A Project Report

on

FLOOD CONTROL AND MANAGEMENT

In partial fulfillment of the requirements of
Bachelor of Technology
in

Civil Engineering
Submitted By

MD MOZMMIL ANSARI (2001292008)

Batch: 2020-24
Under the guidance of

MR. SUDEEP KUMAR PATRO

Page | 0
 Department Of Civil Engineering
Gandhi Engineering College, Bhubaneswar

This is to certify that the work in this thesis report entitled “FLOOD
CONTROL AND MANAGEMENT” submitted by MD MOZMMIL
ANSARI bearing registration no.: 2001292008 in partial fulfillment of
the requirements for the degree of Bachelor of Technology in Civil
Engineering Session 2023-24 in the department of Civil Engineering,
Gandhi Engineering College Bhubaneswar is an authentic work carried
out by him under my supervision and guidance. To the best of my
knowledge the matter embodied in the thesis has not been submitted to
any other University/Institute for the award of any degree.

Guided by: - Prof. SUDEEP KUMAR PATRO

Department of Civil Engineering

Countersigned by: Gandhi Engineering College
Bhubaneswar

HOD
External Examiner (Department Of Civil Engineering)

Page | 1
 DECLARATION

I hereby certify that the work contained in the thesis is original and has been done
by myself under the supervision of my supervisors. The work has not been
submitted to any other Institute for any degree or diploma. I have conformed to the
norms and guidelines given in the Ethical Code of Conduct of the Institute.
Whenever I have used materials (data, theoretical analysis, and text) from other
sources, I have given due credit to them by citing them in the text of the thesis and
giving their details in the references. Whenever I have quoted written materials
from other sources and due credit is given to the sources by citing them. From the
plagiarism test, it is found that the similarity index of the whole thesis within 10%
excluding references/bibliography and own first- author publications, as per the
university guidelines.

Date:20/04/2024

Place:
Bhubaneswar
MD MOZMMIL ANSARI (2001292008)

Page | 2
 ACKNOWLEDGEMENT

It is with a feeling of great pleasure that I would like to express my most sincere heartfelt
gratitude to MR. SUDEEP KUMAR PATRO, Assistant professor, Dept. of Civil
Engineering. GEC, BHUBANESWAR for suggesting the topic for our project report and
for his ready and able guidance throughout the course of our preparing the report. We
thank you Mam, for your help, inspiration andblessings.

We express our sincere thanks to Prof. TRUPTI MALA PATTNAIK, Professor and HOD,
Dept. of Civil Engineering GEC, Bhubaneswar for providing us the necessary facilities in
the department.

We would also express our sincere thanks to laboratory Members of Department of Civil
Engineering, GEC, Bhubaneswar.

GROUP MEMBERS

MD MOZMMIL ANSARI

IMRAN ANSARI

GOURAV BISWAL

SUSHIL KUMAR

Page | 3
 ABSTRACT

Extreme weather events such as droughts, heat waves, dust storms, forest fires,
floods, and landslides, which already disrupt the lives of millions each year, are
expected to increase in frequency and intensity with climate change. The impact
of these sudden events, in addition to the gradual change in climate effects over
time, will put added stress on vital water, sanitation, flood management,
transportation, and energy infrastructure. Responding to the impacts of climate
change presents a major challenge for developing countries lacking adequate
resources, and it is therefore an important focus of the United States Agency for
International Development’s development assistance portfolio. To help address
this challenge, and consistent with Executive Order 13677 – Climate-Resilient
International Development has developed the Global Climate Change,
Adaptation, and Infrastructure Knowledge Management Support Project (a Task
Order under the Architecture and Engineering Indefinite Quantity Contract or
IQC) to articulate best practices in incorporating climate adaption in the
planning and engineering design of infrastructure activities. Under this project,
a suite of knowledge management products has been created, led by the
Overarching Guide: A Methodology for Incorporating Climate Change
Adaptation Infrastructure Planning and Design. The objective of the
Overarching Guide is to support the consideration of climate change risks and
adaptation in infrastructure development activities.

ACRONYMS:
CAPEX  Capital Expenditure
CBA Cost-Benefit Analysis
MCA  Multi-Criteria Analysis
OPEX  Operational Expenditure
USAID  United States Agency for International Development

Page | 4
 CONTENTS
Certificate 1
Declaration 2
Acknowledgement 3
Abstract 4
CHAPTER 1: Flood Control and Management
1.1 Introduction 6
1.2 Structural And Non-Structural Approaches 7
CHAPTER 2: Overview Of Tools And Techniques For Flood Modeling
2.1 Artificial Neural Network (ANN) and Fuzzy Logic 8
2.2 Optimization Methods 9
2.3 Geographical Information System
2.4 Expert System 10
2.5 Multicriteria Decision Making
CHAPTER 3: Literature Review On Flood Modeling And Management
3.1 General 11
3.2 Artificial Neural Network and Fuzzy Logic 12
3.3 Optimization Methods
3.4 Geographical Information System 13
CHAPTER 4: Development Of Long-Term Operating Policy For Flood
Control: A Case Study
4.1 Description of Study Area 14
4.2 Methodology 15
4.2.1 Folded Dynamic Programming 15
4.2.2 Data Collection 17
4.2.3 Computation of Downstream Catchment Contribution 17
4.2.4 Preparation of Block Period 19
4.2.3 Application of Folded Dynamic Programming to 21
Develop Rule Curves for Floods
4.3 Results and Conclusion 24
CHAPTER 5:
Measures Suggested For Evolving A Feasible And Implementable Flood 29
Control Strategy
CHAPTER 6:
Concluding Remarks 30

References 31

Page | 5
 CHAPTER 1

FLOOD CONTROL AND MANAGEMENT

1.1 Introduction
Flood hazard has long been recognized as one of the most recurring, wide-
spread and disastrous natural hazards. In many parts of the Indian subcontinent,
flooding reaches catastrophic proportions during monsoon season. For
centuries, floods in the Ganga, the Brahmaputra and the Godavari basins have
brought number of disasters to the inhabitants in the flood plains apart from
inundating large tracts of fertile land (Kale, 1998). In the last few decades, there
have been many serious attempts to improve the understanding of the causes
and effects of floods for better control and management of floods. Recognizing
the gravity and the extent of the problem, several countries, including India,
have taken up extensive flood management and flood protection works. High
priority has been given to flood forecasting and warning systems on major flood
prone rivers and development of different strategies for flood control and
management. In addition, most of the impacts due to climate change
(Nakicenovic et al., 2000) are expected to increase the frequency of extreme
events of hydrology i.e., floods and droughts in the near future. This will further
necessitate the development of proper measures for flood control and
management. The purpose of this chapter is to study the flood control and
management with special reference to modelling aspects to suggest suitable
management strategies. Initially, various structural and non-structural
approaches are presented for flood control and management. Literature review
of mathematical models and their applications to flood aspects are presented
next. A case study is presented to demonstrate the development operating policy
of a reservoir with flood control as a major criterion. To conclude various
measures are suggested for flood control.

