The SCS-CN Method has been taught in Universities since I was in school in the 1960s. i dont think there have been any substantive updates.  Dam calculations and assumptions are serious business, suggesting at least considering the stream data collected by USGS and others within your physiographic area.  North carolina is blessed with many stream gaging stations, publications on estimating peak flows, access to USGS sream gauging data, hydrographs, etc.  True, many of the gauged strams are not in urban areas, so adjustments may be needed.  On my researchgate, I have one poster that presented the comparison of area normalized flows with a few physiographic areas from Georgia mountains, piedmont, etc.  Most of these streams have little effects of urbanization.  Urbanization with increasd stormwater response will typically increase bankfull flow and cause channel adjustments, typically entrenchment and sometimes gullying and channel network expansion if substrates are fine grained.  Within last decade or so, stormwater regulations are improving retention of stormwaters locally.  I am not sure if the SCS Method considers existing dams upstream and the effects to floodwaters if they fail.  None of these type changes are addressed with SCS MODEL.  I know there are quite a few models like SWAT, WEPP, MIKE with variant subsystems and others that may be applicable to these type problems, but frankly, when NC has abundant real USGS stream data so available, I prefer to use that, or in your instance, at least run the calculations for comparison purposes.  I do not know the degree that the SCS-CN methods considers the intensity of Gulf and Atlantic generated tropical storms.  In 2004, we had Tropical storms Jeanne and Ivan with hydrographs at local stations.  Run the SCS-CN method on a few gauged watersheds of interest to compare results with actual data, land use, etc. factored in.  If thise storms did not affect your area of interest, I would guess you can find some fairly stream flood data for a gauged stream or steams in your physiographic area, perhaps vicinity.  My advice when lives and property are at stake, do a few more calculations and consider local data, and adjust regional data when necessary, as in the Upper Chattooga River, the regional skew flood data forecasted peak flow data was well below my projections considering several years of storm response data I collected at the station.  The difference I surmise, is a rainfall adjustment, as this Upper Chattooga area is one of the highest in eastern US, so physiographic area regional calculations that combine many stations with less rainfall are probably going to underestimate peak flow in highest rainfall area.  If as a professional engineer or hydrologist, you know little about the area under analysis, have not visited and looked at any bankfull flow and if present flood indicators in comparison to your regional stream metrics by physiographic area, I would suggest that there is more to doing a quality job than running models.  In your question, I detect uncertainty and desire to not just get an answer, but do a creditable job.  So this attitude may cause you more work, but in the long run, apt to produce the best results.  As a professional engineer, your name will be signed on the dotted line on dam design projects.  

If not exposed to some of the stream gauging data available, www.usgs.gov,  I think you will be surprised.  Precip data may be available from several sources, but the joint stream and rain gauging that USGS often provides now is helpful, even though rainfall distribution is seldom uniform and may need to consider other data sources.

If you have time to spare, some other models could also be used and compare results among methods.  I dont remember any papers doing this, so perhaps some variant of comparing real stream flood data and discuss the results and limits of various models might be a good graduate degree project.  

Impervious cover estimates from streamstats could be derived from coarse scale land cover data. You could get some better estimates if you have a high-resolution remote-sensed data. 24% is common in moderately urban watershed. For sing CN method, your CN coefficient for the impervious area should be calibrated to your watershed. I recommend running the simulation for the watershed of any gauge station and calibrate the CN and baseflow parameter values. This will lead to better runoff routing. 

The default value for impervious area is 0. Altering this value to 1%, 2% or higher will simply result in higher runoff values in your simulation. By adding a fraction of your sub-basin or basin area as impervious area, it is basically like adding an area which is impermeable to water, with a value of CN 100. There will not be any hydrological losses and all the rainfall volume that falls onto the surface of this area will result into surface runoff.

However, when calculating the CN for a given terrain, based on the land cover and the hydrological group of the soi, it is supposed that the fraction of soils that are characterized as impervious are already taken into account (by dividing the soil surface in permeable, semi-permeable and impermeable). So, in my opinion, be very cautious and conservative on how you use this parameter, for it may result in extremely high volumes of surface runoff. 

Belatedly, data for N. Am. (esp. in the Pac. NW) yield total-impervious areas of 20-25% for 'medium-residential' landscapes, which translates to effective-impervious areas of 10-14%. This isn't as developed as suburban landscapes, so your value of 24.41% is reasonable, but may be an underestimate b/c urban development is prominent in your study area.

-Bob Vadas, Jr.

Since you are in NC, you might try exploring the effect of imperveous areas in the SC physiographic areas using Streamstats, which now used the LiDAR topography and the equations developed by Feaster et al at USGS, which includes an imperveous element. You might select a few sites within a physiographic area in Streamstats that likely have varied drainage size, land use and imperveous areas to test some of your need. The soils data is also available in Streamstats for site selected, so the CN numbers could be assigned for each drainage area evaluated. You might also want to compare the rainfall amount/intensity differences such as with Technical Paper 40 or updated version. The Feaster et al equations are also available for physiographic area, so that also could be considered if you did not want to use Streamstats. There are some SC to NC differences also, and for example some drainage areas of the SC piedmont were more severely eroded and gullied than others, some coastal drainages have more wetlands and storage than others. I would be very leary about relying only on the SCS CN method if other methods are available. Cross checking good, and for dam failure considerations, I would recommend developing envelop curve for past flood data to estimate historic extremes to compare to. It is obvious in SC and NC, the design calculations and mitigation measures used for many small dams was insufficient as evidenced by failures. Some of this may also be due to poor maintenance (see my researchgate report to prevent a small dam failure about 2013).