Advanced Hydrology (HWRE 6031)

Unit Hydrograph

By : Mulu Sewinet
Debre Markos, University
 2.3 Unit Hydrograph
• Different methods have been intensively investigated
for the prediction of flood peak discharges and
discharge hydrographs from rainfall events since the
early 1930s.
• One approach receiving considerable use is called the
unit hydrograph method
• The purpose of this chapter is to:
• Define unit hydrographs and show their utility in
hydrologic studies and design.
• Develop fully the current methods of obtaining,
analyzing, and synthesizing unit hydrographs.
• Present methods for converting unit hydrographs for
one storm duration to other storm durations.
 UNIT HYDROGRAPH contd…..
• The concept of a unit hydrograph was first introduced by
Sherman in 1932. He defined a unit graph as follows:
• The unit hydrograph (originally named unit-graph) of a
watershed is defined as a direct runoff hydrograph (DRH)
resulting from 1 in. (usually taken as 1 cm in SI units) of
excess rainfall generated uniformly over the drainage area at
a constant rate for an effective duration.
• Sherman classified runoff into surface runoff and
groundwater runoff (base flow) and defined the unit
hydrograph for use only with surface runoff.
• Unit hydrograph represents the transformation of a unit
depth of rainfall excess of duration T-h in to a unit depth of
DR and hence is a catchment characteristics as well as the
duration of excess rain.
• The unit hydrograph is a simple linear model that can be
used to derive the hydrograph resulting from any amount of
excess rainfall.
 Unit Hydrograph Contd……
• The term "unit" has to do with the net rain amount of
1.0 inch and does not mean to imply that the duration
of rain that produced the hydrograph is one unit,
whether an hour, day, or any other measure of time.
• The storm duration, X, that produced the unit
hydrograph must be specified because a watershed has
a different unit hydrograph for each possible storm
duration.
• An X-hour unit hydrograph, is defined as a direct runoff
hydrograph having a 1.0 in. volume and resulting from
an X-hour storm having a net rain rate of 1/X in/hr.
• A 2-hr unit hydrograph would have a 1.0-in. volume
produced by a 2-h storm, and a 1-day unit hydrograph
would be produced by a storm having 1.0 in. of excess
rain uniformly produced during a 24-hr period.
Unit Hydrograph Contd……
 Assumptions of Unit Hydrograph (UH)
The following basic assumptions are inherent in this model:
1) The excess rainfall has a constant intensity within the
effective duration (The storms selected for analysis should be of short
duration, produce an intense and nearly constant excess rainfall rate,
yielding a well-defined single-peaked hydrograph of short time base).
2) The excess rainfall is uniformly distributed throughout the
whole drainage area (The unit hydrograph may become inapplicable
when the drainage area is too large to be covered by a nearly uniform
distribution of rainfall. In such cases, the area has to be divided and each
subarea analyzed for storms covering the whole subarea).
3) The base time of the DRH (the duration of direct runoff)
resulting from an excess rainfall of given duration is constant
(The base time is usually short if the direct runoff is considered to include
the surface runoff only; it is long if the direct runoff also includes
subsurface runoff.).
4) The ordinates of all DRH's of a common base time are directly
proportional to the total amount of direct runoff represented
by each hydrograph (Principles of superposition and proportionality
holds for the catchment).
5) For a given watershed, the hydrograph resulting from a given
excess rainfall reflects the unchanging characteristics of the
watershed (The unit hydrograph is considered unique for a given
watershed and invariable with respect to time. This is the principle of time
Unit Hydrograph Contd……
The principles of superposition and
proportionality; together with principle of time
invariant is fundamental to the unit hydrograph
model.

Unit hydrographs are applicable only when channel

conditions remain unchanged and watersheds do
not have appreciable storage.

