HYDRO 2023 INTERNATIONAL

28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

Comparative Study of Geomorphological Instantaneous

Unit Hydrograph (GIUH) models in Banas river basin
G. R. Chauhan1, H. Sharma1 & M. N. Langhi1
1
Institute of Infrastructure Technology Research And Management, Ahmedabad, India
E-mail: [Link].22pc@[Link]

Abstract: Ephemeral streams are often characterized by their limited flow time and flow length. These are often dry and only
carry discharge due to very high precipitation at the upstream. However due to temporal and spatial variations of such rainfall
events, it is very hard to predict the runoff in such cases. Also due to economic reasons the in-feasibility of gauging such streams
with measuring devices results in limited or no data availability which directly affects the calibration of hydrologic models. To
overcome this difficulty one way is to develop a theoretical model which does not rely on past rainfall- runoff data. One such
model is Geomorphological Instantaneous Unit Hydrograph (GIUH) which solely relies on the geomorphology of the catchment
hence omitting the need of past data. This model uses Horton's laws. GIUH is represented by probability distribution function
(PDF) of drop’s travel time. By convoluting PDFs one can determine the response of the basin to the precipitation. Since the
determination of the probability distribution function is very difficult for higher order streams. Many researchers have tried to
combine Nash model of cascade of reservoirs and Clark unit hydrograph model to the GIUH parameters. The present study aims
at developing various GIUH models for Banas River basin which is an arid region of Rajasthan state of India. In order to
determine the runoff generated, three types of GIUH models were developed namely GIUH with PDF, GIUH Nash model and
GIUH Fractal model and they are compared with the hydrograph generated by HEC-HMS Clark model. The hydrographs
generated using conventional Horton’s ratios fails to match the one given by Clark model. However, the models generated by
modified GIUH Nash model and fractal model perform better and show good fit to the Clark hydrograph than the GIUH PDF
model. It is concluded that for better characterization of channel network, modified Horton’s laws give more reliable estimate
than the conventional one.

Keywords: Rainfall runoff modelling, ungauged basin, geomorphology, probability distribution function, unit hydrograph, path
probability

1. Introduction
The lack of rainfall runoff time series data is a major obstacle in forecasting flood from the catchment. Due to
economic reasons, it is not practical to set up a network of rain gauges in developing countries like India. When the
catchment is ungauged it is very difficult to calibrate the unit hydrograph and this has been a major problem in
forecasting flood for such basins. When the stream is ephemeral it becomes more difficult as the nature of climate is
very random. Ephemeral streams are the ones that only receive discharge due to extreme precipitation. Hence most
of the year they are dry. Due to spatial and temporal variation of such precipitation, it is unwise to gauge such
streams. Hence there are no recorded data available for such streams. This restricts the use of conventional synthetic
hydrograph models for such streams. To overcome this Rodriguez- Iturbe et. al (1979) developed a model which
utilizes such recorded data. The proposed model uses the geomorpholy of the catchment to generate an
instantaneous unit hydrograph hence the name given Geomorphological Instantaneous Unit Hydrograph (GIUH).
This model was further modified by Gupta et. al (1980) with rigorous mathematical reasoning. The GIUH model
uses the well-established Horton’s ratios to characterize the basin network.
GIUH is a probabilistic model which uses the probability distribution of travel time estimates of a raindrop
in the basin. Using the network characteristics different paths that a rain drop would take to reach downstream is
determined. Different path probabilities are than convoluted to give a unit hydrograph model. The determination of
path probabilities is in the terms of bifurcation ratio. However, this can be a daunting task if the order of the stream
is higher than 4. This can be simplified if one can relate the peak discharge and peak time at that discharge to
different distribution functions. One of the approaches is to relate them to the Nash model of linear cascades which
uses gamma function as a probabilistic function (PDF). The Nash parameters are calculated from the known peak
discharge and peak time. Similarly, other PDFs can also be implemented to GIUH model.
With the investigations on the geometry of channel networks, Marani (1991) suggested that a better
representation of the network can be found in the fractal nature. By implementing fractal geometry, Kim (2022)
suggested the use of equivalent Horton’s ratios which does not depend on the magnitude of the threshold basin area
while extracting channel network. These ratios are found to be more suitable to represent the channel network.

