J. Indian Water Resour. Soc.

of
Journal Vol.Indian
30 No.Water
1, January, 2010
Resources
Society Vol. 30 No. 1, January, 2010

PROTECTION OF BRIDGE ABUTMENTS FROM SCOUR

V. K. Sarda

ABSTRACT

The failure of bridges due to excessive abutment scour causes high maintenance cost or
even the bridge collapse resulting in interruption of traffic as well as casualties to life. Hence
ability to protect bridge abutment from scour is critical to bridge safety. Review showed that there
are two approaches to scour mitigation viz. bank hardening and flow-altering. Selection of a
countermeasure involves life cycle cost assessment for the particular site along with social and
environmental issues. However, their application for mobile bed and in compound channels is
still to be established.

Key Words : Abutments; Scour protection; Scour countermeasures; Riprap; Spur dyke; Parallel
Walls; Collars; Cable-tied blocks; Geo-bags.

INTRODUCTION In spite of this, the scour at bridge abutments has

Bridge failures due to local scour at bridge
foundations (i.e., bridge abutments and piers) have
heightened interest in scour prediction and scour
countermeasures.
Pier scour and pier scour countermeasures have
been studied by Ettema (1980), Jones (1989), Johnson
(1994), Richardson and Lagase (1999), Mueller and
Landers (1999) and Lagasse et. Al. (2001).On the other
hand, an extensive review, recommendations and design
suggestions for a number of bridge protection devices
are available in Kumar (1998) and Parker et.al. (1998).
Also several comprehensive technical manuals [Federal
Highway Administration] (HEC-18, HEC-20 and HEC-
23) for dealing with the problem of bridge scour and its
protection are in the market.
Literature and data showed that the problem of
scouring at bridge abutments is quite significant. A study
(Richardson and Abed, 1993) carried out in 1973 for
the U.S. Federal Highway Administration concluded that
of the 383 bridge failures, 25% involved pier damage
and 72% involved abutment damage. On the other hand,
of the 108 bridge failures surveyed in New Zealand
during the period of 1960 - 1984, 29 were attributed to
abutment scour (Melville, 1992). It was also mentioned
in this study that 70% of the expenditure on bridge
failures in New Zealand was due to abutment scour.
Life Member, Principal, North West Institute of Technology, Dhudike,
Dist. Moga (Punjab)
Paper No. 1193(f)
received less attention, and countermeasures for
abutment scour are greatly needed.
BRIDGE ABUTMENT SCOUR MECHANISM
The general scour mechanism at piers is well
understood after several comprehensive studies (Ettema,
1980; Kumar, 1998; Melville (1975), in which the
combined action of down flow and the horse shoe
vortices and the wake vortices induced by the presence
of the pier have been found to be responsible for the
scour around the pier.
Recent studies by Wong (1982), Tey (1984), Kwan
(1984, 1988), Kandasamy (1985) and Dongol (1994) of
abutment scour have shown that the scour mechanism
at abutments is very similar to the scour mechanism at
piers. The downflow and the principal vortex at the
upstream corner of the abutment, together with the
secondary vortices and wake vortices at the middle part
and the downstream corner of the abutment, result in
complex interactions between the fluids and the bed
material to cause scour at abutments. Observations of
flow patterns around abutments derived from flow
visualization techniques using dye injection, dye crystals
strategically placed on the sand bed, paper floats, smoke
tunnel experiments by Liu et al. (1961) and Gill (1972)
are summarized in Figure 1.
In addition to the vortex systems, seepage effects
have been found (Hagerty and Parola, 1992) to be very

1

important to interact with turbulent vortices to aggravate
local scour. The fluctuating pressure differences,
induced by the flow separation at abutments, cause
seepage into and out of the abutment foundation. This
results in ejection of sediment particles from the bed
where seepage emerges beside the abutment.
Fig. 1 Flow pattern around an abutment
On the other hand, Molinas et al.(1988)
experimentally studied the shear stress distribution
around vertical wall abutments. It was found that for
Froude numbers ranging from 0.30 to 0.90 and for
protrusion ratios of 0.1, 0.2 and 0.3, the highest values
of shear stresses occurred at the upstream abutment
corner. In this study, shear stresses around vertical wall
abutments were found to be amplified up to a factor of
10 depending upon flow conditions and abutment
protrusion ratios. It was also found that shear stress
amplification due to local effects at the nose region of a
vertical wall is a function of the opening ratio and turning
angle. Shear stress amplification due to channel
restrictions, on the other hand, were found to be a
function of opening ratio, approach Froude number, and
protrusion length. The formulas for estimating shear
stress amplification due to local effects at the nose region
of vertical wall abutments and for estimating shear stress
amplification due to channel restriction were also
proposed. The sum of the two shear stress amplifications
equals the total nose shear stress amplification.
In a study by Ahmad and Rajaratnam (2000) for
flow around a 45° wing-wall bridge abutment, it was
found that the approach flow turns into a complex 3D
skewed flow in the upstream and surrounding regions
of the abutment. The bed shear stress was found to
increase substantially near the abutment, reaching a
peak value of   o  3.63 at the abutment nose, with 
and  o being the shear stress and the approach shear
stress at the bed respectively. It was also found that
the skewing of the flow around the bridge abutment is
greater than flow around bridge piers.
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
SCOUR PROTECTION DEVICES FOR
ABUTMENTS
The successful countermeasures for local scour at
abutments work by either diverging the erosive flow
away from the structure and are called flow altering
countermeasures or making the area around the
abutment more resistant to erosion which are called
bank hardening countermeasures
In HEC-23 (Lagasse et al. 1997), a countermeasure
matrix has been organized to highlight the various groups
of countermeasures and to identify their individual
characteristics. These countermeasures have been
organized into groups based on their functionality with
respect to scour and stream instability.
Flow Altering Countermeasures
They reduce flow’s energy to scour.
Spur dykes
Spur dykes have been studied intensively by Lacey
(1929), Inglis (1949), Laursen (1952, 1962), Garde et
al. (1961), Gill (1970), Cunha (1973), Franco (1982),
Copeland (1983), Rajaratnam and Nwachukwu (1983),
Zaghloul (1983), Brown (1985), Suzuki et al. (1987),
Kwan and Chaudhary (1992), Wu and Lim (1993),
Mayerle (1995), Shields et al. (1995), Tominaga et al.
(1997), Zhang and Du (1997), Soliman et al. (1997) and
Kuhnle et al. (1997, 1998, 1999), as river training or
river rehabilitation structures instead of abutment scour
countermeasures. Spur dike length, alignment with flow,
flow structure around spur dykes, construction materials
and many other parameters have been investigated.
Scour depth predictors around spur dykes and spur dyke
design guide lines have also been provided.
Spur dykes are used to alter flow direction, induce
deposition, or reduce flow velocity [Fig. 2]. Their main
use is to protect banks that contain bridge abutments
from eroding. Spur dykes are commonly used to realign
streams as they approach a bridge abutment. A bridge
abutment may be in danger of being severely eroded
when it is subjected to high velocity flow from a channel
that has changed course due to meandering of the
channel. Spur dykes may also be used to establish and
maintain the alignment of a channel. They have been
used to decrease the length of the bridge required and
reduce the cost and, maintenance of the bridge in
actively migrating braided channels (Lagasse et al.,
2001).
2

Fig. 2 Definition Sketch of a spur dyke
Liu et al. (1961) conducted clear water and live-
bed laboratory experiments on a compound channel using
spur dykes as scour protection device. It was found
that spur dykes can efficiently protect the abutment
provided they are properly designed. Design guidelines
were also discussed. The most effective configuration
to prevent local scour at the abutment consisted of three
spur dykes composed of rock located upstream of the
abutment and two at the corners.
Hard Points
HEC-23 (Lagasse et al., 2001) has proposed hard
points as one of the scour protection countermeasures.
They consist of stone fills spaced along an eroding bank
line, protruding only short distances into the channel. A
root section extends landward to preclude flanking.
Hardpoints are most effective along straight or relatively
flat convex banks where the streamlines are parallel to
the bank lines and velocities are not greater than 3 m/s
within 15 m of the bank line. Hardpoints may be
appropriate for use in long, straight reaches where bank
erosion occurs mainly from a wandering thalweg at
lower flow rate. They would not be effective in halting
or reversing bank erosion in a meander bend unless
they were closely spaced, in which case spurs, retarder
structures, or bank revetment would probably cost less
(HEC-23).
Guide banks or Parallel Walls
Guide banks are earth or rock embankments placed
at abutments to improve the flow alignment and move
the local scour away from the embankment and bridge
abutment [Fig. 3]. The major use of guide banks has
been to prevent erosion by eddy action at bridge
abutments or piers where concentrated flood flow
traveling along the upstream side of an approach
embankment enters the main flow at the bridge (HEC-
23). Spring (1903), Neil (1973), Bradley (1978), Brice
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
Fig. 3 Parrallel wall (Guide banks) countermeasure
and Blodgett (1978), Smith (1984), Richardson and
Simons (1984) and Lagasse et al. (2001) have studied
guide bank orientation, length, crest height, shape and
size, downstream extent and many other concerns.
Design guidelines for guide banks have been
provided by Neil (1973), Bradley (1978), Ministry of
Work and Development (1979), Central Board of
Irrigation and Power (1989) and Lagasse et al. (1996).
Guide banks provide a smooth transition for flow on the
floodplain to the main channel. The effectiveness of
guide banks is a function of river geometry, quantity of
flow on the floodplain, and size of bridge opening
(Richardson and Simons, 1984). By establishing smooth
parallel streamlines in the approaching flow, guide banks
improve flow conditions in the bridge waterway. Scour,
if any, is near the upstream end of the guide bank away
from the bridge.
Guide banks can protect not only bridge abutments
from local scour, but also the approach embankment
because of the still water area behind it. When
embankments span wide floodplains, the flows from high
waters must be aligned to flow smoothly through the
bridge opening. Overbank flows on the floodplain can
severely erode the approach embankment and could
increase the depth of the scour at the bridge abutment.
They can be used to redirect the flow from the
embankment and to transfer the scour away from the
abutment. They also serve to reduce the separation of
flow at the upstream abutment face and maximize the
total bridge waterway area and reduce the abutment
scour by lessening the turbulence at the abutment face
(Lagasse et al., 2001)
There are practically two kinds of guide banks. One
is the American practice, which is to give guide banks
an elliptical form convergent to the opening, whereas
the other one used in Pakistan and India gives that guide
banks should be straight and parallel to the opening with
3

a curved section at the upstream and downstream ends.
Parallel guide banks straighten the flow more effectively
than convergent ones (Richardson and Simon, 1984).
Properly designed guide banks and abutment riprap
provide an acceptable alternative to designing bridge
abutments to withstand the development of the full scour
prism predicted by equations currently available for
estimating abutment scour (Hua et al., 2006). Since they
have excellent performance in the field, guide banks
should be given serious considerations as scour
protection device for abutment scour when developing
plans for repair or replacement of those endangered
bridges. They can eliminate many scour problems
associated with bridge crossings and their use can result
in a worthwhile savings to the highway program (Hua
et al., 2006).
Collars
Collar is a protection device that has been found to
be helpful for arresting scour around bridge pier.
Basically it is a piece of circular shaped steel plate
attached around the abutment [Fig. 4] sitting horizontally
a short distance from the bed. Much literature on
attachment of collars on abutment, as scour protection,
is not available. However, Liu et al (1961) studied flat
horizontal steel collars around wing-wall abutment
ending at the mean channel under clear water flow
conditions in a laboratory flume. It was found that these
collars were able to protect the bridge abutment
efficiently by eliminating secondary vortices that
ordinarily would cause local scour. The minimum collar
dimensions that eliminated local scour were 0.7L  0.23L
[where L is abutment length perpendicular to flow
direction] provided at 0.08y [y is the mean channel flow
depth] below the mean sediment elevation gave the test
results of scour reduction. They also retarded the
development of scour hole.
Fig. 4 Collar device at abutment
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
Bank Hardening Countermeasures
They comprised of various hard materials located
on the bed and banks in the vicinity of abutment to
increase the ability of the bed and bank to resist scour
by flow.
Riprap
Riprap is the most common countermeasure
employed and consists of large rocks arranged flush
with the bed and banks in several layers of thickness. A
number of researchers like Simons and Lewis (1971),
Lewis (1972), Mackey (1986), Simons and Li (1989),
Croad (1989), Pagan-Ortiz (1991) and Eve (1999) have
conducted research on the performance of riprap
protection on abutment slopes and aprons under clear-
water conditions while Melville et al. (2006) studied the
same under live bed conditions. In these studies, failure
of riprap beds has been observed due to (a) dislodging
of the individual rocks due to excessive stream velocity,
(b) dislodging of individual rocks at the edge of the riprap
blanket due to the flow undermining and lifting the rocks
up and into direct contact with the flow, and (c) sinking
of the riprap blanket due to winnowing of the fine bed
material up through the rocks where it is carried away
by the flow [Fig. 5].
Fig. 5 Riprap at abutment
Design consists of the specification of the rock size
to avoid direct dislodging, riprap blanket thickness, the
lateral extent of the blanket to avoid edge failure, the
gradation of riprap, and a filter material to avoid
winnowing of the fines. The size of riprap stone can be
determined from (Pagan-Ortiz, 1991):
0.81
 1.064U 2 y 0.23 
D50    (1)
 S s  1g 
Here D50 is the median riprap size, U is the cross-
sectionally averaged water velocity, Ss is the specific
4
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

gravity of the riprap material, and g is the gravitational include Parker et al. (1998), Jones et al. (1995), Hoe
constant. (2001), Cheung (2002) and Melville et al. (2006). Three
possible failure mechanisms for cable-tied blocks used
The thickness of the riprap blanket, t = 1.5D100 , as scour protection at bridge piers were identified
where D100 is the largest size of the rip rap stone (Parker et al., 1998). These are overturning and rolling
(Lagasse et al., 2001). The lateral extent of the riprap up of the leading edge, uplift of the center of the mat,
blanket can be found form: and winnowing of sediment between the mat and the
Wmin  C1 d s  d b  D50  (2) bridge pier. The latter is more likely if the mat is not
sealed tightly to the pier, or wing wall. Anchoring of the
Here, Wmin is the minimum riprap blanket extent leading edge of the mat is important to prevent
across the channel, C1 is constant equal to 1.68 and overturning and rolling up.
1.19 at the upstream and downstream comers of the
Design issues include primarily the block size, lateral
riprap layer, respectively, ds is the depth of equilibrium
extent, and edge treatment. Block size can be estimated
scour, and db is the depth of the riprap blanket bottom
by the following equation:
below the average channel bed level (Fig. 5). ds can be
found for bedform-dominated cases by: Hb  a cb   2
   Fr (4)
d s  C2 H (3) y   cb   1  p  
Here H is maximum bed-form height and C2 is a Here, Hb is the height of the block, y is the flow
constant values for which are 1.2 and 1.0 for the depth, acb is a constant equal to 0.1,  cb is the block
upstream and downstream comers of the riprap layer, density,  is the fluid density, and Fr is the Froude
respectively (Melville et al., 2006). Otherwise, add other Number.
scour components to ds. The proper gradation of riprap
can be found Table 1. CTB blocks are typically manufactured as a
truncated pyramid shape with a square base and top.
Table 1. Riprap gradation for abutment protection
The spacing between CTB units should be adequate to
Stone Size Percentage of allow the mattress to have a sufficient degree of
Range gradation smaller than flexibility, and that block shape should not inhibit mat
1.5D50 to 1.7D50 100 flexibility. Typically, synthetic filters are used beneath
1.2D50 to 1.4D50 85
CTB mats. Lateral extent of the cable-tied block
1.0D50 to 1.1D50 50
0.4D50 to 0.6D50 15
mattress can be determined from:

To design the filter material the pore space should W = 1.55(ds -db)
be finer than the natural riverbed material. (5)

Cable-tied Blocks Here W is the apron width, ds is the scour depth (=

mat settlement depth) at the outer edge of the mat, and
Cable-tied block consist of a series of blocks linked db is the placement (burial) depth of the mat [Fig. 6].
together with cable to hold them together as a coherent
mat. Where riprap of adequate size is unavailable or
where environmental or geometric constraints preclude
use of riprap, alternatives to riprap, such as cable-tied
blocks, are necessary.
Cable-tied blocks (CTB) comprise concrete blocks
interconnected with metal or non-metallic cables
(Prezedwojski et al., 1995). CTB have the advantages
of ease of construction, minimal encroachment into the
river channel, and a lower weight than riprap per unit
area covered. Previous studies and experiments on the
Fig. 6 Cable tied blocks at abutment
use of CTB for scour protection of bridge foundations

5

To prevent the uplifting of the leading edge of the
mat, the size can be determined from
H b  158  2 n 2
  Fr 0.33 (6)
y  S cb  1  y installation of additional countermeasures as the
Here Scb is the specific gravity of the blocks and n
is the Manning coefficient. Care needs to be taken to
ensure that the leading edge of the mat remains buried.
Geobags
Geobags are bags of pervious material that are filled
with a pervious granular material (sand or gravel) that
are used as bank hardening elements, thereby possessing
enough weight to hold sediment in place, but allowing
the flow of water through them to reduce uplifting
pressure to reduce the likelihood of uplifting of the bag
or winnowing of the fines underneath. The bag material
can be a geosynthetic fabric such as the filter layer of
riprap discussed above. Design considerations include
sizing, linking of bags, angle of placement and placement
extent (Korkut et al., 2006).
Minimum size can be determined by that of
equivalent riprap as mentioned above. The individual
bags should be tied together to help them function as a
single mattress thereby allowing flexibility to conform
to the irregular bed shape. The geobag mattress should
have a maximum slope of 2H: 1 V with a toe extending
a downward length equal to at least 2 bags into the
riverbed.
CONCLUSIONS
Although many studies have been carried out for
each of these countermeasures, there is still need for
further studies of their use in protecting bridge
abutments. For instance, spur dikes have not specifically
been tested and used for abutment protection. Guide
banks have been tested and used for spill-through
abutments on rivers with wide floodplains or rectangular
channel only. For small country bridges with wing-wall
abutments terminating on the main channel bank, studies
are still needed. In addition, there are many controversial
design guidelines and small design issues that need to
be clarified. Collars have been tested for pier scour
protection rather than for abutment protection.
Countermeasures are often damaged or destroyed
by the stream, and stream banks and beds often erode
at locations where no countermeasure was installed.
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
However, as long as the primary objectives are achieved
in the short-term as a result of countermeasure
installation, the countermeasure installation can be
deemed a success (HEC-23). Therefore, to achieve
long-term protection, maintenance, reconstruction, and
responses of streams and rivers to natural and man-
induced changes is needed.
Among armouring countermeasures, riprap and
cable tied blocks are of the most interest by hydraulic
engineers. These two countermeasures have been
investigated by various researchers as erosion control
devices and bank revetments.
However, all these countermeasures are needed to
be tested for oblique flow, for their effectiveness in
cohesive soil and stratified beds. Also no working
relationship for determination of maximum scour depth
with any of the above countermeasures on abutment,
over a wide range of flow conditions and geometries is
available. In addition, information on temporal variation
of scour and location of deepest scour point in the flow
past abutments using most of these countermeasures is
still lacking.
REFERENCES
1. Ahmed, F and Rajaratnam, N.2000. Observations on
flow around an abutment. J. Engineering Mechanics,
ASCE, 125 (1) 51-59.
2. Bradley, J. N. 1978. Hydraulics of Bridge Waterways.
Hydraulic Design Series No.1, U.S. Dept.
Transportation, Federal Highway Administration, 2nd
Ed., Washington, D.C.
3. Brice, J. C. and Blodgett, J. C. 1978. Countermeasures
for hydraulic problems at bridges. Federal Highway
Administration, Report FHW A-RD- 78-162 and Report
FHW A-RD- 78-163.
4. Brown, S. A. 1985. Design of spur-type stream bank
stabilization structures. Final Report. FHWA/RD-84-
101, Federal Highway Administration, Washington, DC.
5. Central Board of Irrigation and Power. 1989. River
behavior management and training. Edited by C. V.J.
Verma, K.R Saxena and M.K. Rao, Publication No. 204,
Vol. 1, Central Board of Irrigation and Power, New
Delhi, India, pp. 469.
6. Cheung, K. M.2002. Cable-tied blocks as a
countermeasure. ME thesis, Civil and Environmental
Engg Dept., The Univ. of Auckland, Auckland, New
Zealand.
6