Page | 6
1.2 Structural and Non-Structural Approaches
Structural methods to safely dispose floods include: (a) construction of
reservoirs for storage of flood water that can be utilized for other purposes once
flood recedes, (b) embankments to retain the flood water far away from the
flood prone areas, and (c) construction and improvements of channels to
adequately discharge the flood waters. Structural measures can be supplemented
with non-structural measures such as floodplain zoning as well as flood
forecasting as a cost-effective strategy. Both structural and non-structural
methods are to be taken into consideration while planning flood mitigation
measures. Salient factors that affect data collection are cost, speed with which
the data can be collected and their accuracy. Data is required about the existing
vegetation/soils, level of urbanization, land use, watershed characteristics,
rainfall intensity and its duration, runoff, degree of flood protection/severity of
flood damage, stage-discharge relationships, climatic records, geology and
geomorphology, depth of groundwater, etc. Database of as many flood events as
possible for different locations is essential to get a comprehensive overview of
flood events over the entire area. This will avoid any ambiguous inferences due
to incomplete data. Additional data can also be collected from secondary
sources such as by interviews and earlier reports. Data thus collected are to be
processed and analyzed in such a manner as to minimize errors which otherwise
affect the accuracy of general analyses as well as calibration and validation of
hydrologic simulation models. Statistical analysis is very helpful in this regard.
The processed data can be utilized for various aspects including development of
maximum probable flood, standard project flood, design flood, flood frequency
analysis, flood sediment, flood envelope curve, stage-discharge curves and
floodplain zoning. The data can also be used for analyzing past flood situations,
managing present flood situation or for forecasting future floods of different
magnitudes.
In India, many agencies are involved in water resources planning, development
and disaster management such as Central Water Commission (CWC), Central
Ground Water Board (CGWB), Indian Meteorological Department (IMD),
National Disaster Management Authority (NDMA), National Water
Development Agency (NWDA), National Institute of Hydrology (NIH), Water
and Land Management Institutes (WALMI) of various State Governments,
National Remote Sensing Centre (NRSC) and State Remote Sensing
Application Centers, Geological Survey of India (GSI), etc. Various
mathematical modeling approaches based on soft computing and related fields
are available for flood control and management (e.g., ASCE, 2000b; Chau et al.,
2005; Jain and Singh, 2006).

Page | 7
 CHAPTER 2

OVERVIEW OF TOOLS AND TECHNIQUES FOR FLOOD MODELING

Modeling for flood control and management has attracted considerable attention
from number of researchers and various tools are developed to combat floods. A
brief overview of some of the tools developed for modeling floods is presented
below.

2.1 Artificial Neural Network (ANN) and Fuzzy Logic

The Back Propagation Algorithm (Rumelhart et al., 1986) is a procedure to train
feed forward ANN models, in which the outputs can be sent only to the
immediate next layers. The selection of a suitable architecture for the problem
on hand can be done in three steps: fixing the architecture, training the network
and testing the network. Main parameters concerned are network architecture,
learning rate, type of activation function, definition of error, number of epochs
etc. The procedures for training the ANN network are described in ASCE
(2000a). Radial Basis Function network (ASCE, 2000a) is a three-layer network
in which the hidden layer performs a fixed non-linear transformation with no
adjustable parameters. This layer consists of a number of nodes and a parameter
vector called a center which can be considered as its weight vector. For each
node, the Euclidean distance between the center and the input vector of the
network input is computed and transformed by a non-linear function that
determines the output of the nodes in the hidden layer. The output layer then
combines these results in a linear fashion. The common activation functions
used in Radial Basis Function are Sigmoidal and the Gaussian Kernel functions.
More details about Radial Basis Functions are available in ASCE (2000a, b).
Applications of ANN with reference to flood modeling are discussed in ASCE
(2000b). Moreover, there are often many situations with components that are
intrinsically vague, called uncertain. Different situations that lead to uncertainty
are unquantifiable information, incomplete information, non-obtainable
information and partial ignorance (Bojadziev and Bojadziev, 1997). In such
cases, fuzzy approach is the most suitable to tackle the vagueness of data. The
advantage of fuzzy approach is that the quantitative as well as qualitative
analysis can be integrated and can be represented using fuzzy membership
functions (Deka and Chandramouli, 2005; Venkatesh et al., 2008). Adaptive
Neuro-Fuzzy Inference System (ANFIS) is also utilized for flood related studies
(e.g., Chau et al., 2005; Akbari et al., 2009).

Page | 8
2.2 Optimization Methods
Traditional optimization methods such as Linear Programming, Nonlinear
Programming, Dynamic Programming, etc. have played a major role in flood
management studies (Mays, 1996; Lund, 2002; Jain and Singh, 2006; Karamouz
et al., 2009). However, some real-world flood management problems present
situations such as a complex search space which are difficult to be solved by
such exact methods. In such cases, more efficient optimization strategies such
as evolutionary algorithms are required (Deb, 2001; Ranjithan, 2005). The
evolutionary algorithms have the following characteristics and advantages:
 Applicability to any problem that can be formulated as a function optimization
task.
 Flexibility of the procedures, as well as ability to self-adapt the search for
optimum solutions.
 Ability to use knowledge and hybridize with other methods.
 Adapt solutions to changing circumstances as compared to traditional methods
of optimization which require a complete restart in order to provide a solution.
 Find multiple optimal solutions simultaneously in multimodal function
optimization problems.

