CEGE 4501 Hydrologic Design

Chapter 7:
Unit Hydrograph and Routing

Ardeshir Ebtehaj
Department of Civil, Environmental, and Geo-Engineering

November 27, 2017

Outline

Linear Reservoir and Convolution

Unit Hydrograph

Unit Hydrograph Derivation

Flow Routing

Unit Hydrograph and Routing Ardeshir Ebtehaj

Linear Reservoir and Convolution I
A linear reservoir is a simple and meaningful representation of a watershed. Continuity equation
for a linear reservoir is written as follows:
n dS(t)
dt = I(t) − Q(t) continuity equation
S(t) = K Q(t) linear storage function

Figure 1: Schematic of a linear reservoir. A reservoir is linear if we assumed that the storage
function S(t) is linearly related to the outflow through the constant K .
where Q(t) [L3 T-1 ] is the outlfow, I(t) [L3 T-1 ] is the inflow, dS(t)/dt [L3 T-1 ] is change in
storage over time, and K represents a time scale that captures the residence time of the reservior
[T]. Now upon substitution, we obtain and solve a first-order ordinary differential equation (ODE):
dQ(t)
K dt = I(t) − Q(t)
dQ(t)
dt = I(t)/K − Q(t)/K

Multiplying both sides by e t/K , one can get:

dQ(t)
e t/K dt + e t/K Q(t)/K = e t/k I(t)/K
Now because of the product rule we can rewrite the above equation as follows:

Unit Hydrograph and Routing Ardeshir Ebtehaj 1

Linear Reservoir and Convolution II


d Q(t)e t/K
dt = I(t) K1 e t/K ,
and thus, we have,
Rt
Rt
d Q(s)e s/K = I(s) 1
K e s/K ds
0 0

Outflow of a linear reservoir:

Z t
−t/K 1 −(t−s)/K
Q(t) = Q0 e + I(s) e ds
K
0

This solution to the differential equation states that the outflow at any time relies on an initial
flow (first RHS term) that goes to zero over time and the integral term (second RHS term) that
represents the effects of the reservoir resident time on the outflow. The integral term is a
convolution integral. Learning the concept of convolution is crucial for understanding the concept
of the unit hydrograph.
Mathematical convolution for two generic functions f (s) and g(t) is defined as follows:
Z +∞ Z +∞

f ∗ g(t) = f (s) g(t − s) ds = f (t − s) g(s) ds

−∞ −∞

Figure 2 contains two toy examples of convolutions to visualize the process. We can see that the
convolution of a function f (t) with a kernel g(s) may result in translation (shift) and
dilution/diffusion of f (t).

Unit Hydrograph and Routing Ardeshir Ebtehaj 2

Linear Reservoir and Convolution III

Figure 2: Toy examples of convolution (Credit: Wikipedia).

Now looking back on the integral in the linear reservoir outflow equation, we can see that it is a
convolutionRoperator as I(s) resembles f (s) and k1 e −(t−s)/K resembles g(t − s) as
∞
I ∗ u(t) = I(s)u(t − s)ds, where u(t) = 1
K u −t/K .
0

If we define the input as the unit impulse function or the Kronecker delta function as follows:
n
0 if s 6= t
I(s) =
1 if s = t
Then, we can define the impulse response function of a linear reservoir as follows:
1 −(t−s)/K
I ∗ u(t) = u(t − s) = e .
K
Similar to the above formalism for a linear reservoir, a watershed has a response function to a unit
pulse of excess precipitation—albeit a more complex one. This response function is known as the
unit hydrograph. More specifically, the unit hydrograph u(t) of a watershed is defined as the
direct runoff hydrograph (DRH) resulting from a unit pulse of excess rainfall, which is defined as a
unit depth of excess rainfall (i.e. 1 in., 1 cm, etc) distributed uniformly over the entire watershed
area for an effective duration of 4t.
Unit Hydrograph and Routing Ardeshir Ebtehaj 3
Linear Reservoir and Convolution IV

Figure 3: The response of a linear reservoir to a unit impulse (left) resembles the response of a
watershed to a unit pulse of rainfall (right), which is a unit amount of rainfall uniformly
distributed over the watershed for an specified duration of 4t

Convolution is a linear operator

f ∗ (a g1 (t) + b g2 (xt)) = a f ∗ g1 (t) + b f ∗ g2 (t),
which means that we can superimpose the impacts of two pulses of rainfall over a watershed
though the linear combination of their individual responses.