The present paper is an investigation on different GIUH approaches applied on the Banas river basin of
India. The approaches used are the conventional GIUH using PDFs, Nash- GIUH model and Nash model modified
using equivalent Horton’s ratios. Again the models were regenerated wherein the network is characterized by the
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

fractal geometry. The generated models are compared with the Clark unit hydrograph generated using HEC-HMS
model. The following sections contain theoretical background of the GIUH models.

2. Horton’s Ratios for network characterization

Strahler's stream ordering definates the topology of channel networks as follows:
• First order streams come from a source.
• Stream order ω +1 is given in case of merging of two streams having similar order.
• The stream at the downstream takes the value of one order higher of the preciding stream.
As per Strahler's scheme, the channel network is designated with the highest order stream, and Horton's ratios are
then calculated as,
N ω−1
R B= (1)
Nω
Lω
R L= (2)
Lω−1
Aω
RA= (3)
A ω−1
where RB, RL, and RA refer to Horton’s ratios. Nω, Lω, and Aω are the number, average length, and average drainage
area of the stream having order ω, respectively.

3. Geomorphologic Instantaneous Unit Hydrograph

GIUH was theorised by Rodriguez Iturbe and Valdes which is a probabilistic approach depending on the drop’s
travel time in the catchment. To determine the runoff, various paths that a randomly selected rain drop would take in
reaching the outlet known as the trapping state is calculated. The path probability functions are then convoluted to
represent the unit hydrograph. The GIUH is interpreted as the probabilistic functionsof the travel times in reaching
the outlet

∂
u ( t )=
∂t (∑ prob (T
Si
Si
)
≤ t ) prob ( S i ) (4)

where, prob ( T Si ≤ t ) denote the cumulative probability distribution with the travel time as a variable in path Si, and
prob(Si) denote the probability of travel path Si for any selected droplet.

Let ci denote the channel state having order i and ri denote the overland segment state of order i and Ω be
the order of the catchment. For a basin of order Ω, the possible transitions are 2Ω −1 . Assuming that initially the
particle will be located at any of the region r. Given that the particle is found in any one of the regions ri, it
undergoes transitions as following:

 From of the state ri, the only transitions are those of the form r i → ci , 1<i<Ω .
 Frohe state ci, the only transitions are those of the form c i →c j, j >1, i = 1, 2, …, Ω.
 State c Ω +1is the last state that is, transitions from the state c Ω +1are none.

For a 4th order network, different transitions possible are as follows.

Path 1: r 1 → c1 → c2 → c3 → c 4
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

Path 2: r 1 → c1 → c2 → c 4
Path 3: r 1 → c1 → c3 → c 4
Path 4: r 1 → c1 → c 4

Path 5: r 2 → c2 → c3 → c 4
Path 6: r 2 → c2 → c 4
Path 7: r 3 → c3 → c 4
Path 8: r 4 → c4

The probability for any path is given by,

prob [ Si ] =θi p ij p jk … p m Ω (5)
where θi, and pij denote the initial and the transition state probabilities respectively. The initial state probability θw
defines the probability of any drop coming from the overland region w. It is determined from the expression given
below.

{
N1 A1
w=1
A
θ w=
( )
q−1
pkw N k A k (6)
A w −∑
k=1 Nw
Nw otherwise
A

The transition probabilities define the repetation cycle of streams having order i flowing into streams
having order j, j >= i+1. The probabilities are calculated for a certain drainage network as the proportion of streams
with order i that merge with the streams having order j to all streams having order i. The transition probabilities are
defied by the following relation when drainage networks can be assumed to have a random topology.