7. Copeland, R. R. 1983. Bank protection techniques using
spur dykes. Hydraulics Laboratory, U. S. Army
Waterways Experiment Station, Vicksburg, Mississippi,
USA.
8. Croad, R. N. 1989. Investigation of the pre-excavation
of the abutment scour hole at bridge abutments. Rep.
No. 89-A9303, Central Laboratories, Works and
Development Services Corporation (NZ) Ltd., Lower
Hutt, New Zealand.
9. Cunha, L. Veigda. 1973. Discussion on “Erosion of
Sand Bed around Spur Dikes.” Journal Hydraulic
Division, ASCE, Vol. 98, No. HY9.
10. Dongol, D.M.S. 1994. Local scour at bridge abutments.
Report No. 544, University of Auckland, School of
Technical Engineering, Department of Civil
Engineering Private Bag, Auckland, New Zealand.
11. Ettema, R. 1980. Scour at bridge piers. Ph. D thesis,
University of Auckland, Department of Civil
Engineering, Report No. 216.
12. Eve, N. J. 1999. Riprap protection of bridge abutments.
ME thesis, Civil and Resource Engineering Dept., The
Univ. of Auckland, Auckland, New Zealand.
13. Franco, J. J. 1982. Summary report: model-prototype
comparison study of dyke systems. Mississippi River.
U.S. Army Engineer Waterways Experiment Station,
Vicksburg, Mississippi, Technical Report HL-82-11, 55.
14. Garde, R. J., Subramanya, K. and Nambudripad, K. D.
1981. Study of scour around spur dykes. Journal of the
Hydraulics Division, ASCE, 87(HY6), 23-37.
15. Gill, M. A. 1970. Bed erosion around obstructions in
rivers. Ph.D thesis, The University of London (Imperial
College of Science and Technology).
16. Gill, M. A. 1972. Erosion of sand beds around spur
dikes. Journal of Hydraulics Division, 98(HY9), 1587-
1602.
17. Hagerty, D. A. and Parola, A. C. 1992. Seepage
influence on stability of bridge abutments. Conf. Proc.
Hydraulic Engg. ASCE, pp. 900.
18. Hoe, D. A. 2001. Cable-tied block protection of bridge
abutments. Project Rep. No. PCRE 01: 18, Dept. of Civil
and Resource Engineering, Univ. of Auckland,
Auckland, New Zealand.
19. Hua, L., Kuhnle, R. A.and Barkdoll, B. D. 2006.
Countermeasures against scour at abutments. National
Sedimentation Laboratory, Rep. No. 49, USDA,
Mississippi, Jan.
20. Inglis, C. C. 1949. The behavior and control of rivers
and canals. Research Publication No. 13, Parts I and
II, Central Waterpower Irrigation and Navigation
Research Station, Poona, India.
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
21. Jones, J. S. 1989. Laboratory Studies of the Effect of
Footings and Pile Groups on Bridge Pier Scour. Proc.
of 1989 Bridge Scour Symposium, FHW A, Washington,
D.C.
22. Jones, J. S., Bertoldi, D. A. and Stein, S. 1995.
Alternative scour countermeasures.” Proc. Hydraulic
Engineering ‘95, ASCE, New York, pp.1819-1823.
23. Johnson, P. A. 1994. Quantification of bridge pier scour
uncertainty. Machine Intelligence and Pattern
Recognition, 17, 407.
24. Kandasarny, J. K. 1985. Local scour at skewed
abutments. School of Engineering, Report No. 375,
University of Auckland.
25. Kandasamy, J. K. 1989. Abutment scour. Rep. No. 458,
School of Engrg., University of Auckland, Auckland,
New Zealand.
26. Khan, K. W. and M. H. Chaudhry, M. H. 1992. Numerical
Modeling of Flow around Spur Dykes. Proc. 4th Int.
Conference of Hydraulic Engineering Software
HYDROSOFT/92, pp 223.
27. Kuhnle, R. A., Alonso, C.V. and Shields, F.D. 1997.
Volume of scour holes associated with spur dikes. Proc.
27th Cong. Int. Assoc. Hydraulic Research Part: B-1,
pp. 418.
28. Kuhnle, R. A., Alonso, C.V. and Shields, F.D. 1998.
Volume of scour holes associated with spur dikes. Proc.
1998 Int. Water Resources Engg. Conf. Part 2 (of 2) v 2,
ASCE, pp.1613.
29. Kuhnle, R. A., Alonso, C.V. and Shields, F.D. 1999.
Geometry of scour holes associated with 90-degree spur
dikes. J. Hydraulic Engg., 125 (9), 972.
30. Kumar, V. 1998. Effect of Down-flow on Scour around
Bridge Pier. J. of The Institution of Engineers (India),
Vol. 79, May, pp 41-47.
31. Korkut, R., Martinez, E. J., Ettema, R. and Barkdoll,
Brian. 2006. Geobag performance as scour
countermeasure for wing wall abutments. J. Hydraulic
Engg., ASCE.
32. Kwan, F. 1984. Study of Abutment Scour. Report No.
328, University of Auckland, School of Engineering,
Department of Civil Engineering Private Bag,
Auckland, New Zealand.
33. Kwan, F. 1988. Study of Abutment Scour. Report No.
451, University of Auckland, School of Engineering,
Department of Civil Engineering Private Bag,
Auckland, New Zealand.
34. Lacey, G.1929. Stable channel in Alluvium. J. of
Institution of Engineers, Paper No. 4736, Vol. 229.
35. Laggasse, P.F., Richardson, E.V. Zevenbergen, L. W.
1996. Design of guide banks for Bridge Abutment
7

Protection. Proc. North American Water and
Environment Congress and Destructive Water, ASCE,
New York, pp. 4188-4197.
36. Lagasse, P. F., Zevenbergen, L.W., Schall, J. D. and
Clopper, P. E. . 2001. Bridge Scour and Stream
Instability Countermeasures. Publication No. FHWA
NHI 01-003, Hydraulic Engineering Circular No. 23,
U. S. Department of Transportation, Federal Highway
Administration. pp. 2.7, 2.9, 4.6, 6.16, 6.18, Design
Guidelines 1, 9, 10.
37. Laursen, E. M. 1952. Observation on the nature of scour.
Proc. 5th Hydraulics Conference, Iowa.
38. Laursen, E.M. 1962. Discussion of “Study of scour
around spur dikes. Journal of the Hydraulics Division,
ASCE, 89(HY3), 225-228.
39. Lewis, G. L.1972. Riprap protection of bridge footings.
Doctor of Philosophy thesis, Colorado State Univ., Fort
Collins, Colo.
40. Liu, M. K., Chang, F. M. and Skinner, M. M.1961. Effect
of bridge construction on scour and backwater. Report
No. CER60-HKL22, Department of Civil Engineering,
Colorado State University, Fort Collins, Colorado.
41. Macky, G. H. 1986. Model testing of bridge abutment
scour protection. Rep. No. 3-86/12, Central
Laboratories, Ministry of Works and Development,
Lower Hutt, New Zealand.
42. Mayerle, R., Toro, F. M. and Wang, S. S. Y. 1995.
Verification of a three-dimensional numerical model
simulation of the flow in the vicinity of spur dikes. J.
Hydraulic Research, 33 (2), 243.
43. Ministry of Works and Development.1979. Code of
practice for the design of bridge waterways. Civil
Division Publication CDP 705/C, Ministry of Works
and Development, Wellington, New Zealand, pp. 57.
44. Melville, B. W. 1975. Local scour at bridge sites. School
of Engineering, Report No. 117, University of Auckland,
New Zealand.
45. Melville, B. W. 1992. Local Scour at Bridge Abutments.
Journal of Hydraulic Engineering, ASCE, Vol. 118,
No.4, April, pp. 615.
46. Melville, B.W., Ballegooy, S. Van, Coleman, S. and
Barkdoll, B. 2006. Scour countermeasures for wing
wall abutments. J. of Hydraulic Engineering, ASCE,
132, (6), 563-574.
47. Melville, B. W., van Ballegooy, R. and van Ballegooy, S.
2006. Flow induced failure of cable-tied blocks. J.
Hydraulic Eng., 132 (3), 235- 245.
48. Molinas, A., K., Kheireldin, K. and Wu Baosheng, Wu.
1988. Shear stress around vertical wall abutments. .J.
Hydraulic Engg,, ASCE, 124 (8), 822.
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
49. Mueller, D. S. and M. N. Landers, M. N. 1999. Portable
Instrumentation for Real-Time Measurement of Scour
at Bridges. Federal Highway Administration
Publication No. FHW A-RD-99-085 (FHW A approval
pending), Turner-Fairbank Highway Research Center,
McLean, VA.
50. Neil, C. R. 1973. Guide to bridge hydraulics. Roads
and Transportation Assoc. of Canada, Univ. of Toronto
Press, Toronto, Canada.
51. Pagan-Ortiz, J. E.1991. Stability of rock riprap for
protection at the toes of abutment located at the flood
plain. Rep. No. FHWA-RD-91-057, Federal Highway
Administration, U. S. Dept. of Transportation,
Washington, D.C.
52. Parker, G., C. Toro-Escobar, C. and Voigt Jr., R. L. 1998.
Countermeasures to protect bridge piers from scour.
Final Report (Project NCHRP 24-7) prepared for
National Co-operative Highway Research Program,
University of Minnesota, Minneapolis, Minnesota,
U.S.A., pp. 402.
53. Przedwojski, B., Blazejeski, R. and Pilarczyk, K. W.
1995. River training techniques. Balkema, Rotterdam,
The Netherlands.
54. Rajaratnam, N. and Nwachukwu, B. A. 1983. Erosion
near groyne-like structures. Journal of Hydraulic
Research, IAHR, 21 (4), 277 - 287.
55. Richardson, E. V. and Abed, L. 1993. Top width of pier
scour holes in free and pressure flow. Proc. Nat. Conf.
Hydraulic Engg. Part 1 (of 2), ASCE, July, pp. 25-30.
56. Richardson, E. V. and Lagasse, P. F. 1999. Stream
Stability and Scour at Highway Bridges. Compendium
of Papers ASCE Water Resources Engineering
Conferences 1991 to 1998.
57. Richardson, E. V. and Simons, D. V. 1984. Use of spurs
and guide banks for highway crossing. Proc.
Transportation Research Record, 2 nd Bridge Engg.
Conf. 2, 184.
58. Shields Jr., F. D. C., Cooper, C. M. and Knight. S. S.
1995. Experiment in stream restoration. J. of Hydraulic
Engg, 121 (6), 494-502.
59. Simons, D. B. and Lewis, G. L. 1971. Flood protection
at bridge crossings. C.S.U. Civil Engineering Rep. No.
CER71-72DBS-GL10, prepared for the Wyoming State
Highway Dept. in conjunction with the U.S. Dept. of
Transportation, Washington, D.C.
60. Simons, D. B. and Li, R. M. 1989. Sizing riprap for the
protection of approach embankments and spur dykes
and limiting the depth of scour at bridge piers and
abutments. Rep. No. FHWA-AZ- 89-260, Arizona Dept.
of Transportation, Vol. 1, Pheonix.
8
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

61. Smith, C. D.1984. Scour Control at Outlook Bridge—a NOTATIONS

Case Study. Canadian Journal of Civil Engineering,
11 (4), 709-716. ac constant equal to 0.1
C1 constant equal to 1.68 and 1.19 at the upstream
62. Soliman, M. M. K., Atria, K.M. A., Kotb, A.M. Talaat
and Ahmed, A. F.1997. Spur dike effects on the river and downstream comers of the riprap layer,
Nile morphology after high Aswan dam. Proc., Congo respectively
Int. Assoc. Hydraulic Research, Part-V, pp. 805. C 2 constant the values for which are 1.2 and 1.0 for
the upstream and
63. Spring, F. J. E. 1903. River training and control of the
guide bank system. Technical Paper No. 153, Railway downstream comers of the riprap layer
Board, Government of India, New Delhi. db depth of the riprap blanket bottom below the
average channel bed level
64. Suzuki, K., Michiue, M. and Hinokidani, O. 1987. Local
ds depth of equilibrium scour
bed form around a series of spur dikes in alluvial
channel.” Proceedings 22nd Congress, IAHR Lausanne,
D 50 the median riprap size
Switzerland, pp. 316-321. D 100 largest size of the rip rap stone
Fr Froude Number.
65. Wong, W. H. 1982. Scour at Bridge Abutments. Report
g gravitational constant
No. 275, Dept. of Civil Engineering, University of
Auckland, Auckland, New Zealand.
H maximum bed-form height
H b height of the cable tied-block
66. Tey, C. B. 1984. Local scour at bridge abutments. Report L abutment length perpendicular to the flow direction
No. 329, University of Auckland, School of Engineering,
n Manning coefficient
Department of Civil Engineering, Private Bag,
Auckland, New Zealand.
p porosity of cable-tied blocks
Scb specific gravity of the blocks
67. Tominaga, A, Nagao, M. and Nezu, I. 1997. Flow S s specific gravity of the riprap material
structures and mixing processes around porous and t thickness of the riprap blanket
submerged spur dikes. Proc. 27th Congress of the Int.
U cross-sectionally averaged water velocity
Assoc. of Hydraulic Research. IAHR, Part B-1, pp. 251.
W apron width
68. Wu, X. and Lim, S.Y. 1993. Prediction of maximum scour Wmin minimum riprap blanket extent across the channel
depth at spur dykes with adaptive neural networks. y flow depth
Civil-Comp93, Part 3: Neural Networks and
 fluid density
Combinatorial Optimization in Civil and Structural
Engineering Civil-Comp93, pp. 61.  cb cable-tied block density
 shear stress at the bed
69. Zaghloul, N. A. 1983. Local scour around spur-dykes.  o approach shear stress
J. of Hydrology, 60, 123-140.
70. Zhang, Y. and Du, X.1997. Limited scour around spur
dyke and the evaluation of its depth. J. Xi’an Highway
Transportation University, 17 (4), 56.

9
 J. Indian Water Resour. Soc.of
Journal Vol.Indian
30 No.Water
1, January, 2010
Resources
Society Vol. 30 No. 1, January, 2010

RAINFALL PATTERN IN NORTHERN KERALA

P. A. Lisha, P. K. Pradeep Kumar and K. V. Jayakumar

ABSTRACT

Available daily rainfall data for 81 years (1901 – 1969, 1990 – 1996 and 2001 – 2005) of
five stations Kannur, Kozhikode, Quilandy, Vythiri and Vadakara in northern Kerala have been
transformed into annual, southwest monsoon, northeast monsoon and non monsoon components
for each year and analyzed for a study of the rainfall behaviour. Regression analysis of south-
west monsoon rainfall and annual rainfall showed a variation in intercept from 734.60 to 1153.86
and in slope from 0.87 to 1.07 with standard error of estimate varies from 282.93 to 389.82.
Correlation analysis shows the variation in coefficient from 0.80 to 0.95. Maximum values of
annual and southwest monsoon rainfall yield a parallel behaviour. Average annual and south-
west monsoon rainfall values also yield a parallel behaviour. Overall southwest monsoon rain-
fall is 76.99% of annual rainfall in the area of study. The variation of annual and southwest
monsoon rainfall show a regular pattern in increase or decrease.

Key words: Kerala monsoons, Regression analysis, RFA, RFSWM, RFNEM, RFNM

INTRODUCTION Floods and landslides are a common occurrence

Kerala experiences rainfall for nearly nine months
in a year in the form of monsoon rain and
thudershowers. The fluctuation in annual rainfall of
Kerala is much less than that compared with other parts
of the Indian sub-continent. The hills and mountains of
the Western Ghats located on the eastern boundary of
the State provide orographic lifting for the southwest
monsoon winds resulting in heavy precipitation over the
western slopes and good rain over midlands and low
lands. The average annual rainfall of the State is 3500
mm. The southwest monsoon (June-September) is the
principal rainy season as the State receives about 70%
of its annual rainfall. Southwest monsoon rainfall as
percentage of annual rainfall decreases from north to
south and varies from 83% in the northern most District
of Kasargod to 50% at southern most District of
Thiruvananthapuram. Northeast monsoon rainfall as
percentage of annual rainfall increases from north to
south and varies from 9% in northern most District of
Kasaragod to 27% in southern most District of
Thiruvananthapuram.
Centre for Water Resources Development and Management
P.O. Kunnamangalam, Kozhikode - 673 571, Kerala
Paper No. 1228
during the rainy seasons in Kerala. Heavy rains during
southwest monsoon period (June-September) and
northeast monsoon (October – December) result in
increased risk of high floods. Rivers in Kerala are
generally narrow and short, which flow down from the
Western Ghats through steep gorges and are therefore
highly susceptible to flash floods which cause immense
damage to the embankments, corps and cattle.
Landslides give rise to blockages, snapping of
communications and damage to the property. It is
essential to have a detailed knowledge on different
patterns of rainfall during different seasons at a
particular region for various planning and execution
purposes. In that light, this particular study has been
carried out to know the different rainfall patterns at
five locations of northern Kerala. The outcome of the
analysis may help various developmental agencies in
their activities.
The interaction between topography and
meteorological elements involves several basic
characteristics of any relief feature. The overall
dimensions and the orientation of a mountain range with
respect to prevailing winds are important for large scale
processes, relative relief and terrain shape are
10

particularly important on a regional scale, while slope
angle and aspect cause striking local differentiation of
climates. The effects that an orographic barrier
produces on air motion depend first on the dimensional
characteristics of the barrier – its height, length, width
and the spacing between successive ridges – and,
second, on the properties of the airflow itself – the wind
direction relative to the barrier, the vertical profiles of
wind and of stability. Each of the three dimensions of a
mountain barrier interacts with a particular atmospheric
scale parameter (Smith, 1979). The air arriving at a
barrier must have sufficient kinetic energy in order to
rise over it against the force of gravity (Stringer, 1972).
Cloud type in mountain areas is primarily determined
by air mass characteristics and is therefore related to
the regional climate conditions. The spatial distribution
of convective upcurrents in mountain regions shows
some pronounced effects of topography. There may
be strong contrasts between shaded and sunny slopes.
Fujita et al (1968) reported rapid cumulus build-up on
the slope of Mt.Fuji between 0845 – 0915 hrs in July as
the solar altitude increased from 47° to 53° and surface
temperature on the rocky slopes exceeded 30° C.
Verma (1993) correlated the monsoon rainfall values
with the gridded surface air temperature over northern
hemisphere land at various time lags of months to identify
teleconnections of monsoon with the northern
hemisphere surface air temperature anomalies. As per
the study, two regions in the higher latitudinal belt of
40°N – 70°N over North America and Eurasia show
positive correlations with temperatures during northern
winter. The region located over northwest India and
adjoining Pakistan show maximum positive correlation
during the pre-monsoon months of April and May.
These relationships suggest that cooler northern
hemisphere during the proceeding seasons of winter/
spring over certain key regions are generally associated
with below normal summer monsoon rainfall over India
and vice versa which could be useful for predictions
for long – range forecasting of monsoon.
Systematic and scientific studies on variation of
precipitation with elevation are limited mainly because
of lack of sufficient information on the amount of
precipitation at higher elevation. This is due to non -
availability of automated recording precipitation gauges
and problem associated with measurement of
precipitation at such higher elevation on a routine basis.
Nearly 35% of the geographical area in India is
mountainous. Of these nearly 58 % is accounted for
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
by the mighty Himalayas, extending from north-west to
east. Besides, the Khasi and Jayantiya hills in the
northeast, the Vindhya and Satpura hills in central India,
the Western Ghats running all along the west coast from
Maharashtra to Kerala and the broken hill ranges of
Eastern Ghats largely determine and guide the Country’s
rainfall pattern during the summer as well as winter.
Isolated hill ranges like, the Aravalis and Nilgiris also
influence the rainfall occurrence in those areas.
Dhar et al (1978) carried out a study of the heavy
rainfall stations in India. For the purpose of the study,
stations with mean annual rainfall of 500 cm were
considered as heavy rainfall stations. Some of the heavy
rainfall stations lie in the Western Ghats and the rest
are located in the hills of northeast India. There are,
however, none in the Himalayan region. There are some
stations in the Darjeeling hills with short period means
over 500 cm. During the onset of the southwest
monsoon, the moisture laden monsoon winds first
approach the Western Ghats and the Khasi Jayantiya
hills and precipitate most of the moisture over these
regions. By the time they approach the Himalayan
regions much of the moisture is lost and, therefore, the
less rainfall in these areas.
Dhar and Bhattacharya (1976) made a study on
the variations of precipitation with elevation in the
Central Himalayas. A relationship between precipitation
and elevation was obtained for the Central Himalayas
using 15 to 20 years data of more than 50 stations.
Variation of rainfall with the elevation showed that there
are two zones of maximum precipitation. One near the
foot of the Himalayas and other at an elevation of 2.0
to 2.4 km. For higher elevations beyond 2.4 km the
precipitation decreases sharply.
Climatologically the onset of the southwest monsoon
over extreme south Kerala is 1 st June. The onset,
however, can take place earlier or later and in some
years there are multiple onsets when the initial onset
takes place too early. Multiple onsets mean that there
is a recession of the monsoon after the initial onset and
another onset of the current that to take place before it
gets established. Between 1901 and 1985, the earliest
onset is 11th May in 1918 and 1955 while the most
delayed onset date is 18th June in 1972 (Menon and
Rajan, 1989). On 47 occasions out of 85 between 1901
and 1985, the onset has taken place between 29th May
and 7th June. During this period, the maximum number
of onset of southwest monsoon occurred on 1st June.
11

Ananthakrishnan et. al (1979), discussed in detail
about the meteorology of Kerala. It is suggested that
the most outstanding feature of the meteorology of
Kerala is the seasonal reversal of the wind circulation
which constitutes the summer and winter monsoons.
This seasonal reversal is linked with the seasonal
reversal of temperature and pressure gradients following
the apparent north-south movement of the Sun in the
course of a year. The seasonal progressions of
temperature, pressure and winds at the surface and in
the upper air are illustrated and the significant points
high lighted. The rainfall time series of 75 rain recording
stations over Kerala for the 80 year period 1901 to 1980
have been statistically examined by Soman et al (1988)
for long term trends. Application of Mann – Kendall
rank statistic test to the time series of annual and
seasonal totals as well as extreme rainfall of 1,2,….,10
day durations revealed a significant decreasing trend in
the rainfall over the eastern high lands and adjacent
areas to the west. This finding is supported by the fact
that the mean rainfall for the second half of the period
is 10 to 20% lower than for the first half over the same
area.
STUDY AREA
Five stations, Kannur, Kozhikode, Quilandy, Vythiri
and Vadakara, in Northern Kerala (Fig.1) are selected
for the study. The locations of the stations in regard to
their respective latitude and longitude are given in Table
1.
Table 1 Details of the raingauge stations
Sl.No. Name of station Latitude Longitude
1 Kannur 11° 52' 75° 22'
2 Kozhikode 11° 15' 75° 47'
3 Quilandy 11° 27' 75° 42'
4 Vythiri 11° 33' 76° 02'
5 Vadakara 11° 36' 75° 35'
DATA
Daily rainfall data of the five stations for 81 years
(i.e., 1901 to 1969, 1990 to 1996 and 2001 to 2005)
have been made available for the present study.
From these data, monthly rainfall data was
calculated and grouped into annual rainfall, southwest
monsoon rainfall, northeast monsoon rainfall and non
monsoon rainfall.
The following are the short forms used in this paper:
RFA - Annual rainfall (January to December)
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
RFSWM - Southwest monsoon rainfall (June to
September)
RFNEM - Northeast monsoon rainfall (October to
December)
RFNM - Non monsoon rainfall (January to May)
RFSWM component has been obtained by adding
the rainfall values of the southwest monsoon months,
viz. June, July, August and September. RFNEM is the
total of the rainfall values of the northeast monsoon
months, viz. October, November and December. RFNM
component is the total rainfall values of non monsoon
months, viz. January, February, March, April and May.
All values of RFA, RFSWM, RFNEM and RFNM are
given in mm.
Fig. 1 Locations of the selected raingauge stations on the relief
map of northern Kerala
ANALYSIS
RFSWM – RFA relationship
It is widely believed that the annual rainfall (RFA)
of India, as a whole is a reflection of the total southwest
monsoon rainfall (RFSWM). On the average, the
RFSWM is around 77% of the RFA of the country. To
examine this contention in Northern Kerala, the 81 year
RFSWM – RFA data of each station is tested for
regression and correlation. The regression constants
of intercept and slope and also the correlation
coefficients are found and given in Table 2. The
standard errors of estimate about the regression line
are also calculated for each station and given in Table
2.
12
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Table 2 RFSWM – RFA relationship values

Parameter Kannur Kozhikode Quilandy Vythiri Vadakara

Intercept 771.52 1153.86 788.79 868.91 734.60
Slope 0.95 0.87 0.99 0.98 1.07
Std. error of estimate 308.59 334.58 312.54 282.93 389.82
Correlation coefficient 0.88 0.80 0.91 0.95 0.86

The RFSWM has been found to correlate well with increases with RFSWM and hence correlation coeffi-
the RFA in Vythiri (correlation coefficient = 0.95) while cient is positive at all stations. The regression lines for
the correlation is less in Kozhikode (correlation coeffi- each station are as shown in Figures 2a – 2e.
cient = 0.80). Also, as RFSWM is a part of RFA, RFA

KANNUR
KOZHIKODE
6000
6000

5000
5000
y =771.52+0.95x
RFA (y) in mm

r = 0.88
RFA (y) in mm
4000 4000 y =1153.86+0.87x
r = 0.80
3000 3000

2000 2000
Observed
Observed
Linear Linear
1000 1000
1000 2000 3000 4000 5000 1000 2000 3000 4000

RFSWM (x) in mm RFSWM (x) in mm

Fig. 2a Relationship between RFSWM and RFA at Kannur Fig. 2b Relationship between RFSWM and RFA at Kozhik

QUILANDY VYTHIRI
6000 9000

8000

5000
7000
y =788.79 + 0.99x y =868.91 + 0.98x
RFA (y) in mm
RFA (y) in mm

r = 0.91 6000 r = 0.95

4000

5000

3000
4000

3000
2000
Observed Observed
2000
Linear Linear
1000 1000
0 1000 2000 3000 4000 5000 1000 2000 3000 4000 5000 6000 7000 8000

RFSWM (x) mm RFSWM (x) in mm

Fig. 2c Relationship between RFSWM and RFA Fig. 2d Relationship between RFSWM and RFA at Vythiri
at Quilandy

13
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

VADAKARA 3918.1 mm at Kozhikode both during the year 1961.