Unit Hydrograph and Routing Ardeshir Ebtehaj 4

Linear Reservoir and Convolution V

Figure 4: Since we aim to model the watershed response to pulses of rainfall thorough convolution,
we can linearly combine the response functions due to linearity of the convolution operator. For
example, the above schematic shows that Q(t) = 1 × u(t − s1 ) + 0.5 × u(t − s2 ).
Therefore, having the unit hydrograph, one can obtain the outflow of a watershed as convolution
of excess rainfall pe (t) with the unit hydrograph u(t) as follows:

Z t

Q(t) = pe (s)u(t − s)ds.

0

Unit Hydrograph and Routing Ardeshir Ebtehaj 5

Unit Hydrograph I

Figure 5: Convolution of Excess Rainfall Hyetograph (ERH) with m = 1 . . . m intervals with the unit
hydrograph of a basin, which is characterized at l = 1 . . . L intervals, results in a Direct Runoff
hydrograph (DRH) at n = 1 . . . N points.
Since, hydrologic data is not continuous, we must use discretized versions of the convolution
equation we discussed previously as follows:

min(n,M)
X
Qn = pm un−m+1 n = 1, ..., N and N = M + L − 1.
m=1

Unit Hydrograph and Routing Ardeshir Ebtehaj 6

Unit Hydrograph II
For the example shown in Figure 5, we have

Qn = p1 un + p2 un−1 + p3 un−2 ,
where M = 3, L = 6, N = M + L − 1 = 8. Therefor, the above convolution can be expressed as
a linear system of equation as follows:

Q 1 = p1 u1 + 0 + 0
Q 2 = p1 u2 + p2 u1 + 0
Q 3 = p1 u3 + p2 u2 + p3 u3
Q 4 = p1 u4 + p2 u3 + p3 u2
Q 5 = p1 u5 + p2 u4 + p3 u3
Q 6 = p1 u6 + p2 u5 + p3 u4
Q 7 = 0 + p2 u6 + p3 u5
Q 8 = 0 + 0 + p3 u6

Thus one can arrange all the rainfall values in a matrix form as follows:

Unit Hydrograph and Routing Ardeshir Ebtehaj 7

Unit Hydrograph III
Q1  p1 0 0 0 0 0 
Q2 p2 p1 0 0 0 0 u1
Q3  p3 p2 p1 0 0 0  u2
Q4  0 p3 p2 p1 0 0  u3 
  
 = 
Q5  0 0 p3 p2 p1 0  u4 
Q6  0 0 0 p3 p2 p1  u5
Q7 0 0 0 0 p3 p2 u6
Q8 0 0 0 0 0 p3
and solve it for ui values to obtain the unit hydrograph from the observed flow of a precipitation
excess event. We can see that in the above linear system of equations we have more equations
that unknown. Therefore, calculation of the unit hydrograph is often an overdetermined problem.

Unit Hydrograph and Routing Ardeshir Ebtehaj 8

Unit Hydrograph IV
As previously explained, the unit hydrograph (UH) is the response function of a linear hydrologic
system to a unit pulse of rainfall, which can be convolved with the excess rainfall to simulate the
direct runoff. Below is an example of a 6-hour unit hydrograph being convolved with hourly
rainfall averaged into 6-hour bins. The example shows the principle of hydrograph superposition.,
It is important to note that by construction, the base time of a DRH resulting from a certain
duration ERH is constant, therefore UHs vary with storm duration and for a given watershed, the
UH is an unchanged characteristics of a watershed for a specific duration of excess rainfall.

Figure 6: Example of 6-hr Unit Hydrograph convolution (Credit: The COMET Program).

Unit Hydrograph and Routing Ardeshir Ebtehaj 9

Unit Hydrograph Derivation I
In a general form the convolution of excess rainfall and the unit hydrograph can be expressed as
follows:
Pmin(n,M)
Qn = pm un−m+1
m=1

Qn : Direct runoff [L3 T-1 ] , n = 1, ..., N

pm : Excess rainfall [L], m = 1, ..., M

un−m+1 : Unit hydrograph [L2 T-1 ], l = n − m + 1.