This condition is violated when the drainage area

contains many reservoirs, or when the flood
overflows into the flood plain, thereby producing
considerable storage.
 Unit Hydrograph Contd……
• Law of proportionality: If a solution f(Q) is multiplied by a
constant c, the resulting function cf(Q) is also a solution. For
example, if 6.5 mm of effective rainfall fall on a catchment area
in T h, then the hydrograph resulting from that effective
rainfall is obtained by multiplying the ordinates of the TUH by
6.5
• Law of Superposition: If two solutions f1(Q) and f2(Q) of the
equation are added, the resulting function f1(Q) +f2(Q) is also a
solution of the equation. The total hydrograph of direct runoff
due to n successive amounts of effective rainfall (for instance
R1 and R2) is equal to the sum of the n successive hydrographs
produced by the effective rainfall (the latter lagged by T h on
the former).
• Time invariance assumption: assumes that the effective
rainfall-surface runoff relationship does not change with time,
i.e., the same TUH always occurs whenever the unit of
effective rainfall in T h is applied on the catchment.
 Unit Hydrograph Estimation
1) From simple Rainfall excess events
a) When the hydrograph of DR is known along with
information about the rainfall excess event
Procedure
i. From the rainfall hyetograph estimate the intensity of
precipitation and the duration of the event
Let the intensity be I cm/hr and the duration of storm
T hr
ii. Calculate depth of precipitation as I*T =k cm.
iii. Divide each ordinate of the DRH by k. The resulting
hydrograph of direct runoff will now correspond to a
depth of 1cm and duration T hrs.
Unit Hydrograph Estimation Contd……
b) Hydrograph of direct runoff and the area of the
catchment is known
Procedure
i) From the hydrograph of DR calculate the total
volume of DR by direct numerical integration let
this value be V.
ii) Divide the estimated volume of DR V by the
catchment area A to get the depth of rainfall
excess K
i.e K= V/A
iii. Divide each ordinate of DRH by k to get a
hydrograph of direct runoff now corresponding to
a depth of D.R equal to 1cm.
 Derivation of Unit Hydrograph
Steps in developing a unit hydrograph for an isolated storm are:
• Separate the base flow from the total runoff
• From the ordinates of the total runoff hydrograph deduct the corresponding
ordinates of base flow; to obtain the ordinates of DR
• Divide the ordinates of the direct runoff hydrograph by total direct runoff
volume in inches, and plot these results versus time as a unit graph for the
basin
• The surface runoff volume is determined as a depth of flow by numerical
integration: i.e
Where, d = depth of surface runoff in mm; Δt = uniform time interval in hours
at which the ordinates of the surface runoff are measured; ΣQ = sum of all
ordinates of surface runoff hydrograph in m3/s; A = catchment area in Km2
• Divide the ordinates of the surface runoff hydrograph by the runoff depth d
due to the ordinates of the unit hydrograph.
• Finaly , the unit hydrograph for effective rainfall of duration T, the TUH, is
plotted,
Example: Derive a unit hydrograph for the 1715ac drainage area by
using the total direct runoff hydrograph given in Fig. below.
 Solution
1. Separate the base or groundwater flow to get the total
direct runoff hydrograph. A common method is to draw a
straight line AC that begins when the hydrograph starts an
appreciable rise and ends where the recession curve
intersects the base flow curve.
2. The depth of direct runoff over the watershed is calculated
using

Where DR is the average height of the direct runoff ordinated

using a chosen time period ∆𝑡 (in this case ∆𝑡 : 1. 0 hr) .
3. Compute ordinates of the unit hydrograph by using
𝑄
𝑄𝑢 = 𝑠 where D= 1.415
𝐷
Where 𝑄𝑠 =the magnitude of a hydrograph ordinate of direct
runoff
𝑄𝑢 =the ordinate of the unit hydrograph having a volume of 1
in. at some instant of time
Table: determinatoin of a 2 .hr unit hydrograph from an isolated
storm
Unit hydrographs for different rainfall durations

• When a unit hydrograph of a given excess-rainfall

duration is available, the unit hydrographs of other
durations can be derived.