{
N i+1 ( N i−2 N i+1 ) E[ j , Ω]
2 + j=i+1

( )
Ni Ω
N i ∑ E [k , Ω]
pij = k=i+1
(7)
( N i−2 N i+1 ) E[ j , Ω]
otherwise

(∑ )
Ω
Ni E [k , Ω]
k=i+1

where E [j, Ω] is the expectation values for links having order i in a network having order Ω, and can be calculated
as

(∏ )
i
N j −1−1
E [ i, Ω ]=N i (8)
j=2 2 N j−1

It is assumed that a drop's waiting time in an i th order state is a variable with an exponential distribution and
parameter i, where i is the reciprocal of the average waiting time. Thus, after considering both overland flow time
and stream flow time, λ−1i is the average time taken by a drop in i th state. Given that there are far fewer drops that
drain directly by overland flow, the value of the time spent as overland flow seems to be somewhat less significant
than that of the stream. As a result, i is taken as a function of LΩ. and is taken as the reciprocal of the average waiting
time in ith order streams.
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

V
λ Ω=
LΩ
(9)

where V denotes the velocity in the basin and LΩ is the length of stream having highest order. The mean waiting
time for lower order streams can thus be expressed as,

Ω−i
λ i=λ Ω R L
(10)

where RL denotes the Horton’s length ratio. The travel times are then estimated by inversing λi.

1
Tx= (11)
i
λi

The GIUH for the watershed is derived using the probabilistic travel times for stream and overland flows as
u ( t )= ∑[
f x oi (t)∗f x i(t)∗f x j (t)∗f x k (t)∗…∗f xΩ (t) ⋅ prob [ Si ] ] (12)

The probability distribution function (PDF) f x is used for the travel time of overland and stream flows, the
i

GIUH is calculated from Equation (2) as,

−t
1 T
fx= e xi
(13)
i
Tx i

where T x idenote average time of travel for state i.

The GIUH is conveniently can be defined by a triangular hydrograph approach where the peak discharge
(qp) and peak time (tp) of the IUH are required. The peak discharge and the peak time are calculated using the
formula

0.43
1.31 R L V
q p=
LΩ
(14)

( )( )
0.55
LΩ RB −0.38
t p=0.44
V RA
( RL) (15)

where qp is peak discharge, tp is peak time, RB, RL, RA are Horton’s ratios, LΩ denote the length of highest order
channel , and V is velocity. Multiplication of equations (14) and (15) resultd in a dimensionless parameter which
does not depend on velocity.

( )
0.55
RB 0.05
q q ⋅t p =0.5764
RA
( R L) (16)

4. Nash GIUH model

The Nash model (Nash, 1957) is derived assuming that the instantaneous inflow can be routed through a sequence of
linear reservoirs having similar storage constants to get the IUH. The second reservoir's inflow is regarded as the
first reservoir's outflow, and so on. The Nash model uses the parameter dimensionless reservoir numbers (n) and the
hourly storage constant (k) for the derivation of IUH. The Nash IUH model's governing equation is written as:
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

−t

()
n−1
1 t
u ( t )= ek
k Γ (n) k
(17)
where, u(t) denotes the IUH ordinates1, t denotes the time and, Γ is the gamma function.
Linking the GIUH's qp and tp parameters to the Nash IUH model's scale (k) and shape (n) parameters will
yield the GIUH's full shape. To obtain the time to peak discharge, t p, the first derivative of Equation (17) is equated
to zero. Consequently, by from the natural logarithm of both sides of Equation (17), its derivative with respect to t,
and simplification gives,
∂
∂t
ln [ u ( t ) ] =
k
+[
−1 (n−1)
t ] (18)

To obtain tp, equation (18) is equated to 0, which on simplifying results in

t=t p=k (n−1) (19)
Substituting value of tp in equation (17), qp is obtained as
1 −(n−1) (n−1)
q p= e (n−1) (20)
k Γ (n)
Multiplying equation (19) and (20),
(n−1) −(n−1)
q q ⋅t p = e (n−1)(n−1 ) (21)
Γ (n)
From equation (16) and equation (21),

( )
0.55
(n−1) −(n−1) R
e (n−1)(n−1)=0.5764 B 0.05
( RL ) (22)
Γ (n) RA
The Newton-Rapshon technique of optimization is used to solve Equation (22) and yield the Nash model
parameter n, which is the only remaining undetermined term. Equations (15) and (19) and the knowing the
parameter n the Nash model parameter k is calculated for the velocity V as follows:

( )( )
0.55
L
k =0.44 Ω
V
RB
RA
−0.38
( R L) ( n−11 ) (23)

Using Equation (17), the calculated values of n and k determines the whole shape of the GIUH-based Nash
model.