Q1 = p1 u1
Q2 = p2 u1 + p1 u2
Q3 = p3 u1 + p2 u2 + p1 u3
. . . .
. . . .
. . . .
QM = pM u1 + pM−1 u2 + ··· + p1 uM
QM+1 = 0 + pM u2 + ··· + p2 uM + p1 uM+1
. . . . . .
. . . . . .
. . . . . .
QN−1 = 0 + 0 + ··· + 0 + 0 +pM uN−M + pM−1 uN−M+1
QN = 0 + 0 + ··· + 0 + 0 +0 + pM uN−M+1

Matrix Derivation
The above system of linear equation which is results from a discrete convolution operation can be
rewritten in matrix form as follows:
q = [Q1 , Q2 , . . . , QN ]T , u = [u1 , u2 , . . . , uL ]T , L=N−M+1

Unit Hydrograph and Routing Ardeshir Ebtehaj 10

Unit Hydrograph Derivation II
 
p1 0 0 0 ... 0 0 0 ... 0 0 0
 pp23 p1
p2
0
p1
0
0
...
...
0
0
0
0
0
0
...
...
0
0
0
0
0
0

 
 .. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

. . . . . . . . . . . . 

P = pM pM−1 pM−2 pM−3 ... p1 0 0 ... 0 0 0


0 pM pM−1 pM−2 ... p2 p1 0 ... 0 0 0 
. . . . . . . . . . . .

. . . . . . . . . . . . 
. . . . . . . . . . . . 
0 0 0 0 0 0 0 0 0 0 pM pM−1
0 0 0 0 0 0 0 0 0 0 0 pM
where q is an N-by-1 column vector, P is an N-by-L matrix, and u is an L-by-1 column vector.
For example, the above matrix form in Figure 5 can be expressed as follows:
Pu = q
p1 0 0 0 0 0   Q1 
p2 p1 0 0 0 0 u1 Q2
p3 p2 p1 0 0 0 u2 Q3 
0 p3 p2 p1 0 0  u3 
   Q4 
 · =  
0 0 p3 p2 p1 0  u4  Q5 
0 0 0 p3 p2 p1  u5 Q6 
0 0 0 0 p3 p2 u6 Q7
0 0 0 0 0 p3 Q8
where M = 3 is an L = 6 and N = M + L − 1 = 8.

Unit Hydrograph and Routing Ardeshir Ebtehaj 11

Unit Hydrograph Derivation III
The matrix form can be easily implemented into software library for numerical linear algebra. As
previously explained, the equation Pu = q is an over-determined linear system, since we have
more equations than unknowns. Two of the main methods used to solve this system of equations
for u are:

1. Forward Substitution: Write out the first L equations, which are in upper triangular form,
solve the first equation and sequentially substitute and solve the remaining equations. In this
formalism only the first L equations are used and this questions remains unanswered that what
happens if we involve other existing equations in our calculation. In other words, because
equations are more than unknowns there are multiple solutions to this problem.
2. Least Squares: Using linear algebra, we can derive a formula to calculate the least squares
estimate of the unit hydrograph ordinates as follows:
Pu = q
T T
P Pu = P q
T

−1 T
u= P P P q

where (.)T is the matrix transposition operator, which flips a matrix over its diagonal. Note that
by least squares, we mean that the solution has minimum error variance amount all other
possible solutions.
The above matrix formulation of the linear system can be solved efficiently using a technical
programing language such as MATLAB. Note that, in the above formulation, we multiply both
sides with PT to obtain a square matrix PPT , which might be invertible. We need to note that
P is often a rectangular matrix with more rows than columns (N  L) and can not be inverted.

−1
The matrix PT P PT is called pseudo inverse of P. In MATLAB, computation of the unit
hydrograph can be simply implelmented using eaither of the following commands:
u = P\q or u = pinv(P)*q. The first command is preferred, though.
Unit Hydrograph and Routing Ardeshir Ebtehaj 12
Unit Hydrograph Derivation IV
S-Hydrograph Method
As we mentioned the UH is a time invariant characteristic of a watershed in response to unit
pulse of excess rainfall with an specific duration. However, since the UH is obtained trough a
linear process, we can change a unit hydrograph form one duration to another. To that end, we
need to derive the S-hydrograph based on the principle of superposition.
By definition, the S-hydrograph is the direct runoff hydrograph of a watershed to a continuous
train of pulses of rainfall with duration ∆t.
In other words, the S-hydrograph g(t) = [u(t) + u(t − ∆t) + u(t − 2∆t) + . . .] is the summation
of an infinite number of unit pulse of excess rainfall with duration ∆t as illustrated in Figure 7.

Figure 7: Conceptual representation of the S-hydrograph.

