Chapter -4

SYNTHETIC UNIT
HYDROGRAPH (SUH)
MODEL
 Chapter 4

SYNTHETIC UNIT HYDROGRAPH (SUH) MODEL

4.1 Introduction
Unit hydrograph (UH) remains as a basic tool in the hands of hydrologist since
Sherman (1932) introduced it to represent the hydrologic response of ungauged
watershed through which effective rainfall is transformed to direct runoff. The UH is a
surface runoff hydrograph resulting from one unit of rainfall excess uniformly
distributed spatially and temporally over the watershed for the entire specified rainfall
excess duration ( Chow1964). The concept of UH has undergone many changes over
time and is termed as instantaneous unit hydrograph (IUH), geomorphologic
instantaneous unit hydrograph (GIUH), synthetic unit hydrograph (SUH) based on the
duration, stream and watershed properties respectively. In a purely ungauged watershed,
the paucity of observed rainfall runoff data sparkled the idea of synthetic unit
hydrograph (SUH) concept that are derived from watershed characteristics rather than
rainfall-runoff data. The examples of some of traditional methods of SUHs as proposed
by Snyder (1938), SCS(2002), Taylor and Schwarz (1952), Gray (1961), Espey (1974)
are available to hydrologists and these are region specific, simple, easy for development
and requires less data.

Due to similarity in the shape of statistical distributions and conventional unit

hydrograph, several attempts have been made by different researches to derive the SUHs
by using the probability distribution functions (PDFs). Use of probability distribution
function removes the subjectivity of manual fitting of the points other than the salient
points of unit hydrograph. Nash (1959) and Dooge (1959) developed the general equations
of the instantaneous unit hydrograph (IUH) in the form of two-parameter Gamma
distribution. Later on, two parameter (2p) Gamma, three parameter (3p) Beta distribution
have been successfully utilized in deriving SUH for Turkey and Indian catchments
(Haktanir and Sezen 1990, Bhunya et al. 2004) respectively and the parameters of these
distributions are derived by the least square approach, non dimensional approach along
with suitable optimization techniques or any other suitable criteria. Jena and Tiwari (2006)
modeled the parameters of SUH using the geomorphologic parameters (channel and basin
parameters) for two watersheds of India deriving the data from Survey of India maps using
GIS techniques. They presented a nonlinear model correlating UH parameters with
geomorphologic parameters having higher degree of correlations with UH parameters.
Rosso (1984) adopted a parametric approach to define the form of IUH by relating the
Nash model parameters with the Horton ratios used by Rodrigues-Iturbe and Valdes
(1979). Bhunya et al. (2008) further extended Rosso (1984) and derived expression for
two-parameter Weibull distribution in describing the shape of the IUH and made a
comparison between two-parameter Weibull distribution and two parameter gamma
distribution using field data.

4.2 Synthetic Unit Hydrograph (SUH)

Hydrologic models require rainfall-runoff data. These models are not applicable to
ungauged watersheds that lack such data. In case of conceptual models, it requires either
rainfall-runoff data or the physically measurable catchment characteristics in order to
determine the parameters of the model. The model parameters are determined through an
optimization process subjective to an objective function. These are then employed to
predict runoff from another rainfall event that is not used in the optimization. Other
approaches include that establish linear or nonlinear relationships between model
parameters and physically measurable watershed characteristics. These relationships are
then assumed to hold good for other ungauged watersheds having similar hydrologic
characteristics. So synthesizing rainfall-runoff relationships between individual runoff
hydrograph and the watershed characteristics have given birth to the idea of Synthetic Unit
Hydrograph (SUH) those are most appropriately applicable to ungauged catchments.

Some of the traditional methods of SUHs can be referred chronologically are the
work of Snyder (1938), Taylor and Schwarz (1952), SCS (1957), Gray (1961), Espey (
1974) and a few later on. Their simplicity and ease in development has popularized these
synthetic or artificial unit hydrographs because of lesser data requirement and for giving a
smooth and unique curve corresponding to one unit runoff volume, which is essential for
unit hydrograph derivation. These methods utilize a set of empirical equations relating the
physical characteristics of watershed to the few salient points of the hydrograph such as
peak flow rate (Qp), time to peak (tp), time base (tb), and UH width at 0.5 Qp and 0.75 Qp

53
i.e., Wo.5 and Wo.75, respectively. However, in the SUH development a great degree of sub-
jectivity is involved in fitting the remaining points on the SUH. In addition, simultaneous
adjustments are required for the area under the SUH to be the unity corresponding to unit
rainfall-excess.