2.3 Geographical Information System

Geographical Information Systems (GIS) technology provides tools for
effective and efficient storage and manipulation of remotely sensed information
and other spatial as well as non-spatial information (e.g., Engman and Gurney,
1991; De Vantier and Feldman, 1993; Singh and Fiorentino, 1996; Vieux, 2001;
Islam and Sado, 2002; Nagesh Kumar, 2002). The strength of GIS results from
its ability to analyze data representing a particular point, line or polygon as all
features of any landscape can be reduced to one of these three spatial categories.
Remote sensing images, effectively integrated within GIS, can be used to
facilitate measurement, mapping, monitoring and modeling activities. Building
a GIS database includes import or entry of data from different sources and
digitizing the data from different source documents. GIS analyses allow the user
to perform a wide variety of investigations such as (DeMers, 1997; Heywood et
al. 1998; Clarke, 1999; Verbyala, 2002):
 Proximity analyses, neighborhood operations (for example identifying objects
within a certain neighborhood fulfilling specific criteria).
 Determine the relationships between datasets within such a neighbourhood.
 Temporal operations and analyses.
 Generation of new information by combining several data layers and attributes.
Clark (1998) suggested three different approaches for models using GIS for
flood management: (i) Data process in the pre-modeling phase; (ii) direct GIS
support for flood modeling; and (iii) integrating GIS and flood models for post-
processing of data such as for flood risk mapping.

Page | 9
2.4 Expert System
Expert System is a branch of Artificial Intelligence which uses the knowledge
and inference procedures to solve problems (Ignizio, 1991; Simonovic, 1991;
Varas and Von Chrismar, 1995; Nagesh Kumar, 2006). These are designed to
carry information in the form of knowledge base (rules) inferred from experts
and provides this knowledge to users for further decision making. Expert
Systems are getting prominence due to their ability to provide consistent
answers for repetitive decisions, processes and tasks; hold and maintain
significant levels of information; centralize the decision making process; create
efficiencies and reduce time needed to solve problems; combine multiple human
expertise, reduce the human errors and review transactions that human experts
may have overlooked (Simonovic, 1991).

2.5 Multicriteria Decision Making

Multicriteria Decision Making (MCDM) is an emerging area of decision
making where the role of all stake holders (all agencies directly or indirectly
involved in the decision making including those affected) can be considered in a
structured environment. Levy (2005) discussed about flood risk management for
which MCDM was used for selecting a suitable flood control strategy with
inputs from models such as GIS, hydrological models and real-time flood
information systems. Costa et al. (2004) evaluated flood control options for the
catchment of Livramento creek in the peninsula of Setúbal, in Portugal in
Multicriterion context using MACBETH approach. Yu et al. (2004) proposed
multi-person multiobjective decision-making model for the problems of flood
control operation for Fengman Reservoir located in Songhua River Basin in
China. Hajkowicz and Collins (2007) present an excellent review of the
applications of multicriteria analysis to water resources planning and
management.

Page | 10
 CHAPTER 3
LITERATURE REVIEW ON FLOOD MODELING AND MANAGEMENT
A brief review of the literature related to the general aspects of floods and
different tools used for flood modeling and management are presented in this
section.

3.1 General
Seth (1998) discussed about flood estimation, flood routing, flood inundation,
flood mapping, floodplain zoning and flood insurance. Garde (1998) suggested
non-structural methods of flood control including the installation of temporary
openings or permanent closure of existing openings in structures, raising the
existing structures, installation of flood warning systems with an appropriate
evacuation plan, tax incentive to encourage proper use of flood plains, adoption
of appropriate development policies for facilities in or near floodplain land and
flood insurance. Lekuthai and Vongvisessomjai (2001) proposed Anxiety-
Productivity and Income interrelationship approach (API) to quantify the
intangible damage in monetary terms. Hall et al. (2003) discussed about
Integrated Flood Risk Management. Mohapatra and Singh (2003) discussed
about flood management in India. They presented some special flood problems
in India such as river bank/bed erosion, sediment transport by rivers, dam break
flows, urban drainage, flash floods, floods due to snow melt etc. They also
discussed current status of flood management in India such as progress of
structural measures and nonstructural measures. They recommended legislation
for floodplain zoning, flood cushion in reservoirs, flood insurance, flood data
center, community participation, international cooperation etc. Carsell et al.
(2004) quantified the benefit of a flood warning system. Simonovic and Li
(2004) discussed sensitivity of the Red River Basin Flood Protection System,
Canada, to climate variability and change. Kumar and Chatterjee (2005)
developed regional flood frequency relationships for the estimation of floods of
various return periods for gauged and ungauged catchments in North
Brahmaputra region of India. Sanders et al. (2005) studied national flood
modeling for insurance purposes. The study focused on the flood risk
information needs of insurers and how these can be met. Data requirements of
national and regional flood models are also addressed in the context of accuracy
of available data on property location. Insurance implications of recent flood
events in Europe and the issues surrounding insurance of potential future events
are also analyzed. Holz et al. (2006) stressed the potential of Information and
Communication Technology (ICT) for flood management and discussed the role
of various stake holders. Slutzman and Smith (2006) presented a methodology
to estimate flood response in urban watersheds for assessing the performance of
systems of small flood control reservoirs for extreme floods.

Page | 11
3.2 Artificial Neural Network and Fuzzy Logic
Solomatine and Xue (2004) compared ANN with M5 model tree machine
learning approach and observed that accuracy of M5 trees is similar to that of
ANN. Deka and Chandramouli (2005) studied river flow prediction using a
fuzzy neural network (FNN) model and validated on the River Brahmaputra.
Wu et al. (2005) demonstrated application of ANN for watershed-runoff and
streamflow forecasts and concluded that ANN-hydrologic forecasting models
can be considered as an alternate and practical tool for streamflow forecast.
Ahmad and Simonovic (2006) developed intelligent DSS including ANN and
expert systems to assist decision makers during different phases of flood
management. Akbari et al. (2009) presented fuzzy rule-based (FRB) control
model using Takagi-Sugeno fuzzy system and a model modification algorithm.