6000 Considering the maximum RFNEM at the five stations,
maximum was 1227.3 mm at Quilandy during the year
5000 y =734.60 + 1.07x 1943 and the lowest was 808.7 mm at Vythiri during
r = 0.86 the year 1902. In the case of maximum RFNM at the
five stations, maximum was 1642.6 mm at Vadakara
RFA (y) in mm

4000
during the year 1942 and the lowest was 1010.7 mm at
3000
Kannur during the year 2004.
Similarly, observing minimum RFA at the five
2000
Observed
stations, the highest value was 2437.5 mm at Vythiri
Linear
during the year 1993 and the lowest value was 1056.9
1000
mm at Quilandy during the year 1964. Among minimum
1000 2000 3000 4000 5000
RFSWM, the highest value was 1637.8 mm at Vythiri
RFSWM (x) in mm
during the year 1918 and lowest value was 889.8 mm
Fig. 2e Relationship between RFSWM and RFA at Vadakara at Quilandy during the year 1964. In the case of
minimum RFNEM at the five stations, the highest value
Maximum and Minimum of RFA, RFSWM,
was 116.0 mm at Vadakara during the year 2003 and
RFNEM and RFNM
the lowest value was 58.0 mm at Quilandy during the
The maximum and the minimum of RFA, RFSWM, year 1963. Considering minimum RFNM, the highest
RFNEM and RFNM observed at the five stations during value was 72.9 mm at Kozhikode during the year 1917
the period of study are given in Table 3 in mm along while the stations Quilandy, Vythiri and Vadakara had
with the respective years of occurrence. only 0 mm rain, all during the year 1990.

Considering maximum RFA at the five stations, Figure 3 represents the variation in the maximum
maximum was 8233.2 mm at Vythiri during the year and minimum values of RFA, RFSWM, RFNEM and
1961. The lowest value among maximum RFA was RFNM at the five stations.
4962.3 mm at Kozhikode during the same year 1961.
From the figure it is clear that maximum RFA and
Among maximum RFSWM at these stations, maximum
maximum RFSWM show a parallel behaviour which
was 6959.5 mm at Vythiri and the lowest value was
occurred in the same years.

Table 3 Maximum and minimum of the respective groups and its year of occurrence
Type Group Kannur Kozhikode Quilandy Vythiri Vadakara
RFA 5660.5 4962.3 5227.0 8233.2 5846.9
YEAR 1961 1961 1959 1961 1907
RFSWM 4663.3 3918.1 4349.5 6959.5 4486.7
MAX YEAR 1961 1961 1959 1961 1907
(mm) RFNEM 920.6 951.5 1227.3 808.7 970.5
YEAR 2002 1932 1943 1902 1932
RFNM 1010.7 1206.0 1163.8 1100.0 1642.6
YEAR 2004 1933 1955 2004 1942
RFA 1497.3 2311.9 1056.9 2437.5 1597.1
YEAR 1968 1911 1964 1993 2003
RFSWM 1270.9 1069.3 889.8 1637.8 1379.2
MIN YEAR 1968 1918 1964 1918 1918
(mm) RFNEM 79.2 68.8 58.0 62.7 116.0
YEAR 1968 1906 1963 1968 2003
RFNM 11.7 72.9 0.0 0.0 0.0
YEAR 1934 1917 1990 1990 1990

14
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

9000

8000

7000

R a in f a ll in m m
6000

5000

4000

3000
M A X R FA
2000 M A X R FSWM
M A X R FNEM
1000
M A X R FNM

0 M IN RFA
M IN RFSWM
Kannur Kozhikode Quilandy Vythiri Vadakara
M IN RFNEM
STATIONS M IN RFNM

Fig. 3 Variation in maximum & minimum of RFA, RFSWM, RFNEM and RFNM at the five stations

It is significant to observe that the maximum values Table 4. and fig. 4 The percentages of RFSWM to
of non monsoon rainfall exceed maximum RFNEM RFA at these stations are also given in the Table.
except at Quilandy. These values also exceed minimum
RFA at Quilandy and Vadakara and exceed minimum The percentage value of the RFSWM to RFA ranges
RFSWM except at Kannur and Vythiri. These values from 73.33% at Kozhikode to 81.37% at Vythiri. The
exceed minimum RFNEM at all the five stations. overall average percentage comes to 76.99%.
Considering the maximum values it is evident that non The overall average values of RFA, RFSWM,
monsoon rainfall values cannot be neglected towards RFNEM, and RFNM were found to be 3494.4 mm,
the assessment of rainfall resources in this region. 2698.2 mm, 409.9 mm and 386.3 mm respectively.
Averages and percentages
Figure 4 shows variation in the averages of RFA,
The average RFA, RFSWM, RFNEM and RFNM RFSWM, RFNEM and RFNM at the five stations.
values observed at the five stations are given in
4500

4000

3500
R a in f a ll in m m

3000

2500

2000

1500

1000

500 AVE RFA

AVE RFSWM
0
Kannur Kozhikode Quilandy V ythiri Vadakara AVE RFNEM

STATIONS AVE RFNM

Fig. 4 Variation in average of RFA, RFSWM, RFNEM and RFNM at the five stations

Table 4 Average RFA, RFSWM, RFNEM and RFNM and % of RFSWM w.r.t. RFA
Ave. RFA Ave. RFSWM Ave. RFNEM Ave. RFNM %
Station
(mm) (mm) (mm) (mm) RFSWM / RFA
Kannur 3247.3 2601.0 347.4 298.8 80.10
Kozhikode 3166.8 2322.3 441.3 403.1 73.33
Quilandy 3255.3 2485.2 414.4 355.6 76.34
Vythiri 4278.5 3481.6 436.9 360.0 81.37
Vadakara 3524.2 2600.7 409.7 513.8 73.80
Average 3494.4 2698.2 409.9 386.3 76.99

15
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Fig. 5 Variation in annual rainfall at the five stations

Fig. 6 Variation in southwest monsoon rainfall at the five stations

Fig. 7 Variation in northeast monsoon rainfall at the five stations

Fig. 8 Variation in non monsoon rainfall at the five stations

16
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

The variations in RFA, RFSWM, RFNEM and RFNM their respective values at all the stations indicative of a
over the 81 years at the five stations are presented in uniform pattern.
Figures 5, 6, 7 and 8 respectively. The lines showing
the variations in the annual and southwest monsoon Rainfall frequency distribution
rainfall values at the five stations over the 81 years The frequency distribution of annual rainfall at the
broadly present a regularity in increase or decrease in five stations is presented in Table 5. The rainfall values
and ranges are given in mm.
Table 5 Rainfall frequency distribution (RFA)
Kannur Kozhikode Quilandy Vythiri Vadakara
<1000 0 0 0 0 0
1000 - 2000 1 0 2 0 1
2000 - 3000 27 33 30 3 18
3000 - 4000 45 40 37 28 43
4000 - 5000 7 8 10 39 16
5000 - 6000 1 0 2 8 3
>6000 0 0 0 3 0

From Table 5, it can be seen that 70 to 81 of the during 1968 at Kannur and 1597.1 mm during 2004 at
years (ie, 86.4% to 100%) are covered by annual rains Vadakara. The 3000-4000 mm range of the annual
ranging from 2000 mm to 5000 mm. For the entire rainfall values covers all the five stations during more
zone, there are only 3 rainfall values exceeding 6000 than 28 years of the 81 years of observations. Figure 9
mm, viz. 6022.6 mm, 6420.1 mm and 8233.2 mm all at shows the frequency distribution of total annual rainfall
Vythiri during the years 1923, 1924 and 1961 at the five stations.
respectively. There are four annual rainfalls which
Table 6 illustrates the frequency distribution of the
are less than 2000 mm, viz. 1912.9 mm during 1963 and
occurrence of southwest monsoon rainfall at the five
1056.9 mm during 1964 both at Quilandy, 1497.3 mm
stations.
Table 6 Southwest monsoon rainfall frequency distribution
Kannur Kozhikode Quilandy Vythiri Vadakara
<1000 0 0 1 0 0
1000 - 2000 11 23 17 2 9
2000 - 3000 53 50 48 22 56
3000 - 4000 16 8 13 43 14
4000 - 5000 1 0 2 10 2
5000 - 6000 0 0 0 3 0
>6000 0 0 0 1 0
50
45
O F YE A R S

40
35
O .YEAR

30
25
20
NO.NOF

15
10 Kannur
5 Kozhikode

0 Quilandy
<1000 1000 - 2000 2000 - 3000 3000 - 4000 4000 - 5000 5000 - 6000 >6000 Vythiri
RAINFALL RANGE Vadakara

Fig. 9 Frequency distribution of annual rainfall at the five stations

17
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

During the southwest monsoon season of the study than 6000 mm, i.e. 6959.5 mm at Vythiri during the year
period there is only one incident having the value less 1961.
than 1000 mm, i.e.889.8 mm at Quilandy during the year
1964. Also, there is only one RFSWM value higher Figure 10 shows the frequency distribution of
southwest monsoon rainfall at the five stations.

60

50
F YE A R S

40
. OYEAR

30
NO
NO. OF

20

Kannur
10
Kozhikode
0 Quilandy
<1000 1000 - 2000 2000 - 3000 3000 - 4000 4000 - 5000 5000 - 6000 >6000 Vy thiri
RAINFALL RA NGE
Vadakara
Fig. 10 Frequency distribution of southwest monsoon rainfall at the five stations

Figure 10 gives the fact that in all the years there is than 6000 mm.
southwest monsoon rainfall ranging from 1000 to 5000
mm at all the stations except at Quilandy having rain The frequency distribution of northeast monsoon
less than 1000 mm and at Vythiri having rain higher rainfall at the five stations is given in Table 7.

Table 7 Northeast monsoon rainfall frequency distribution

Kannur Kozhikode Quilandy Vythiri Vadakara

<200 15 4 11 5 9
200 - 400 36 32 25 30 33
400 - 600 26 29 36 33 29
600 - 800 3 14 7 12 7
800 - 1000 1 2 1 1 3
>1000 0 0 1 0 0

All the stations have RFNEM value less than 200 mm. mm, i.e. 1227.3 mm at Quilandy during the year 1943.
Also, all the stations have covered within 78 to 80 years
of total 81 years in the range between 0 mm and 800 The frequency distribution of northeast monsoon
mm. There is only one RFNEM value higher than 1000 rainfall at the five stations is given Fig. 11.

40

35
O F YE A R S

30

25
O . YEAR

20
NO.NOF

15

10 Kannur
5 Kozhikode
0 Quilandy
<200 200 - 400 400 - 600 600 - 800 800 - 1000 >1000 Vy thiri
RAINFALL RANGE
Vadakara

Fig. 11 Frequency distribution of northeast monsoon rainfall at the five stations

18
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Table 8 Non monsoon rainfall frequency distribution

Kannur Kozhikode Quilandy Vythiri Vadakara

<200 36 18 20 14 14
200 - 400 28 36 37 46 18
400 - 600 5 8 12 9 26
600 - 800 7 11 6 6 9
800 - 1000 4 6 4 5 8
>1000 1 2 2 1 6

50
45
O F YE A R S

40
35
O . YEAR

30
25
N OF

20
NO.

15 Kannur
10
Kozhikode
5
Quilandy
0
Vy thiri
<200 200 - 400 400 - 600 600 - 800 800 - 1000 >1000
RA INFALL RANGE Vadakara

Fig. 12 Frequency distribution of non monsoon rainfall at the five stations

Table 8 indicates the frequency distribution of non
monsoon rainfall at the five stations, of which the
graphical representation is Fig. 12.
During non monsoon period Kannur, Quilandy and
Vythiri cover 69 years out of the total 81 years (85% of
the years) having the range less than 600 mm rainfall
while Kozhikode and Vadakara cover 62 years (76.5%)
of the 81 years and 58 years (71.6%) of the 81 years
respectively by the same range.
CONCLUSIONS
The study was carried out to find out the rainfall
distribution at the five stations of northern Kerala during
different periods; annual, southwest monsoon, northeast
monsoon and non monsoon periods. The analysis was
done making use of 81 years of daily rainfall data.
Regression analysis of the southwest monsoon
rainfall with respect to annual rainfall of 81 years yield
intercept values varying from 734.60 to 1153.86 and
slopes varying from 0.87 to 1.07. For each relation,
standard error of estimate was also calculated.
Correlation analysis of RFSWM – RFA gives the
coefficients varying between 0.80 and 0.95. This shows
very close relationship between them.
The highest of the maximum annual rainfall (RFA)
observed was 8233.2 mm at Vythiri during the year 1961
and the lowest of the minimum non monsoon rainfall
(RFNM) observed was 0 mm at Quilandy, Vythiri and
Vadakara all during the year 1990.
Maximum annual rainfall and southwest monsoon
rainfall show a parallel behaviour and fall during the
same years.
The overall average RFA, RFSWM, RFNEM and
RFNM were found to be 3494.4 mm, 2698.2 mm, 409.9
mm and 386.3 mm respectively. Also overall southwest
monsoon rainfall was 76.99% of the overall annual
rainfall.
The variations of annual rainfall and southwest
monsoon rainfall at the five stations for the 81 years
show a parallel behaviour in increase and decrease of
values.
The frequency distributions vary depending upon
the groups of the study. There is only one RFSWM
value higher than 6000 mm and only one RFNEM value
higher than 1000 mm.
ACKNOWLEDGEMENTS
The authors express their sincere and heartful

19

thanks to Mr.C.K.Gopinathan for all the help and support
extended for the study. The immense help rendered by
Ms.E.Deepa during the analysis of data and computer
works is also gratefully acknowledged.
Due acknowledgement is given to India
Meteorological Department for the source of data used
in this study.
REFERENCES
1. Ananthakrishnan, R., Parthasarathy, B. and Pathan,
J.M., 1979. Meteorology of Kerala. Contribution to
Marine Sciences Dedicated to Dr. C.V. Kurian, pp. 60-
125.
2. CWRDM, 1995. Water Atlas of Kerala. Centre for Water
Resources Development and Management, Kozhikode.
3. Dhar, O.N. and Bhattacharya, B.K., 1976. Variation of
rainfall with elevation in the Himalayas- a pilot study.
Indian Journal of Power and River Valley Development,
26, 179-185.
4. Dhar, O.N., Nandal, B.N. and Ghose, G.C., 1978. Heavy
rainfall stations of India. Indian Journal of Power and
River Valley Development, 25, 123-134.
5. Fujita, T., Baralt, G. and Tsuchiya, K., 1968. Aerial
measurements of radiation temperatures over Mt. Fuji
and Tokyo areas and their application to the
determination of ground and water – surface
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
temperatures. Journal of Applied Meteorology, Vol.7,
pp. 801-816.
6. Gulezian, R.C., 1979. Statistics for decision making.
W.B.Saunders Company, Philadelphia.
7. Jagannadha Sarma, V.V., 2005. Rainfall pattern in the
coastal zone of Krishna Godavary Basin, Andra
Pradesh, India. Journal of Applied Hydrology, Vol.
XVIII, (1 & 2), 1 – 11.
8. Menon, P.A. and Rajan, C.K., 1989. Climate of Kerala.
Classic Publishing House, Cochin.
9. Muthuchami, A. and Ravikumar, P.V., 1992. The inverse
relationship between the activity of monsoon systems
and the intensity of southern hemispheric equatorial
trough. Vayu Mandal, 22 (1-2), 40-45.
10. Smith, R.B., 1979. The influence of mountains on the
atmosphere. Advances in Geophysics, 21, 87-130.
11. Soman, M.K., Krishna Kumar, K. and Nityanand Singh,
1988. Decreasing trend in the rainfall of Kerala.
Current Science. 57 (1), 7-12.
12. Stringer, E.T., 1972. Foundations of climatology. W.H.
Freeman and Co., San Francisco, pp. 141-167.
13. Verma, R.K., 1993. Variability of Indian summer
monsoon, relationship with surface air temperature
anomalies over northern hemisphere. Mausam, 44,
191-198.
20
 J. Indian Water Resour. Soc.of
Journal Vol.Indian
30 No.Water
1, January, 2010
Resources
Society Vol. 30 No. 1, January, 2010

SUSTAINABLE MANAGEMENT OF THE WATER STRESSED AQUIFERS IN

SABARMATI BASIN GUJARAT, INDIA

R. C. Jain1 and A. K. Sinha2

ABSTRACT

The Sabarmati basin is underlain by a multi-aquifer system, with prolific yields but the
excessive groundwater pumping over many years has resulted in de-watering of a substantial
aquifer thickness and reduction in pressure heads as reflected in long term decline in ground
water levels. The ground water use in the basin being practiced currently is considered
unsustainable as the ground water draft far exceeds the replenishable recharge. For providing
sustainability to ground water withdrawal structures and keeping in view the increasing thrust on
development of ground water resources, there is an urgent need to address the problem of ground
water depletion through both supply side as well as demand side management interventions .
Based on the in depth analysis of hydrogeology, long term water level trends , aquifer parameters
and the availability of surplus runoff a strategy for managed aquifer recharge (MAR) has been
formulated to augment the ground water resources of the water stressed aquifers. Typical designs
of different types of structures suitable for artificial recharge have been included based on the
lessons learnt. The paper presents a matrix of recommendations for implementing MAR in the
basin through various stakeholders. For demand side management it is essential that ground
water development is controlled and regulated through appropriate legal and administrative
measures in view of the fast emerging aquifer depletion in large part of Basin.

Key words : long term decline, managed aquifer recharge, unsustainable, management inter-
ventions, virtual water.

INTRODUCTION added to the capital cost of construction of the tube

Sabarmati river basin is located in the western part
of India (Fig.1). The basin in its western ,central and
southern parts is underlain by a multi-aquifer system,
with prolific yields but the excessive groundwater
pumping over many years has resulted in de-watering
of a substantial aquifer thickness and reduction in
pressure heads. Agriculture production has been
seriously affected due to conspicuous decline in water
levels caused by over- exploitation of groundwater
(Jain, et al,2001).
The basin is more vulnerable for such adverse
effects due to low rainfall reliability and recurrence of
droughts.
The detrimental effects of over exploitation of
groundwater i.e., decline in water levels have not only
1
West Central Region, Central Ground Water Board, Ahmedabad.
2
Deptt. of Geology, University of Rajasthan, Jaipur.
Paper No. 1239
wells and the energy bill but have also added to annual
cost of maintenance of tube wells (INREMF,2001).
Conventional dug wells and dug-cum-bored wells are
already, gradually going out of operation and in many
areas the dug wells have dried up since many years.
Further, during drought years sustained ground water
withdrawals due the pumpage of groundwater far
exceeds that of normal years resulting in permanent
de-watering of aquifers (Jain, et al,2000).
In fact because of economic factor, farmers are
largely growing the cash crops in the area and are taking
high risks in construction of the deep tube wells even
with a reduced life span due to lowering in water levels
and consequent reduction in discharge, higher and higher
maintenance cost. Large financial investments made in
construction of tube wells need to be protected from
implied hazards of over exploitation of ground water.