The SUH models can be put under three categories.

1. Deterministic SUH Model: Snyder (1938), SCS(2002), Taylor and Schwarz

(1952), Espey(1974), Jena and Tiwari (2006), CWC (1997)

2. Probabilistic SUH Model: Nash (1959) Gray (1961), Dogee (1959), Croley (1980),
Haktanir & Sazen(1990), Bhunya(2003,2004,2007,2008)

3. Geomorphologic SUH Model: Rodriguez-Iturbe and Valdes(1979), Gupta et

al.(1980), Rosso(1984), Yen & Lee (1997), Jain et al. (2003), Sahoo et al.(2006),
Bhunya et al.(2008)

Details of these models are well described in standard textbooks and hence need not be
reproduced here except the models practiced by CWC (1997), which is used for further
analysis and Jena & Tiwari (2006) that takes into account various morphologic parameters
for SUH computation. Some of the other SUHs given by researchers those are mostly
referred to in this field are also given.

4.2.1 Central Water Commission (1997)

Central Water Commission (CWC 1997), India has presented synthetic unit
hydrographs for different sub zones to be used for determination of design flood of 25, 50
and 100 year return period pertaining to small and medium catchments. It provides seven
points of the 1hr SUH, which are tp (time to peak from centre of rainfall in hr), q p ( peak
discharge in m3/sec per sq.km catchment area, W50 ( width at 0.5Qp in hr), W75 ( width at
0.75Qp in hr), WR50 ( width at 0.5Qp in hr of the rising limb), WR75 ( width at 0.75Qp in hr
of the rising limb) and tb ( time base in hr). These UH parameters are related to
geomorphologic parameters such as catchment area ‘A’ (sq.km), length of the main stream
L (km), length of the main stream from a point near the centre of gravity of catchment to
the outlet Lc (km) and the equivalent slope S (m/km) of the main channel. The following

54
equations are in practice for preparation of 1 hr SUH for the Mahanadi Sub-Zone (3d) for
small and medium catchments.
0.261
LL c 0.725
tp 1.757 (4.1a), qp 1.260 t p (4.1b)
S
1.104 1.125
W50 1.974 q p (4.1c), W75 0.961 q p (4.1d)
0.829 0.932
WR 50 1.150 q p (4.1e), WR 75 0.527 q p (4.1f)
0.826
tb 5.411 t p (4.1g) Qp qp. A (4.1h)

The above relations are regionalized and recommended relations for computations
of UHs for small and medium catchments of varying size up to 1000 sq. km and more even
up to 5000 sq. km in the sub-zone.

4.2.2 Jena & Tiwari (2006)

Jena and Tiwari (2006) modeled the SUH parameters utilizing the
geomorphologic parameters (channel as well as basin parameters) of two medium sized
agricultural watersheds in India. These basin parameters were obtained from topo sheets
prepared by the Survey of India using Geographical Information System (GIS) techniques
and drainage network updated from Indian remote sensing satellite digital images. A
correlation matrix between UH parameters and geomorphologic parameters was generated
and geomorphologic parameters having higher degree of correlation with UH parameters
were selected. Regression equations were developed between individual UH parameters
and the geomorphologic parameters obtained previously. Then non-linear regression
models were developed between UH parameters and the above-selected geomorphologic
parameters. The developed models were tested in Tarafeni watershed in India and its sub-
watersheds using different statistical tests for different rainfall events. These models were
suggested to be suitable for small and medium agricultural sub tropical sub humid basins
having similar geohydrological conditions. Jena & Tiwari (2006) have given the following
equations for the Unit hydrographs.

55
 1 Hr Unit hydrograph:

tp = 1.688 Lb 0.270 Lc0.280, R2 = 0.943 (4. 2.a)

tb = 2.837 Lc 0.396 A 0.098 Lca 0.283 Lb -0.155 , R2 = 0.996 (4. 2.b)

qp = 1.224 S1 -1.911 Ns 1.396 Lca 0.488 , R2 = 0.997 (4.2.c)

2 Hr Unit hydrograph:

tp = 2.099 Lca 0.546, R2 = 0.932 (4.2d)

tb = 3.791 Lc 0.257 A 0.069 Lc 0.269, R2 = 0.997 (4.2e)

qp = 8.273 S1 4.412 Ns -3.089 Lca 2.137 Lb -3.362 , R2 = 0.990 (4.2f)

where tp, tb, qp are as defined earlier and Lb = longest dimension of the basin parallel to the
principal drainage line, Lc = length of main channel, Lca = distance from basin centroid to
outlet, A = catchment area of the watershed, S 1 = total stream length, Ns= total number of
stream. The other salient points of the UH as per Snyder method are adopted here.