3.3 Optimization Methods

Needham et al. (2000) presented a mixed integer linear programming model for
a reservoir system analysis of three projects on the Iowa and Des Moines
Rivers. They reviewed operating procedures for the flood of 1993 and
illustrated how much damage could have been reduced if inflows could be
predicted months in advance or if the existing operating rules were more averse
to extreme flood events. Lund (2002) proposed two-stage linear programming
formulation which provides an explicit economic basis for developing
integrated floodplain management plans. Chau et al. (2005) employed the
genetic algorithm based artificial neural network (ANN-GA) and the adaptive-
network-based fuzzy inference system (ANFIS) for flood forecasting.
Travis and Mays (2008) introduced discrete dynamic programming technique to
determine the ideal locations and geometry of retention basins within a
watershed. Karamouz et al. (2009) presented two optimization models based on
GA. The first model is to determine economical combination of permanent and
emergency flood control options and the second one is to determine the optimal
crop pattern. Wei and Hsu (2008) developed a generalized multipurpose multi-
reservoir optimization model based on feed-forward back-propagation neural
network for basin scale flood control in which one constraint is linear channel
routing.

Page | 12
3.4 Geographical Information System
Rao et al. (1998) discussed role of GIS and remote sensing in flood mapping,
flood damage assessment, identification of erosion-prone areas and flood risk
zone mapping in India. Noman et al. (2003) presented a process for delineating
flood plains accurately from a digital terrain model (DTM) and hydraulic model
in an automatic environment, providing the flexibility of incorporating
professional judgment in the process. Liu and Smedt (2005) employed Wetspa
for calculating flood discharges. Jain et al. (2005) applied remote sensing
techniques for the Koa catchment, Bihar. Jain et al. (2006) used NOAA-
AVHRR data for mapping of flood-affected area during the year 2003 for the
River Brahmputra. Khan et al. (2007) quantified the impact of flood and rain
events on spatial scale for a case study of shallow water-table levels and salinity
problems in the Murray irrigation area using GIS. Bahremand et al. (2007)
employed Wetspa model within GIS to predict flood hydrographs and spatial
distribution of hydrologic characteristics in a watershed. Machado and Ahmad
(2007) assessed the flood hazard caused by Atrato River in Quibdo, northwest
of Colombia using statistical modeling techniques, hydraulic modeling using
HEC-RAS and GIS. Three flood hazard maps for return periods of 10, 20 and
50 years are generated. It is concluded that results can be useful for evacuation
planning, estimation of damages and post-flood recovery efforts. Venkatesh et
al. (2008) presented key issues associated with uncertainty in flood inundation
mapping and proposed an integrated approach in producing probabilistic flood
inundation maps. Li and Zhang (2008) compared three GIS-based distributed
hydrological models for flood simulation and forecasting.

Page | 13
 CHAPTER 4

DEVELOPMENT OF LONG-TERM OPERATING POLICY FOR

FLOOD CONTROL: A CASE STUDY
In this section, a case study of Hirakud reservoir for the development of long-
term optimal operating policy for flood control using Folded Dynamic
Programming (FDP) is presented.

4.1 Description of Study Area

The project under the present study, Hirakud single reservoir system is situated
in Mahanadi River basin. The Mahanadi basin lies mostly in Madhya Pradesh
and Orissa States. It is bounded on the north by Central India Hills, on the south
and east by Eastern Ghats and on west by the Maikela range, the south east part
of Deccan Plateau. The basin is situated between 80° 30¢ and 86° 50¢ East
longitudes and 19° 20¢ and 23° 35¢ North latitudes. It is roughly circular in
shape with a diameter of about 400 km with an exit passage of 160 km in length
and 60 km in breadth. Area of this basin is 1,41,600 km2 and is broadly
divisible into three distinct zones, the upper plateau, the central hill part flanked
by Eastern Ghats, and the delta area. Hirakud dam across the Mahanadi River is
located in the second zone. The Mahanadi River originates in Raipur district of
Madhya Pradesh at an elevation of 442 m above mean sea level. The total
length of this east flowing river from its origin to its outfall into the Bay of
Bengal is 851 km of which 357 km is in Madhya Pradesh and remaining 494
km in Orissa state. After a run of 450 km from its starting point, the river
reaches Sambalpur district of Orissa, where the Hirakud dam is built on the
main river. Below the dam, the river gets water mainly from two sub-basins,
Ong and Tel, in addition to free catchment along the river. The river flows down
to Naraj, the head of delta and finally joins the Bay of Bengal. The catchment
area up to Naraj is 1,32,200 km2. On the downstream of Naraj, the river divides
into several branches, namely, Birupa, Chitrotpala, Devi, Kushabhadra,
Bhargabi, Daya etc. and runs 80 km before discharging into the Bay of Bengal.
The multipurpose Hirakud reservoir is utilized mainly for three purposes, flood
control, irrigation, and hydropower production. There is expectation from
Hirakud reservoir to control flood at coastal delta area by limiting the flow at
Naraj within 25,500 m3/s. There are three head regulators, which can draw
128.8 m3/s from the reservoir for irrigation purpose. Areas of 0.16 and 0.11
million hectares are irrigable from the reservoir during Kharif and Rabi seasons,
respectively. The total installed hydropower capacity of the project is 307.5

Page | 14
MW, out of which 235.5 MW can be produced from the seven units of Hirakud
hydropower station, and 72 MW from three units of Chipilima hydropower
station, located further d/s of the Hirakud dam. The water, used for power
generation at Hirakud, flows from the Hirakud hydropower station to the
Chipilima hydropower station through a power channel of 22.4 km long. After
generating power at Chipilima, water flows back into the river. Flood control is
the first preferred objective for this reservoir with hydropower generation and
irrigation as the secondary objectives. This reservoir is situated 400 km
upstream of confluence of Mahanadi River with the Bay of Bengal. There is no
other flood controlling structure downstream of Hirakud reservoir. During
monsoon season, the coastal delta part, between Naraj and the Bay of Bengal, is
severely affected by floods. This flood-prone area gets water from the Hirakud
reservoir and from rainfall in the downstream catchment. Naraj is situated at the
head of the delta area, where the flow of the Mahanadi River is measured. The
flow of the Mahanadi
River at Naraj is used as
an indicator of
occurrence of flood in
the coastal delta area by
the Hirakud authority.
As the Hirakud
reservoir is on the
upstream side of delta
area in the basin, it
plays an important role
in alleviating the
severity of the flood in
this area. This is done
by regulating release
from the reservoir. The
schematic diagram of
Hirakud project is
shown in Fig. 1.