Unit Hydrograph and Routing Ardeshir Ebtehaj 13

Unit Hydrograph Derivation V
As a result, the unit hydrograph for ∆t 0 duration, which must be an integer multiplier of ∆t,
can be obtained as follows:

0

1  0

u ∆t = g(t) − g(t − ∆t )
∆t 0

0
Figure 8: Derivation of the unit hydrograph with duration ∆t = n∆t, where n ∈ N, from an
S-hydrograph that is generated from summation of unit hydrographs with duration ∆t.

Unit Hydrograph and Routing Ardeshir Ebtehaj 14

Unit Hydrograph Derivation VI
Synthetic Unit Hydrographs
Thus far, the methods we have described only apply for the basin with gage measurements of the
streamflow, however, many basins are ungauged. Therefore, engineers have analyzed many unit
hydrographs from gauged basins to obtain a universal functional representation for the unit
hydrograph that can be generalized to ungauges basins. This representation of the unit
hydrograph is often called synthetic unit hydrograph (SUH) that can be applied to ungauged
basins. These synthetic methods typically scale a general UH shape or calculate key ordinates
based on easily determined watershed and excess precipitation characteristics. There are many of
these methods, such as Snyder’s or Clark’s SUH, however, we will cover the very popular NRCS
(SCS) dimensionless synthetic hydrograph method. The dimensionless SCS hydrograph and its
triangular representation are shown in Figure 7 and tabulated in Table 1.

Figure 9: Dimensionless curvilinear and equivalent triangular SCS unit hydrographs (USDA, 1986)

Unit Hydrograph and Routing Ardeshir Ebtehaj 15

Unit Hydrograph Derivation VII

Figure 10: Ratios for dimensionless SCS synthetic unit hydrograph and mass curve.

Unit Hydrograph and Routing Ardeshir Ebtehaj 16

Unit Hydrograph Derivation VIII
The SCS hydrograph is defined by three parameters: (1) Time of concentration, (2) Time to
Peak, and (3) Peak Discharge.

Time of Concentration

As we previously discussed the time of concentration (tc ) for a watershed is the time for
a water parcel to travel from the farthest point in the watershed to the outlet. The SCS
recommends two methods for calculation of tc :
B The lag method: defines the time lag as the time in hours from the center of the
mass of the rainfall excess to the peak discharge, which can be computed as follows
L0.8 (S + 1)0.7
tL =
1900Y 0.5
where Y is the average slope of the watershed in %, L denotes the hydraulic length
of the main channel in feet, and S represents the potential maximum retention. The SCS
method suggests the following relationship for the concentration time
5
tc =
tL
3
B The upland or velocity method defines the concentration time as follows:
L
tc = [hr]
3600V
where L and V are hydraulic length and velocity in feet and ft/s, respectively. The
velocity of the upland method can be obtained from Figure 8 based on the average slope
of the watershed.

Unit Hydrograph and Routing Ardeshir Ebtehaj 17

Unit Hydrograph Derivation IX
Clearly, if we discretized the the stream into k segments, the concentration time is the sum of
travel time for different stream segments as we discussed before,
1 k Li
tc = Σ [hr]
3600 i=1 Vi

Figure 11: Velocities for upland method for estimation of tc (USDA, 1986).

Unit Hydrograph and Routing Ardeshir Ebtehaj 18

Unit Hydrograph Derivation X
Time to Peak

Time to peak (tp ) is the time from the beginning of the rainfall to the time of the peak
discharge, which can be expressed as a function of the time lag tL as follows:
tR
tp = + tL
2
where tp is in hours, tR is the duration of the rainfall excess in hours, and tL is the lag
time in hours. When tR is not known, the SCS method suggests
tR = 0.133tc
and because tL = 0.6tc , one can obtain the following the following relationship for
approximating the time to peak:
0.133tc
tp = + 0.6tc = 0.67tc
2

Peak Discharge

From continuity equation, we know that the area under the unit hydrograph is equal to
the volume of the excess unit rainfall—example 1 [cm] or 1 [in] all over the watershed.
Here, the unit of qp in SCS method is the unit of discharge per depth of rainfall excess
such as [ m3 s-1 cm-1 ] or [ cfs in-1 ].

