA and imperv storage values) with small storms, and pervious ness areas(soil parameters and perv storage values)with large storms.

If impervious runoff can cross a pervious surface before it gets to a storm drain or creek (NDCIA), it has a chance to be infiltrated into the soil. When enough NDCIA crosses the pervious area, the infiltration rate will decrease (decay rate) and pervious surface runoff will increase (i.e. acts more like a pervious surface eventually). You calibrate your infiltration (soils) parameters to adjust for the NDCIA area that crosses the pervious area.

How big is your storm?

Dave Briglio

I've been following this thread for a couple of days. If I understand the theme, there are at least two types of impervious area, connected and non-directly connected (to the surface runoff regimen). The initial runoff calibration using the impervious area serves to define these two types. The question remains then what happens to the NDCIA, do quantity and quality parameters from this component contribute to the pervious or subsurface regimen? Assuming that this is a true distillation of the thread, my question is how could such a potentially large component of the quality and quantity picture be omitted from RUNOFF's algorithms? Brett, excuse the question, but if SWMM does not dynamically compensate for the real amount of directly connected impervious area shouldn't your compensation factor be larger than one or the inverse of what you mentioned. I am bordering on being thoroughly confused, what am I saying, I AM CONFUSED.

Sipho M. Munyaradzi
The Ohio State University

Sorry for my stupid answer to the same question. I'm better at understanding
french.

Janek Milewski
milewski@libertysurf.fr

I posted the original question, but didn't mean to cause confusion. That's why the subject of the message was "simple RUNOFF question". This ambiguity brings me back to my position that SWMM is only good for urban areas that are almost completely served by a runoff collection network directly connecting to all impervious lots. Otherwise someone please tell us how the NDCIA is handled in RUNOFF and why it is a sufficient computational procedure. For the watershed that I am modeling 60% of the entire watershed seems to be NDCIA.

Kaveh Zomorodi, Ph.D.
kzomorodi@nier.org

In the spirit of the simple question, I have a simple answer...

Runoff from impervious area that is not directly connected must flow over pervious area before it reaches the conduit network. That's the definition of "not directly connected". In that case, it will infiltrate as if it never fell on the impervious surface and its overland flow roughness will also be that of the pervious surface. Depression storage is a more difficult issue. You could just wave your hands and call it a calibration parameter, or assume that it's the same as pervious surface.

In short, if 60% of the watershed is NDCIA, then it should be modeled as 60% pervious, regardless of how much pavement there is.

Bill Frost
KCI Technologies, Inc.

Kaveh,

I have been reading this long drawn out discussion for several days and have some input by way of a couple of questions.

1) Are you comfortable with the actual impervious area value of 75%? Has this value been physically measured from aerial mapping or similarly accurate measurements?

2) Can you, from the same source, estimate the NDCIA?

3) Regarding the storm event to which you are calibrating:
Can this be considered a major storm event, where the DCIA is the major controlling parameter? Or, could this be considered a more minor storm event, where the surface retention storage, overland roughness, and infiltration parameters have a much larger bearing on the results of the peak discharge? Do you also have enough information from the actual storm event to be able to compare the volume runoff? How do your modeling results compare in terms of both peak discharge and total runoff?

I find it relatively inappropriate to assume that the whole of your calibration would revolve around the DCIA / NDCIA parameter. It seems much more appropriate to make a physically represented calculation of the DCIA and then question each of the modeling parameters from there. It concerns me that rainfall falling on a rooftop or other impervious surface, which is then transported over a pervious surface, is simply treated as pervious area: (a) the surface storage will be different; and (b) the infiltration for the 'true' pervious area will infiltrate, whereas the NDCIA will not necessarily be able to infiltrate in reality, if that pervious area has already satisfied the maximum infiltration with it's own rainfall. On the flip side, it is certainly more appropriate to route the NDCIA using the roughness values for the pervious area.

Please, let us all know how this issue comes to completion.
Christopher L. Doherty, P.E.
cdoherty@ayres-fc.com

In SWMM, there is no NDCIA. You can either lump it with the pervious area where is treated as though it were pervious, or is included with the impervious area and is treated as though it were impervious. Logically, neither of these is strictly correct as runoff from the NDCIA actually spills onto the pervious area and thus generates more runoff than you would otherwise get from the equivalent pervious area but less than you would get from an impervious area.

In a perfect world we would represent the NDCIA explicitly but this is more difficult that might appear at first. ILLUDAS and SWMHYMO are two models that try to do so. Rainfall on the NDCIA is added to the rainfall on the pervious area as though it were uniformly distributed over the pervious area. I believe this would be a useful change to SWMM. In the meantime, Brett's approximation of reducing the effective infiltration in proportion to the ratio of NDCIA to pervious area, seems to me to have much the same effect and seems like a sensible approach.

I would submit that this is still only an approximation of the real world, as for example a roof downspout that drains to the lawn does not spread water over the entire lawn surface but rather tends to drain to low areas of the lot which the owner, naturally, tries to drain to the street. Thus the imperviousness and infiltration parameters are not strictly deterministic as we would like to assume but may need to adjusted (hopefully by calibration) to reflect actual runoff conditions. In short, there is room for judgement. In our work we have found that taking half of the NDCIA as pervious and half as impervious seems to work reasonably well.

Every model is an approximation of the real world and I for one am glad this is the case. It reminds me that we are imperfect beings living with imperfect knowledge in an imperfect world, which leaves a lot more room for intuition, experience, and creative thinking, and makes likfe a lot more interesting if sometimes frustrating.

Larry Bodnaruk, P.Eng.
bodnarukl@edmonton.associated-eng.com

Kaveh,

I think the truth is more like this: SWMM does not explicitly model NDCIA. Forgetting snowmelt for the moment, a Runoff subcatchment is divided into 3 areas (see attached figure):

1) Pervious Area
2) Directly Connected Impervious Area with detention
3) Directly Connected impervious area without detention

You have to divide your subcatchment area into these three categories. As Bill was saying, NDCIA is normally assigned to the pervious area (1).

Even though the manual simply refers to 'percent impervious area', it really means 'percent DCIA' and should not include NDCIA. The manual is correct in stating that the flow from one area is not routed over another area as there is no explicit NDCIA area (not to mention the fact that there is no 'routing' at all within a subcatchment).

The third area is included to give some immediate runoff and is assumed to be 25% of the DCIA by default (can be modified with the optional B4 line).

See http://www.chi.on.ca/tutorial4.html for more details on SWMM surface routing.

Sincerely,
Rob James
Computational Hydraulics Int.

I would like to thank every body who contributed to this discussion. I repeated sensitivity with an intense rainfall (1.5" over 3 hours) and still pervious parameters did not come out to be important. Any how, I conclude from the Rob's and other responses that there is no such thing as NDCIA in SWMM. Thanks again,

Kaveh Zomorodi, Ph.D.
kzomorodi@nier.org

Before we beat this thing to death, I would like to suggest that we (me?) make the following modification to SWMM: allow a subcatchment to drain to the surface of another subcatchment. In this case the runoff from a subcatchment can be added to the input volume of a selected subcatchment. I realize that this approach has it's drawbacks (i.e. surface routing still isn't possible - the volume of inflow from the upstream subcatchment is instantly distributed over the entire downstream subcatchment). I have long thought about doing this as a way of modeling run-on. Any comments?

Sincerely,
Rob James
Computational Hydraulics Int.