21

Fig. 1 Location map of Sabarmati Basin
During the early sixties the tube wells were drilled between
60 and 100 m and their water levels ranged between 10 and 15
m bgl. These tube wells have also gone out of use. The tube
wells being drilled now are between 250 and 300m depth with
the water levels ranging between 80 to more than 120 m bgl.
(GOG,2002 ).
The adoption of high yielding crops for boosting the
agricultural production ,fodder production to meet the
demand for dairy cattle and the impetus in rural electrification
has resulted in substantial increase in ground water pumping.
Earlier through conventional open well with discharge 4-6
m3/hr. an area of 0.4 to 0.6 ha could be irrigated. Now with
energisation of open wells and construction of deep tube
wells the average ground water abstraction rates have gone
as high as 90 to 112 m3/hr. with command areas ranging from
40 to 60 ha ( Jain,2003).
Magnitude of ground water depletion
Because of increased pace of development during
last three decades for meeting the increasing demands
for irrigation, drinking water supplies and industrial needs
the dependence on ground water has tremendously
increased (Govt. of Gujarat,2005). The increase in the
ground water draft has been mainly due to increased
pumping for meeting the ever increasing water demand
of irrigation, drinking water and industrial sector. This
is clearly evident from a perusal of few representative
hydrographs shown in figure 2 which very clearly depict
the alarming declines ranging from more tahn 5 to 15 m
in the shallow aquifers and from more than 20 up to 60
m in the deeper aquifers during the last one decade or
so. (Jain,2008)
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
Groundwater Resource Availability Scenario
In the Sabarmati River basin the net ground water
availability is 2342.59 MCM/yr. The gross ground water
draft has increased from 1580 MCM/yr as in
1991(CGWB,1995) to an alarming level of 2415 MCM/
yr as in 2004 (CGWB,2006). As per the latest ground
water resource computations a negative balance
of (-)72.37 MCM/yr is indicated. The stage of ground
water development has increased from nearly 60% of
replenishable recharge in 1991 in the basin to >103%
of replenishable recharge in 2004 over a period of last
one and a half decade. A comparative depiction of
change in stage of ground water development in
important talukas in the basin is shown in Fig.3.
Based on overall stage of ground water
development of 103.09 % of replenishable recharge ,
the basin is categorised as OVER EXPLOITED .The
present status of development is unsustainable as the
aquifers are under stress .the stage of ground water
development varies widely from as low as 28% of
replenishable recharge as in Anand taluka in the southern
part of the basin to nearly 230% replenishable recharge
of in Mansa taluka in western part of the basin (2004).
Groundwater depletion – Management options
The Sabarmati river basin has the problem of
intensive ground water extraction / overexploitation
leading to long term decline in water levels in the
northern, central and eastern part ,while in the southern
part the level of ground water extraction is comparatively
low due to availability of ample surface water irrigation
facilities through the imported water from the Mahi
River Basin. At present the gross irrigated area in the
Sabarmati basin is 7,89,500 ha., which will increase to
12,43,500 ha by the year 2025 (ICID,2005).
The total mean annual surface water resources in
Gujarat part of the basin are estimated at 2018.20 MCM
including 150 MCM from neighbouring states. Further
as per Plans of the Gujarat Govt. surface water potential
to the extent of 1587.13 MCM is committed , leaving
nearly 431.07 MCM untapped , which accounts for 21%
of mean annual runoff in the Sabarmati River
basin(CDO,2003).
For providing sustainability to ground water
withdrawal structures in Sabarmati River basin and
keeping in view the increasing thrust on development
of ground water resources for meeting the growing
22
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Fig. 2 Hydrographs of select observation wells in Sabarmati Basin

23

Fig. 3 Stage of ground water development in Sabarmati River
basin during 1991-2004.
demands of water in agriculture, industrial and domestic
sectors, there is an urgent need to adopt both the supply
side and demand side management interventions to
augment and conserve the depleting ground water
resources in the active recharge zone.
Managed aquifer recharge - Supply side
management
Natural replenishment of ground water reservoir is
slow process. It is unable to keep pace with the
excessive continued exploitation of ground water
resources . In order to augment the natural supply of
ground water, managed aquifer recharge ( MAR) is
an important and frontal management strategy which
can provide succor to address the problem of ground
water depletion (Jain& Sinha,2004). MAR is the
process by which the ground water reservoir is
augmented at a rate exceeding that under natural
conditions of replenishment. The recharge efforts are
basically augmentation of natural movement of surface
water into ground water reservoir through suitable civil
structures under suitable hydrogeological and hydrologic
conditions. As the rainfall occurrence in the basin is
limited to about 10 to 22 days during three months
monsoon period,the natural recharge to ground water
reservoir is restricted to this period only. The Managed
aquifer Recharge aims at increasing the recharge
period in the post-monsoon season by about 3 more
months . This results in providing sustainability to ground
water development during the lean season.
The efficacy of artificial recharge schemes depends
largely on the source water availability and capability
of ground water reservoir to accommodate it ,which
requires detailed knowledge of geological and
hydrological features of the area for site selection and
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
design of artificial recharge structures(Kovalevsky, et
al, 2004). While assessing the availability of source water,
which is one of the prime requisites for ground water
recharge, careful consideration has to be given to
committed monsoon storages, so that the harvesting of
non committed surplus monsoon run off from the area
does not have adverse environmental impact like drying
up of existing lakes/ponds/reservoirs due to reduction
in inflows from their catchments. Detailed knowledge
of dimensional data of the aquifer viz. their thickness
and lateral extent is necessary for evaluation of the
storage potential. The availability of sub-surface storage
space and its replenishment capacity governs the extent
of recharge.
Identification of areas feasible for managed aquifer
recharge in Sabarmati Basin
The basin is having total area of 22,260 sq.kms in
Gujarat. The geological formations in the basin are
alluvium, igneous, metamorphic and other soft rocks. In
the major part of the basin alluvium consisting of re-
worked aeolian sand, silt, kankar and clay forms the
principal aquifer (Fig. 4).
Fig. 4 Hydrogeological map of Sabarmati River basin
The unsaturated thickness of rock formations,
occurring beyond eight metres below ground level has
been considered to assess the requirement of water to
build up the sub-surface storage by saturating the entire
thickness of the vadose zone to 8 metres below ground
level. The upper 8 m of the unsaturated zone is not
considered for MAR; in view of the prevailing
hydrological, hydrogeological, geomorphic and
24

topographic conditions. The post-monsoon depth to
water level represents a situation of minimum thickness
of vadose zone available for recharge which can be
considered with reference to surplus monsoon run off
in the area.
The average post monsoon water level (1998-2007)
ranges between less than 8 and 35 mbgl in the Sabarmati
basin (Fig 5). The quality of ground water is good in
Fig. 5 Average post monsoon depth to ground water level (1998-
2007) in Sabarmati Basin
Fig. 6 Ground water quality in Sabarmati Basin during
pre-monsoon 2007
Fig. 7 Areas feasible for artificial recharge to ground water in
Sabarmati basin
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
general except the western part which is saline having
TDS more than 2500 mg/l (Fig 6). Based on the quality,
post monsoon water level and the declining trend the
feasible areas for managed aquifer recharge are
identified in the northern and western parts of the basin(
Fig.7). The south, central and the western part of
basin having post monsoon decadal average depth to
water level less than 8mbgl and saline ground water
are not considered feasible for MAR.
It is estimated that an area of 14,517 Sq.km. having
depth to water table more than 8 m bgl, spread over all
the geological formations is feasible for managed aquifer
recharge through different type of recharge structures
at appropriate sites. The total volume of vadose zone
of this area is 41,348 MCM. Further in an area of 464
sq.kms the declining trend of water level is observed
and the volume of vadose zone in this area is 66.93
MCM. Thus the total volume of de-saturated zone
feasible for recharge in the basin is 41,414 MCM.
Considering percentage of clay in alluvium, coefficient
replenishment of different formations, it is estimated
that about 3,534 MCM of water is required for recharge
to bring the water level up to 8 m bgl .
The non-committed monsoon flow of 431 MCM/
annum could be used as source water for managed
aquifer recharge for augmenting the storage in the
stressed aquifers (UNDP,1988). This surplus runoff can
be utilised for MAR through proven techniques of
recharge like injection wells, dug wells, check dams,
spreading channels and percolation tanks in the area
(Jain, 2003). Based on an average recharge rate of
0.088 MCM/annum for a check dam/percolation tank,
0.015MCM/annum for injection wells and 0.019MCM/
annum for dug wells it is estimated that 13,300 Dug
wells, 4,700 injection tube wells/connector wells, 500
percolation tanks and 1,400 check dams are needed for
transferring the non-committed runoff to the stressed
aquifer. This will cost about Rs. 1400 Million. All the
structures are not required to be constructed afresh, In
fact , a large number of existing dug wells /tube wells
can be provided with suitable siltation/filtration chambers
for removing the physical impurities in source water
derived from monsoon flows. Similarly the deepening
of existing ponds for increased storage and percolation
into the aquifer would also be very useful , as creation
of new surface storages adds more to the evaporation
losses than to percolation to the underlying aquifers .The
spreading channels are feasible in the limited area of
common recharge zone at the foot hill of Arvalli’s in
25
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

RCC M 20 GRADE SLAB

6000 MM

2800 MM
400 MM DIA
INJECTION
CONNECTING PIPE WELL
150 MM DIA WITH 2600 MM
CONTROL VALVE
150 MM DIA PVC
CASING PIPE
SOURCE WATER UPPER 10 M
6250 MM
DRAIN BLANK & LOWER
35 M STAINER
PCC 1:3:6, 150 MM THICK (6KG / CM2)
SCUM BOARD

ANNULUS
WITH 10 TO 20
MM Ø PEBBLE
IN SPACE

1500 X 2000 X 1500 MM SIZES FILTER

CHAMBERS, ONE FILLED WITH
COARSE SAND & ANOTHER FILLED
WITH PEBBLES OF 10 TO 20 MM DIA 15 MM DIA AIR
SIZE. ALL THE WALLS ARE B. M. VENT PIPE 30 M
WALLS IN C. M. (1:3), 230 MM THICK. LONG WITH 5 MM
PROVIDE 3 PIPES 150 MM DIA WITH DIA HOLES IN
PIPE CAP HAVING HOLES ON ENTRY LOWER 10 M
SITE. LENGTH

SECTIONAL VIEW

400 MM DIA
6000 MM INJECTION WELL

1500 MM 150 MM DIA PVC

2000 MM 1500 MM
CASING PIPE
UPPER 10 M
BLANK & LOWER
2500 MM 2000 MM 35 M STAINER
(6KG / CM2)

15 MM DIA AIR VENT PIPE

MAN HOLE COVER 30 M LONG WITH 5 MM DIA
HOLES IN LOWER 10 M
PLAN LENGTH

Fig. 11. Typical design of an injection well in alluvial area

27
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

north eastern part of the area. Typical design of various
type of structures is given in Fig.8,9,10 &11.
As the predominant contribution of recharge to
deeper aquifers is through the vertical leakage from
the aquitards underlying the shallow aquifers, the
recharge of surplus monsoon water in the shallow aquifer
will benefit both the phreatic as well as deeper aquifers
.This will go a long way in arresting the rate of decline
in ground water levels and providing sustainability to
the fast depleting stressed aquifers (Jain & Sinha,2002).
Demand side management - Regulation and
control of ground water development-
As the available ground water resources are finite
and aquifer depletion is fast emerging in large part of
Sabarmati basin ,it is essential that ground water
development is controlled and regulated through
appropriate legal and administrative measures. To begin
with any additional abstraction from the deeper aquifers
should be banned, because the ground water resource
of deeper aquifers get replenished over large no. of
years unlike the shallow aquifer which gets replenished
every year(World Bank,1999). The in-storage reserves
of ground water in the deep multi-aquifer system should
be treated as strategic resource to be utilised in times
of emergency only (UNDP,1976). However, exception
can be made for drinking water supply ,where a limited
quantity of withdrawal can be permitted not exceeding
safe yield of the aquifer. Fortunately, Gujarat state has
a State Ground Water Authority in place, but the need
of the hour is to control and regulate the ground water
abstraction by effective implementation of the ground
water legislation.
In addition to the requirement of water for dairy
development, fodder for the cattle is also grown which
leads to over exploitation of ground water. If the fodder
requirement could be met from agriculture production
in the central and southern part of Gujarat having low
stage of ground water development, which can then be
transported to over exploited areas .Thus a large
quantity of ground water can be conserved in the form
of virtual water.
More emphasis should be laid on adoption of
micro-irrigation technology like drip irrigation etc,
which will help in conservation of the precious resource.
Strategy for implementation of MAR
By far the most critical response to stressed aquifers
in the Sabarmati Basin demands exploring synergies
from a variety of players for the basin wide recharge
program. Evolving a groundwater recharge strategy
needs to begin with an appreciation of the variety of
actors that can contribute through different kinds of
recharge structures as suggested in the Table 1.
Public agencies with strong science and engineering
capabilities need to play a major role in constructing
and managing large recharge structures (Planning
Commission,2007). However, an intelligent strategy can
also involve millions of farmers and householders and
thousands of their communities—each of whom can
contribute small volumes to recharge dynamic
groundwater. The Table 1 indicates who can play what
role to achieve the purpose of resource augmentation
under prevailing conditions.

Table 1. Actors and their roles in implementation of MAR

Aquifers Key players Numbers Recharge Location of

Actors affected of actors Volumes/ structures
who can structure
contribute
Small structures for Dynamic Individual millions 100-5000 m3 Private farm
recharging wells and groundwater in farmers and urban lands and
roof-water harvesting alluvial and citizens homes
structures hard-rock areas
Check dams, Dynamic Communities Tens of 100,000- Common-
percolation tanks, groundwater in using a common thousands 5,000,000 m3 property or
Sub-surface dykes, etc alluvial and aquifer system government
hard-rock areas land
Large structures on Confined Public agencies few 0.1 to 1 km3 Government
government land for aquifers: large with hydro- or more waste lands or
recharge to confined alluvial aquifers geology forest lands
aquifers expertise;

28
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

In view of the criticality of groundwater recharge,
a Special Purpose Vehicle (SPV) needs to be created
for overseeing private and NGO-implemented
groundwater recharge programs as well as for
executing, operating and maintaining large-scale
groundwater recharge program. Besides the scientific
talent such a SPV needs to build engineering and
management capacity needed for the purposes on hand.
In our country water policy has so far focussed on
what governments and government agencies can do.
Now, it needs to target networks of players, each with
Table 2. Outline of an MAR Strategy for Sabarmati River Basin
Key actors Arid alluvial Hard rock
aquifer areas aquifer areas
Farmers Dug wells, Dug wells, farm
roof-water ponds, roof-
harvesting; water harvesting;
other private other private
recharge recharge
structures structures
NGOs, local Percolation ponds, check dams,
communities sub-surface dykes on streams
Groundwater Recharge canals to transport
recharge surplus flood waters for recharge
SPV in groundwater-stressed areas e.g.,
Sujalam- Sufalam in North
Gujarat
Large recharge structures in
recharge zones of confined
aquifers
distinct capabilities and limitations. If MAR is to be a
major response to provide succour to the stressed
aquifers in the Sabarmati River Basin , the state needs
to evolve an integrated groundwater recharge strategy
(Jain,& Sinha, 2004)and work with role and space for
various players to contribute as outlined in the
Table 2.
Acknowledgements : The author acknowledges the
Chairman, Central Groundwater Board, Govt. of India
for the permission to publish this paper .
Roles that need to be played by State Govt, Recharge SPV,
other public agencies
Vigorous Information, Education, Communication campaign to
promote MAR to stressed aquifers through dug wells
Technical support in construction of desiltation/filtration pits
for recharge, and periodic desiltation of wells
Financial incentives and support to farmers adopting MAR
Technical and financial support to local communities, NGOs for
construction and maintenance
Supportive policy environment and incentive structures
Support for building local institutions for groundwater recharge
Create a Special Purpose Vehicle to execute, operate and
maintain large-scale recharge structures
Build and operate large-scale recharge structures in upstream
areas of confined aquifers. e.g. at the base of Aravalli’s in North
Gujarat

REFERENCES 6. Government of Gujarat, 2005. Report of the Committee

1. Central Designs Organisation, 2003. Report on
Integrated river basin planning, development and
management of Sabarmati River basin. Govt. of Gujarat.
2. Central Ground Water Board, 2002. Master Plan for
Artificial Recharge to Ground Water in India, Ministry
of Water Resources, Government of India.
3. Central Ground Water Board, 2006. Dynamic Ground
Water Resources of India, Ministry of Water Resources,
Government of India.
4. Central Ground Water Board, 2007. Ground water Year
Book, Tech. Report, CGWB, WCR, Ahmedabad.
5. Government of Gujarat, 2000.White paper on water in
Gujarat , Narmada Water Resources and Water Supply
Department.
on Estimation of Ground water resources and
Irrigation Potential in Gujarat. Narmada and Water
Resources Department.
7. Indian Natural Resource Economics and management
Foundation, 2001. Integrated water resources
management in Sabarmati Basin: Some issues and
options.
8. International commission on Irrigation and Drainage,
2005, Water Resources Assessment of Sabarmati River
Basin, India.
9. Jain,R.C., Nagar,A., Jain,P.K. and Krishna, V.S.R.,2000.
Declining Water levels in deep aquifers of Gujarat. Tech.
Report ,CGWB, WCR, Ahmedabad.
10. Jain, R.C., Jain,P.K., Jain,A.K. and Krishna,V.S.R.,2001.
Hydrogeological framework of Gujarat State with

29

special emphasis on issues related to Ground Water
Development and Management. Workshop on
Sustainable Development and Management of Ground
Water Resources in Gujarat . Tech. Report ,CGWB, WCR,
Ahmedabad .
11. Jain,R.C. and Sinha,A.K.,2002. Intensive use of Ground
water in North Gujarat, India- Challenges and
opportunities. Proceedings of the SINEX Symposium
,Barcelona, Spain,2002. Published in “Groundwater
Intensive Use ” Sahuquillo, A.,Capilla,J., Martinez-
Cortina,L., Sanchez-Villa,X.,(Editors),Publication No.
SP 7 of the International Association of
Hydrogelogists.2005.
12. Jain,R.C. , 2003. Hydrological opportunities and
limitations of Artificial Recharge to Ground Water in
Gujarat. Publication of The International Water
Management Institute, Anand Centre, Gujarat.
13. Jain,R.C. and Sinha,A.K. , 2004. Hydrogeological
framework of the over-exploited multi-aquifer system
in North Gujarat, India –Problems and prospects of
sustainability. Proceedings  of  the  32nd  International
Geological Congress, Italy,2004.
14. Jain,R.C. and Sinha,A.K., 2008. Groundwater
management in Sabarmati river basin - Perspectives
& Options. International Groundwater Conference on
Groundwater Dynamics and Global Change , March
11-14, 2008.
15. Jain, R.C., 2009. Hydrogeological Framework,
Groundwater Resources and Strategies for sustainable
management in Gujarat. Proceedings of the National
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
Seminar on Water for Future, Organised by Narmada
and Tapi river Basin Organisation, Central Water
Commission, March 4-5,2009. Ahmedabad, Gujarat,
India.
16. Jain, R.C., 2009.Emerging challenges for sustainable
ground water management in India. Journal of Applied
Hydrology, Vol. XXII, No.1, 2009.
17. Jain, R. C., 2009. Trends and sustainability of
groundwater in highly stressed aquifers of Gujarat,
India . Trends and Sustainability of Groundwater in
Highly Stressed Aquifers .Proc. of Symposium JS.2 at
the Joint IAHS & IAH Convention, Hyderabad, India,
September 2009. IAHS Publ. 329, 2009.
18. Kovalevsky, Kruseman,G.P.and Rushton,K.R.,2004.
Ground Water Studies -An international guide for
hydrogeological investigations, Unesco-IHP-VI,Series
on Ground water No.3,p.430.
19. Planning Commission , 2007. Report of the Expert group
on Ground water Management and Ownership. Govt.
of India.
20. United Nations Development Programme, 1976.
Ground water surveys in Rajasthan and Gujarat.
Technical Report.
21. United Nations Development Programme, 1988. Pilot
project on artificial recharge to ground water in
Mehsana and coastal area of Gujarat. Terminal report-
summary of findings and recommendations.
22. World Bank, 1999.Ground water Regulation and
Management , Water sector report. .New Delhi.
30
 J. Indian Water Resour. Soc.of
Journal Vol.Indian
30 No.Water
1, January, 2010
Resources
Society Vol. 30 No. 1, January, 2010

RUNOFF ESTIMATION BY DISTRIBUTED CURVE NUMBER TECHNIQUE

USING REMOTE SENSING AND GIS

Susanta Kumar Jena1, Kamlesh Narayan Tiwari2 and Ashish Pandey3

ABSTRACT

A study was undertaken to compare composited and distributed Curve Number (CN)
techniques for estimation of runoff from a medium size agricultural watershed namely Tarafeni
(West Bengal). Indian remote sensing satellite digital images were used for classifying the land
use/ land cover of the study watershed. The study watershed was divided into grids and weighted
CN was calculated for individual grids by intersecting the grid with CN coverage. Using distributed
CN technique runoff was simulated at each grid and averaged at the outlet of the watershed and
finally compared with the values obtained from traditional compositing technique. The different
initial abstractions (Ia) of 0.2S and 0.3S (S is the maximum potential retention) have also been
taken in different simulations and results were compared with the observed runoff. The study
revealed that there is definite increase in runoff value estimated by distributed CN technique over
composited technique. The amount of increase in runoff value calculated is more for the case of
Ia=0.3S than Ia=0.2S. The percent increase in runoff is very high for small events, moderate for
medium and low for high rainfall events. The results from statistical analysis, which includes
percent deviation, model efficiency, coefficient of determination, coefficient of residual mass,
root mean square error and student’s t-test for significant difference, show that distributed CN
with Ia=0.2S estimated runoff values are closely matching the observed runoff.

