Comparison of Routing Methods in HMS and GSSHA

CEEn-531 Dr. Nelson

HMS Results

1. Muskingum parameters:

Length of reach L = 13260.07 (From WMS, Muskingum Cunge dialog length of routing reach)

Velocity of flow V = 5.16 ft/sec (WMS channel calculator)

K = L/V = 0.714hr

Number of time steps = K/time step = 0.714*60/6 ≈ 8

To determine X:

Slope = 0.0056 (From WMS, Muskingum Cunge dialog)

z = 2

b = 15 ft

n = 0.05

Q₀ = 125cfs

B = 20.464 ft (WMS Channel calculator)

V = 5.16 ft/sec (WMS channel calculator)

DX = L / number of time steps = (13260.07/8) = 1657.509 ft

X = 0.487

Old Parameters

K                                = 2.26 hrs

X                                = 0.386

No. of Time steps     = 22

n                                 = 0.15

v                                 = 3ft/sec

L                                 = 24410 ft

2. Muskingum Method Results

Muskingum Base model

K = 0.714 hrs

X = 0.487

Number of subreaches = 8

X sensitivity

K = 0.714 hrs

X = 0 and 0.5

Number of subreaches = 8

K Sensitivity

X = 0.487

K = 0.357 and 1.071hr

Number of subreaches = 8

You can see that increasing K increases the lag in the hydrograph, though not significantly.  Also fluctuation of X between .5 and 0 is changes the results from translation to attenuation.  However, because of the relatively short travel times for the routed hydrographs there is not much attenuation (not time to develop it).

Exaggerated Travel Times

You can see that extending the travel time (larger K) not only increases the lag (even the double peak), but it also increases the attenuation because now there is more time for the hydrograph to be affected by storage factors in the reach.

3. Muskingum Cunge Method Results

Cross section shape = Trapezoidal and Deep

Channel shape had very little effect on the routed hydrograph.  It probably would have been interesting to run this with identical cross section areas as I suspect what little differences there are exist because of the difference in available storage within the reach (more cross sectional area means more storage volume).

4. Comparison between Muskingum and Muskingum Cunge Methods of Routing

Both models compare, but the Muskingum Cunge as defined with it's roughness clearly lags and attenuates the hydrograph more than the chosen X and K values of the Muskingum method.  Volumes should be the same.

5. Comparison between No routing, Two sub basin (Muskingum Cunge), 4 Sub basins, 8 sub basins, MODClark and GSSHA

This simulation has the affect of making the model more distributed. As more sub-basins are added routing is converted from "overland" as estimated by lag time or Tc to channel.  The channel routing methods as you would expect clearly transmit the water faster than the overland methods.  It is a little hard to compare the GSSHA method to the others, without know a little more about the drainage density (to be tested later).  The MODClark runs show the difference between the runoff volume computed with Ia=.2 and a value of Ia=.75 which is very high, but approximates the same runoff volume as the other models.  However, it once again illustrates that overland dominated runoff takes longer and spreads out the volume.

GSSHA Results

1. High Density Model Vs Low Density

The following Green and Ampt parameters are used

 

Unlike the HMS models, differences in GSSHA stream densities have large affects on runoff volumes.  This is because with a smaller drainage density the water stays on the overland plane longer and therefore has more opportunity to infiltrate.  If the water gets in the channel "quicker" because of the higher density then it will route directly to the outlet without losses (at least the way this model was developed). 

2. Trapezoidal Vs Natural cross sections (Low Density model)

Channel geometry had very little affect on runoff in GSSHA, suggesting that a good estimate is adequate for hydrologic runoff purposes.  Channel geometry can have a more important role where water surface elevation in the channel is an important factor.

3. Comparison of the effects of Manning's n for the stream (Low density/Trapezoidal section)

Manning's n values can have a rather large effect on the runoff time.

*****