4.2.3 Haktanir & Sezen (1990)

Haktanir & Sezen explored the suitability of two-parameter Gamma and three-
parameter Beta distributions as synthetic unit hydrographs for Anatolia catchments in Tur-
key. The analytical expressions for scale-adjusted Gamma and Beta distributions as SUH
are expressed as

(i) Gamma SUH

1 n 1 t K
Q G (t ) A 0.36 t K e (4.3)
K n

where QG(t)= the flow rate of the gamma SUH at time t in m3/s/cm; A = watershed area in
sq.km; t = time in hours; n is the number of linear reservoirs, and K is the storage
coefficient of the reservoirs in units of hours.

56
 (ii) Beta Synthetic Unit Hydrograph
p r 1
Q B (t) A 0.36 t r 1
b t Bb p 1
(4.4)

where B is given as B r p r p (4.5)

and QB(t) is the flow rate of Beta SUH at time t in m3/s/cm; r and p are the shape
parameters; b is the scale parameter in hours. The parameters of both distributions are
estimated by using classical Newton iterative algorithm. They found that both the
distributions fit reasonably well to observed unit hydrographs.

4.2.4 Bhunya et al.(2003)

Bhunya et al. (2003) introduced a simplified version of two-parameter Gamma

distribution to derive a synthetic unit hydrograph more conveniently and accurately than
the popular Snyder, SCS, and Gray methods. The analytical form of the model is expressed
as

n 1 t
1 t k
q( t) e (4.6)
k n k

where q (t) is the depth of runoff per unit time per unit effective rainfall. The
parameters n and k are often termed as the shape and scale parameters, respectively of the
model. It is noteworthy that parameter n is dimensionless, and k has the unit of time. The
area under the curve defined by Equation (4.6) is unity. Thus the rainfall-excess and direct
surface runoff depths are equal to unity.

The authors defined a non- ptp , where q p and tp are peak flow rate
and time to peak flow rate respectively. The expression for

(n 1) (n 1) e (n 1)
(4.7)
1)

Since the exact solution of n in terms of from Equation (4.7) is not possible, they
developed simpler relationships between n and to obtain the simplified versions of
Gamma distribution. The developed relationships are given as

1.75
n = 5.53 +1.04 for 0.01< < 0.35; R2 = 1 (4.8)

57
 1.998
n = 6.29 +1.157 for > 0.35; R2 = 1 (4.9)

Finally, the scale parameter K can be derived as

tp
k (4.10)
n 1

The parameter n and dimensionless

characteristics of the watershed, but also on its storage characteristics.

4.2.5 Bhunya et al. (2007)

Bhunya et al. (2007) used four pdfs, i.e., 2 parameters Gamma, 3 parameters Beta, 2
parameters Weibull, and single parameter Chi-square distribution to derive SUH. For the
reasons of non-availability of an explicit parameter estimation procedure, probability
density function parameters were determined using the least square approach or any other
optimization procedures with suitable error criteria. The authors developed simple
analytical and numerical relationships to compute the distribution parameters, and checked
their validity using simulation and field data. Though all the considered PDFs describe the
UH shape well, the major disadvantage with them, except Beta-distribution, is their
inability to yield a fixed tb-value. The Gamma, Weibull, and Chi-square distributions yield
tb approaching infinity when q approaches zero in the upper tail. Beta and Weibull
distributions possess the ability to skew on both sides (positive and negative) similar to an
UH encountered in practice, they are the most flexible and preferred distributions.

4.2.6 Bhunya et al. (2008)

Bhunya et al. (2008) derived the expressions for the shape parameter (n) and scale
parameter (k) of Equation (4.6) in terms of the Horton’s ratio as

0.9625
R
n 2.312 B R 0L.088 1.04, 0.01 0.35 (4.11)
RA

1.099
R
n 2.118 B R 0L.099 1.157, 0.35 (4.12)
RA

RB, RA, RL are the Horton’s bifurcation, area and the length ratios respectively.