• If other durations are integral multiples of the given

duration, the new unit hydrograph can be easily
computed by application of the principles of
superposition and proportionality.
Example: Develop hourly ordinates of the total hydrograph resulting from
a 4-hr design storm having the following excess amounts

Table: Discharge of rates for the 2-hr unit hydrograph

Solution. The 4-hr duration of the design storm is an integer

multiple of the unit hydrograph duration
• The first 2-hr storm segment has 1.0 in. of net rain and thus
reproduces a unit Hydrograph
• The second 2 -hr storm segment has 2.0 in. of net rain (in 2
hr);thus its ordinates are twice those of a 2-hr unit
hydrograph.
Figure derivation of total runoff hydrograph using a 2-hr unit hydrograph
 Applications
1) Given a T-hr Unit hydrograph a DRH for an ER of T-hr
duration and depth other than unity can be obtained by
simply multiplying the unit hydrograph ordinates by the
indicated ER depth (Proportionality). Time bases of all
DRHs obtained in this way is equal to that of the UHG
2) DRHs produced by a storm consisting of succession of
effective rainfall events of the same duration but varying
depths can be estimated by
i) Scaling up/down of the U.H ordinates for the catchment, in
proportion to the depth of ER of each event
ii) Lagging the “scaled unit hydrograph” by appropriate
amounts in conformity with the causal nature
iii) Summation of simultaneous occurring ordinates is yield the
DRH for the composite storm
 UH for different rainfall durations contd….
• Thus the composite DRH ordinates
are a linear composition of the unit
hydrograph ordinates, while the
composite hydrograph time base is
the sum of the unit hydrograph
time base plus composite storm.
 2.4 S.Hydrograph/ S-curve Method
• Defined as the response of
the basin or the DRH for a
basin receiving rainfall excess
uniformly over its entire area
for an indefinitely long
duration and at unit intensity
(1cm/hr)
• It may be depicted as follows
• A short duration rainfall
excess generally result in a
normal single peak DRH
(curve B)
• Rainfall excess for extended
durations results in DRH
which increases with time
until it reaches a constant
equilibrium value Qe. This is
depicted by curve A
 S.HYDROGRAPH contd……
• Point of inflation on curve
A: time at which entire
drainage area contributes to
flow.
• At this point Max storage of
watershed is only partially
complete as rain continues,
max storage is reached after
which discharge Qe remains
constant for catchment area
A km2 and intensity of
excess rainfall i cm/hr
1∗10−2 𝑚3
• 𝑄𝑒 = ∗𝐴∗ 106 = 2.778𝐴
3600 𝑠
S.HYDROGRAPH contd……
 S- Curve hydrograph contd…….
Steps
• Construct s-curve from the given UH of known time D-
hrs

• Advance or offset the position for the desired duration

Do hrs

• Find the difference of the ordinates of the original S-

curve and offset the hydrograph

• Divide this difference by (Do/D) to get the ordinates of

the new UH of Do hrs duration
 Example S- Hydrograph
• Given the following 2-hr hydrograph
Time (hrs) 0 2 3 4 5 6
Q(cfs) 0 100 250 200 100 50

• Solution. The 2-hr unit hydrograph is

the runoff from a 2-hr storm of 0.5
in./hr. The S-hydrograph is formed
from a net rain rate of 0.5 in./hr
lasting indefinitely as shown in Fig. Its
ordinates are found by adding the 2-hr
unit-hydrograp runoff rates from each
contributing 2 -hr block of rain.
 S- Hydrograph contd……
• To find a 3-hr hydrograph, Time 1st 2-hr 2nd 2-hr 3rd 2-hr 4th S-
(hr) hydrograph
the S-curve is lagged by 3-hr
0 0 0
and subtracted as shown in 1 100 100
2 250 0 250
Fig.
3 200 100 300
• This results in a hydrograph 4 100 250 0 350
5 50 200 100 350
from a 3-hr storm of 0.5 6 0 100 250 0 350

in./hr, or 1.5 in. total. Thus 7 50 200 10 350

0
the ordinates need to be 8 0 100 25 350
0
divided by 1.5 to produce
the 3-hr unit hydrograph:
S- Hydrograph contd……

Time S-hydrograph Lagged s- difference 3-hr unit

(hr) hydrograph hydrograp
h
0 0 0 0
1 100 100 67
2 250 250 167
3 300 0 300 200
4 350 100 250 167
5 350 250 100 67
6 350 300 50 33
7 350 350 0 0
 2.5 Synthetic Unit Hydrograph
All the methods of deriving a UH discussed previously
require observations of rainfall and runoff.
However, for some drainage basins rain gages and/ or
streams flow gages are not, therefore, rainfall – runoff
data are not available.
For those basins, some kind of techniques to generate
UH (without using rainfall – runoff records) is needed.
The UH so derived is called synthetic Unit Hydrograph.
Three types of synthetic UHs:
1. Those relate hydrograph characteristics (peak discharge,
time base, time-to-peak, etc.) to basin characteristics
(Snyder, 1938; Gray, 1961)
2. Those based on dimensionless UH ((Soil Conservation
Service, 1972)
3. Those based on models of watershed storage (Clark,
1943).
 Synthetic-Unit Hydrograph (Synder,1938)