Key words: Distributed curve number, GIS, Remote sensing, Runoff, Statistical test

INTRODUCTION One of the most widely used techniques for

Watershed management, which is a comprehensive
term meaning the rational utilization of land and water
resources for optimal production and minimum hazard
to natural resources can solve the problems of soil
erosion and degradation in quality and quantity of water
resources. In many countries, non-availability of data
on amount and rate of runoff, sediment yield, nutrient
loss, etc., is the major handicap to start with watershed
development programme. There are numerous methods
to estimate runoff for ungauged watersheds, which is
the most important input required for design of hydraulic
structures, flood forecasting and erosion prediction and
to link with water quality problems etc. in a watershed
management programme.
1 This is because many practicing professionals have not
Directorate of Water Management (Formerly WTCER) (ICAR),
Chandrasekharpur, Bhubaneswar-751023
2
Dept. of Agricultural & Food Engineering, Indian Institute of
Technology Kharagpur-721302
3 for an alternate approach are widely publicized. Runoff
Dept. of WRD&M, Indian Institute of Technology Roorkee
Paper No. 1243
estimating direct runoff depths from storm rainfall is
the United States Department of Agriculture (USDA)
Soil Conservation Service’s (SCS) (now the Natural
Resources Conservation Service) Curve Number (CN)
method (Mishra and Singh, 2003). The SCS method
has been used by many researchers to determine the
rainfall runoff relationship (Stuebe and Johnston, 1990;
Sharma et al., 2001; Sharma and Kumar, 2002; Mishra
et al. 2004; Pandey and Dabral, 2004; Pandey and Sahu,
2004; Pandey et al. 2005; Mishra et al. 2005; Jain et al.
2006). Prior to the widespread use of computers,
averaging techniques were necessary to reduce
calculations in manual runoff analysis. Although high-
speed personal computers have reduced the tedious of
repetitive calculations, compositing is still widely used.
yet been able to take full advantage of new computer
technology, and because accepted approaches are
usually slow to change unless significant advantages
estimation using traditional composite technique is still
31

in use worldwide. The increased availability of digital
spatial databases and the use of GIS in recent times
has helped in improving the practicality of estimating
runoff depth using distributed curve numbers. However,
if distributed CN approaches are to be adopted, it is
important to understand how results from this approach
compare with results from the traditional composited
CN technique. Keeping this in mind, this study was
undertaken to estimate runoff depth using different CN
methods and with different values of initial abstraction
and results thus obtained were compared with the
observed runoff from a medium agricultural watershed.
THEORETICAL CONSIDERATIONS
The SCS-Curve Number method (Mishra and Singh,
2003.) is based on the water balance equation and two
fundamental hypotheses. The first hypothesis states that
the ratio of the actual amount of direct runoff to the
maximum potential runoff is equal to the ratio of the
amount of actual infiltration to the amount of the potential
maximum retention. The second hypothesis states that
the amount of initial abstraction is some fraction of the
potential maximum retention. Expressed mathematically,
the water balance equation and the two hypotheses,
respectively, are
P  Ia  F  Q (1)
Q F
 (2)
P  Ia S
I a  S (3)
where, P = total precipitation (mm), Ia= initial
abstraction (mm), F = cumulative infiltration excluding
Ia (mm), Q = direct runoff (mm), and S = potential
maximum retention or infiltration (mm).
The current version of the SCS-CN method
assumes  equal to 0.2 for usual practical applications.
As the initial abstraction component accounts for
surface storage, interception, and infiltration before
runoff begins,  can take any value ranging from 0 to
. Combining equations (1) and (2), one can write an
equation for Q as follows:
(P  I a ) 2
Q (4)
P  Ia  S
Substituting Ia by 0.2S, the equation in its standard
form is
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
(P  0.2S) 2
Q (5)
(P  0.8S)
25400
S  254 (6)
CN
where CN is the curve number which depends upon
land use, hydrologic soil group and antecedent soil
moisture condition.
Narayana (1993) recommended  equal to 0.3 for
most of the regions in India except for the regions having
black clay soils where  = 0.1 seems more appropriate.
MATERIALS AND METHODS
The study was undertaken in the Tarafeni watershed
in Midnapore district of West Bengal in India. The
watershed lies within 22037 to 22047 N latitude and
86038 to 86048 E longitudes. The study area falls under
subtropical humid zone with an annual rainfall of 1350
mm. The area of the watershed is 158.06 km2 and the
relief varies from 110 m to 290 m above mean sea level.
The shape of the watershed is almost circular and it is
covered by Survey of India (SOI) toposheet number:
73/J-9, 10, 13 and 14 (1:50000 scale). The outlet of river
Tarafeni is considered at the Tarafeni barrage, which
comes under Kangsabati irrigation project, West Bengal.
Hydrologic data
The “Irrigation and Waterways Department” of
Government of West Bengal measures the hydrologic
data such as runoff at the outlet of the watershed every
one hour. An automatic rain gauge and another non-
recording rain gauge were installed at the outlet of the
watershed from which daily rainfall as well as eventwise
rainfall amount and intensity with respect to time were
found out. The data are available from the year 1995
onwards.
Land use/ land cover data
Indian Remote Sensing satellite digital image (IRS-
1A, LISS-II-B1, row: 20, path: 52 with spatial resolution
of 36.25 m) of year 1989 (dates of pass 21st February
and 13th November) as well as (IRS-1D, LISS-III, row:
107, path: 56 with resolution 23.5 m) for the year 2000
(dates of pass 22nd February and 14th November) were
imported to the ERDAS/ IMAGINE image processing
software. These images were rectified and
geometrically corrected with respect to already rectified
and mosaic topo-sheets of that area. This process of
32

rectification was done by transferring the coordinates
of permanent features like road and rail crossings;
bridges, barrages etc. on the river; as well as road-
canal or rail-canal crossings present on rectified topo-
sheets to satellite images. The three basic colours of
red, blue and green were assigned to different bands of
image, which produced false colour composite (FCC)
image. Then the Tarafeni watershed portion was subset
from the full image. The classification of FCC was
carried out with ERDAS/ IMAGINE software using
maximum likelihood classifier (MLC) algorithm and
ground truth information. All the three levels of land
use/land cover classification system (NRSA, 1995) were
adopted for classification. A total of eight classes were
considered for classification. The classified images in
raster format was changed to vector format (polygons)
using ERDAS/IMAGINE software. Each polygon had
different attributes, like its area, perimeter and land use
/land cover class. Next, this coverage was intersected
with the watershed grid coverage to find out grid wise
land use /land cover values. The resolution of the grid
was kept at 400 m × 400 m. In individual grids, there
were several polygons having different attributes.
Soil coverage
Soil map of the study area was procured from All
India Land Use and Soil Survey, and it was scanned,
rectified and geometrically corrected. Then different
soil polygons representing different types of soil texture
were digitized using ARC/INFO software. Different
attributes of soil such as texture, hydrologic soil groups,
etc., were assigned to these polygons.
Curve number coverage
The generated land use coverage and the soil
coverage were merged using UNION command of
ARC/INFO software. The resultant coverage contained
attributes of both the coverage. The antecedent moisture
condition was assumed to be AMC II. Using ARC/
INFO software, all the polygons having a particular land
use and a hydrologic soil group were selected at a time
and then curve numbers were assigned to these polygons
using conditional assignment of attributes. Thus, a curve
number coverage is generated in which different
polygons have different curve number values. Similarly
curve number coverages were developed for other
AMC conditions. Finally, these coverages were
intersected with watershed grid. For each grid, the curve
number was determined by using weighted area method.
If a grid contains n number of polygons having different
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
curve number values corresponding to different land
uses, then the weighted curve number for that grid was
found out by using the formula
CN1  A1  CN 2  A2  ...  CN n  An
CN  (7)
A
where A1, A2 , ……, An correspond to area of each
polygon having CN value CN 1 , CN 2 ,…, CN n
respectively, and A = area of each grid
=A1 +A2 +…..+An .
Composited CN method
In composited method, CN was estimated using a
procedure in which area weighted average curve
numbers are calculated either for the entire area under
consideration, or for a small number of relatively
homogeneous sub-areas. Composite curve numbers
were determined by overlaying land-use and soils maps
to delineate polygons with unique land use and
hydrologic soil group (HSG) combinations within the
watershed or sub-watershed being studied. A CN value
was then assigned to each grid and the area-weighted
average was calculated to determine the composite CN
for the watershed.
Distributed CN method
In the distributed approach, runoff depth was
estimated for each individual grid cell or polygon in the
watershed, based on the land use and soil conditions at
that location. There was no CN averaging, rather
separate CN values were determined for each cell or
polygon and separate runoff values were calculated for
each cell or polygon. These runoff values were then
averaged to find out the total runoff depth for the whole
watershed.
Simulation studies
Four different simulations with different
combinations of CN methods and initial abstraction
values were considered in this study. In simulation I,
runoff was estimated using composited CN method and
initial abstraction (Ia) value as 0.3S, for simulation II it
was distributed CN with initial abstraction (Ia) value as
0.3S, simulation III was composited CN with Ia as 0.2S,
and for simulation IV it was distributed CN with Ia as
0.2S. Then runoff values obtained by these four
combinations were compared with the observed runoff
values for different rainfall events. The runoff values
obtained by all these four methods were then statistically
33
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

evaluated to find out the superior method for estimating that of equation (8). The E values can vary from 0 to 1,
runoff. The statistical methods used are explained in with 1 indicating a perfect fit. The model efficiency, E
the following paragraphs. has been widely used to evaluate the performance of
hydrologic models (Wilcox et al., 1990). Model efficiency
Evaluation criteria of different CN methods represents an improvement over the R 2 for model
The different curve number approaches were evaluation purposes, in that it is sensitive to differences
evaluated using different statistical techniques as in the observed and model simulated means and
suggested by ASCE Task Committee (1993). The variances.
following performance evaluation criteria were adopted Another technique is determination of coefficient
in this study. of residual mass (CRM) (Hack-ten Broeke and
The coefficient of determination (R2) describes the Hegmans, 1996). CRM statistics gives the degree to
proportion of the total variance in the observed data which the prediction is overestimated or underestimated.
that can be explained by the model. It ranges from 0.0 Positive value of CRM indicates that the model
to 1.0, with higher values indicating better agreement, underestimates the measured or observed and a negative
and is given by: value of CRM indicates a tendency to overestimate.
The expression of CRM can be given as:
2
  N N
 N 

 
Oi  O avg Si  Savg    Oi   Si
  i 1 i 1
2 i 1 CRM 
R  N (11)
0.5 
 N 0.5
 N
2  (8)  Oi

  Oi  Oavg  
2

  Si  Savg  
 
i 1
  i 1   i 1  
Another measure of the accuracy of prediction
capability of a model is the root mean square error
where, Oi = the ith observed data, Oavg = mean of (RMSE), defined as (Thomann, 1982):
the observed data, Si= the ith simulated value, Savg =
the mean of simulated value, and N = total number of N
events.  O i  Si 2 (12)
i 1
RMSE 
The percent deviation, Dv is one “goodness of fit” N
criterion (Martinec and Rango, 1989), which is
expressed by RMSE describes the difference between the model
simulations and observations in the units of the variable.
V  V
D v (%)   100 (9)
V Based on the probability concept, pairwise
comparison for significant difference between the
where V is the observed value of variable and V  is means of observed and model simulated values were
the simulated value. Smaller value of Dv indicates better done by using Student’s t statistics and given by:
model prediction. For a perfect model, Dv is equal to D av N 0.5
zero. Another basic goodness of fit criterion is the model tc  (13)
efficiency, E (Nash and Sutcliffe, 1970) which is given
S D 
by
where D av is the average of the differences
N between the model simulated (Si) and observed (Oi)
 O i  Si 2 values, N is the number of observations and SD is the
E  1  i 1 sample standard deviation of the difference values.
N (10) Following hypotheses is tested for determining that
 O i  O avg 2 whether the difference between the observed and
i 1
predicted values is significant or not.
The notations of above equation are the same as H0: Dav = 0.0 and H1: Dav  0.0

34
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Criteria for acceptance: if |tc|  t0.975, N-1 accept H0. was then calculated at each grid level and averaged to
The calculated value of student’s t (tc), less than that of find out the runoff at the outlet. Runoff was calculated
the t-tabulated value at N-1 degree of freedom and 5% using both composited and distributed CN techniques
level of significance (i.e. probability of type-I error/ using two different values of Ia : 0.2S and 0.3S. The
significance level of the test () = 0.025 for two tailed results are presented in figure 1.
‘t’-test), indicates that the difference between means
is not significant at 5% significance level. The rejection It is observed from figure 1 that there is definitely
of null hypothesis at 5% significance level indicates that an increase in runoff value estimated by distributed CN
there exists significant difference between the means technique over composited CN technique for both the
of observed and simulated values. cases of initial abstraction values of 0.2S and 0.3S.
However, the percentage increases in runoff incase of
RESULTS AND DISCUSSION 0.3S is comparatively higher than that of 0.2S case.
Hence for choosing which CN technique should be used
The results of land use/ land cover classification for estimation of runoff in a watershed, it is desirable to
are presented in Table 1. The classification accuracy compare runoff values obtained by different techniques
for the Tarafeni watershed for the year 2000 is 91.2% with the observed values. The results are presented in
and for 1989 is 89.3%. From Table 1, it is observed that Figure 2 and Table 2.
in the Tarafeni watershed dense forest, degraded and
fallow land have been reduced, open forest area 30.0

increased from 13.3% to 19.3%, waste land developed 25.0 Composited,0.3S

to 2.3% area of total watershed from nil. 20.0

Distributed, 0.3S
Composited,0.2S
Runof f , mm

Distributed, 0.2S
The distributed CN technique was applied to the 15.0
Observed

Tarafeni watershed. The gridwise curve number 10.0

coverage was generated both for seedbed preparation 5.0

stage and growing stage for both the years. The runoff 0.0
25.0 30.0 32.6 34.0 36.0 38.0 40.0 43.0 45.5 48.6 51.0 63.0 70.0 77.0 80.0
Rainf all, mm
40
(% )

35
Fig. 2. Funoff estimated by different methods
seed bed (2000)
in r u n o f f

30 growing (2000)
It is observed from figure 2 that runoff estimated
s erunoff

25 seed bed (1989)

growing (1989)
using both composite and distributed CN techniques
20
considering Ia=0.3S yields lower runoff compared to
In c r e ain

15
Increase

10
the observed one in all the events considered for rainfall
5
amounts varying from 25 mm to 80 mm during 1999 to
0
2001. Runoff estimated using composite technique for
30 40 50 60 70 80 90 100 110 120 130 Ia=0.2S yields comparable results with the observed
Rainfall (mm)
runoff for rainfall events of more than 63 mm of rainfall.
(a) Ia=0.2S However runoff estimated using distributed technique
140 and Ia=0.2S gave almost same results as observed.
in r u n o ff ( % )

120 seed bed (2000) To support the above findings, statistical analysis
growing (2000)
100 was carried out to find out superiority of one approach
In c r e ains erunoff

seed bed (1989)

80
growing (1989)
over the other in terms of closeness of predicted and
60 the observed values. The results are presented in Table
Increase

40 2. It is observed from the table that percent deviation

20 of runoff estimated using composited method for Ia=0.3S
0
from the observed runoff varies from 21.27% to 99.88%.
30 40 50 60 70 80 90 100 110 120 130 The lowest deviation is for the largest event (80 mm on
Rainfall (mm) 20th July 1999) and the highest deviation of 99.88% is
(b) Ia= 0.3S for the smallest event (25 mm on 1st September 2000)
Fig. 1 Increase in runoff estimated by distributed CN over considered for this study. Similarly for distributed CN
composited CN technique for Ia=0.2S and Ia=0.3S. method with Ia=0.3S, percent deviation varies from

35
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Table 1. Land use/ land cover statistics of the Tarafeni watershed

Land use/ land cover Year 1989 Year 2000 Change in

Area (ha) % Area (ha) % area (%)
Dense forest 2715.0 17.18 1874.9 11.86 -5.32
Open forest 2109.6 13.35 3046.6 19.28 +5.93
Degraded forest 3337.0 21.11 2932.7 18.55 -2.56
Cropland 1961.6 12.41 2207.0 13.96 +1.55
Fallow land 4029.8 25.5 3815.4 24.14 -1.36
Wet land 605.0 3.83 639.4 4.05 +0.22
Waste land 0.0 0.0 367.9 2.33 +2.33
Water bodies 1048.1 6.63 922.13 5.83 -0.80
Total 15806.1 100 15806.1 100

Table 2. Evaluation of different CN techniques for estimation of runoff (t- critical value for two tail t distribution=2.145)

13.91% to 63.64%. This deviation is lower as compared
to the composited method with same initial abstraction
value. For initial abstraction of 0.2S the composited
method estimated runoff with percent deviation varying
from 0.15% to 63.95%. Though it has very low percent
of deviation for larger events, the deviation is greater
for small events. For the distributed CN method and
Ia=0.2S, the percent deviation varies from 16.86% to
23.03%. Though this method does not give definite trend
to predict runoff, the percent deviation is lowest in
comparison to all other methods considered in this study.
From the other statistical analysis results presented
in Table 2, it is observed that the coefficient of residual
mass (CRM) is positive and it varies from 0.399 to 0.103
for composited methods (for both Ia=0.2S and 0.3S)
and also for distributed method (Ia=0.3S). In distributed
CN approach for Ia=0.2S, the CRM is very low (-0.022)
which reflects the closeness of prediction with a very
little tendency of over prediction. Model efficiency varies
from 0.699 to 0.990. The highest value being for
distributed CN with Ia=0.2S. Coefficient of determination
(R2) value is very high for all the cases. This is due to

36

the fact that runoff estimated using different
methodologies use the same equation. RMSE is only
0.73 mm for distributed CN (Ia=0.2S), whereas it is as
high as 4.12 mm for composited (Ia=0.3S). From the
above results, it is seen that percentage deviation and
RMSE are lowest for runoff values estimated using
distributed CN technique with Ia=0.2S. In students’ t-
test, it is found that calculated t value for first three
techniques yield higher value than the critical table value
(t0.975,14 ) for two tail distribution. This proves the
superiority of distributed CN technique (Ia=0.2S) over
all other techniques as the calculated t value of –1.191
is lower than the critical table value of 2.145.
30
Distributed CN, Ia=0.2S
25 1:1 line
simulated runoff (mm)
20
15
10
5
0 for the performance of hydrological models. Water
0 5 10 15 20
observed runoff (mm)
Fig. 3. Observed versus simulated runoff using distributed CN
technique with Ia=0.2
Figure 3 shows a graph of observed versus
simulated runoff using distributed CN technique for
Ia=0.2S. It is clearly seen that the data points are very
close and clustered around the 1:1 line. Hence it can be
inferred that for the Tarafeni watershed distributed CN
technique considering initial abstraction as 0.2S is a
better predictor of runoff using SCS CN technique.
CONCLUSIONS
In Indian watersheds, usually composite CN
technique is used with initial abstraction as 0.1S for
black clay soil and 0.3S for rest of the soils (Narayana,
1993), But from the above investigation, it is found that
these techniques yield lower runoff values compared
to the observed value. The pertinent observations from
the study is that there is a definite increase in runoff
value estimated by distributed CN technique over
composited and the amount of increase is more for the
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
case of Ia=0.3S than Ia=0.2S. The percent increase in
runoff is very high for small events, moderate for
medium and low for high rainfall events. Hence,
distributed CN technique with initial abstraction as 0.2
times of potential maximum retention of soil should be
used for runoff estimation in medium or large agricultural
watersheds having large variation in land use. Further,
remote sensing and GIS can be effectively used to
accurately estimate runoff through distributed curve
number technique.
REFERENCES
1. ASCE task committee. 1993. Criteria for evaluation of
watershed models. J. Irri. and Drain. Engg., Vol. 119,
No. 3, pp. 429-442.
2. Hack-ten Broeke M J D; Hegmans J H B M. 1996. Use of
soil physical characteristics from laboratory
measurements or standard series of modeling
unsaturated water flow, Agril. Water. Manag. Vol.29,
No. 2, pp. 201-213.
3. Jain, M.K., Mishra, S.K. and Singh, V.P. 2006.
Evaluation of AMC-dependent SCS-CN-based models
using watershed characteristics,’ J. Water Resources
Management, 20(4): 531-555.
4. Martinec J; Rango A. 1989. Merits of statistical criteria
25 30 Resour. Bull. AWRA, Vol. 25, No. 2, pp. 421-432.
5. Mishra, S.K. and V.P., Singh, 2003. Soil Conservation
Service Curve Number (SCS-CN) Methodology, Kluwer
Academic Publishers, Dordrecht, The Netherlands,
ISBN 1-4020-1132-6.
6. Mishra, S.K., Jain, M.K., and Singh, V.P. 2004.
Evaluation of the SCS-CN-based model incorporating
antecedent moisture, J. Water Resources Management
(WARM), (18): 567-589.
7. Mishra, S.K., Jain, M.K., Bhunya, P.K. and Singh, V.P.
2005.Field applicability of the SCS-CN-inspired
Mishra-Singh general model and its variants,’ J. Water
Resources Management, 19(3): 37-62.
8. Narayana, V. V. D. 1993. Soil and water conservation
research in India, Indian Council of Agril. Research,
Krishi Anusandhan Bhawan, Pusa, New Delhi.
9. Nash J E; Sutcliffe J V. 1970. River flow forecasting
through conceptual models, Part-1: a discussion of
principles, J. Hydrol., Vol. 10, No. 3, pp. 282-290.
10. NRSA. 1995. Integrated mission for sustainable
development. technical guidelines, National Remote
Sensing Agency, Dept. of Space, Balanagar, Hyderabad,
pp. 86-87.
37

11. Pandey, A. and Dabral, P.P. 2004. Estimation of runoff
for hilly catchment using satellite data and GIS. J.
Indian Soc. Remote Sensing 32 (2): 235-240.
12. Pandey, A. and Sahu, A.K. 2004. Estimation of runoff
using IRS-1 B LISS-II data. Indian J. Soil Cons. 32 (1):
58-60.
13. Pandey, A., Chowdary, V.M., Mal, B. C. and Dabral, P.
P.2005. Estimation of surface water potential of
agricultural watershed using geographic information
system. Asian J. Geoinformatics. 5 (4): 29-36.
14. Sharma, D. and Kumar, V., 2002, Application of SCS
model with GIS data base for estimation of runoff in an
arid watershed, Journal of Soil and Water
Conservation, 30 (2) , 141-145.
15. Sharma, T., Satya Kiran, P.V., Singh, T.P., Trivedi, A.V.
and Navalgund, R.R., 2001, Hydrologic response of a
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
watershed to landuse changes: A remote sensing and
GIS approach. International Journal of Remote
Sensing, 22 (11), 2095-2108.
16. Stuebe, M. M. and Douglas M. J., 1990, Runoff volume
estimation using GIS technique, Water Resources
Bulletin, American Water Resources Association, 26(4),
611-620.
17. Thomann, R. V. 1982. Verification of water quality
models, J. Environ. Engg. Div., 108 (5), 923-940.
18. SCS (Soil Conservation Services), 1985, National
Engineering Hand book, section 4: Hydrology
(Washington, D.C., Soil Conservation Services, USDA).
19. Wilcox B P; Rawls W J; Brakensiek D L; and Weight J R.
1990. Predicting runoff from rangeland catchments: a
comparison of two models, Water Resour. Res., 26, 2401-
2410.
38
 J. Indian Water Resour. Soc.of
Journal Vol.Indian
30 No.Water
1, January, 2010
Resources
Society Vol. 30 No. 1, January, 2010
SEDIMENT RUNOFF MODELING USING ARTIFICIAL NEURAL NETWORKS

Archana Sarkar1, M. Mohan Raju2 and Anil Kumar2

ABSTRACT

The magnitude of sediments transported by rivers is a major concern for the water resources
planning and management. The methods available for sediment estimation are largely empirical,
with sediment rating curves being the most widely used. In this study, Artificial Neural Network
(ANN) technique has been applied to model the sediment-discharge relationship of a river. Daily
data of sediment concentration and discharge of Pranhita River (a sub-basin of Godavari River)
in India have been used. A comparison has been made between the results obtained using ANNs
and sediment rating curves. The sediment load estimations in the river obtained by ANNs have
been found to be significantly superior to the corresponding classical sediment rating curve
ones.