58
Using the Equation (4.11) or (4.12) in Equation (4.6) and Equation (4.10), the complete
shape of the SUH can be determined. They have also derived similar equations for the
Weibull distribution using the Horton’s ratio. The performance of the proposed methods
was tested describing a synthetic unit hydrograph (SUH) for limited data conditions of two
real catchments. This method exhibited better result than Rosso (1984).

4.3 Objectives

Central Water Commission (CWC) India’s regionalized Equations 4.1(a-h) for

computation of UH parameters are being followed by all practicing hydrologists in the
subzone for watersheds of any size for finding design solution of water resources problem.
The peak flood (Qp), time to peak (tp) and time base (tb), the three important parameters of
the UH may either be underestimated or over estimated in case of the same equations are
adopted indiscriminately irrespective of size of watershed, since flood propagates and
attenuates through many natural channels of varying capacity and this phenomena is termed
as flood routing. In natural channels the runoff from a single catchment comes across many
lateral inflows as it goes down to the outlet. The time, position and sectional parameters of
the routing reaches and the flood plain control the behavior of the runoff. The peak flow
(Qp), time to peak (tp) and the time base (tb) are influenced to greater extent by the above.
These three basis parameters of the routed hydrograph may not agree with the parameters
suggested by CWC when the catchments are lumped. So the present objective is to evaluate
the performance of the SUH concept by making comparison between UH suggested by
CWC and the Routed UH and make recommendation for improvement for designing any
water resources structures.

The solution to the above problem is attempted through the application of a

synthetic unit hydrograph suggested by Central water Commission (CWC), India to a real
world catchment in the Mahanadi Subzone. The preference of a synthetic unit hydrograph
to a flood hydrograph is based on the concept that unit hydrograph is also a flood
hydrograph arising out of unit rainfall and in absence of any measured rainfall, runoff and
sectional data, the problem can be solved. The problem is attempted with the following
objectives.

59
 1) Considering the unit hydrograph of each sub catchment as the gauged hydrograph
just in the upstream of the confluence point i.e. tributaries that meet the main
channel of the watershed (corresponding node), the discharge is routed to an
arbitrary downstream outlet.

2) Determine the hydrograph at the outlet after routing the concurrent discharges
independently from the entire sub catchments by Muskingum linear routing method
and add together based on the principle of superposition. The outflow hydrograph,
hereafter, referred as Routed UH.

3) Compare the Routed UH of the outlet with the CWC UH at the same outlet
considering the catchment as a whole. The comparison will provide adequate
insight to the present problem.

4.4 Hydrologic River Routing

Two approaches, i.e., hydraulic routing and hydrologic routing are mainly
employed to route flood in natural channels. In hydraulic routing the flow is considered to
be gradually varied and Saint-Venant equations govern the flow wave numerically. The
complete solutions of the Saint-Venant equations are difficult to solve since the magnitude
of different terms in the momentum equation are widely varying. Muskingum two
parameter flood routing is one such hydrologic routing method which assumes that the
storage in a channel reach through which a flood is being routed is proportional to a
weighted sum of inflow into and outflow from the reach. O’Donnell (1985) proposed a
three parameter Muskingum model to account for lateral inflow expressed as a component
of the main channel inflow. However, if inflow increases substantially in the form of
tributary flows, then the problem becomes very difficult to solve. Under such cases it is
essential that (i) Muskingum routing method should be applied by terminating the routing
reaches at a confluence augmenting the main channel flow by the tributary flow for the
next reach (NERC-1975), (ii) superposing after routing as per Muskingum method, where
stream flow from each tributary is routed to the outlet of the catchment and is summed
assuming that the flood wedges from tributaries propagates independently without being
disturbed by downstream conditions (Choudhury 2007), (iii) transforming the multiple
gauged inflows into a single equivalent inflow at a characteristic point in the watershed and

60
then routing the equivalent inflow to the catchment outlet. The characteristic point is such
that the inflow given by the summation of all gauged concurrent channel inflow is
equivalent to the network inflow gauged independently at various gauging sites
(Choudhury, 2002).