5. Equivalent Horton’s ratios of channel network

Horton’s laws of drainage composition are used to describe the shape of the channel network. Combining them with
Strahler scheme of stream ordering it can be used to describe the hierarchical network structure. Horton’s ratios are
found to have a similar nature as with fractal dimension of complex phenomenas such as channel network. Kim
(2022) developed equivalent Horton’s ratios to better describe channel network as
α
2
R Be= (24)
λ

( )
α 1
2 2α
R¿ = (25)
λ

=( )
α 1
2 α
R Ae (26)
λ
where
ln R B
α=
ln R A
(27)

[ ]
α
N Ω 2 AΩ
λ=
RB A
(28)
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

6. Study Area
River Banas runs through the east-central Rajasthan state of India. It starts in the Aravali range's Khamnor hills and
ends up in Rajasthan. The overall catchment area is around 51,779 km2 in size.

7. Derivation of GIUH models

The SRTM-DEM having a spatial resolution of 30 m was used to delineate the channel network. To extract drainage
channels and other geomorphological parameters, the QGIS 3.22.8 software's pcraster and hydrology tools were
employed. Table 1 displays the basin's morphology. The peak discharge and the peak time are determined to be
0.111 hour-1 and 10.398 hour, respectively, from the Horton's ratio.

Table 1. Geomorphology of the Basin

Number Drainage Average Average
Length of Stream Stream
Stream of area of stream stream Bifurcation
streams, Li length area ratio,
order, i streams, streams, Ai length area ratio, RB
(km) ratio, RL RA
Ni (km2) (km) (km2)
1 44 211.575 32096.295 4.808 729.461
2 29 130.095 11104.347 4.486 382.908
3.347 1.905 1.002
3 14 62.461 3368.573 4.461 240.612
4 1 41.313 859.189 41.313 859.189

7.1. GIUH PDF model

It is assumed that in the process of reaching its outlet and produce the unit hydrograph, a unit of rainfallhaving
several noninteracting raindrops, applies uniformly and instantly to a watershed. To get to the watershed outlet, each
drop landing on the overland segment travels sequencely from lower to higher order channels. The water drops
travel along a certain route from one state to another state before arriving at a watershed outlet. Table 2 lists various
routes taken by water drops as they move from an overland plane through channels to the watershed's outlet, along
with the likelihood that they will take each route.

Table 2. Path-wise probability distribution

Path (Si) Path route Path probability, P(Si)
1 r 1 → c1 → c2 → c3 → c 4 0.036875
2 r 1 → c1 → c2 → c 4 0.044209
3 r 1 → c1 → c3 → c 4 0.064785
4 r 1 → c1 → c 4 0.026073
5 r 2 → c2 → c3 → c 4 0.144378
6 r 2 → c2 → c 4 0.028714
7 r 3 → c3 → c 4 0.108768
8 r 4 → c4 0.388392

7.2. GIUH Nash Model

From the known values of peak discharge and peak time, the Nash parameters n and k came out to be 8.81 and 1.13
respectively.
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

7.3. GIUH Fractal Model

The equivalent Horton’s ratios are found out as suggested by Kim, using the fractal geometry to represent the
network. These new ratios are then used to generate the Nash parameters. The equivalent Horton’s ratios along with
computed Nash parameters are given in Table 3.

Table 3. Equivalent Horton’s ratios and Nash parameters

RBe RLe RAe n k
3.347 1.002 1.001 8.425 1.4

8. Results and Discussion

The GIUH is generated from the geomorphology of the basin as defined of Horton’s law. The travel time
distributions for channel flows are computed using Eq. (11). The hydrographs are generated for the three approaches
and the performance is compared with the HEC-HMS Clark model as shown in Fig. 2. From the figure it is evident
that the GIUH is an effective tool for producing the runoff model for an ungauged basin solely from the
geomorphological characteristics. On comparison with HEC-HMS Clark model, the Hydrographs generated using
GIUH Nash model and GIUH fractal model show good fit while the GIUH PDF model shows large deviations from
the HEC-HMS Clark model.

Figure 2. Comparison between different GIUH models and HEC-HMS Clark model

9. Conclusion
The applicability of GIUH to produce runoff hydrograph is evaluated by applying it to the Banas river Basin. Three
type of GIUH are produced, first with the help of probability distribution function of travel time, GIUH Nash model
and GIUH fractal model. In comparison with HEC- HMS Clark model, the UH produced with the help of Nash
GIUH model and GIUH fractal model show better fit than the UH produced with the help of GIUH PDF model. The
results indicate that GIUH can be utilized as a tool for determining discharge for ungauged basins.