Page | 15
4.2 Methodology

4.2.1 Folded Dynamic Programming

Before discussing the algorithm of FDP, it is necessary to explain the way of
finding maximum and minimum possible storages at the beginning of each time
period, hereafter called time step, of the whole operating horizon for
reservoir(s). The operating horizon, the duration of reservoir operation, is
considered as stage and the storage of reservoir is considered as state variable in
DP formulation. In physical terms, the storage state variable can be at any point
between the dead storage level and full reservoir level. In the FDP algorithm,
the entire storage state space at each time period is required to be divided into
four equal state increments to form five grid points. The storage can be changed
from any grid point of one time step to any grid point of adjacent time step.

Step-wise procedure of the proposed algorithm (FDP) is explained below:

I. Depending on the natural inflow, release capacity, and boundary
conditions of storage, the minimum and maximum possible storage
values for each reservoir (in case of multiple reservoir system) at every
time step of operating horizon are found out.
II. Considering the minimum and maximum possible storages as two
extreme grid points, three middle grid points are determined adopting
uniform state increments. This means that the possible storage space at
each time step is divided into four equal state increments to get five grid
points. So, there are 5M grid points in M dimensional problem (M
reservoirs in multiple reservoir system) at each time step. State increment
is different for different time steps as well as for different reservoirs. The
mesh of these grid points for the whole operating horizon of all the
reservoirs forms the corridor.
III. Conventional DP is run through this corridor to find the trajectory, P,
which gives maximum objective function value, F.
IV. For finding the trajectory for next iteration, if this trajectory (P) is either
the minimum or maximum storage value, i.e., extreme grid points at any
time step, these points are changed to the next interior grid points to form
the revised trajectory. This revised trajectory will be within the corridor
for the next iteration.
V. In the next iteration, the state increment is halved at each time step. A
fresh corridor is formed by taking two state increments or grid points on
each side of the trajectory. Then conventional DP is run through this
corridor as in Step 3 to find the best trajectory, P¢, whose objective
function value is F'.
VI. The iterations are continued with half value of state increments of the
previous one at each time step. There can be two stopping rules. First, the
decrement of state increment at a time step stops, where state increment

Page | 16
 happens to be less than a predefined value. The iteration stops, when
decrement of state increment process stops at each time step. Second, the
iteration stops, when F" < x is satisfied, where F" = (F' – F)/F and x is a
predefined ratio. In the present case, second stopping rule is applied.

Further details of FDP could be found in Baliarsingh (2000), and Nagesh

Kumar and Baliarsingh (2003).
Various phases are formulated to derive the long-term operating policy for the
reservoir. These include data collection, computation of downstream catchment
contribution, preparation of block period, formulation of basis for rule curves
for floods and the application of Folded Dynamic Programming (FDP) to
evolve rule curves for reservoir management during floods. These are described
below.

4.2.2 Data Collection

Data of sixty-eight floods such as peaks of inflow hydrograph, peak of outflow
from reservoir during each flood, and peak of flow hydrograph at Naraj, spread
between 1958 and 1995, without downstream catchment contribution, are
available in the Hirakud data book (Patri, 1993). The outflow from the reservoir
is a combination of release for power and spill from reservoir. The distance
between the Hirakud reservoir and Naraj is 320 km. Hence, apart from the
measured runoff of Ong and Tel basins (Fig. 1), there is a considerable amount
of lateral flow coming from the free catchment between Hirakud and Naraj
joining the main river, which was not measured and therefore is arithmetically
assessed. The representative inflow into the reservoir and downstream
catchment contribution series are used subsequently for evolving the rule curve.

4.2.3 Computation of Downstream Catchment Contribution

The downstream catchment contribution during the sixty-eight historical floods
is calculated from the known values of outflow from the reservoir and flow at
Naraj. The hydrograph at upstream section of the Hirakud-Naraj River reach is
the outflow of the Hirakud reservoir, which is the combination of release for
power generation and spill from reservoir. This outflow collects the downstream
catchment contribution with it while traveling down to Naraj. As the
downstream catchment contribution was not measured, this data is not available
for the sixty-eight floods during 1958 and 1995. However, this quantity is
calculated theoretically, by subtracting the routed quantity of outflow of
Hirakud reservoir from the flow at Naraj. The relationship between the peak
outflow from Hirakud reservoir and its routed quantity at Naraj is obtained from
the data of nine floods between 1992 and 1995. In this regard, the reach of the
river from Hirakud to Naraj is considered for application of Muskingum routing
method. Muskingum equation for the flood routing is
QHt = C0 RPSt-1 + C1 RPSt + C2 QHt-1 (1)
Page | 17
where RPSt = ordinate of inflow (outflow from reservoir) hydrograph at the
upstream section of Hirakud- Naraj routing reach at beginning of routing
interval t; QHt = ordinate of outflow (routed quantity of outflow from reservoir
at Naraj) hydrograph at the downstream section of routing reach at beginning of
routing interval t; C0, C1, C2 = coefficients of Muskingum equation; and t =
routing interval.
The values of the coefficients of Muskingum equation are arrived at as C0 =
0.471, C1 = 0.117, and C2 = 0.412. Time interval of the ordinates of the
hydrographs for routing procedure is chosen as 24 hours (Subramanya, 1994).
For all the nine floods considered, the initial flow at Naraj is taken to be same as
the initial value of outflow from the reservoir and routing procedure is carried
out to get the relationship between peak outflow from reservoir and its routed
quantity at Naraj.
The percentage of attenuation varies from one flood to another. The floods are
characterized by peak outflow from reservoir. It is observed from this analysis
that there is no distinct correlation between the percentage of attenuation and
peak outflow from the reservoir. It is also observed that attenuation varies from
26 to 8% among all the floods. The higher the percentage of attenuation
considered for routing, the lower would be the routed outflow quantity of
Hirakud reservoir. Hence, the calculated downstream catchment contribution
will be more. The rule curve, obtained for this higher downstream catchment
contribution will also serve for the lower value of downstream catchment
contribution. Therefore, 26% of attenuation is considered for the calculation of
downstream catchment contribution, which is used for finding the representative
flows. All the floods whose peak is more than 5,000 m3/s are taken into
consideration for this purpose. Applying 26% attenuation, the peak flow at
Naraj from the outflow from Hirakud reservoir alone is calculated which is
assumed to occur simultaneously with the total peak flow at Naraj. This value is
subtracted from the total peak flow at Naraj to get the peak flow of downstream
catchment contribution.

