This is a question about the runoff calculation when we use the curve number option. As we know curve numbers have built-in reflection of land use imperviousness (i.e. paved parking lots: CN=98; Industrial land use on A soil: CN=81). But in SWMM we also need to input the %imperv for a sub-catchment. So, when we use the curve number method for soil infiltration, are we double-counting the effects of imperviousness. Or it might be the case that one method overrides the other when it comes to the curve number option. I did some test case and found out that the curve number value is less influential than the %imperv, but just want to be clear on this.

The original question can be simplified as: Should the CN we input in SWMM be a composite CN for a subcatchment or the CN for pervious land only?

Based on my re-read of the SWMM5 manual (no luck) and the source code, the answer I have now is the CN should be for the pervious land only. Please correct me if my understanding is wrong. Below is a section of source code from "subcatch.c," and within the function of "getSubareaRunoff()."

   // --- compute infiltration loss rate for pervious subarea
    //     (inflow consists of runon + precip - evap)
    infil = 0.0;
    if ( i == PERV )
    {
        if ( Subcatch[j].infil  == j )
        {
            infil = infil_getInfil(j, InfilModel, tStep,
                    (subarea->inflow - surfEvap), subarea->depth);
        }
        if ( infil > availMoisture - surfEvap )
        {
            infil = MAX(0.0, availMoisture - surfEvap);
        }

So as the code shows, the infiltration calculation is carried out to pervious area only (the CN is used later in the function of "infil_getInfil()"). Thus, the CN should be for the pervious land only instead of a composite CN for the whole subcatchment. Unfortunately this is not highlighted either in the manual or the annotation of the CN input window.

Please correct me if the above observation is wrong.

In SWMM 5 you should input the CN for pervious land area only since the CN is ONLY used in SWMM 5 to calculate the infiltration loss. If the CN is only used for infiltration purposes it should have been obvious that entering a composite CN is not required or even a good idea. I think the problem is that users are so used to the concept of the CN being used to generate the runoff that using the SCS unit hydrograph that a model that only uses the CN for infiltration confuses the modeling issues. If you use the CN method in SWMM 5 the Runoff from the watershed is still generated exactly the same way as if you used Green Ampt or Horton infiltration.

Your discussion thread did prompt one idea that would help resolve such questions in the future. It would be a good idea if the subcatchment summary table told you how much runoff was coming from the pervious and impervious area as well as the total runoff from the subcatchment for calibration purposes and making it easier to understand the output of the program. Many others have suggested this improvement in the past.

------------------------------------------------------------------------------------
                     Total     Total     Total     Total     Total     Peak   Runoff
                    Precip     Runon      Evap     Infil    Runoff   Runoff    Coeff
Subcatchment            mm        mm        mm        mm        mm      CMS
------------------------------------------------------------------------------------
Wat1                10.000     0.000     0.009     9.893     0.065    0.003    0.006
Wat2                10.000     0.000     0.009     9.893     0.051    0.004    0.005
Wat3                10.000     0.000     0.009     9.893     0.051    0.004    0.005
Wat4                10.000     0.000     0.009     9.893     0.051    0.004    0.005
------------------------------------------------------------------------------------
System              10.000     0.000     0.009     9.893     0.053    0.014    0.005

I totally agree with you, especially with the experiences from one previous project. Two flow stations are located in one urbanizing watershed, with one monitoring the urban portion and the other one monitoring the less-developed portion. When calibrating SWMM to the two stations at the same time, model performances at the two locations are split (i.e. over-prediction at one location and under-prediction at the other). If we could check the detailed runoff from pervious/impervious surfaces, then the calibration process would definitely benefit from that.

Another issue might be useful in the next release (just my 2 cents) could be adding the adjustment of CN according to antecedent soil moisture conditions. I think it is fine to have the infiltration rate in an event remain constant in the current CN option (not as the decreasing infiltration rates in Horton and Green-Ampt options), since the CN option is more like long-term simulation. But the adjustment of CN based on antecedent soil moisture conditions is appropriate and do not involve much coding. Please point out for me if I miss something here.

One thing to keep in mind about is how the CN method works (assuming SWMM still uses the original SCS method from NEH-4). CN is widely accepted because of its simplicity and directness, but it did not produce a time- and rate-dependent accounting of runoff. Runoff produced by the method is calculated as total runoff from total rainfall, and for a storm of any particular depth and distribution, the process calculates the runoff for the cumulative rainfall at each point along the way, with incremental runoff being calculated by subtracting cumulative totals at subsequent times.

The result is that runoff is independent of rainfall rate, and for a given CN, 6-inches of rain falling in an hour will produce only the same runoff as 6-inches distributed evenly over 24 hours. Of course, if SCS (NRCS) has changed this in more recent times, I would appreciate learning about it.

Please see my posting of 1/24/2007 that explains how the SWMM 5 Curve Number infiltration process works.

https://www.openswmm.org/Topic/3391/curve-number-for-infiltration-calculation

Instead of the CN being adjusted, the moisture retention capacity S is reduced during wet time periods (by the incremental infiltrated volume) and is increased during dry periods using a first-order recovery model (where the maximum retention is Smax = (1000/CN) - 10, and is the value used for the initial storm event). At the start of the next discrete storm event, the current capacity value S is used in an incremental form of the Curve Number infiltration equation to compute infiltration as a function of cumulative rain volume during the event. This process thus allows the CN infiltration method to account for antecedent conditions (S decreases during the current storm and increases during the dry period before the next storm).

In the first releases of SWMM 5 we used the varying S in the incremental CN infiltration equation within the storm event, but in later releases decided to keep it constant at the value at the start of the storm for the entire event so as to remain more faithful to the original SCS procedure.

I posted yesterday that the SCS method:

"... calculates the runoff for the cumulative rainfall at each point along the way, with incremental runoff being calculated by subtracting cumulative totals at subsequent times. ...runoff is independent of rainfall rate, and for a given CN, 6-inches of rain falling in an hour will produce only the same runoff as 6-inches distributed evenly over 24 hours. ... "

Having said that the SCS method doesn't make any accounting for rate of rainfall or declining infiltration, I want to be sure that this isn't an unsupported conclusion. I haven't heard of any changes, and assume that this remains true, given below that:

"... in later releases decided to keep it constant at the value at the start of the storm for the entire event so as to remain more faithful to the original SCS procedure."