• To develop a unit hydrograph to a catchment,

detailed information about the rainfall and the
resulting flood hydrograph are needed.
• But, such information would be available only
at a few locations; or the data would normally
be very scanty.
• In order to construct a unit hydrographs for
such areas, empirical equations of region
validity which relates the salient hydrographs
characteristics to the basin characteristics are
available.
 Synder method contd….
• Snyder (1938), based on a study of a large number of
catchments in the Eastern United States developed a set of
empirical equations for synthetic unit HG in those areas.
• These equations are in use in the USA, and with some
modifications in many other countries.
• Snyder defined a standard unit hydrograph as one whose
rainfall duration tr is related to the basin lag tp by

• The first of the Snyder's equation relates the basin lag tp,
defined as the time interval from the mid point of the unit
rainfall excess to the peak of the unit hydrographs
 Synder method contd….
For a standard unit hydrograph he
found that:
1. The basin lag is:
• Where tp is in hours, L is the
length of the main stream in
kilometers (or miles) from the
outlet to the upstream divide, Lc
is the distance in kilometers
(miles) from the outlet to a
point on the stream nearest the
centroid of the watershed area,
C1 = 0.75 (1.0 for the English
system), and Ct is a coefficient
derived from gaged watersheds
in the same region.
Synder method contd….
 Synder method contd….
• But, Linsley et al. found that the basin lag tp is
better correlated with the catchment parameter

and S = basin slope and the above equation

was modified as

• Where, CtL and n (~0.3) are basin constants

Synder method contd….

2) The peak discharge per unit drainage area in m3/s.km2

(cfs/mi2) of the standard unit hydrograph is

 where C2 = 2.75 (640 for the English system) and Cp is a

coefficient derived from gaged watersheds in the same
region.
• To compute Ct and Cp for a gaged watershed, the values of
L and Lc are measured from the basin map.
• From a derived unit hydrograph of the watershed are
obtained values of its effective duration tR in hours, its
basin lag tpR in hours, and its peak discharge per unit
drainage area, qpR in m3/s.km2.cm (cfs/mi2 .in for the
English system).
• If tpR = 5.5tR , then tR =tr , tpR = tp , and qpR = qp , and Ct and
Cp are computed by the above Eqs.
Synder method contd….
• If tpR is quite different from 5.5tR , the standard
basin lag is

• When an ungagged watershed appears to be

similar to a gaged watershed, the coefficients Ct
and Cp for the gaged watershed can be used in the
above equations to derive the required synthetic
unit hydrograph for the ungagged watershed.
3) The relationship between qp and the peak
discharge per unit drainage area qpR of the
required unit hydrograph is
 Synder method contd….
4) The base time tb in hours of the unit hydrograph can be
determined using the fact that the area under the unit
hydrograph is equivalent to a direct runoff of 1 cm.
Assuming a triangular shape for the unit hydrograph, the
base time may be estimated by

Where C3 = 5.56 (1290 for the English system).

5) The width in hours of a unit hydrograph at a discharge
equal to a certain percent of the peak discharge qPR is
given by

where Cw = 1.22 (440 for English system) for the 75-percent

width and 2.14 (770, English system) for the 50-percent
width.
 Synder method contd…..
• Example: From the basin map of a given watershed,
the following quantities are measured: L = 150 km, Lc =
75 km, and drainage area = 3500 km2. From the unit
hydrograph derived for the watershed, the following
are determined: tR = 12 h, tpR = 34 h, and peak
discharge = 157.5 m3/s-cm. Determine the coefficients
Ct and Cp for the synthetic unit hydrograph of the
watershed.
2.6 RESPONSE FUNCTIONS OF HYDROLOGIC SYSTEMS
• The amount of water stored in a hydrologic system such as
reservoir, S may be related to the rates of inflow (I) and
outflow (Q) by the integral equation of continuity. i.e