Page | 18
4.2.4 Preparation of Block Period
As the major objective of the Hirakud reservoir is flood control, the reservoir
operation is considered only for the monsoon season (floods occur only in this
season) in the present study. Hence, for control purpose, the reservoir is made
empty at beginning of monsoon and is made full by the end of monsoon to
utilize the water for conservative purposes during the following non-monsoon
period. The monsoon period for reservoir operation is taken as 1st July to 30th
September. The reservoir is operated for filling up from the dead storage level
of 179.83 m on 1st July to the full reservoir level of 192.02 m by 1 st October
every year. There are 92 days in the duration from 1st July to 30th September.
Hereafter, 1st July is termed 1st day and 30th September 92nd day. The outflow
from reservoir on the 92nd day will take more than a day to reach Naraj.
The outflow from reservoir on 92nd day will collect the downstream catchment
contribution of 93rd day to form the flow at Naraj on 93rd day. The predicted
value of downstream catchment contribution of 93rd day is to be known, to take
necessary reservoir operation decision for 92nd day. Therefore, the
representative downstream catchment contribution series is generated for 93
days and the representative inflow into the reservoir series is generated for 92
days. Nine blocks of 10 days each are considered for placing one flood in each
of the blocks. Though there is no chance of occurrence of flood in each of the
10 days block period simultaneously, it is assumed so, to get the rule curve and
to tackle the flood during any duration individually. The peaks of inflow
hydrograph to the Hirakud reservoir of the historical sixty-eight floods are
arranged in nine different block periods and are shown in Table 2.
Similarly, the peaks of calculated downstream catchment contributions are
shown in Table 3. The peaks of individual block period with respect to the peak
of the whole monsoon season in terms of percentage for both these components
are also shown in Tables 2 and 3. During transit from the Bay of Bengal on to
the land surface, the monsoon brings rain first to the d/s catchment of Hirakud
reservoir. Then it takes another day to travel to interior area, contributing to the
inflow into the reservoir. That is why, the two representative hydrographs are
made to lag by a day. The patterns of these two representative hydrographs in
each block period are kept same as that for probable maximum flood. In this
process, the hydrograph of inflow into the reservoir is made to lag by a day
from the hydrograph of downstream catchment contribution. Here, flow on any
day means the flow at the beginning of that day. After getting the pattern of the
above two series, it is necessary to obtain the flow rate of each day for various
possible peaks. When the total flow reaches 9,000 m3/s of inflow hydrograph at
Naraj, it is treated as flood by the Hirakud authority. The peak of maximum
probable flood is 80,000 m3/s.

Page | 19
Table 2. Peaks of inflow hydrograph into the reservoir (thousand m3/s) of historical sixty-
eight floods (1958-1995) during various block periods

Table 3. Peaks of calculated downstream catchment contribution (thousand m3/s) of the

historical sixty-eight

Page | 20
Therefore, an operating policy is to be developed for peaks of inflow lying
between 9,000 and 80,000 m3/s. The various hypothetical series for inflow into
the reservoir are calculated with peaks of 9, 23, 37, 51, 60, 74, and 80 thousand
m3/s. Similarly, 3, 9, 14, 20, and 25.5 thousand 3 m/s of peaks of downstream
catchment contribution series are considered. Upper bound is taken as 25,500 3
m/s, which is required to keep the flow at Naraj below this value. It can be
observed from Fig. 2 that the data of inflow into the reservoir is generated for
92 days and the highest peak (37 thousand m3/s) occurs on the 4th day.
Similarly, the data of downstream catchment contribution is generated for 93
days and the highest peak (25.5 thousand 3 m/s) occurs on the 53rd day (Fig. 3).

4.2.5 Application of Folded Dynamic Programming to Develop Rule Curves

for Floods
As the major objective of the Hirakud reservoir is flood control, operating
policy is developed only for the monsoon season, i.e., from 1st July to 30th
September. The minimum and maximum possible storages at each time step are
dependent on the inflow into the reservoir, maximum release capacity during
that time step and the maximum permissible release from reservoir for flood
control purpose at Naraj, which varies with respect to time. The initial storage
state on 1st July is taken as the dead storage and final desired storage state on
1st October is the full capacity of reservoir. Hence, forward pass alone is
sufficient to find out the maximum possible storage state, starting from initial
time step and backward pass alone is sufficient to find out the minimum
possible storage state starting from last time step. In case of Hirakud reservoir
system, release for power and spill from the reservoir join back the Mahanadi
River downstream floods (1958-1995) during various block periods of the
Hirakud dam and ultimately contribute to the flow at Naraj.

Fig. 2 Representative series of inflow into the reservoir with peak of 37 thousand m3/s.

Page | 21
Fig. 3 Representative downstream catchment contribution series for Hirakud
reservoir with peak of 25.5 thousand m3/s.

