Hi Stuti,

[I assume all your calculations are verified to be correct]

It is odd, but those two results are not identical (and maybe significantly off!). The reason is SCS curve number method is not a completely physical based method. CN is a factor which has lump effect of infiltration rate, slope, land cover, etc. Therefore it is an empirical method and heavily relies on "calibration" and experiment. The composite CN for the large watershed not necessarily shows you the suitable CN.

About your question: What should you do to improve the results?

All rainfall-runoff models has imperfection as they have couple of "up scaling" in their assumptions. But if you use a more physical based method you would get better answer. I suggest you use Kinematic Wave Routing with one advanced infiltration function (not constant infiltration, something which includes time variation and infiltration variation based on saturation).

Hope this helps,

Kaveh

Just consider that: the system is very heterogeneous (infiltration and slope), and nonlinear (infiltration and runoff flow) reducing that to a linear equation with handful of parameters is a very large simplification. Ideally (I know it is not possible to do that for city) use real rainfall on the grid and solve Richards' equation to find runoff part and route that part (any routing method, more advance the better)

When you divide a watershed into subunits, there will be some pure hydrologic units, and some composite subwatersheds. For gaining stream systems, some rainfall is absorbed in headwaters or side slopes, and delivered downstream at some distance (variable). Sometimes, water in forested headwaters is mostly absorbed, with little runoff, but interflow and groundwater eventually can show up downstream, it not used for ET, deep storage, contained in lakes, ponds and wetlands. In addition, rainfall may not be uniformly distributed across watershed. To apply the SCS CN approach, it is best to have it validated, and adjusted if needed. Also factoring in the type of channel system could influence results, with gully networks favoring runoff, depletion of groundwater table and sediment delivery, and braided sections favoring aggradation of sediment, broader floodplain accessnd and enhanced storage in water table, sometimes to the point of swamping downstream valley areas (Stanley Trimble on his findings in SC USA Piedmont). When possible validate SCS method with real stream data such as USGS streamgauges, and see how it functions relative to size of the hydrologic unit.

Please suggest me some of rainfall- runoff model which required same input parameters as in SCS-CN methods, which help me to validate my work.

I would suggest using the HEC-HMS model. This model has the flexibility to handle the parameters that you have in each watershed area.

Mohammad Alkhairat

TR-55 is SCS (or curve number) method, which she already used

Thank you all for your suggestion @ Kaveh Zamani, Mohammad Alkhairat, Wesam

Mohammad Ali and William F. Hansen.

I would be surprised if the sum of runoff from 82 subwatersheds was the same as the total runoff. You were not specific as to whether you meant peak runoff, daily runoff, or annual runoff/water yield, and the answer will depend on the temporal context of runoff. If you mean peak runoff, then the answer is a definite no. The larger the watershed, the lower the peak runoff per unit area. If you mean daily or annual runoff, then groundwater and hypereic flows become important. I have seen watershed with “loosing“ streams where the total runoff is less than the sum of the individual subwatersheds because some of the stream flow becomes ground water. I have carried out studies on other watersheds where there was almost no runoff measured from small subwatersheds because all of the excess water came out further down the watershed as base flow, and so the total runoff was much greater than the sum of the subwatersheds. Be wary of any curve number or other model that only considers surface runoff, as watersheds are much more complicated than that. To understand the science behind your question, you must be able to incorporate surface runoff, lateral flow or other subsurface flows, such as pipe drains, and base flow. The main hydrologic model I use is the WEPP model, but it has its limitations to large watersheds, and does not have the ability to consider the complex surface water/ground water interactions that become important in larger watersheds. I am not aware of a hydrologic model that does address this complexity. Such conditions may best be modeled by some method that includes a surface water model and a groundwater model.

Thank you Mr William J. Elliot to make my concept clear.

Hello Stuti!

As you know that the estimation of Runoff depends on different parameters. But in your case in all 82 subwatersheds, what was you input parameters for rainfall. is it same for all. if not you can check for the local monitoring stations and would also recommend as Mr. wesam has told about HECHMS.

You have to take into account that there is an evaporation in the main chute after sub-basin taxation. An example in the Sobadinho lake, on the São Francisco River (Brazil), has an evaporation equal to 10% of the flow if we consider that there are 10 more dams and that this river is over 1000 km long. approach that 25% value found.

you should map the lateral evolution of the difference between subcatchment flows and sum of subcatchment flows.

You could then look where the differences are higher and lower and better refine the search for explanation

Poster Surface Runoff estimation using the SCS-CN method for Kandi ...