4.5 Muskingum Flood Routing

The linear Muskingum method of flood routing (McCarty 1938) based on a simple
storage-discharge relationship in river system, is extensively used in river engineering (Gill
1979). The method performs better in natural channels where inertia effects and
downstream influences are very small and where the model parameters are approximately
chosen for representing the hydraulic behavior of the system (Chang et al. 1983). The linear
storage relationships are expressed as:

dSt
Continuity: It Ot (4.13b)
dt

and Storage is expressed as follows: St k xIt (1 Ot ) (4.13b)

where St, It, Ot are concurrent amount of storage, inflow and outflow respectively at a
given time t and k is the storage constant expressing the ratio between storage and
discharge in a given reach and x is a dimensionless weighting factor that varies between 0
and 0.5 for natural rivers with an average around 0.2. This weighting factor describes the
relative importance of inflow and outflow to storage. The storage constant, k, equates
closely to the flow travel time through the river reach (McCuen 1998). If k and x are
known, routing is performed by using:

Ot CiI t C 2It t C3O t t (4.14)

in which C1, C2, C3 are routing coefficients given by:

0.5 t kx
C1
1 x k 0.5 t (4.15a)

0.5 t kx
C2 (4.15b)
1 x k 0.5 t

61
 1 x k 0.5 t
C3 (4.15c)
1 x k 0.5 t

with the condition that C1+ C2+ C3 = 1 (4.16)

In Equation (4.14), t is the time step and It- t and Ot- t are the inflow and outflow
discharges at time t- t respectively. Once the coefficients are determined, Equation (4.14) is
repeatedly used to determine the outflow Ot at the outlet point at any time.

The values of k and x are derived from observed upstream and downstream
hydrographs extracted from historical flow records. These methods are broadly represented
in five classifications: (a) graphical methods; (b) least square methods; (c) methods of
moments; (d) direct optimization; and (e) those based on Saint-Venant equations. Yoon and
Padmanabhan (1993) identified three methods for linear parameter estimation. The
algorithm included forward and backward optimization using t- statistics, an outlier
filtering estimation method and a quadratic programming algorithm.

The graphical method is generally satisfactory (Chow 1964, Linsley et al. 1975,
Viessman et al. 1972, Wilson 1990), but time consuming. The objective selection criterion
exists for choosing the appropriate value of x and therefore the method requires a level of
subjective interpretation to determine a value that optimizes the linear relationship
(Gelegenis and Serrano 2000, Yoon and Padmanabhan 1993, Chang et al. 1983). The least
square scheme is based on minimizing the sum of squares of the deviations between
observed storage and computed storage for a given inflow and outflow hydrograph (Gill
1978, Birkhead and James 1997, Al-Humoud and Essen 2006). The methods of moments
are similar and are based on relating the first and second moments of the instantaneous unit
hydrograph(IUH) of the Muskingum reach to the Muskingum routing parameters, k and x
(Dooge 1973). The method of direct optimization is based on minimizing the difference
between observed and computed hydrograph to determine directly the routing coefficients
of the Muskingum model without explicitly estimating k and x (Gelegenis and Serrano
2000). This was modified by Cunge (1969) including the effects of geometrical and
resistance properties of the river reach in the original Muskingum method to develop the
Muskingum-Cunge (M-C) model. In the M-C flood routing procedure, the necessity of

62
calibration that characterizes the Muskingum method is not required and routing
parameters, k and x, are obtained from hydraulic properties of the reach using:

X 1 Q
k and x 1 (4.17)
C 2 BSC X

where parameters k and x are as described above and C is the flood wave celerity,
longitudinal channel distance increment, Q is the discharge and B is the top width of the
flow section at downstream outlet.

Perumal et al. (2007) presented a methodology for developing the rating curve at
ungauged site in a restrictive sense that considers the routing parameters (C1, C2, C3 in
Equation 4.14) based on variable parameter Muskingum stage hydrograph (VPMS) routing
method. These parameters vary at every routing time s
discharge and the recorded stage hydrograph. But when the catchment is ungauged in the
ideal sense i.e. neither stage hydrographs nor the discharges are available, such method can
make the present analysis more completed.

In the present study, the synthetic unit hydrograph (SUH) as suggested by Central
Water Commission (CWC), India for sub-zone 3(d) is considered as inflow hydrograph and
the components are determined using the relations Equation 4.1(a-h). The value of k is
determined by first part of Equation (4.17) and x is estimated using the following simple
equation (Wilson et al. 1990, Subramanya 2000, Sadeghi & Singh 2010):

0.5Vm
x (4.18)
1.7 Vm

1 2 / 3 1/ 2
where Vm R S is the average velocity of the flow in m/sec determined by
n
Manning’s equation with the variables as R= Hydraulic radius (m), S = average slope and n
= Rugocity coefficient.