INTRODUCTION fitting methods are unable to model the non-linearity in

The sediment outflow from a catchment is induced
by processes of detachment, transportation and
deposition of soil materials by rainfall and runoff. The
assessment of the volume of sediments being
transported by a river is required in wide spectrum of
problems such as the design of reservoirs and dams;
hydroelectric power generation and water supply;
transport of sediment and pollutants in rivers, lakes and
estuaries; determination of the effects of watershed
management; and environmental impact assessment.
Keeping this in view, in the present study, runoff-
sediment modeling has been carried out for Pranhita
River, a major tributary of Gopdavari River. Pranhita
sub-basin system, which conveys the combined waters
of Penganga, Wardha and Wainganga influences the
Godavari river system to the maximum possible extent
(with 34% drainage area i.e., 1,09,100 km2 area) by
means of rainfall, runoff and sediment transportation.
For estimating the sediment concentration/yield,
there exist various models and techniques, such as
sediment rating curves, erosion modeling, etc. The
models vary from a simple regression relationship to
complex simulation models. As the sediment-discharge
relationship is not linear, conventional statistical tools
used in such situations such as regression and curve
1
National Institute of Hydrology, Roorkee-247667, Uttarakhand,
India
2
Department of Soil and Water Conservation Engineering, College of
Technology, G. B. Pant University of Agriculture and Technology,
Pantnagar-263145, Uttarakhand, India.
Paper No. 1181
the relationship. On the other hand, the application of
physics-based distributed process computer simulation
offers another possible method of sediment prediction.
But the application of these complex software programs
is often problematic, due to the use of idealized
sedimentation components, or the need for massive
amounts of detailed spatial and temporal environmental
data, which are not available. Simpler approaches are
therefore required in the form of ‘black-box’ modeling
techniques. Neurocomputing provides one possible
answer to the problematic task of sediment transfer
prediction. In recent years, artificial neural networks
(ANNs) which are simplified mathematical
representation of the functioning of the human brain
have been widely used in runoff and sediment yield
modeling. Three layered feed forward ANNs have been
shown to be a powerful tool for input-output mapping
and have been widely used in water resources
engineering problems (ASCE Task Committee, 2000).
The application of ANN approach for modeling
sediment-discharge process is very recent, and has
already produced very encouraging results. In a
research project by Rosenbaum (2000), ANN technique
has been used to predict sediment distribution in Swedish
harbors. Baruah et al., (2001) developed neural network
models of Lake surface chlorophyll and sediment
content from LandsatTM imagery in order to assess
the water quality of the lake Kasumigaura in Japan and
found that back propagation neural network with only
one hidden layer could model both the parameters better

39

than conventional regression techniques. Jain (2001)
used the ANN approach to establish an integrated stage-
discharge-sediment concentration relation for two sites
on the Mississipi River and showed that the ANN results
were much closer to the observed values than the
conventional technique. Nagy et al., (2002) applied
ANN technique to estimate the natural sediment
discharge in rivers in terms of sediment concentration
and addressed the importance of choosing an appropriate
neural network structure and providing field data to that
network for training purpose. Sarkar et al., (2004)
applied ANN technique to model the sediment-discharge
relationship of Kosi River of Bihar in India and found
that sediment load estimations in the river obtained by
ANNs were superior to the sediment rating curve ones.
Kerem et al., (2006) estimated the river suspended
sediment using ANN algorithm and compared with the
conventional sediment rating curve and found the
superiority of the ANN application. Kisi (2007) and Rai
& Mathur (2008) have reported good application
efficiency of ANNs in the sediment yield modeling when
compared with the conventional modeling techniques.
In the present study, two techniques, namely,
sediment rating curve and artificial neural networks
(ANNs), have been applied for modeling the sediment-
discharge relationship for Pranhita River basin and a
comparison of these techniques has been made.
SEDIMENT RATING CURVES
Sediment rating curves are widely used to estimate
the sediment concentration being transported by a river.
A sediment rating curve is a relation between the
sediment concentration and river discharge. Sediment
rating curves may be plotted showing average sediment
concentration or load as a function of discharge averaged
over daily, monthly, or other time periods. Rating curves
are developed on the premise that a stable relationship
between concentration and discharge can be developed
which, although exhibiting scatter, will allow the mean
sediment yield to be determined on the basis of the
discharge history. A problem inherent in the rating curve
technique is the high degree of scatter, which may be
reduced but not eliminated. Concentration does not
necessarily increase as a function of discharge
(Ferugson 1986).
Mathematically, a rating curve may be constructed
by log-transforming all data and using a linear least
square regression to determine the line of best fit. The
log-log relationship between load and discharge is of
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
the form:
C = aQb (1)
And the log-transformed form will plot as a straight line
on log-log paper:
log C = log a + b log (Q) (2)
Where, C = sediment concentration (or load), Q =
discharge, a & b are regression constants.
ARTIFICIAL NEURAL NETWORKS (ANNs)
An ANN is a computing system made up of a highly
interconnected set of simple information processing
elements, analogous to a neuron, called units. The
neuron collects inputs from both a single and multiple
sources and produces output in accordance with a
predetermined non-linear function. An ANN model is
created by interconnection of many of the neurons in a
known configuration. The primary elements
characterizing the neural network are the distributed
representation of information, local operations and non-
linear processing. Fig.1 shows the general structure of
a three layer back propagation ANN.
h1
wh11 wo11
x1
y1
x2 y2
Input . Output
. .
. : yno
. h4
xni
.
. wono,ni
whnh,ni
hnh
Input layer Hidden layer Output layer
i =1, 2, 3……ni j=1, 2, 3 …….nh k=1, 2, 3…….no
Fig 1. Structure of a multi-layer feed forward artificial
neural network model.
The main principle of neural computing is the
decomposition of the input-output relationship into series
of linearly separable steps using hidden layers (Haykin,
1994). Generally there are four distinct steps in
developing an ANN-based solution. The first step is
the data transformation or scaling. The second step is
the network architecture definition, where the number
of hidden layers, the number of neurons in each layer,
and the connectivity between the neurons are set. In
the third step, a learning algorithm is used to train the
network to respond correctly to a given set of inputs.
Lastly, comes the validation step in which the
performance of the trained ANN model is tested through
40

some selected statistical criteria. The theory of ANN
has not been described here and can be found in many
books such as Haykin (1994).
STUDY AREA, DATA AVAILABILITY AND
SELECTION OF INPUT/OUTPUT VARIABLES
In the present study the runoff-sediment modeling
has been carried out for Pranhita sub-basin system,
which conveys the combined waters of Penganga,
Wardha and Wainganga. The hydrological data for the
study has been collected at Tekra site on Pranhita river
(Fig. 2). After the Tekra site, Pranhita river joins the
main Godavari in Andhra Pradesh.
The daily data of sediment concentration and
discharge were available at the Tekra site for four water
Fig. 2 Pranhita river system and hydrological study location (Tekra site)
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
years (June 1, 2000 – May 31, 2004) constituting a total
of 1461 patterns. Out of this, 730 patterns were used
for training, 365 patterns for testing and 366 patterns
for validation.
The first step in developing any model is to identify
the input and output variables. The output from the
models is the sediment concentration at time step t; Ct.
It has been shown by many authors that the current
sediment concentration can be mapped better by
considering, in addition to the current value of discharge,
the sediment and discharge at the previous times.
Therefore, in addition to Qt, i.e., discharge at time step
t, other variables such as Qt-1, Qt-2, and Ct-1, Ct-2, were
also considered in the input.
41
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

RATING CURVE ANALYSIS Q = Discharge in the River Pranhia at Tekra in

Based on the sediment rating curve technique given
by equation (1), the sediment rating equation between
sediment load and discharge for Pranhita River at Tekra
site for the training period is
C = (4.94E-04)Q0.770
Where, C = Sediment concentration in the River
Pranhia at Tekra in gm/l at time t
Cumec at time t
DESIGN AND TRAINING OF ANN MODELS
Various combinations of input data considered for
training of ANN in the present study are given in Table
(3) 1. However, the input-output variables of ANN-1 have
been used for the conventional sediment rating curve
analysis.

Table 1: Various ANN Runoff-Sediment Models

ANN Architecture Output Input Variables

Model Variable
ANN-1 [1 – 2 – 1] Ct Qt
ANN-2 [3 – 3 – 1] Ct Qt, Qt-1, Ct-1
ANN-3 [5 – 4 – 1] Ct Qt, Qt-1, Qt-2, Ct-1, Ct-2
ANN-4 [7 – 6 – 1] Ct Qt, Qt-1, Qt-2, Qt-3, Ct-1, Ct-2, Ct-3
ANN-5 [9 – 7 – 1] Ct Qt, Qt-1, Qt-2, Qt-3, Qt-4, Ct-1, Ct-2, Ct-3, Ct-4
ANN-6 [1 – 2 – 5 – 1] Ct Qt
ANN-7 [3 – 3 – 5 – 1] Ct Qt, Qt-1, Ct-1
ANN-8 [5 – 4 – 5 – 1] Ct Qt, Qt-1, Qt-2, Ct-1, Ct-2
ANN-9 [7 – 6 – 5 – 1] Ct Qt, Qt-1, Qt-2, Qt-3, Ct-1, Ct-2, Ct-3
ANN-10 [9 – 7 – 5 – 1] Ct Qt, Qt-1, Qt-2, Qt-3, Qt-4, Ct-1, Ct-2, Ct-3, Ct-4

Where, C=Sediment Concentration at Tekra in g/l,
Q=Discharge at Tekra in cumecs, t represents the time
step in days
A back-propagation ANN (BPANN) with the
generalized delta rule as the training algorithm has been
employed in this study. The ANN package Neural
Power (NP), 2003, downloaded from the Internet has
been used for the ANN model development. The
structure for all simulation models is three and four layer
BPANN which utilizes a non-linear sigmoid activation
function uniformly between the layers. Nodes in the
input layer are equal to number of input variables, nodes
in hidden layer are varied from the default value by the
NP package for various number of input nodes above
to approximately double of input nodes (Zhu et al., 1994)
and the nodes in the output layer is one as the models
provide single output. According to Hsu et al. (1995),
three-layer feed forward ANNs can be used to model
real-world functional relationships that may be of
unknown or poorly defined form and complexity.
Therefore, three-layer networks were tried in this study.
However, four layer ANN models were also tried for
comparison purpose.
The modeling of ANN initiated with the
normalization (re-scaling) of all inputs and output with
the maximum value of respective variable reducing the
data in the range 0 to 1 to avoid any saturation effect
that may be caused by the use of sigmoid function. All
interconnecting links between nodes of successive layers
were assigned random values called weights. A constant
value of 0.15 and 0.8 respectively has been considered
for learning rate  and momentum term  selected after
hit and trials. The range tried for learning rate  and
momentum term  were 0.10-0.40 and 0.7-0.9
respectively. The quick propagation (QP) learning
algorithm has been adopted for training of all the ANN
models. QP is a heuristic modification of the standard
back propagation and is very fast. The network weights
were updated after presenting each pattern from the
learning data set, rather than once per iteration. The
criteria selected to avoid over training was generalization
of ANN through cross-validation (Haykin, 1994). For
this purpose, the data were divided into training, testing
and validation sets. Training data (730 patterns) were
used for estimation of weights of the ANN model and
testing data (365 patterns) for evaluation of the
performance of ANN model during training. Training
was stopped when the error for the testing dataset
started increasing i.e., when R started decreasing for
the testing dataset as the software does not calculate
RMSE automatically during testing/validation phase. In

42
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

this way, the training and testing datasets have been
used to assess the performance of various candidate
model structures, and thereby choose the best one. The
particular ANN model with the best performing
parameter values was chosen and the generalized
performance of the resulting network has been measured
on the validation data set (366 patterns) to which it has
never before been exposed. The performance of all
the ANN models have been tested through three
statistical criterion, viz, root mean square error (RMSE),
correlation coefficient (R) and Coefficient of
Determination (DC).
RESULTS AND DISCUSSION
The comparative performance of various ANN
models and rating curve analysis in terms of RMSE, R
and DC are given in Table 2. It can be seen from Table
2 that the RMSE values are generally low (less than
0.075g/l) for all the ANN models except ANN-1 and
ANN-9 models, during training. However, RMSE is
lowest for ANN8 (0.151g/l) during testing and for ANN9
(0.162g/l) during validation. Whereas, RMSE of the
rating curve model is very high, i.e., 0.472g/l, 0.359g/l
and 0.424g/l during training, testing and validation
respectively.
It can be seen from Table 2 that the correlation (R)
values are high (more than 0.90) for all the ANN models
during all the three phases. The performance of ANN-
9 model is the best in R statistic with R values of 0.959
and 0.954 during testing and validation respectively. The
performance of the rating curve model is the worst when
compared with the ANN models. The R values for rating
curve model are 0.923, 0.870 and 0.867 during training,
testing and validation respectively. These values are
even lower than the worst ANN model, i.e., ANN-1
model.

Table 2: Comparative Performance of Various ANN models and Rating Curve

ANN Network Training Testing Validation

Model Architecture RMSE R DC RMSE R DC RMSE R DC
(g/l) (g/l) (g/l)
ANN1 [1-2-1] 0.075 0.978 0.957 0.228 0.907 0.743 0.284 0.912 0.667
ANN2 [3-3-1] 0.067 0.983 0.966 0.170 0.951 0.856 0.217 0.946 0.805
ANN3 [5-4-1] 0.063 0.985 0.969 0.174 0.949 0.850 0.223 0.945 0.794
ANN4 [7-6-1] 0.060 0.986 0.973 0.168 0.953 0.860 0.217 0.943 0.806
ANN5 [9-7-1] 0.064 0.985 0.969 0.173 0.944 0.852 0.210 0.940 0.819
ANN6 [1-2-5-1] 0.071 0.980 0.962 0.229 0.903 0.740 0.284 0.919 0.675
ANN7 [3-3-5-1] 0.070 0.981 0.963 0.159 0.955 0.874 0.198 0.949 0.837
ANN8 [5-4-5-1] 0.067 0.984 0.966 0.151 0.953 0.887 0.217 0.920 0.805
ANN9 [7-6-5-1] 0.076 0.978 0.957 0.187 0.959 0.904 0.162 0.954 0.891
ANN10 [9-7-5-1] 0.056 0.987 0.974 0.169 0.950 0.859 0.210 0.948 0.818
SRC -- 0.472 0.923 0.680 0.359 0.870 0.870 0.424 0.867 0.668

In the coefficient of determination (DC) statistic,
all the ANN models except ANN-1 perform well. In
DC statistic also, ANN-9 model performs the best during
testing and validation phases. The performance of rating
curve model in DC statistic has gone down drastically
with DC values as low as 0.68, 0.87 and 0.668 during
training, testing and validation respectively.
It can be seen that ANN-9 model is the best
performing model in two statistical and hydrological
criteria during testing and validation phase. ANN( model
also has low RMSE values of 0.076, 0.187 and 0.162
during training, testing and validation respectively. A
model which performs better in the validation phase is
the best model as it is the most generalized model.
Therefore, ANN-9 model is the model representing
sediment-discharge relation of the Pranhita River at the
gauging site Tekra. The performance of the rating curve
model is average in the R criteria but drastically poor in
other two criteria. It is because the estimated sediment
series (from sediment rating curve model) follows a
good general trend as that of the observed sediment
series which gives high R values, but there is a significant
difference in the numeric values of observed and
estimated sediment concentration due to which the
RMSE and DC values are very poor. The performance
of the corresponding ANN model with only discharge
as input, i.e. ANN-1 is also better as compared with
the sediment rating curve technique.
The temporal variation of the observed sediment
concentration and the estimate using the conventional

43
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

technique and ANN (ANN-9) for the training, testing significant mismatch with the observed curve, especially
as well as validation period is plotted in Fig 3. It is seen during the validation phase which conforms to the low
from the graph that ANN estimates very closely follow coefficient of determination of rating curve technique.
S e d i m e n t c o n c e n t r a t i o n ( g /l )

the observed curve, whereas conventional approach has

14
Obs erv ed
12 Es tiamte d (A N N )
10 Es timate d (SR C )

8
6
4
2
0
1 101 201 301 401 501 601 701
-2
T im e (d ay s )
S e d i m e n t c o n c e n t r a t i o n ( g /l )

(a) Training

5
Observed
4.5 Est imated (A N N )
4 Est imated (SR C )
3.5
3
2.5
2
1.5
1
0.5
0
-0.5 1 101 201 301
T im e (d ays )
S e d i m e n t c o n c e n t r a t i o n ( g /l )

(b) Testing

3.5
Observed
Est imated (A N N )
3
Est imated (SR C )
2.5

1.5

0.5

0
1 101 201 301
-0.5 T im e (d ays )

(c) Validation
Fig. 3. Comparative performance of observed, estimated (ANN) and estimated (rating curve) sediment concentration series

44

CONCLUSIONS
In the presented study ANN technique has been
utilized for modeling the sediment-discharge process in
a river. The primary aim of the presented study is to
illustrate the capability of ANN technique for modeling
sediment load in rivers. To achieve the objectives, a
case study has been done utilizing four years of daily
data of the Tekra gauging site of Pranhita river (a sub-
basin of Godavari River) in India for analysis. Based
on the selected performance evaluation criteria, ANN-
9, i.e., four layered back propagation ANN model with
7 input variables, 6 nodes in the first hidden layer, 5
nodes in the second hidden layer and one output variable
represents the best model simulating the sediment-
discharge relationship of the Pranhita River at Tekra
gauging site. The results of ANN have been compared
with those of the conventional sediment rating curve
approach. ANN results have been found to be much
closer to the observed values than the conventional
technique.
The study demonstrates that ANN technique can
be successfully applied for development of reliable
relationships between sediment and discharge in a river
when other approaches cannot succeed due to the
uncertainty and the stochastic nature of the sediment
movement. Moreover, ANN technique has preference
over the conventional methods as ANNs can accept
any number of effective variables as input parameters
without omission or simplification as commonly done in
conventional methods. The presented ANN model is
designed by using only field river data, and it has no
boundary conditions in application. The only restriction
is that the model cannot estimate accurately the
sediment load for data out of the range of training pattern
data. Such a problem can easily be overcome by feeding
the training patterns with wide range data.
REFERENCES
1. ASCE. Task Committee on Application of Artificial
Neural Networks in Hydrology. 2000, “Artificial Neural
Networks in Hydrology. II: Hydrologic Applications”,
Journal of Hydrologic Engineering, ASCE, Vol.5 (2),
pp.124-137.
2. Baruah, P. J., Tamura, M., Oki, K. and Nishimura, H.
2001, “Neural Network Modelling of Lake Surface
Chlorophyll and Sediment Content from LandsatTM
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
Imagery”, Proc. 22nd Asian Conf. on Remote Sensing,
5-9 November 2001, Singapore.
3. Ferugson, R.I. 1986, “River Loads Underestimated by
Rating Curves”, Water Resources Research, Vol. 22(1),
pp.74-76.
4. Haykin, S. 1994, “Neural Networks - a Comprehensive
Foundation. Macmillan”, New York.
5. Hsu, K., Gupta, H.V., & Sorooshian, S. 1995, “Artificial
Neural Network Modelling of the Rainfall-Runoff
process”, Water Resources Research, vol. 31(10),
pp.2517-2530.
6. Jain, S.K. 2001, “Development of Integrated Sediment
Rating Curves using ANNs”, J. of Hydraulic
Engineering, ASCE, Vol. 127(1), pp.30-37.
7. Kerem, H., Cigizoglu, and Alp, M. 2006. Generalized
regression neural network in modeling river sediment
yield. Advances in Engineering Software archive,
37(2), pp. 63-68.
8. Kisi, O. 2007. Development of streamflow-suspended
sediment rating curve using a range dependent neural
network. International Journal of Science and
Technology.
9. Nagy, H.M., Watanabe, B., and Hirano, M., 2002,
“Prediction of Sediment load Concentration in Rivers
Using Artificial Neural Network Model”, J. of
Hydraulic Engineering, ASCE, Vol. 128(6) pp.588-595.
10. Rawat J.S and Rawat M.S. 1994, “Accelerated erosion
and denudation in the nana kosi watershed, Central
Himalaya, India, Part I: sediment load”, J. of Mountain
Research and Development, Vol. 14(1), pp 25-38.
11. Raymo M.E. and Ruddiman W.F. 1992, “Tectonic forcing
of Late Cainozoic Climate”, Nature, Vol. 359, pp. 117-
122.
12. Rosenbaum, M 2000, “Harbours- Silting and
Environmental Sedimentology (H-SENSE), Final
Report, Dept. of Civil & Structural Engineering”, The
Nottingham Trent University, Nottingham, UK. http://
hjs.geol.uib.no/HSense/
13. Sarkar, A., Kumar, R., Singh, R.D., Thakur, G and Jain,
S.K. 2004, “Sediment-Discharge Modelling in a River
using Artificial Neural Networks”, Proc. Int. Conf.
ICON-HERP, Oct 26-28, 2004, IIT, Roorkee, India.
14. Zhu, M., Fujita, M., and Hashimoto, N. 1994,
“Application of Neural Networks to Runoff Prediction,
“Stochastic and Statistical Method in Hydrology and
Environmental Engineering, vol. 3, K.W. Hipel et al.,
eds., Kluwer, , The Netherlands, pp.205-216.
45
 J. Indian Water Resour. Soc.of
Journal Vol.Indian
30 No.Water
1, January, 2010
Resources
Society Vol. 30 No. 1, January, 2010

SINGLE RESERVOIR OPTIMIZATION USING HYBRID GENETIC

ALGORITHM

Nitin M Mohite1 and Sandeep Narulkar2

ABSTRACT

In this paper a hybrid genetic algorithm approach is proposed wherein the GA acts as a
base level search algorithm that generates the initial values for solving the problem with a con-
strained optimization algorithm. The adaptation of the hybrid approach has eliminated the prob-
lems associated with the GA as well as the constrained optimization. The hybrid method proves to
be superior to the pure GA in finding a good solution quickly. The problem is applied to a single
reservoir planning problem for Parbati-Kalisindh-Chambal Interlinking Project in India.