10. REFERENCES
Rodríguez-Iturbe, I., Vald´es, J.B. (1979). “The geomorphologic structure of hydrologic response.” Water Resource
Res. 15 (6), 1409–1420.
 HYDRO 2023 INTERNATIONAL
28th International Conference on Hydraulics, Water Resources, River and Coastal Engineering
21-23 December 2023
NIT WARANGAL, INDIA (राष्ट्रीय प्रौद्योगिकी संस्थान वारंगल, भारत)

Gupta V. K., Waymire E., Wang C. T. (1980). “A representation of an instantaneous unit hydrograph from
geomorphology.” Water Resource Res 16(5):863–870.
Kim J.C. (2022). “Equivalent Horton’s ratios of channel network within the framework of variable source basin
area.” Journal of Hydrology: Regional Studies. 39.
Nash J. E. (1957). “The form of instantaneous unit hydrograph.” Int Assoc Sci and Hydrology 45(3):114–121
Marani A., Rigon R. and Rinaldo A. (1991). “A note on fractal channel networks.” Water Resource Res., 27, 3041-
3049.
Abrahams A. D. “Channel networks: A geomorphological perspective”. Water Resources Research,
20:161–188,1984.
Allam M. N. and Balkhair K. S. “Case study evaluation of the geomorphologic instantaneous
unit hydrograph.” Water Resources Management, 1:267–291, 1987.
Bhaskar N. R., Parida B. P., and Nayak A. K. “Flood estimation for ungauged catchments
using the giuh”. Journal of Water Resources Planning and Management, 123:228–238, 1997.
Bhunya P. K., Berndtsson R., Singh P. K., and Hubert P. “Comparison between Weibull and gamma
distributions to derive synthetic unit hydrograph using Horton ratios”. Water Resources Research, 44, 2008.
Boyd M. J. “A storage-routing model relating drainage basin hydrology and geomorphology”. Water Re-
sources Research, 14:921–928, 1978.
Cudennec C., Fouad Y., Gatot I. S., and Duchesne J. “A geomorphological explanation
of the unit hydrograph concept”. Hydrological Processes, 18, 2004.
de Moura M. M, Beskow S., da Silva Terra F., de Mello C. R, DA
Cunha Z. A., and Cassalho F. “Evaluation of geomorphological approaches combined with digital eleva-
tion models for the Nash’s instantaneous unit hydrograph”. Journal of South American Earth Sciences,
107:103153, 2021.
Dıaz-Granados M., Bras R. L., and Valdes J. B. "Incorporation of channel losses in the geomor-
phologiciuh", 1986.
Fleurant C., Kartiwa B., and Roland B. “Analytical model for a geomorphological instantaneous unit
hydrograph”. Hydrological Processes, 20, 2006.
Gandolfi C. and Bischetti C. B. “Influence of the drainage network identification method on
geomorphological properties and hydrological response”. Hydrological Processes, 11:353–375, 1997.
Gupta V. K. and Waymire E. C. “On the formulation of an analytical approach to hydrologic
response and similarity at the basin scale”. Journal of Hydrology, 65:95–123, 1983.
Horton R. E. “Erosional development of streams and their drainage basins; hydrophysical approach
to quantitative morphology”. Geological Society of America Bulletin, 56:275–370, 1945.
Kim J. C. “Equivalent horton’s ratios of channel network within the framework of variable source
basin area”. Journal of Hydrology: Regional Studies, 2022.
Kumar A. “Geomorphologic instantaneous unit hydrograph based hydrologic response models for un-
gauged hilly watersheds in India”. Water Resources Management, 29:863–883, 2015.
Kumar R., Chatterjee C., Lohani A. K., Kumar S., and Singh R. D. “Sensitivity
analysis of the giuh based clark model for a catchment”. Water Resources Management, 16:263–278, 2002.
Lee K. T. and Yen B. C. “Geomorphology and kinematic-wave–based hydrograph derivation”.
Journal of Hydraulic Engineering, 123:73–80, 1997