63
4.6 Material and Methods

4.6.1 Study Area

The Brutang watershed of Mahanadi Sub Zone 3(d) between Longitude 20 022’ to
20023’ N and Latitude 84081’to 84048’ E is selected as the study area. Details of the study
are described in Chapter 3, Section 3.5.1 and Table- 4.1.

Table 4.1 Location of Study Area

Study Sub-Zone Longitude Latitude Mean Area

Area Rainfall (km2)
(mm)
Brutang Mahanadi, 20 022’ to 20 023’ N 84081’to 1171 799.83
River Baitarani,Brahmani 84 048’ E

4.7 Model Formulation

This study considers SUH parameters suggested by CWC, India with respect to
their application in various sizes of catchments in the Mahanadi Sub Zone 3(d). The
formulation of the models involves various steps as described below:

1) Study area map is prepared using topographical maps (1:50,000) provided by

Survey of India. The watersheds of each tributary that meet the main channel
including the head catchment are delineated (Fig.4.1) from the prepared map in the
GIS environment using Arc GIS 9.3. Under this step total 75 no sub-catchments are
identified from node 0 to 117 (Outlet at Takara) and the size of the sub-watershed
varies from less than1 sq.km to 130 sq.km.

2) The sub-catchment area (A), its length of the tributary (L), length of the tributary up
to the centre of gravity (Lc) of the sub-catchment and the slope of the tributary (S)
and are also determined. From the prepared map of the study area, the main channel
is identified. The distance of each reach (
(node-0) to the outlet (node- 117) is measured.

3) The UH is plotted after computing the parameters (Qp, tp, tb, WR50, WR75, W50, W75)
using Equation 4.1(a-h). The discharges ordinates of the UH at 1 hr time step are

64
 computed from the UH plot and these are the inflow discharge ordinates at
corresponding time steps.

4) The UH of each sub-catchments is checked for unit rainfall. Adjustments are made
in the recession limb without disturbing Qp, tp and tb such that the area under the
UH becomes unity. In this process as Qp, tp and tb are not disturbed and only slight
variation occurs in parameters like W50 and W 75, hence the shape of the UH is
acceptable.

5) Qt, the discharges at time t as determined in the previous step no 2 are considered to
be the measured discharges just in the upstream point of the confluence point of the
tributary and the main stream (corresponding node).

6) Subsequently, Muskingum linear routing equations (Equation 14 & 15a-c) are

applied to determine the hydrographs at any downstream outlet at Node-4, 10, 29,
56, 73, 78, 99, 117 (Figure 4.2 a-f & Figure 4.1) by considering the concurrent
discharges (Qt) from the tributaries. The parameters k and x cannot be determined
using Equation (4.17) as no sectional parameters are available; since the study area
is un-gauged i.e. no downstream flow data nor sectional parameters are available for
using Equation (4.17). However, Equation (4.17) can be used if the shape of the
channel section at the outlet point is assumed. In the present case the channel
section at the outlet is assumed to be parabolic and wide.

For a parabolic section, according to Viessman et al. (1989)

11
celerity, C Vm (4.19)
9

V m being the average velocity as described under Equation (4.18). In the present
case the value of n is assumed to be 0.350 for natural channels.

For a wide parabolic section, the hydraulic radius (R) is given by (Koegelenberg et
al. 1997)

2yW 2
R (4.20)
3W 2 8y 2

65
 Where W= width of the flow section= 4 .71 Q (Lacey’s equation cited by Punmia

& Pandey 1981 and MH Tewolde 2005),

3 /5
Qn
y= depth of flow= (4.21)
0.508 WS

and Q, S , n are as described earlier.

7) After determining k, x in previous steps, the routing coefficients are calculated

using Equation (4.15a-c). The upstream hydrograph from each sub-watershed is
then routed to the downstream outlet point independently using Equation (4.14).
The routed concurrent discharges from each watershed are added to get the UH at
the outlet. It is based on the principle of superposition after routing, which signifies
each flow is independent to propagate to the outlet without being influenced by
downstream condition and no lateral flow except the tributary flow plays its role.

8) Then the UH at the outlet is compared with corresponding UH suggested by CWC

(Equation. 4.1 a-h) considering the watershed as a whole.