Unit Hydrograph and Routing Ardeshir Ebtehaj 19

Unit Hydrograph Derivation XI
Peak Discharge (continued)

Therefore base on conservation of mass of the depth of the rainfall and the triangular
shape of the proposed SCS unit hydrograph, we have
1 qp
1= (tp + tr ).
2 A
which can be rearranged as  
2 A
qp =
1 + tr /tp tp

For the SI unit, in the above equation the area should be in [m2 ] and time is in second.
To convert the above formula to be used for watershed area in square kilometers and
hourly time scale, we need to to multiply the above equation by 104 /3600. Moreover,
the SCR method suggests tr = 1.67tp and thus
cA
qp =
tp
where c = (10000/3600) × (2/2.67) = 2.08. The c value in English unit is 483.4, where
A should be in square miles.

Unit Hydrograph and Routing Ardeshir Ebtehaj 20

Unit Hydrograph Derivation XII
Construction of SCS synthetic UH:

1: Compute the time of concentration (tc ) using the lag method or the upland-velocity
method.
2: Compute the time to peak (tp = tR /2 + 0.6tc ) and then the peak discharge qp = cA/tp .
3: Compute the the base time of the SUH as tb = tp + tr , which is tb = 2.67tp for
the triangular hydrograph and tb = 5tp for the curvilinear one and the recession time
tr = tb − tp .
4: Compute the SHU ordinates and plot them. For the the triangular SUH, the ordinates can
be directly computed while for the curvilinear one the dimensionless ratios (t/tp , q/qp )
in Figure 8 shall be used.

Unit Hydrograph and Routing Ardeshir Ebtehaj 21

Flow Routing I
Flow routing is the procedure of determining the hydrograph at a point from hydrographs at one
or more points upstream. As water enters a channel, either from hillslopes or upstream channels,
the flow velocity shifts the location of the peak flow and surface roughness and storages along the
water path attenuate the magnitude of the peak. The flow routing is referred to a class of
methods that calculates the redistribution (widening) and translation of upstream hydrograph as
a function as the water flows downstream.

Figure 12: Left: Schematic of redistribution and translation from Chow et al. (1988). Center: The
storage accumulation and release performed by routing. Right: Conceptualization of channels as
storage reservoirs (Credit: Th COMET Program)

Unit Hydrograph and Routing Ardeshir Ebtehaj 22

Flow Routing II
There are two main types of routing:
1. Lumped (Hydrologic) Routing: Flow is calculated at a particular point as a function of time
alone.
2. Distributed (Hydraulic) Routing: Flow is calculated as a function of time and space.
For this course, we focus on the simpler lumped modeling.
Within lumped modeling, there are multiple methods depending on what system we are trying to
model. We are going to focus on two common methods for flow routing in reservoirs (level-pool)
and stream channels (Muskingum).
Derivation of routing methods begins with the continuity equation:
dS(t)
= I(t) − Q(t)
dt
In routing, I(t) is known inflow hydrograph, while S(t) and Q(t) are unknown. Just as in our
discussion of linear reservoirs, we need a storage-discharge relationship, S = f (Q).
There are two types of storage-discharge relationships: (1) invariable reservoirs and (2) variable
reservoirs. The invariable are those in which there is a one-to-one relationship between the
outflow and the reservoir storage. However, for variable reservoirs, the relationship between the
outflow and storage capacity is not unique and follows a loop when the reservoir is filled or
emptied. This nonlinear relationship mostly occurs in long and narrow reservoirs where the water
velocity is significant and produces backwater effects in the reservoir.

Unit Hydrograph and Routing Ardeshir Ebtehaj 23

Flow Routing III

Figure 13: Invariable reservoir (left) that is wide and deep compared to its length. The flow velocity
is low and the water surface is almost horizontal. A variable reservoir (right) is a long, narrow and
shallow reservoir with high flow velocity such that the water surface profile is markedly curved due
to the backwater effect.
Level Pool Routing
Level pool routing typically applies to man-made hydraulic structures like reservoirs, dams and
stormwater ponds common in many watersheds. The key assumption is a horizontal water surface
and negligible velocity in the reservoir. This allows an invariable relationship between storage
(water surface height) and discharge, because any change in storage results in a change in the
uniform water surface of the reservoir and in turn height of water above the flow control devices.
Beginning with the continuity equation dSdt = I(t) − Q(t) and thus dS = I(t)dt − Q(t)dt, one
can discretize it as follows assuming that the changes in inflow and outflow are linear over over a
small timestep, 4t:

Unit Hydrograph and Routing Ardeshir Ebtehaj 24

Flow Routing IV

Ij + Ij+1 Qj + Qj+1
Sj+1 − Sj = 4t − 4t
2 2

Figure 14: Discretization of the reservoir continuity equation over a small time interval.