Keywords : Optimization of single reservoir; Genetic algorithms, Hybrid methods

INTRODUCTION reservoir system operation and compared the

The water resources planning and management
problems, particularly the reservoir design and operation
problems are quite complex and dimensionally large.
Simulation and optimization techniques play an important
role in providing a good insight in solving these problems.
Numerous researchers have solved the problems of
optimum reservoir system planning and operation during
past five decades using Simulation, Linear Programming
(LP), Dynamic Programming (DP), Nonlinear
Programming (NLP) etc. along with variants and
combination of these techniques. An exhaustive and
extremely useful state-of-the-art review of simulation
and optimization methods applied to reservoir systems
planning and operation problems can be seen in the
works of Yeh (1985), Simonovic (1992), Wurbs (1993),
and Labadie (2004) etc. In the recent era various
evolutionary search algorithms i.e. the Genetic
Algorithms (GA), Ant Colony Optimization (ACO),
Particle Swarm Optimization (PSO) etc. have taken
the fore seat.
The Genetic Algorithm is a global stochastic search
technique based on the Darwinian Survival-of-the-
Fittest principle (Holland 1975). GA models have been
applied successfully to a wide range of reservoir design
and optimization problems of moderate size. East and
Hall (1994) and Fahmy et al. (1994) applied GA to the
1
Research Scholar, Dept of Civil & Applied Mechanics S.G.S.I.T.S,
Indore
2
Professor, Dept. of Civil & Applied Mechanics S.G.S.I.T.S, Indore
Paper No. 1216
performance of the GA approach with that of DP. Both
the works showed the significant potential of GA in
water resources system optimization, and clearly
demonstrated the advantage of GA over standard DP
in terms of computational requirement. Oliveira and
Loucks (1997) used a GA model to evaluate operating
rules for multi reservoir systems, demonstrating that GA
can be used to identify effective operating policies.
The breakthrough research work on GA applications
to reservoir planning and operation problems was
presented by Wardlaw and Sharif (1999). They
evaluated GA technique through the solution of the
classic 4 reservoir problem and the 10 reservoir problem
with extended time horizons explaining various aspects
of the applicability of the technique. They evaluated
several combinations of GA parameters viz. the binary
coded and real coded representation, selection,
crossover and mutation etc. Sharif and Wardlaw (2000)
applied the technique to a multiple reservoir systems
for a case study in Indonesia by considering the existing
development situation in the basin and two future water
resources development scenarios. Results of GA were
compared with those obtained from discrete differential
dynamic programming. They concluded that genetic
algorithm results are closer to the optimum.
Many other works reported in the literature including
Hilton and Culver (2000) comparing Additive Penalty
Method (APM) and Multiplicative Penalty Method
(MPM) for constraint handling in non linear problems
with GA technique, Ahmed and Sarma (2005) comparing

46

GA and Stochastic Dynamic Programming (SDP) to
derive optimal operating policy, Nageshkumar et al.
(2006) stating a GA model for obtaining an optimal
operating policy and optimal crop water allocation from
an irrigation reservoir, Jothipraksh and Ganesh Shanti
(2006 and 2009) demonstrating application of GA to
multipurpose reservoir system, Momtahen and Dariane
(2007) comparing Real Coded Genetic Algorithm with
that of conventional model involving explicit stochastic
optimization (ESO) and implicit stochastic optimization
(ISO) approach, Kangrang and Chaleeraktrakoon
(2007) discussing a problem of deriving optimal
operating rule curves by using genetic algorithms
connected simulation models, Chouhan and Shrivastava
(2008) applying preference based approach using GA
based model for optimal operation of a reservoir system
maximizing irrigation in the command area for two
seasons and inter reservoir transfers in one of the
seasons, Hashemi et al. (2008) showing the applicability
of GA model in MATLAB environment for operation
of a multipurpose reservoir in stochastic frame work
etc. can be considered to be the benchmark works in
the reservoir system planning and operation. Recent
work of Dariane and Momtahen (2009) demonstrating
an application of direct search genetic algorithm (DSGS)
for multi- reservoir system operation can also be
considered as important from the view point of large
scale reservoir system problems.
Despite the fact that the GA Technique is one of
the most suitable technique for direct search and as an
optimization model, its applicability to large scale
problems has certain limitations as pointed out by many
researchers. Kapelan (2002) has thoroughly
summarized the advantages and limitations of GA. He
mainly points out that despite the GA Technique having
good capability of an efficient exploration of large,
complex, multi-modal search spaces with a less chance
to get trapped into the local optimum, their use in terms
of CPU time is usually very expensive. He further adds
that there is no guarantee that the global optimum will
be found, even though good solutions are usually found.
Moreover, in the constrained problems if a penalty
function is not chosen carefully, it may significantly affect
GA search performance. He also noticed that GA
suffers from the slow finishing problem. Premature
convergence is another GA disadvantage. One of the
greatest drawbacks of GAs is that they require a high
number of function evaluations to achieve convergence.
Each function evaluation entails a full extended-period
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
simulation of the system, which is a computationally
expensive process. The net result is that GA optimization
is time consuming (Van Zyl et al., 2004). Similar
conclusions were drawn by Chun et al. (2008).
The limitations and disadvantages of the GA
techniques can be alleviated through the use of Hybrid
Genetic Algorithm (HGA). The basic concept of HGA
is coupling of two models, the GA and a constrained or
an unconstrained nonlinear optimization model. A large
number of such hybrid algorithms have been reported
viz. a combination of GA and Sequential Quadratic
Programming SQP (Marco, 1996, 1998), GA and
Generalized Reduced Gradient (GRG) (Yen, 2005), GA
coupled with Chaos Optimization Algorithm (COA)
(Chun, 2008). These combinations have made the
complex optimization problems simpler and easily
applicable. Kapelan (2002) narrates following
advantages of HGA over GA as follows:
1. Increased computational efficiency typically
manifested as reduced CPU search time necessary
to find an optimal solution, or to reach some pre-
specified objective function value.
2. The optimal objective function value found by HGA
is better or similar to the one found by GA, indicating
better (or similar) model fit. The variation of best
fitnesses from multiple HGA runs is usually smaller
than the variation of best fitnesses determined from
multiple GA runs.
3. More accurate (less uncertain) calibration
parameter values are typically determined with
HGA than GA.
The applications of HGA to water resources
problems are of recent origin except a few studies.
Marco Franchini (1996) and Marco Franchini et al.
(1998) applied HGA for calibration of conceptual rainfall
runoff modeling. Hsiao and Chang (2002) solved an
optimum pumping problem with the use of HGA.
Kapelan (2002) applied the HGA to a pipe network
calibration problem. Van Zyl et al., (2004) applied GA
and two popular search methodologies the Hooke and
Jeeves and Fibonacci methods to solve the problem of
water distribution system network problem. Espinoza
et al (2005) have developed new Self Adaptive Hybrid
Genetic Algorithm (SAHGA) to solve optimal ground
water remediation problem. They compared the results
with simple genetic algorithm and Non Adaptive Hybrid
Genetic Algorithm (NAHGA). They concluded that the
selection of the local search algorithm to be combined
47

with the simple genetic algorithm is critical to algorithm
performance.
Reis et al. (2005) have applied hybrid genetic
algorithm (GA-LP) technique for multi-reservoir
operation planning problem. The results were compared
with the stochastic dual dynamic programming (SDDP).
Reis et al. (2006) have applied hybrid genetic algorithm
method using genetic (GA) and linear programming (LP)
to determine operational decisions for a reservoir system
over the optimization period. Ebrahimi et al. (2007) used
the G.A approach and four combined models of G.A
and wavelet transforms to optimize the operation of
the Vanyard dam reservoir. Chung et al. (2008) proposed
a novel Chaos Genetic Algorithm (CGA) and applied it
to optimal operation of hydropower station reservoir
system in China. They combined chaos optimization
algorithm (COA) and genetic algorithm (GA) to
overcome premature local optimum and increase the
convergence speed of genetic algorithm. Li and Wei
(2008) developed a GA – Simulated Annealing Method
for operation planning of a multiple reservoir system.
The objective was to maximize generation output from
the 3-reservoir systems over each 12-month operating
periods. The developed algorithm was stated to be faster
and better in comparison to GA. Yuan et al. (2008) have
proposed a new real value encoding self-adaptive chaotic
genetic algorithm to solve hydrogenation scheduling
problem. The new crossover operation with probability
distribution function and a self-adaptive chaotic mutation
operator combined chaotic dynamic character with
artificial neural network theory are used. The result
shows that the model gives better quality solution.
THE MODEL
Though Hybrid GA model is considered to perform
better than the GA model, proper selection of the search/
optimization algorithm to supplement GA is still a
question. In most of the applications the local search
part of the algorithm is problem specific. However, the
reservoir planning and operation problems pose a
constrained optimization problem and one should use a
constrained optimization model for the search of the
optimum solution. The best constrained non linear model
is the Quadratic Programming (QP) model if the
objective function is strictly quadratic and the constraints
are linear. The convergence of the QP to a satisfactory
local optimum shall be fast and improved if this model
is coupled with a GA model which provides a good initial
point for the QP model.
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
In the present study the GA is applied first to produce
the proper starting point. Then Quadratic Programming
(QP) technique is used through optimization tool box in
MATLAB Environment. The MATLAB software is
extremely useful and user friendly mathematical tool to
solve various problems. In early seventies MATLAB
software was written to solve the problem in the field
of matrices, linear algebra and numerical analysis in
Stanford and New Mexico University. The important
feature of this software is a matrix that doesn’t need to
specify its dimensions. So the solution of the problem is
as easy as writing them. There are several different
toolboxes available in MATLAB. One of the toolbox is
for the applications of the Genetic Algorithms. The
invocation of the GA is through GA tool which accepts
the objective functions and the constraints in the form
of various matrices. The GA tool also specifies a wide
range of parameters to be chosen with the GA
application and also the convergence criteria. The
genetic algorithm terminates when one of the following
exit conditions are met: (1) No improvement in the search
direction (2) no improvement in the solution for more
than pre defined maximum number of iterations
(generations) (3) maximum time in seconds the genetic
algorithm runs before stopping. The GA results are then
considered as an initial starting point for QP. The QP
toolbox is accessible form: Start/Toolboxes/
Optimization/Optimization tool (optimtool). More details
of the use of these functions can be found in MATLAB
User’s guide (2007b). A flow diagram of the HGA
method is given in Fig. 1
(GA Toolbox using GA and QUADPROG options
of MATLAB 7.0).
The proposed model is applied to a reservoir
planning model. The reservoir is a component of an
inter basin water transfer system in west central India.
The details of the system and the model etc. are as
follows.
MODEL FORMULATION
The objective of the model is to optimize the capacity
of a single reservoir that serves multiple purposes such
as Municipal and Industrial water supply, Irrigation and
Transfer of water to other reservoir. The present
objective is to minimize the capacity of a reservoir while
maximizing the other objectives. The objective function
is:
48
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Initial Population
Generation no (G=1) TB = Relative weight indicating the benefit coefficient
Specification of
Parameters of unit volume of water transfer in Millions/MCM
TRt = Monthly transfer of water from the reservoir to
Evaluation of other reservoir in period t in MCM
Objective function
(Fitness function)
TTt = Monthly transfer target of water from the reservoir
to other reservoir in period t in MCM
Selection

The objective function is subjected to following

Crossover
constraints
Mutation
The reservoir continuity equation
Elitism
The continuity equation for reservoir storage at the
end of time t , S t 1 is calculated as
If any:
1. Meet specific value
for objective function
Start QP:
Use GA final solution as
(1  at ) S t 1  (1  at ) S t  I t  IRt  TRt  SPt  Lt (2)
2. Meet stall generation QP required initial point
limits
Where
Solve the problem with
QP
S t = Reservoir storage at the beginning of time period t
Converge to
I t = Inflow in the reservoir in time period t
desired
solution? SPt = Spillage in time period t
Lt = the fixed evaporation loss at site = et * Ao , et is
Final solution
the evaporation rate in period t , Ao = water surface area
Fig.1 Flow chart of hybrid genetic algorithm
at the top of the dead storage level, at  aet / 2, and
a is the surface area per unit active storage.
Minimize
Y Y
The S t , Irrt , Tt , S t are in units of Million cubic meter
A * (C p ) 2  B * (Cc ) 2   Birr * ( IRt  IDt ) 2   TB * (TRt  TTt ) 2
t 1 t 1 (1) (x106 m3) and et is in units of millimeters (mm). The
method of computation of evaporation is as per the
Where procedure stated in Loucks et al. (1981).
u = Total no of years of planning
Bounds on Capacity Constraints
T = Total No. of time steps used for planning in each
year The reservoir capacity is constrained by upper
Y = Total No. of time steps in the time horizon (= u x T) bound
A = Relative weight indicating the unit cost of capacity C p  C pu (3)
of reservoir.
C p = Live storage capacity of the reservoir in MCM C pu = The upper bound on the capacity in MCM
B = Relative weight indicating the unit cost of capacity C c  C cu (4)
of canal.
C c = The canal Capacity in Cumec. C cu = The upper bound on the Canal capacity in Cumec.
Birr = Relative weight indicating the benefit coefficient
of unit volume of water supplied to irrigation in Millions/ Reservoir capacity constraint
MCM S t 1  C p  0 (3)
IRt = Monthly irrigation released at the reservoir in
period t in MCM The storage in the reservoir under consideration in
any time period t cannot exceed the live storage
IDt = Monthly target irrigation demand at the reservoir
capacity.
in period t in MCM

49

Canal capacity constraint
The water released from reservoir for irrigation and
transfer form the reservoir during t time cannot exceed
the canal capacity.
 * ( IRt  TR t )  C c (4)
t  1, 2,.., T ;  = factor of conversion of MCM to
cumec (m3/s)
THE CASE STUDY
The model formulated above is applied to the
Patanpur reservoir, which is first proposed reservoir in
Parbati-Kalisindh-Chambal (P-K-C) link project of the
most ambitious river interlinking project envisaged by
the Government of India. The reservoir is a multipurpose
reservoir having municipal and industrial water supply,
irrigation and water transfers to other reservoir for the
purpose of inter-basin water transfer as the objectives.
The proposed reservoir is located between the North
latitude 230 42’ and 240 2’and East longitudes of 770 9’
and 770 15’. The climate of the area is subtropical and
semi-arid. The maximum and minimum annual rainfall
in the basin varies from 356 mm to 1270 mm. The
proposed active storage capacity of the reservoir is 110
MCM and catchment area 5312 Km2. In the present
study the main objective is to optimize the live storage
capacity of the reservoir and the link canal capacity
maximizing the fulfillment of the objectives of the inter-
basin transfer.
MODEL APPLICATION
The model formulated above was applied to the
Patanpur reservoir which is one of the reservoirs of
Parbati-Kalisindh-Chambal (P-K-C) link project. The
present values of cost parameter of dams and canals
and annual benefit of irrigation and transfer of water
have been computed form the available data (Pre-
Feasibility Report, 2004). The unit cost of structure was
calculated by dividing its design project cost by the design
capacity as per NWDA. The rate of annual interest
was taken as 5% of the respective capital cost. The
annual operation and maintenance cost is assumed to
be 10% of the total cost of project. The cost parameter
(including all costs i.e. Construction cost + Operation
and Maintenance cost) of the reservoir, canal and the
benefit parameters of irrigation and transfers shown in
Table 1. The water demands in various reservoirs and
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
transfer targets are shown in Table 2 and 3 respectively.
The target annual irrigation yield was at 124 million cubic
meters (MCM) and municipal requirement at 3 MCM.
An inflow record of 20 years was used to demonstrate
the application. The transfer targets were varied from
the NWDA specified values and are specified as D1,
D2, D3 and are mentioned in Table 3. As per the priorities
for water demand satisfaction set by the NWDA the
highest priority municipal and industrial demand was
fulfilled in all the seasons by subtracting it from the
natural inflows. The irrigation demands as well as the
transfer demands were weighed according to the
expected benefits. The upper bound on the reservoir
capacities were varied from the NWDA specified
values and many trial runs were taken to generate
minimum capacity of the reservoir fulfilling the monthly
irrigation targets. The upper bound on the canal capacity
was also varied in similar manner.
For model computations, the MATLAB software
was used since it is easy to use and can accommodate
large scale optimization problems. The GA component
of the model had many options due to various GA
Parameters viz. the population size, number of
generations, the crossover and mutation probabilities etc.
In general a population size of 500 and a generation
number of about 1000 was sufficient for a good degree
of optimum result as an initial value for QP algorithm.
The crossover probabilities were taken in the range of
0.7 to 1 and the mutation probabilities are taken in a
range .001 to 0.3. Better results are obtained at a
crossover probability at 0.75 and the mutation probability
at 0.002. Further the problem was solved through QP
using optimization toolbox in the MATLAB Environment.
RESULTS AND DISCUSSION
The results of application of the Hybrid Model
proposed in the present study for various target transfer
values are presented in Table 4. It is observed that in all
20 years the M & I demands along with the irrigation
targets are fulfilled. In comparison to the planned
capacity of the reservoir and canal, based on simulation
studies carried out by NWDA for a given set of
hydrologic record, present model generated better
values. With an increase in the target transfer values,
the canal capacities are increased in proportion to the
highest volume of transfer chosen in a specific month.
The reliability of the transfer target met is around 80%
for all the cases. Also there is an increase in the net
benefits from the proposed system.
50
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

CONCLUSIONS evolutionary algorithm is good for solving complex

The present study demonstrates a successful
application of hybrid GA approach for solving a problem
of optimal planning for a single reservoir. The GA model
applied to the problem terminated to suboptimal solutions
in all the solutions employing various combinations of
the GA parameters viz. the population size, number of
generations, the crossover and mutation probabilities etc.
The direct application of the quadratic programming to
the same formulation though generated similar results
as the hybrid model; the starting point to be chosen was
a tricky aspect. Moreover, the number of iterations
required was more in such cases. A major conclusion
was thus drawn that though, GA model as an
problems, its application to a large dimensional
constrained problem has some limitations. The worth
of quadratic programming method to purely quadratic
formulations such as presented in the study, has been
established since early stages of research in optimization.
But with the use of a hybrid approach, the computational
efficiency of the QP has increased considerably. The
results obtained in the present study indicate that there
is a scope of optimization of the proposed capacities of
the reservoirs in the Inter basin Transfer Project to
achieve better water management as well as better
environmental conditions. In the present study the
optimization has been achieved through trials of different
operating policies.

Table 1 Cost and benefit Parameter values of Patanpur Table 2 Water demands and transfer targets for Patanpur
reservoir reservoirs (MCM)
Parameters Patanpur Month Water Supply Irrigation
June 0.25 8.68
Cost of Dam 7.594
July 0.25 19.84
Cost of canal 27.4
August 0.25 13.64
Benefits from Irrigation 5.06
September 0.25 8.68
Benefits from Transfer 1.478
October 0.25 6.2
November 0.25 8.68
The cost and benefits of all the variables are in December 0.25 13.64
Million Rs. / MCM. MCM indicates Million Cubic January 0.25 18.6
February 0.25 17.36
Meters and Rs. Specify the Indian Rupee. The cost of
March 0.25 3.72
canal is in Million Rs. / cumec. Cumec indicates Cubic April 0.25 2.48
Meter per Second. May 0.25 2.48
Annual 3 124

Table 3 Transfer targets for Patanpur reservoirs (MCM)

Month NWDA D1 D2 D3
June 6 6 6 6
July 67 80 100 100
August 131 400 450 475
September 24 250 230 205
October 57 5 5 5
November 4 5 5 5
December 4 5 5 5
January 4 1 1 1
February 4 1 1 1
March 4 1 1 1
April 4 1 1 1
May 4 1 1 1
Total 313 756 806 806

51
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Table 4 Summary of results of optimization for Patanpur reservoir

Reservoir Storage Patanpur % Reduction
Capacity (MCM)
Adopted in planning 110
NWDA Policy 110 0
D1 101.5 7.73
D2 97 11.81
D3 97 11.81
Canal capacity (cumec)
Adopted in planning 199.32 % Reduction
NWDA Policy 59.5475 70.12467
D1 171.9895 13.71187
D2 192.8895 3.226219
D3 203.3395 -2.01661
Total Transfer From Patanpur for 20 Years (MCM)
Maximum Value Actual Reliability of
Transfers Target Met
NWDA Policy 6260 5253.951 83.9
D1 15120 12019.6 79.41
D2 16120 12625.99 83.5
D3 16120 12675.44 83.84
Net Benefit from Proposed System (Mill. Rs.)
NWDA Policy 17748.35 % Increase
D1 24738.09 39.38
D2 25076.21 41.28
D3 24862.63 40.08

REFERENCES 6. Ebrahimi, L., Barani, G. A., and Ebrahimi, H. 2007.

1. Ahamed, J.A., and Sarma, 2005. Genetic algorithms for
optimal operating policy of a multi-purpose reservoir.
J. Water Resor. Manage., 19(2), 145-161.
2. Chauhan, S., and Shrivastava, R.K., 2008. Optimal
operation of multipurpose reservoir using genetic
algorithms. J. of Indian Water Resources Society., 28(4),
9-17.
3. Chun, C. T., Wang, W. C., Mei, W.D., and Chau, K. W.
2008. Optimizing hydropower reservoir operation
using genetic algorithm and chaos. J. Water Resources
Manage., 22 (7), 895-909.
4. Dariane, A., and Momtahen, S. 2009. Optimization of
multireservoir systems operation using modified direct
search genetic algorithm. J. Water Resour. Plan.
Manage. ASCE 135(3), 141-148.
5. East, V., and Hall, M.J. 1994. Water resources system
optimization using genetic algorithms.” Proc., 1st Int.
Conf. on Hydroinformatics, Baikema, Rotteram,
Netherlands, 220-231.
Optimization of reservoir operation using combined
genetic algorithm and wavelet transform methods.
American J. of Agricultural and Biological sciences, 2
(3), 153-158.
7. Espinoza, F.P., Minsker, B. and Goldberg, D. 2005.
Adaptive hybrid genetic algorithm for groundwater
remediation design. J. Water Resour. Plan. Manage.
ASCE 131(1), 14-24.
8. Fahmy, H.S. King, J.P, Wentzel, M.W. and Seton J.A. 1994.
Economic optimization of river management using
genetic algorithms. Paper no. 943034, ASCE Int.
Summer meeting, Am. Soc. Of Agricultural Engrs., St.
Joseph, Mich.
9. Hashemi, M. S., Barani, G.A., and Ebrahimi, H. 2008.
Optimization of reservoir operation by genetic
algorithm considering inflow probabilities (Case
study: The Jiroft Dam Reservoir). J. of Applied Sciences
8(11), 2173-2177.
10. Hilton, A.B. and Culver, T.B. 2000. Constraint handling
for genetic algorithm in optimal remediation design. J.
Water Resour. Plan. Manage. ASCE 126(3), 128-137.