𝑑𝑠
=𝐼−𝑄 (2.6.1)
𝑑𝑡

 The amount of storage rises and falls with time in response

to (I) and (Q) and the amount of storage at any time can be
expressed by a storage function as:

𝑑𝐼 𝑑 2 𝐼 𝑑𝑄 𝑑 2 𝑄
𝑠=𝑓 𝐼, , 2 , … … … . . , 𝑄, ,…. 2.6.2
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 2

 The continuity equation (2.6.1) and the storage function

equation (2.6.2) solved simultaneously by integral, approach
where the input I and output Q are both functions of time.
 Response Functions Systems contd…
𝑑𝑄 𝑑2 𝑄 𝑑 𝑛−1 𝑄 𝑑𝐼 𝑑2 𝐼
𝑠 = 𝑎1 𝑄 + 𝑎2 + 𝑎3 2 + ⋯+ 𝑎𝑛 𝑛−1 + 𝑏1 𝐼 + 𝑏2 + 𝑏3 2 +
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑 𝑚−1 𝐼
⋯ + 𝑏𝑚 𝑚−1 (2.6.3)
𝑑𝑡
 where a1, a2, a3…an, b1,b2, b3, ….bm are constants and
derivatives of higher order and also time invariant
𝑑𝑠
• Differentiating eqn. (2.6.3), substituting the result for
𝑑𝑡
in eqn. (2.6.1), and rearranging yields
𝑑𝑛 𝑄 𝑑 𝑛−1 𝑄 𝑑2 𝑄 𝑑𝑄 𝑑𝐼
𝑎𝑛 𝑛 + 𝑎𝑛−1 𝑛−1 + ⋯ + 𝑎2 2 + 𝑎1 + 𝑄 = 𝐼 − 𝑏1 −
𝑑𝑡
2
𝑑𝑡 𝑚−1
𝑑𝑡𝑚 𝑑𝑡 𝑑𝑡
𝑑 𝐼 𝑑 𝐼 𝑑 𝐼
𝑏2 2 − ⋯ − 𝑏𝑚−1 𝑚−1 − 𝑏𝑚 𝑚 (2.6.3)
𝑑𝑡 𝑑𝑡 𝑑𝑡
Equation 2.6.3 is written in compact form as
Response Functions Systems contd…
•𝑁 𝐷 𝑄=𝑀 𝐷 𝐼
𝑑
• Where 𝐷 = and N D and M D are the
𝑑𝑡
differential operators
𝑑𝑛 𝑑 𝑛−1 𝑑
𝑁 𝐷 = 𝑎𝑛 𝑛 + 𝑎𝑛−1 𝑛−1 + ⋯+ 𝑎1 +1
𝑑𝑡 𝑑𝑡 𝑑𝑡
and
𝑑𝑚 𝑑 𝑚−1 𝑑
𝑀 𝐷 = 𝑏𝑚 𝑚 + 𝑏𝑚−1 𝑚−1 + ⋯+ 𝑏1 +1
𝑑𝑡 𝑑𝑡 𝑑𝑡
• For Q yields
𝑀 𝐷
𝑄 𝑡 = I(t)
𝑁 𝐷
• The function M(D)/N(D) is called the transfer function
of the system; it describes the response of the output
to a given input sequence.
 Response Functions of Linear systems
• The solution for the transfer function of hydrologic
systems follows two basic principles for linear system
operations
1. Principle of proportionality: If a solution f(Q) is
multiplied by a constant c, the resulting function cf(Q)
is also a solution
2. Principle of superposition: If two solutions f1(Q) and
f2(Q) of the equation are added, the resulting function
f1(Q) + f2(Q) is also a solution of the equation
• The particular solution adopted depends on the input
function and on the specified initial conditions or values
of the output variables at t = 0.
 2.7 Impulse Response Function
The response of a linear
system is uniquely
characterized by its impulse
response function.
If a system receives an input
of unit amount applied
instantaneously (a unit
impulse) at time τ, the
response of the system at a
later time t is described by
the unit impulse response
function u(t — τ); t — τ is
the time lag since the
impulse was applied [Fig.a]. Fig.2.7.1 (a) Unit impulse response function
Impulse Response Function contd….
Example of a response to an impulse:
The response of a guitar string when it is plucked
The response of the shock absorber in a car after
the wheel passes over a pothole.
If the storage reservoir is initially empty, and then
the reservoir is instantaneously filled with a unit
amount of water, the resulting outflow function
Q(t) is the impulse response function.
 Impulse Response Function contd….
• Example on superposition
principles of linear system
operation
• if two impulses are applied,
one of 3 units at time τ1
and the other of 2 units at
time τ2 the response of
the system will be 3u(t -τ1 )
+ 2u(t -τ2 ), as shown in Fig.
(b). Fig. 2.7.1 (b) The response to two impulses is fou
by summing the individual response functions.
 Impulse Response Function contd….
• The amount of input entering the system between times τ
and τ + dτ is I(τ) dτ.
• For example, if I(τ) is the precipitation intensity in inches
per hour and dτ is an infinitesimal time interval measured in
hours, then I(τ)dτ is the depth in inches of precipitation
input to the system during this interval.
• The direct runoff t-τ time units later resulting from this
input is I(τ)u(t -τ)dτ.
• The response to the complete input time function I(τ) can
then be found by integrating the response to its constituent
impulses:
𝑡
•𝑄 𝑡 = 0
𝐼 τ 𝑢 𝑡 − τ 𝑑τ (2.7.1)
• This expression, called the convolution integral, is the
fundamental equation for solution of a linear system on a
continuous time scale
Impulse Response Function contd….