This quantity is to be regulated at reservoir during flood to restrict the flow at

Naraj below non-damaging flow (25.5 thousand m3/s). Therefore, this factor is
to be taken care of while finding minimum and maximum possible storages.
Accordingly,
Smax,t+1 = Smax,t + (It – IRt) – RPSt
Smin,t+1 = Smin,t + (It – IRt) – RPSt
SDSL £ Smax,t £ SFRL
SDSL £ Smin,t £ SFRL
RPSMIN £ RPSt £ RPSMAX

Where, Smax,t = maximum possible storage at beginning of the time period t;

Smin,t = minimum possible storage at beginning of the time period t; It =
inflow into the reservoir during time period t;
IRt = release for irrigation during time period t; RPSt = release for power and
spill, if any, during time period t;
SDSL = reservoir storage at dead storage level; SFRL = reservoir storage at full
reservoir level;
RPSMIN = minimum required release from reservoir through spillway, sluice,
and powerhouse;
and RPSMAX = maximum release capacity of reservoir (summation of
maximum release through spillway, sluice, and power house).

The difference between non-damaging flow at Naraj and downstream catchment

contribution is the maximum permissible release, which can be made from
reservoir to keep the flow at Naraj within nondamaging flow. For calculating
the permissible release from reservoir for tth day, downstream contribution of

Page | 22
(t+1)th day is considered to take into account the lag time between Hirakud
reservoir and Naraj. It is observed in the calculations of percentage of
attenuation from the historical nine floods that the percentage of attenuation
varies from a minimum of 8% to a maximum of 26%. If the percentage of
attenuation is more than 8%, then the routed quantity of flow at Naraj
corresponding to outflow from reservoir will be less than the flow with 8%
attenuation, without affecting the objective of filling up the reservoir. The result
obtained by 8% attenuation will hold good for the flood, whose percentage of
attenuation is more than this value. Therefore, a minimum value of 8%
attenuation is used for this purpose. The factor 1/ (1 – 0.08) is used to account
for the attenuation of routing the outflow from reservoir to Naraj. RPSt in any
period should be such an amount that the flow at Naraj will be less than non-
damaging flow (NDF

where NDF is the non-damaging flow at Naraj (i.e., 25.5 thousand m3/s), and
DCt is the downstream catchment contribution during time period t.
The forward pass uses Eqns. (2), (4), (6), and (7) and the backward pass uses
Eqns. (3), (5), (6), and (7). Releases for irrigation and downstream catchment
contribution in each time duration are the known components. Non-damaging
flow at Naraj is 25.5 thousand m3/s, as adopted by Hirakud reservoir authority.
Flood control objective is to try and keep the flow at Naraj, comprising release
from dam with proper routing and downstream catchment contribution, to be
within 25.5 thousand m3/s (Baliarsingh, 2000).
The level of 179.83 m [dead storage level corresponding to a volume of 1.816
Thousand Million Cubic Meter (TMCM)] of the reservoir on 1st July is
expected to go to a level of 192.02 m (full reservoir level corresponding to a
volume of 7.197 TMCM) by 1st October. So, both minimum and maximum
possible storages on 1st July are 1.816 TMCM. Similarly, minimum and
maximum possible storages on 1st October are 7.197 TMCM. For finding
maximum possible storages of the remaining days of monsoon, Eqns. (2), (4),
(6), and (7) are used, starting from 1st July to 1st October. For finding minimum
possible storage, Eqns. (3), (5), (6), and (7) are used in backward direction, i.e.,
from 1st October to 1st July. The limits of release for power and spill (RPSt) for
any time duration are given by Eqns. (6) and (7). In case of Hirakud reservoir,
RPSMIN is zero. Hence, RPSt can vary from a minimum value of 0 to a
maximum value of (NDF – DCt+1)/(1 – 0.08) or RPSMAX, whichever is less. It
is always tried to keep RPSt as 0 in Eqns. (2) and (3). Two values, minimum
and maximum possible storages are different for different combinations of
inflow into reservoir and d/s catchment contribution. In the process, if the

Page | 23
maximum possible storage is more than full reservoir level in the forward pass
[Eqn. (2)], only then the RPSt is allowed to be positive. Similarly, if the
minimum possible storage is less than the dead storage level in backward pass
[Eqn. (3)], only then RPSt will be positive. If it is not possible to keep these two
factors within the limits by utilizing the RPSt to its full extent, then that
particular combination of inflow into the reservoir and d/s catchment
contribution cannot be handled by Hirakud reservoir to keep the flow at Naraj
within non-damaging flow. When this combination of inflow into the reservoir
and d/s catchment contribution occurs, the flow at Naraj will exceed non-
damaging flow. One sample of minimum and maximum possible storages for
the entire monsoon season are shown in Table 4 corresponding to a
representative inflow with peak of 37 thousand m3/s and a representative
downstream catchment contribution with a peak of 25.5 thousand m3/s.

4.3 Results and Conclusion

In this chapter, the rule curves corresponding to flood conditions are found out
by using Folded Dynamic Programming (FDP). It is observed that the
combination of 51 thousand m3/s as peak inflow into reservoir and any value up
to 25.5 thousand m3/s as peak downstream catchment contribution can be
handled by the Hirakud reservoir, but combination of 60 thousand m3/s inflow
into the reservoir and any value of downstream catchment contribution cannot
be handled. The minimum and maximum possible storages for the combination
of 3, 9, 14, 20, 25.5 thousand m3/s as peak downstream catchment contribution
and 9, 23, 51 thousand m3/s as peak inflow into reservoir are computed. It is
observed from the analysis that there is very little change in minimum and
maximum possible storages for the variation in downstream catch'ment
contribution for any specified value of inflow into the reservoir. Then the
process was tried with the peak inflow into the reservoir from 51 to 60 thousand
m3/s at an increment interval of three thousand m3/s. It is observed that the
combination of 51, 54, 57 thousand m3/s as peak inflow into reservoir with
25.5, 20, 14 thousand m3/s, respectively as peak downstream catchment
contribution can be handled by the Hirakud reservoir and are considered as
critical combinations for flood situation. The minimum and maximum possible
storages for these three critical combinations are shown in Fig. 4. Any
combination with more flow rate than these three critical combinations will
either create flow at Naraj to be more than non-damaging flow and/or overflow
of reservoir.
Now, FDP is applied to these critical combinations of inflow into the reservoir
and d/s catchment contribution to get rule curves for flood conditions. The
process of FDP is applied with the objective function (maximization) and
constraints applicable to the Hirakud reservoir which are as follows.