52

11. Holland, J.H. 1975. Adaptation in natural and artificial
systems. MIT Press, Cambridge, Mass.
12. Jothipraksah and Ganesh Shanti 2006. Single
reservoir operation using genetic algorithm.” J. Water
Resources Manage. 20 (6), 917-929.
13. Jothipraksah and Ganesh Shanti 2009. Comparison of
policies derived from stochastic dynamic programming
and genetic algorithm models. J. Water Resources
Manage. 23 (8), 1563-1580.
14. Kangrang, A. and Chaleeraktrakoon, C. 2007. Genetic
algorithm connected simulation with smoothing
function for searching rule curves. American J. of
Applied Sciences. 4(2), 73-79
15. Kapelan, Z. 2002. Calibration of water distribution
system hydraulic models. Ph D Thesis
16. School of Engineering and science, University of Exeter,
Exeter, UK
17. Labadie, J. W., 2004. Optimal operation of
multireservoir systems: state-of –the-art- Review. J.
Water Resour. Plan. Manage. ASCE 130(2), 93-111.
18. Li, X., and Wei, X. 2008. An improved genetic algorithm-
simulated annealing hybrid algorithm for the
optimization of multiple reservoirs. J. Water Resources
Manage. 22 (8), 1031-1049.
19. Momtahen, Sh., and Dariane, A.B. 2007. Direct search
approaches using genetic algorithms for optimization
of water reservoir operating policies. . J. Water. Resour.
Plan. Manage. ASCE 133 (3), 202-209.
20. Nagesh Kumar, D., K. Srinivasa Raju, and Ashok, B.
2006. Optimal reservoir operation for irrigation of
multiple crop using genetic algorithms. J. Water. Resour.
Plan. Manage. ASCE 132 (2), 123-129.
21. Marco, F. 1996. Use of a genetic algorithm combined
with a local search method for the automatic calibration
of conceptual rainfall-runoff model. Hydrological
Science Journal, 41 (1), 21-39.
22. Marco, F. Galeati, G. and Berra, S. 1998. Global
optimization techniques for the calibration of
conceptual rainfall-runoff models. Hydrological
Science Journal, 43 (3), 443-458.
23. Oliveira, R., and Loucks, D.P. 1997. Operating rules for
multireservoir systems. Water Resour. Res., 33 (4), 839-
852.
J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
24. Preliminary study of Parbati-Kalisindh Link 2004.
National Water Development Agency, New Delhi, F.R.
P/9/04 vol., I & II.
25. Reis, F.R., Walters, G.A., Savic, D. and Chaudhry,
F.H.2005 .Multi-reservoir operation planning using
hybrid genetic algorithm and linear programming (GA-
LP): an alternative stochastic approach. J. Water
Resources Manage. 19 (6), 831-848.
26. Reis, L. F. R., Bessler, F. T., Walters, G.A. and Savic, D.
2006. Water supply reservoir operation by combined
genetic algorithm-linear programming (GA-LP)
approach. J. Water Resources Manage. 20 (2), 227-
255.
27. Simonovic, S. P. 1992. Reservoir system analysis:
Closing gap between theory and practice. J. Water.
Resour. Plan. Manage. ASCE 118(3), 262-280.
28. Sharif, M., and Wardlaw, R. 2000. Evaluation of genetic
algorithms for optimal reservoir system operation. J.
Water Resour. Plan.Manage. ASCE 15(2), 25-33.
29. Van, Z., Savic, D.A., and Walters, G. 2004. Operational
optimization of water distribution systems using hybrid
genetic algorithm. J. Water Resour. Plan.Manage. ASCE
130(2), 160-170.
30. Wardlaw, R., and Sharif, M. 1999. Multireservoir system
optimization using genetic algorithm: case study. J.
Water Resour. Plan.Manage. ASCE 14(4), 255-263.
31. Wurbs, R. A. 1993. Reservoir-system simulation and
optimization models. J. Water
32. Resour. Plan. Manage. ASCE 119(4), 455-472.
33. Yeh, W. W.G. 1985. Reservoir management and operation
models: A state-of-the-art review. J. Water Resour. Res.,
21(12), 1797-1818.
34. Yen, M. T., Frank, T.C., Tsai, A.M., and Yeh W. G. 2005.
Optimization of water distribution and water quality
by hybrid genetic algorithm. J. Water Resour. Plan.
Manage. ASCE 131(6), 431-440.
35. Yuan Xiaohui., Yongchuan, Z. and Yanbin, Y.2008.
Improved self-adaptive chaotic genetic algorithm for
hydrogenation scheduling. J. J. Water Resour. Plan.
Manage. ASCE 134(4), 319-325.
53
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

REVIEWERS OF IWRS JOURNAL FOR THE YEAR - 2008 - 2009

A timely and critical review of the manuscripts that are submitted is very important for publication of the journal. Such
a review also provides useful and necessary feedback to the author. The Editor and Joint Editor of IWRS Journal are thankful
to the following experts who have reviewed the various manuscripts that were submitted for possible publication in the IWRS
Journal during the year 2008-2009.

Dr. A.A.Kazmi, Deptt. of Civil Engg., IIT Roorkee
Dr. A.K.Rastogi, DEC, IIT Bombay
Dr. A.K.Lohani, Sci. NIH Roorkee
Dr. A.K.Vashisht, GBPUT, Pantnagar
Dr. Anupama Sharma, Sci. NIH Roorkee
Ms. Archana Sarkar, Sci. NIH Roorkee
Dr. Ashok Mishra, Deptt. of Agril. & Food Engg., IIT, Kharagpur
Dr. Ashish Pandey, Deptt. of WRD&M, IIT Roorkee
Dr. Arnab Bandyopadhyay.NIH, Guwahati
Prof. B.C.Mal HOD, Deptt. of Agril. & Food Engg., IIT, Karagpur
Dr. D.C.Singal, Deptt. of Hydrology, IIT, Roorkee.
Prof. Deepak Khare, Deptt. of WRD&M, IIT, Roorkee
Mr. Deepak Jhajaria, Deptt. of Hydrology, IIT Roorkee
Mr. Dilip Durbudhe, Deptt. of Hydrology, IIT Roorkee
Prof. G.S.Rajput, College of Agri. Engg., JNKVV, Jabalpur
Prof. G.C.Mishra, Deptt. of WRD&M, IIT, Roorkee
Dr. Himanshu Joshi, Deptt. of Hydrology, IIT Roorkee
Dr. J.V.Tyagi, Sci. NIH Roorkee
Prof. K.L. Mishra, College of Agril. Engg., JNKVV, Jabalpur
Prof. K.C.Patra, HOD, Deptt. of Civil Engg., NIT Rourkela
Dr. K.D.Sharma, NRAA, New Delhi
Prof. K.N.Tiwari, Deptt. of Agril. & Food Engg., IIT, Kharagpur
Dr. K.P.Tripathi, CSWCRTI
Dr. M.J.Kaladhorkar, CSWCRTI, Karnal
Dr. M K Jain, Deptt. of Hydrology, Roorkee.
Dr. M.P.Tripathi, IGAU, Raipur
Prof. M.K.Hardha, College of Agri. Engg., JNKVV, Jabalpur
Dr. M.K.Goel, Sci. NIH Roorkee
Dr. M.L.Kansal, Deptt. of WRD&M, IIT, Roorkee
Prof. M.Perumal, Deptt. of Hydrology, IIT, Roorkee
Prof. M.P.Sharma, AHEC, IIT Roorkee
Dr. N.C.Ghosh, Sci. NIH Roorkee
Dr. N..K.Gontia, Junagadh Agri. University, Junagarh
Dr. N. Panighahi, NIH, Roorkee
Dr. Nayan Sharma, Head, Deptt. of WRD&M, IIT Roorkee
Dr. Omkar Singh, NIH, Roorkee
Dr. P.K.Gupta, SAC, Ahamadabad
Dr. Pramod Kumar, Deptt. of Civil Engg., IIT Roorkee
Dr. P.K.Bhunya, Sci. NIH Roorkee
Dr. P.K.Garg, Civil Engg. Deptt., IIT Roorkee
Dr. P.P.Dabral, Deptt. of Agril. Engg., NERIST, Nirjuli
Prof. R.K.Nema, College of Agri. Engg., JNKVV, Jabalpur
Dr. R.D.Garg, Deptt. of Civil Engg., IIT Roorkee
Dr. Ravi Galkati Deptt. of Agril. Engg., NERIST, Nirjuli
Dr. Rakesh Kumar, NIH, Roorkee
Dr. R.P.Pandey, Sci. El, NIH Roorkee
Er. R.K.Jain, CWC, New Delhi
Dr. Surjeet Singh, NIH, Roorkee
Dr. Sanjay K.Jain, NIH, Roorkee
Dr. S.K.Mishra, Deptt. of WRD&M, IIT Roorkee
Dr. S.K. Tripathi, Professor , Deptt. of WRD&M, IIT, Roorkee.
Dr. S.R.Bhakar, Deptt. of SAWE, CTAE, Udaipur
Dr. S. Moulik, Deptt. of Agril. & Food Engg., IIT, Kharagpur
Dr. S.K.Jain , Deptt. of WRD&M, IIT, Roorkee
Dr. U. C. Kothayari, Deptt. of Civil Engineering, IIT Roorkee.
Prof. V.K.Pandey, Agril. Engg., IGAU, Raipur
Dr. Virendra K. Chaubey, Sci. NIH Roorkee
Dr. Vijay Kumar, Sci, NIH Roorkee

54
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Journal of
Indian Water Resources Society

Volume 30 Number 1 January, 2010

CONTENTS

Page

1. Protection of Bridge Abutments from Scour 1

V. K. Sarda

2. Rainfall Pattern in Northern Kerala 10

P. A. Lisha, P. K. Pradeep Kumar and K. V. Jayakumar

3. Sustainable Management of the Water Stressed 21

Aquifers in Sabarmati Basin Gujarat, India
R. C. Jain And A. K. Sinha

4. Runoff Estimation by Distributed Curve Number Technique 31

Using Remote Sensing And GIS
Susanta Kumar Jena, Kamlesh Narayan Tiwari and Ashish Pandey

5. Sediment Runoff Modeling using Artificial Neural Networks 39

Archana Sarkar, M. Mohan Raju and Anil Kumar

6. Single Reservoir Optimization using Hybrid Genetic Algorithm 46

Nitin M Mohite and Sandeep Narulkar

55
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

ISSN O970-6984
Journal of
Indian Water Resources Society
Volume 30 Number 1 January, 2010

President

Er. A. K Bajaj
EDITORIAL
Chairman, CWC, New Delhi Dear Friends,

These days, one frequently hears about climate

change in a wide range of forums: conferences,
Editor seminars, meetings, newspapers, television, etc.
Any weather event which may have slight deviation
Prof. Sharad Kumar Jain from the general trend is attributed to climate
Deptt. of WRD & M change. Thus, climate change is blamed for an intense rainfall event and
IIT, Roorkee - 247 667 so is drought in an area. Hot weather is certainly the result of global
warming but big snow storms are also due to changes in climate. Certainly
this way of thinking is not correct and we need to address the issue with
Joint Editor a rational approach that is based upon proven theory and is in conformity
with the observed data.
Dr. Ashish Pandey
Assistant Professor Climate change is also the subject of a large number of conferences
Deptt. of WRD & M wherein this topic is being discussed from different perspectives. Although
IIT, Roorkee - 247 667 much noise is being generated in some of these events, hopefully there
Er. Avinash Agarwal will be many useful ideas and outcomes. These conferences also provide
an opportunity for interactions between people from academics, research
institutes, government organization, and political leaders. Such
interactions should eventually lead to water management that is based
IWRS as a body accepts no resoponsibility for on sound scientific principles and is sustainable.
the statements made by the individuals/authors.
The National Action Plan on climate change launched by the Government
of India in the year 2008 has proposed comprehensive steps to tackle
Further. views expressed by authors need not this problem through eight national missions. National Water Mission is
necessary be the views of the organisation to one of these. Besides, water sector has important role to play in other
missions: Enhanced Energy Efficiency, Sustainable Habitat, Sustaining
which they belong.
the Himalayan Ecosystem, Sustainable Agriculture, and Green India. A
good aspect of these missions is that many programs and activities in
the water sector in India which should have been initiated two-three decades
Reprints of any portion of this publication may be earlier are proposed under these. In that sense, whether the climate change
made, provided that reference thereto is quoted. takes place or not, proposed actions will be good for the nation. The
action plan was launched with much fanfare but typical to the Indian way
of working, the actual progress so far is not very encouraging. One can
This Journal is for private circulation only. only hope that there will be actions on the ground which will change (for
good) the way water is managed in this country.

Principal Office at : The topic of climate change was also under focus due to some alarming
statements in the Fourth Assessment Report of IPCC which had predicted
that the Himalayan glaciers will disappear by the year 2035. This error
Deptt. of Water Resources
was an aberration in the report which continues to be a valuable source of
Development & Management information for scientists and policy makers.
Indian Institute of Technology Roorkee
ROORKEE - 247 667
Uttarakhand, India (SKJain)

56
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010
Registered with the Registrar News Paper of India, under Registration of Books Act. 1867 No. 61811/95

INDIAN WATER RESOURCES SOCIETY

PAST PRESIDENTS
Sri R. Ghosh, Former Chairman, CWC New Delhi 1980-82,1990-92
Sri Pritam Singh, Former Chairman, CWC New Delhi 1982-84
Sri J. F. Mistry, Former Secretary, Gujarat Irrigation Departmem, 1984-86
Dr. Mahesh Verma, Former Professors, Head, WRDTC, UOR, Roorkee, 1986-88
Dr. M. A. Chitale, Former Secretary, MOWR, GDI, New Delhi, 1988-90
Sri A. B. Joshi, Former Chairman, CWC New Delhi 1993-95
Dr. M. S. Reddy, Former Secretary, MOWR, GOI, New Delhi, 1995-97, 1997-99
SriZ. Hasan, Former Secretary, MOWR, GOI, New Delhi, 1999-2001
Sri A. K. Goswami, Former Secretary, MOWR, GOI, New Delhi, 2001-05
Er. M. Gopalakrishnan,Secretary General, ICID (2004 - till date) New Delhi, 2005-2009
EXECUTIVE COMMITTEE
Members Co-opted Members
President - Er. A.K.Bajaj Dr. Z. Ahmad
Executive Vice-President (HQ) - Dr. Ram Pal Singh Prof. Brijesh Chandra
Er. A.S.Dhingra Prof. Gopal Chauhan
Vice President (Executive Office) - Er. R.C.Jha Er. M. L.Garg
Vice President - Er. S.K.Kumar Er. S.S.Doifode
Er. V. K.Kanjilia
Secretary - Dr. R.P.Pandey Dr. Arun Goel
Sri R. N.Kalra
Joint Secretary - Mr O.P.Khanda Sri. Som Dutt Gupta Er. R. K.Khanna
Treasurer - Dr. R.D.Garg Er. Madhu Chandra Jain Er. M.S. Menon
Joint Treasurer - Mr. Amrinder Kumar Er. N.K.Jain Prof. S. K.Mazumder
Editor - Dr. S. K. Jain Dr. R. N. Khare Er. Jagdish Mohan
Joint Editor - Dr. Ashish Pandey Prof. D.T.Shete Dr. R. D.Singh
Er. Ravindra K.Sinha Prof. C.P.Sinha
Ms. Jaya Sood

ACTIVE LOCAL CENTRES

Centre Chairman Convenor
Ahmednagar Sri M.S.Mundhe Sri A. S. Garudkar
Bangalore Sri D. Satya Murty Sri Rajan Nair
Bhopal Sri M.M.Mahodaya Sri B. O. Joshi
Chennai Sri G. Ganapathi Subramanian Sri R. Subramanian
Delhi Er. R. C. Jha Sri R. K. Khanna
Gandhi Nagar Sri J. B.Patel Sri D. H. Patel
Guwahati Er. P. Naog
Hissar Dr. D. K. Kataria Dr. Pratap Singh
Indore Er.A. K. Sojatia Dr. R. K. Srivastava
Jaipur Sri S. K. Jain
Kolkata Sri P K. Bose Dr. Kalyan Bhar
Lucknow Sri Jagdish Mohan Sri J. M. Gupta
Meerut Er. S. K. Kumar Mr. Vijendra Kumar Tyagi
Mysore Sri M. N. Narse Gowda Sri N. Srinivas Murthy
Nagpur Er. S. S Doifode Sri S. G. Deshpande
Patna Prof. C. P. Sinha Ms. Arti Sinha
Pune Sri S. G. Shirke Sri R.K. Suryawanshi
Roorkee Er. P. K. Bhargav
Shimla Er. C. P. Mahajan Er. C. M. Walia

Printed &. Published by Dr. R. P. Pandey, Secretary, IWRS on behalf of the Indian Water Resources Society.
Department of Water Resources Development & Management, Indian Institute of Technology, Roorkee-247 667, (UK)
India Telephone : Office PBX (01332-276220, 286690)
Printed at : Jain Printing Press, Main Bazar, Roorkee, Ph. : 262722, 269999
Secretary : E-mail : secretaryiwrs@gmail.com
Editor : E-mail : editor.iwrs@gmail.com
Notional Price for Individual Members Rs. 10/-
57
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

ISSN O970-6984

Journal of
Indian Water
Resources Society

Volume 30 Number 1 January, 2010

58
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

Journal Article :
Rao, S. S., 1979. Flood Control Regulation of Reservoirs. Journal of Indian Water Resources Society, 1 1(2), 19-26.

Chapter in Book or Paper in a Proceedings :

Singhal, S. P. and Kasliwal, R., 1991. Calibration of Conceptual Catchment Models , in Applied Water Resources Planning,
P. V. Narasimha (Editor). Tata McGraw Hill Publishing Co. New Delhi.

References to unpublished work should not be listed. Papers approved for publication can be listed. In doubt, see recent
of IWRS journal to locate appropriate Styles.

Tables :
Tables should be carefully prepared and should be used to replace text, not duplicate it. They should be numbered
consecutively and should have a brief descrptive title. Use a separate page for each table, and include all tables after the
References section.

Figures :
Use a separate page for each figure. Each figure should have a title and be numbered. The figure will be placed in the
paper as soon as possible after it is first mentioned in the text. The authors shoudl take care to provide good quality figures that
can be sent directly for printing or that can be reduced to fit into journal’s one-column ot two-column format. The original figures
should be sent with the final manuscript. Figures should not be larger than twice paper size. Please send only black & white
figures. No colour figures would be accepted.

Appendices :
All appendices must have a title (for example Appendix 1 : Description of the Model). The Appendices are to be placed
after the References section.

Mathematical Symbols :
Avoid using the mathematical symbols that are difficult to read. If the length of an equation is likely to exceed one-
column width (8 cm), break the equation appropriately. Try not to make the paper too mathematical.
Please visit www.iwrs.org.in for membership application form and previous issues of the IWRS Journal

JOURNAL RECEIPT FORM

JOURNAL RECEIPT FORM
Dear Fellows/Members,
It is to bring to your notice that many copies of Journals are coming back to us undelivered because the address of
the recipient has changed. This is causing heavy finalcial loss on account of postage charges and yet the journal does not
reach to the members. Therefore, you are requested to fill up the following response form and send the same to Secretary,
IWRS, Water Resources Development and Management, Indian Institute of Technology, Roorkee - 247 667 or E-mail
at secretaryiwrs@gmail.com
1. Name (Capital) ........................................................................................................................................................................................
2. Life Membership/Fellowship No. .......................................................................................................................................................
3. Would you like to continue to receive the Journal Yes / No
4. Postel Address ............................................................................................................................................................................
............................................................................................................................................................................
............................................................................................................................................................................
............................................................................................................................................................................

Signature of Member/Fellow

59
 J. Indian Water Resour. Soc. Vol. 30 No. 1, January, 2010

JOURNAL OF INDIAN WATER RESOURCES SOCIETY

GUIDELINES FOR AUTHORS
GENERAL INFORMATION

The Indian Water Resources Society (IWRS) is a multi-disciplinary organization dedicated to the advancement of the
science and technology of water resources development and management. The IWRS Journal publishes original papers on
scientific, engineering and socio-economic aspect that are clear, concise and presented in a style readily understood by an
international audience. Before publication, the papers are reviewed by the Editor/ reviewers. Detailed instructions for preparing
manuscripts are included in the following. A review of previous issues will be helpful in organizing the manuscript.

The text of the paper should be well organized and presented in a logical manner. The appropriate use of subheadings,
especially in long sections, enhances the readability and quality of paper. Repetition of data given in tables and figures should be
avoided. The paper should be written so that it will be of interest to persons in the wide variety of disciplines represented by
IWRS’s membership. Complex sentences and the excessive use of highly technical jargon are prime offenders in detracting the
readers. The articles should have well-defined objectives, discussions, applications and conclusions that are easily understood.
The papers not complying w these guidelines will be returned to the authors for improvement or publication elsewhere.

The acceptance or rejection of a paper is based on appraisal by the reviewers who examine the paper for its originality,
relevance, adequacy and conciseness of the presentation. Depending on the results of reviewer, a manuscript may be returned to
the authors for revision. If the authors suitably revise the manuscript, it is accepted for publication. If the review comments are
extensive, the authors should include a reply to the reviewer, in addition to making changes to the manuscript.

Submit two copies of the manuscript typed double spaced on A4 size paper, with 2.5 cm margin on all sides along with
clear and legible figures and tables to the Editor, Indian Water Resources Society, Water Resources Development & Management
Department, Indian Institute of Technology, Roorkee-247 667. Uttarakhand. (Add line numbers in the text of your manuscript).
The references and abstract must also be double-spaced. Include a CD-ROM with text, tables, and figures. Single-spaced
manuscripts, those clumsily typed, or those with inadequate marginal space or unreadable illustrations and tables will be re-
turned. Retain the original manuscript and illustrations and send the same to IWRS when your manuscript is approved for
publication.

PREPARATION OF MANUSCRIPT

Title page :
The title should be short, informative and should clearly reflect the content of the paper. The title should not exceed 15
words in length. Terms like Preliminary Investigations, Contributions to, Studies on etc., should be avoided.

The title page should also include the name(s) of the author(s); The affiliation and mailing address of each author; The
e-mail address, telephone and fax numbers of the corresponding author.

Abstract :

The abstract should be concise, and complete in itself, without reference to the text of the paper. It should state the
general problem and objective(s), summarize the results, and state general implications. The abstract should be in single para-
graph of not more than 200 words. Five to Six keywords supplied by the author should appear on a line following the abstract.

References :

Cite references to published literature in the text sorted by author(s) and year, for example, Prasad (1994). Avoid using a
numbering system. List all references in alphabetical order in the Reference section. Give the complete title of the reference and the
source. Please follow the sample styles shown below

Books :
Gupta, S.K., 1999. Engineering Hydrology. Tata Mc Graw-Hill Publishers, New Delhi.

Continued on Inside Back Cover

60