I(t) is known at all times while initial storage (Sj ) and outflow (Qj ) are specified as initial
conditions, so we can re-arrange the equation into knowns and unknowns as follows:

Level Pool Routing Equation

h i h i
2Sj+1 2Sj
+ Qj+1 = Ij + Ij+1 + − Qj (1)
4t 4t

Unit Hydrograph and Routing Ardeshir Ebtehaj 25

Flow Routing V

Figure 15: Photos of a reservoir, dam and a pond with hydraulic structures as flow controls

To use this equation,we need the elevation-storage and elevation-outflow relationships of the
reservoir to create a storage-outflow function S = f (Q). The elevation-storage function S = f (H)
is simply constructed from the geometry of the reservoir or pond. The elevation-discharge
Q = f (H) is slightly more difficult to obtain, however, the outflow of storage facilities is often
controlled by hydraulic structures such as sluice gates, weirs, pipes, and orifices which typically
have well-defined equations based on water head (See Figure 15).

For example, for an uncontrolled ogee crest

spillway the outflow as a function of total
water head on the crest is:
3/2
Q = C0 LH
L : is the effective length of the crest
C0 : is the discharge coefficient
Q : is discharge in [cfs].

Figure 16: Discharge relationship for vertical-faced ogee crest (US

bureau of reclamation 1987). The figure is for English system and ha
is the water surface upstream from weir drawdown.

Unit Hydrograph and Routing Ardeshir Ebtehaj 26

Flow Routing VI
Once the storage-outflow function is calculated, as schematically shown in Figure 17, one can
calculate the RHS of equation 1 and look up the LHS from the aforementioned table.
Additionally, there are more accurate methods to calculate level-pool routing, such a Runge-Kutta
methods, if the assumption of inflow and outflow varying linearly at small timesteps is not
acceptable.

Figure 17: Change of storage during a routing period and development of the storage-outflow
function for level pool routing on the basis of storage-elevation-outflow curves (Chow et al. 1988).

Unit Hydrograph and Routing Ardeshir Ebtehaj 27

Flow Routing VII
Hydrologic River Routing (Muskingum Method)
In order to calculate the routing for streams in a watershed, we can no longer assume that the
water surface is level, since there is not negligible velocity in a stream and there are backwater
effects. Therefore, we need to define a different storage-outflow equation, S = f (Q).
The Muskingum method assumes that storage in a channel is the sum of a prism and a wedge
storage. The prism storage is a linear reservoir defined as S = KQ where K is a constant
representing approximately the travel time through the reach. The wedge storage is defined as
S = KX (I − Q), where X is a weighting factor ranging from 0 to 0.5 and thus,
S = KQ + KX (I − Q) = K (XI + (1 − X )Q)
This wedge storage is representing a flood wave as it propagates through a stream channel as
shown in Figure 18.

Figure 18: Schematics of wedge storage and floodwave propogation assumed in Muskingum method
(Credit: The COMET Program)

Unit Hydrograph and Routing Ardeshir Ebtehaj 28

Flow Routing VIII
We can discretize this storage equation as follows
Sj = K (XIj + (1 − X )Qj ) Sj+1 = K (XIj+1 + (1 − X )Qj+1 )
and thus:
Sj+1 − Sj = K {[(XIj+1 + (1 − X )Qj+1 ] − [XIj + (1 − X )Qj ]} (2)
Recall we previously defined the change in storage as:
Ij + Ij+1 Qj + Qj+1
Sj+1 − Sj = 4t − 4t (3)
2 2
Combining equations 2 and 3 one can obtain the following flood routing equation:

Muskingum Equation for Stream Routing

Qj+1 = C1 Ij+1 + C2 Ij + C3 Qj

4t − 2KX 4t + 2KX 2K (1 − X ) − 4t
C1 = C2 = C3 =
2K (1 − X ) + 4t 2K (1 − X ) + 4t 2K (1 − X ) + 4t

where C1 + C2 + C3 = 1. Typically, the streams being analyzed are broken up into N subreaches
when performing this calculation for stability purposes. The U.S. Army Corps of Engineers
recommends the following condition to ensure solution stability:
1 K 1
≤ ≤
2(1/X ) N4t 2X

Unit Hydrograph and Routing Ardeshir Ebtehaj 29