Fig. 2.7.2 The relationship between continuous and discrete convolution.

Impulse Response Function contd….
• For most hydrologic applications, solutions are
needed at discrete intervals of time, because the
input is specified as a discrete time function, such as
an excess rainfall hyetograph.
• To handle such input, two functions namely
1. The unit step response function and
2. The unit pulse response function are needed,
1.Step Response Function
 A unit step input is an input that goes from a rate of 0
to 1 at time 0 and continues indefinitely at that rate
thereafter [Fig. 2.7.3 (b)].
Impulse Response Function contd….
• The output of the system, its unit step response
function g(t) is found from (eqn 2.7.1) with I(τ) = 1
for 𝜏 ≥ 0, as
𝑡
•𝑄 𝑡 =𝑔 𝑡 = 0
𝑢 𝑡 − τ 𝑑τ (2.7.2)
• If the substitution Ɩ= t -τ is made in eqn. (2.7.2)
then dτ = -dƖ, the limit τ= t becomes Ɩ= t — t = 0,
and the limit τ = 0 becomes Ɩ= t - 0 = t. Hence,
0
𝑔 𝑡 =− 𝑡
𝑢 Ɩ 𝑑Ɩ
𝑡
𝑔 𝑡 = 𝑢 Ɩ 𝑑Ɩ (2.7.3)
0
 Impulse Response Function contd….
2. Pulse Response Function
• A unit pulse input is an input of unit amount occurring in
duration Δ𝑡. The rate is I(τ) = 1/ Δ𝑡, 0 < τ < Δ𝑡, and zero
elsewhere.
• The unit pulse response function produced by this input
can be found by the two linear system principles.
• By the principle of proportionality, the response to a unit
step input of rate 1/ Δ𝑡 beginning at time 0 is
(1/Δ𝑡)𝑔(𝑡).
• If a similar unit step input began at time Δ𝑡 instead of at
0, its response function would be lagged by time interval
Δ𝑡, and would have a value at time t equal to
(1/Δ𝑡)𝑔(𝑡 − Δ𝑡).
 Impulse Response Function contd….
• Then, by using the principle of superposition, the
response to a unit pulse input duration Δ𝑡 is found by
subtracting the response to a step input of rate 1/ Δ𝑡
beginning at time Δ𝑡 from the response to a step input
of the same rate beginning at time 0, so that the unit
pulse response function h(t) is
1
ℎ 𝑡 = g 𝑡 − g(𝑡 − Δ𝑡 (2.7.4)
Δ𝑡
𝑡 𝑡−∆𝑡
1
ℎ 𝑡 = 𝑢 Ɩ 𝑑Ɩ − 𝑢 Ɩ 𝑑Ɩ
∆𝑡 0 0

𝑡
1
ℎ 𝑡 = 𝑢 Ɩ 𝑑Ɩ (2.7.5)
∆𝑡 𝑡−∆𝑡
 Impulse Response Function contd….

