HMS Results
1. Transformation Method Comparisons
-
SCS Lag time: 98.31 mins
-
Clark Lag time: 98.28 mins
The SCS method uses the same dimensionless hydrograph for all hydrographs with the only variable being the lag time. However, the Clark method provides the possibility of accounting for natural watershed storage because of the "R" storage coefficient. Thus you see the distinct possibility of reducing the time to peak and spreading the volume of runoff out over a longer duration. It should be noted here that the volume is computed from the loss method equations and therefore should be equal for the two methods.
2. Sensitivity of R in Clark Model
As should be expected peak flow goes down with larger storage coefficient and goes up with a smaller coefficient. The volume once again remains constant, but the distribution is greatly affected by the amount of storage (lag) in the the runoff as represented by the Clark coefficient.
3. Lag Time Sensitivity in Clark Model
-
SCS Lag time: 98.31 mins
-
Tulsa 50% Urban: 90.63 mins
-
USGS: 130.122 mins
-
Taylor: 1196.07 mins
Most equations provide fairly similar results, however as can be seen from these results some equations can give much, much larger values. The Taylor method relies heavily on average watershed slope and Judy's Branch is rather shallow. It is always a good idea to check the assumptions for different lag time and time of concentration equations. When in doubt if an equation applies it won't hurt to try a few different ones. Note that the volume of each hydrograph is the same.
4. Initial Abstraction variation in MODClark Model (100 by 100 grids)
As would be expected an increase in initial abstractions causes a decrease in runoff volume, and peak runoff. Time to peak is not affected. There is a prevailing consensus that the .2S is too large a value when dealing with low return period storms (common storms), and that a value of .05S is more appropriate. This exercise gives you an idea of the relative differences that can result from different Ia values.
5. Grid Resolution Variation in MODClark
It would appear that once you reach a reasonable grid cell, there is not much difference in the computed runoff.
GSSHA Results
1. Comparison of how HMS and GSSHA treat transformation
GSSHA has the ability to model retention storage, and if we were to do that the results may look a little more like Clark than SCS. One big difference in GSSHA is that the volume is less than the two HMS runs (which have the same volume). This is likely due to the fact that surface infiltration is computed continuously in GSSHA. As such, water is tracking across each grid cell with the opportunity to infiltrate more through time. If we were to decrease roughness we would likely see an increase in runoff.
2. Determination of Time of concentration
In an idealized sense we can get an idea of the time of concentration of a watershed using GSSHA. In this example we have not modeled infiltration, which also has an effect, but we can see that in fact Tc is quite sensitive to variable intensity. This is another good reason to only use the Rational Method for short duration, small watersheds (there is less variation). In most models we assume that Tc is the same no mater the storm intensity, or antecedent conditions.
3. Creating Unit Hydrographs
Both unit hydrographs represent 1 inch of water over the watershed, but the 3 hour unit hydrograph illustrates how a longer duration spreads the runoff out over time.
4. Comparison of UH methods in SCS, Clark, MODClark and GSSHA
This summarizes the different transformation methods, including MODClark. The MODClark model has significantly more volume than the others because the SCS method was used for losses rather than Green and Ampt. You should notice that different lag times and storage coefficients produce quite different result for the same amount of runoff.
*****