4.8 Results and Discussion

The results show that values of UH parameters computed with CWC and routing
method differ. The relation of these parameters with catchment characteristics L, Lc and S
also differ with methods and are presented in Table-4.2 (a-d). The parameters are better
correlated with the catchment characteristics in case of routing method than CWC method;
the R2 values are the indicators of such relationships. The Routed UH parameters are
presented in Table 4.2(b). It can be observed from Table 2(b) that the peak discharges in
case of node 4 to 10 (Catchment area upto 238.718 sq.km) is more than that calculated by
CWC- UH. The catchment up to node 10 is hilly with steeper longitudinal slope of the
main channel. From node 10 to node 78 with catchment area between 238 to 583 sq.km
with moderate longitudinal slope where the peaks by CWC UH exceed the peaks of Routed
UH . From node 99 onwards the peak again starts rising up because the slope of the channel
becomes very mild. Since the size of the present study area is limited to799 sq.km, the
present concept could not be projected further, but logically because of milder slope of the
channel as the catchment size increases, the peak of the Routed UH will remain higher than

66
the CWC UH. Of course, not only the size but also other geomorphologic parameters also
govern this variation. For smaller size watersheds with steep slopes, it can be always
recommended to be governed by Routed UH. It is obvious that as the peak flow Qp
increases, the time to peak tp decreases as evident from Table-1(a-d). This shows that the
peak flow occurs in a shorter time period than the CWC UH. The peak discharge Qp and
the time to peak tp being the important parameters of the flood hydrograph as far as the
flood analysis is concerned, the recommended UH parameters of CWC should be suitably
adjusted considering the routing effects.

In both the cases, the UHs are drawn considering the discharge ordinates at start of
the hydrograph, point at W50 and W75 on the rising limb, point of peak flow, point of W50
and W75 on recession limb and at the end of hydrograph.. The rest points are locally
adjusted in order to bring the area under UH to unity which is a bit difficult exercise.
Assuming the intermediate points at W50 and W75 insignificant, any number of intermediate
points between t= 0, and tb can be considered in order to generate a smooth curve with the
criterion that the area under the curve should be unity. In order to improve upon the
problem of manual fitting, the two parameter Gamma and the three parameter Beta
distribution as suggested by Bhunya et al. (2007) are used to fit in any number of
intermediate points. The equivalent 2p Gamma curve of the Routed UH is determined using
the Equations (4.6 to 4.10). The equivalent 3p Beta curve of the Routed UH is determined
as per Bhunya et al. (2004) using the following 3p Beta equation (Haktanir & Sezen, 1990).

p r 1
tr 1 b t
qt (4.22)
B r, p r b p-1

where p, r are the non-dimensional shape parameters and b is the scale parameter of the
Beta curve and B( r, p-r ) if the Beta function.

Bhunya et al. (2004) defined the following non-dimensional terms to define the shape of a
Beta curve.

tb
and q p .t p (4.23)
tp

67
 governs the
skewness and quantifies the hydrograph peakedness and influences the hydrograph shape.

Using Marquardt Algorithm they have derived the following relations among these non-
dimensional terms.

1 .874 0 .954
p 2 .5355 0.0481 2.961 10 .7277 (4.24)

2 p
r 1 (4.25)

And b = t b (4.26)

Now using Equation (4.22 – 4.26), it is possible to determine the equivalent 3p Beta UH
using the statistic of Routed UH. Since Beta UH is flexible in its shape, care has been taken
to maintain the its peak ( qp ) and time to peak ( tp ) closer to that of Routed UH.

The 2p Gamma curve of the Routed UH indicates that both the curves are almost
identical except that the peak gets slightly shifted to the right and with slight increase in tp.
The corresponding power relations of different UH parameters with L, Lc, and S are given
in Table-4.3 and R2 values indicate better correlation. The study shows that Routed UH is
an improvement over CWC UH and the equivalent 2p gamma UH removes the practice of
manual fitting of the various points of the UH to get unit area. Similar study also generates
the equivalent 3p Beta curve of the Routed UH. In this case, it indicates that the peak gets
reduced as indicated in Table- 1(d) and tb gets extended. The UH parameters are relatively
poorly related to L, Lc and S as compared to 2p Gamma Curve. But 3p Beta Curve enjoys
the freedom of both positive and negative skewness and is applicable for all type of
situations. The Figure (4.3 a-h) gives a graphical representation of CWC, Routed, 2p
Gamma and 3p Beta UH curve for different nodes.