Fig 2.7.3 Response functions of a linear system. The response functions in (a), {b), and (c) are on a
continuous time domain and that in (d) on a discrete time domain.
Impulse Response Function contd….
• Example. Determine the impulse, step and pulse
response functions of a linear reservoir with
storage constant k(S = kQ).
• Solution: Please refer Chow et. al (1988) page
no.(208-209).
2.8 Linear System in Discrete Time
• The impulse, step, and pulse response functions have
all been defined on a continuous time domain.
• Now let the time domain be broken into discrete
intervals of duration ∆𝑡.
• There are two ways to represent a continuous time
function on a discrete time domain, as

a pulse data system or

as a sample data system.

Linear System in Discrete Time contd…
• The pulse data system is used for precipitation and
the value of its discrete input function for the mth
time interval is
𝑚∆𝑡
𝑝𝑚 = (𝑚−1)∆𝑡
𝐼 𝜏 𝑑𝑡 𝑚 = 1,2,3 … . (2.8.1)
• 𝑝𝑚 is the depth of precipitation falling during the
time interval (in inches or centimeters).
• The sample data system is used for streamflow and
direct runoff, so that the value of the system output
in the nth time interval (𝑡 = 𝑛∆𝑡) is
𝑄𝑛 = 𝑄 𝑛∆𝑡 𝑛 = 1,2,3 … … . (2.8.2)
• 𝑄𝑛 is the instantaneous value of the flow rate at the
end of the nth time interval (in cfs or m3/s).
 Linear System in Discrete Time contd…
• The effect of an input pulse of duration ∆𝒕 beginning at
time (m -1) ∆𝒕 on the output at time t = 𝐧∆𝒕 is
measured by the value of the unit pulse response
function h[t - (m -1) ∆𝑡 ] = h[n ∆𝑡 −(m - 1) ∆𝑡 ] = h[(n -
m + 1) ∆𝑡 ],
𝑛−𝑚+1 ∆𝑡
1
ℎ 𝑛 − 𝑚 + 1 ∆𝑡 = 𝑢 Ɩ 𝑑Ɩ (2.8.3)
∆𝑡 𝑛−𝑚 ∆𝑡
• On a discrete time domain,
• The input function is a series of M pulses of constant
rate: for pulse m, I(τ) = Pm/∆𝑡 for (m - 1) ∆𝑡 < τ< m ∆𝑡,
• I(τ) = 0 for 𝜏 > 𝑀∆𝑡.
 Linear System in Discrete Time contd…

• Consider the case where the output is being calculated

after all the input has ceased, that is, at 𝑡 = 𝑛∆𝑡 >
𝑀∆𝑡

• The contribution to the output of each of the M input

pulses can be found by breaking the convolution
integral (2.7.1) at 𝑡 = 𝑛∆𝑡 into M parts:
 Linear System in Discrete Time contd…
𝑛∆𝑡
𝑄𝑛 = 0
𝐼 𝜏 𝑢 𝑛∆𝑡 − 𝜏 𝑑𝜏
𝑝1 ∆𝑡 𝑝2 2∆𝑡
= 𝑢𝑛∆𝑡 − 𝜏 𝑑𝜏 + 𝑢
𝑛∆𝑡 − 𝜏 𝑑𝜏 + ⋯
∆𝑡 0 ∆𝑡 ∆𝑡
𝑝𝑚 𝑚∆𝑡
+ 𝑢 𝑛∆𝑡 − 𝜏 𝑑𝜏 + ⋯
∆𝑡 𝑚−1 ∆𝑡
𝑝𝑀 𝑀∆𝑡
+ 𝑢 𝑛∆𝑡 − 𝜏 𝑑𝜏 (2.8.4)
∆𝑡 (𝑀−1)∆𝑡

𝑝𝑀
• where the terms , m = 1,2, . . . , M, can be brought outside the
∆𝑡
integrals because they are constants.
• In each of these integrals, the substitution Ɩ = 𝑛∆𝑡 − 𝜏 is made,
so dτ = −𝑑𝑙 ,
• the limit τ = 𝑚 − 1 ∆𝑡 becomes Ɩ = 𝑛∆𝑡 − 𝑚 − 1 Δ𝑡 = (𝑛 −
𝑚 − 1)∆𝑡 and the limit 𝜏 = 𝑚∆𝑡 becomes Ɩ = 𝑛 − 𝑚 Δ𝑡
Linear System in Discrete Time contd…

• The nth integral in (2.8.4) is now written

𝑚∆𝑡 (𝑛−𝑚)∆𝑡
𝑝𝑚 𝑝𝑚
𝑢 𝑛∆𝑡 − 𝜏 𝑑𝜏 = −𝑢 Ɩ 𝑑Ɩ
∆𝑡 𝑚−1 ∆𝑡 ∆𝑡 (𝑛−𝑚+1)∆𝑡

𝑝𝑚 (𝑛−𝑚+1)∆𝑡
= 𝑢 Ɩ 𝑑Ɩ
∆𝑡 (𝑛−𝑚)∆𝑡

= 𝑝𝑚 ℎ 𝑛 − 𝑚 + 1 ∆𝑡 (2.8.5)
Linear System in Discrete Time contd…
• By substituting from (2.8.2). After making these
substitution for each term in (2.8.4),
𝑄𝑛 = 𝑝𝑚 ℎ (𝑛∆𝑡) + 𝑝𝑚 ℎ (𝑛 − 1)∆𝑡 + ⋯

+𝑝𝑚 ℎ 𝑛 − 𝑚 + 1 ∆𝑡 + ⋯

+𝑝𝑀 ℎ 𝑛 − 𝑀 + 1 ∆𝑡 (2.8.6)

• It is a convolution equation with input Pm in pulses and

output Qn as a sample data function of time.
 2.9 Discrete Pulse Response Function
• The continuous pulse response function h(t) may be
represented on a discrete time domain as a sample data
function U where
• 𝑢𝑛−𝑚+1 = ℎ 𝑛 − 𝑚 + 1 ∆𝑡 (2.9.1)
• It follows that 𝑢𝑛 = ℎ 𝑛∆𝑡 , 𝑢𝑛−1 = ℎ (𝑛 − 1)∆𝑡 , and
𝑢𝑛−𝑀+1 = ℎ (𝑛 − 𝑀 + 1)∆𝑡 .
• Substituting into (2.8.6) ,
• the discrete time version of the convolution integral is
𝑄𝑛 = 𝑝1 𝑢𝑛 + 𝑝2 𝑢𝑛−1 + ⋯ + 𝑝𝑚 𝑢𝑛−𝑚+1 + ⋯ + 𝑝𝑀 𝑢𝑛−𝑚+1
𝑀

= 𝑝𝑚 𝑢𝑛−𝑚+1 (2.9.2)
𝑚=1
Discrete Pulse Response Function contd….
• Equation (2.9.2) is valid provided n > M;
• if n < M, one would only need to account for the first
n pulses of input, and these are the only pulses that
can influence the output up to time 𝑛∆𝑡 . In this case,
(2.9.2) is rewritten
𝑛

𝑄𝑛 = 𝑝𝑚 𝑢𝑛−𝑚+1 (2.9.3)
𝑚=1
 Discrete Pulse Response Function contd….
• Combining (2.9.2) and (2.9.3) gives the final discrete
convolution equation for a linear system.

𝑛≤𝑀

𝑄𝑛 = 𝑝𝑚 𝑢𝑛−𝑚+1 (2.9.5)
𝑚−1

The notation n < M as the upper limit of the

summation shows that the terms are summed for m =
1 , 2 , . . . , n for n < M, but for n > M, the summation is
limited to m = 1,2,. . . ,M.
 Example
• Suppose there are M = 3 pulses of input: P1, P2 , and
P3.
• For the first time interval (n= 1), there is only one term
in the convolution, that for m = 1;
Fig. Application of the discrete convolution equation to the output from a linear
system.
Application of convolution eqn.
• The discrete convolution equation allows the
computation of direct runoff Qn given excess
rainfall Pm and the unit hydrograph Un-m +1

𝑛≤𝑀

𝑄𝑛 = 𝑝𝑚 𝑢𝑛−𝑚+1
𝑚=1

The reverse process, called deconvolution, is needed to derive a

unit hydrograph given data on Pm and Qn.