4.9 Conclusion

The study attempts the application of synthetic unit hydrograph approach for
determination of design flood in the small catchments. The following conclusions are
inferred from the analysis results. The peak discharge of the routed UH is higher than the
CWC UH in the hilly catchment having steeper slope. Similar results are also observed

68
when the catchment area increases and with milder slope in plane. CWC UH shows higher
peaks than the Routed UH for catchment having areas broadly between 200 sq.km to 600
sq.km. But for the catchment area lower than this limit with steeper slope and for larger
catchment area with milder slope, the CWC UH does not hold good. Based on the above
findings it is recommended that the peak discharge of the CWC UH should be corrected by
introducing a suitable multiplying factor while designing water resources structures in hilly
as well as plane areas. The CWC UH is synthetically prepared considering data of some
bridge catchments. The geomorphologic parameters (L, Lc,S) of these bridge catchments
governs the UH relations. The present study can be further refined by considering more
number of catchments of varying size and longitudinal channel slope along with other
morphologic parameters such as basin perimeter, basin length, drainage density, basin relief
(Jena & Tiwari, 2006) etc. The present approach if applied to larger catchments will
provide the upper limit of the size of catchment to which the synthetic unit hydrograph
concept is applicable for design flood estimation.

Figure 4.1 Sub-catchments of Brutang up to Takara

69
Figure 4.2(a) Sub-catchment at Node- 10 Figure 4.2(b) Sub-catchment at Node- 29

56

Figure 4.2(d) Sub-catchment at Node-

Figure 4.2(c) Sub-catchment at Node- 56 73

70
 Figure 4.2(e) Sub-catchment at Node- 78 Figure 4.2(f) Sub-catchment at Node-
99

UH at Node-4 90 UH at Node-10
60
80
CWC UH CWC UH
50 Routed UH 70 Routed UH
2p Gamma UH
3p Beta UH 60 2p Gamma UH
40 3p Beta UH
50
30 40

20 30
20
10
10

0 0
0 5 10 15 20 25 0 5 10 15 20 25 30

Time (tb) in hr Time (tb) in hr

Figure 4.3(a) CWC, Routed, 2p Gamma, 3p Beta Figure 4.3(b) CWC, Routed, 2p Gamma, 3p
UH Beta UH

71
 180
UH at Node-29 UH at Node-56
160
140 CWC UH
Routed UH 140 Routed UH
120 2p Gamma UH 2p Gamma UH
120 3p Beta UH
3p Beta UH
100 100
80 80

60 60
40
40
20
20
0
0 0 10 20 30
0 20
Time (tb) in hr
Time (tb) in hr

Figure 4.3(c) CWC, Routed, 2p Gamma, 3p Beta UH Figure 4.3(d) CWC, Routed, 2p Gamma, 3p Beta UH

180 UH at Node-73 160 UH at Node-78

160 140
CWC UH CWC UH
140 Routed UH Routed UH
2p Gamma UH
120
2p Gamma UH
120 3p Beta UH 3p Beta UH
100
100
80
80

60 60

40 40
20 20
0
0 5 10 15 20 25 30 35 0
0 5 10 15 20 25 30 35
Time (tb) in hr
Time (tb) in hr
Figure 4.3(e) CWC, Routed, 2p Gamma, 3p Beta UH
Figure 4.3(f) CWC, Routed, 2p Gamma, 3p Beta UH

72
 200 UH at Node-99 180 UH at Node-117
180 160
160 CWC UH CWC UH
Routed UH 140 Routed UH
140 2p Gamma UH 2p Gamma UH
3p Beta UH 120 3p Beta UH
120
100
100
80
80
60 60

40 40
20 20
0 0
0 5 10 15 20 25 30 35 0 10 20 30 40 50

Time (tb) in hr Time (tb) in hr

Figure 4.3(g) CWC, Routed, 2p Gamma, 3p Beta Figure 4.3(h) CWC, Routed, 2p Gamma, 3p Beta UH
UH

Peak Discharge (Qp) vs Catchment Area Plot (A)

200
180
160
140
CWC UH
120
ROUTED UH
100
80
60
40
20
0
0 200 400 600 800 1000
Catchment Area in sq.km (A)

Figure 4.4 Qp vs A plot for Brutang Catchment

73