Page | 24
The backward recursive relationship is
Maximize pt(St) = max [Bt(St, PRt) + ft + 1(St + 1)]
PRt PRcap and t = 0, 1, …, T–1
where pT (ST) = 0
subject to the following constraints:
St+1 = St + CF [(It + It+1)/2 – (IRt + IRt+1)/2]
– CF [(PRt+ PRt+1)/2 + (SRt + SRt+1)/2] t = 0, 1, ..., T–1

Table 4. Minimum and maximum possible storages for the combination of 37 thousand m3/s
as peak of inflow into the reservoir and 25.5 thousand m3/s as peak of downstream catchment
contribution

Page | 25
Bt (St, PRt) =[Ht (PRt + PRt+1)]/(13800*2) t = 0, 1, ..., T–1
Ht = fs [(St + St+1)/2] t = 0, 1, ..., T–1
[1/(1-0.08)] (PRt + SRt) (NDF – DCt+1)
RPSMIN £ PRt + SRt £ RPSMAX
Smin,t St £ Smax,t
PRt ≥ 0; SRt ≥ 0

where, pt(St) = total power produced in mega watts for the remaining
periods of t, t+1, …, T–1 with St as initial storage state at the beginning of time
period t; St = storage state in TMCM at beginning of time period t; PRt is the
rate of release for hydropower in m3/s [ft3/s only in Eqn. (11)] at beginning of
time period t; PRcap = turbine capacity in m3/s; CF = conversion factor to
convert from rate to volume of any variable; It = rate of inflow into the reservoir
in m3/s at beginning of time period t; IRt = release rate for irrigation in m3/s at
beginning of time period t; SRt = rate of spill from dam in m3/s at beginning of
time period t; Bt (St, PRt) = power produced in mega watts during time period t;
Ht = average head available for power production in feet during time period t;
fs = relationship of storage-elevation curve; NDF = non-damaging flow at Naraj
in m3/s; DCt = downstream catchment contribution at beginning of time period
t; RPSMIN = minimum required release from reservoir; RPSMAX = maximum
release capacity of reservoir; Smin,t = minimum possible storage at the
beginning of time period t; and Smax,t = maximum possible storage at the
beginning of time period t. There are T time periods in the operation horizon.
Here, T = 92 days, which is designated as 0, 1, ...., 91. In Eqn. (8), pt(St) is the
objective function, which is to be maximized. Equation (9) shows the objective
function value at t = T. This is taken as 0 here, but any value other than zero will
not affect the result. Mass balance of the reservoir is answered by Eqn. (10).

Page | 26
Equation (11) shows the amount of hydropower produced within time period t,
which is a function of average release for power and the head of water available
to the turbine. This equation is used by Hirakud authority, in which actual
turbine efficiency is incorporated. In this equation alone, PRt is expressed by
the unit of cusecs and Ht by the unit of feet. The average reservoir storage level
during time period t can be found out from the reservoir storage by Eqn. (12).
This is obtained from the elevation-storage table of Hirakud reservoir (Patri,
1993). The quantity released for power comes back to Mahanadi River and joins
with spill and d/s catchment contribution to make up the total flow at Naraj.
This combined release from reservoir (release for power and spill) is restricted
by Eqns (13) and (14) taking due consideration of lag time and attenuation to
keep the flow at Naraj below the non-damaging flow, i.e., 25.5 thousand m3/s.
Equation (15) shows the restriction of reservoir storage state, which should be
in between minimum and maximum possible storages. Equation (16) shows the
non-negative constraints on release for power and spill from the dam. As
discussed earlier, the combinations of 51, 54, 57 thousand m3/s as peak inflow
into reservoir and 25.5, 20, 14 thousand m3/s, respectively as peak downstream
catchment contribution form the critical combinations for flood situation. The
rule curve for critical combination of representative series (inflow into the
reservoir and downstream catchment contribution) for flood condition is shown
in Fig. 5. The continuous wave like pattern, seen in the rule curve is because of
the hypothetical inflow series, where a flood is assumed during every 10 days
block period. In actual flood condition, the chance of occurrence of these types
of series may not be there, but there is every chance of occurrence of single
flood in the corresponding 10 days block period as shown in the historical sixty-
eight floods. To make the reservoir ready to tackle these floods, should be
brought at least equal to the level corresponding to the trough of each wave. If
the reservoir is at a level more than this and that particular flood occurs, either
the reservoir will overflow or flow at Naraj will be more than 25.5 thousand
m3/s. The final rule curve for flood condition corresponds to the trough of the
waves of rule curves, obtained from the critical combination of two flow series.
It may be observed from Fig. 5 that the combination of 57 thousand m3/s of
inflow into reservoir and 14 thousand m3/s for downstream catchment
contribution is the most critical among the above three critical combinations of
flow series. It is always necessary to keep the reservoir at the lowest level for
tackling these combinations. Hence, the rule curve, corresponding to this
combination, is chosen for finding the final rule curve for flood condition. The
level starts from 179.83 m at 0th time period. The lowest level among all
troughs of the whole season occurs in 73rd time period and corresponds to
184.3 m. Hence, there is no point in increasing the level beyond 184.3 m before
73rd time period. At the end of monsoon season, i.e., at end of 92nd time
period, the reservoir level is 192.02 m. The levels for all intermediate time
periods are obtained by linear interpolation.

Page | 27
This is the final rule curve for flood condition. Although in this particular case
study, only one lowest value of the troughs (corresponding to 73rd time period)
is used for arriving at the final rule curve, in a general case, all the troughs may
have to be used if there is a gradual increase in the lowest value of troughs with
increase in time. As such, this represents the most conservative attitude, which
is vital for the present study area because the main objective of reservoir is
flood control. The following conclusions could be drawn from the present case
study:
 FDP is a useful tool to develop the optimal operation policy of a reservoir
system for flood control.
 The long-term operating policy developed for the Hirakud reservoir
system can be actually adopted for better utilization of the system.
 In case of need to change the operating policy due to change in hydrology
regime or increased risk aversion, the methodology developed in this
study can be used to develop a new rule curve.

Page | 28
 CHAPTER 5

MEASURES SUGGESTED FOR EVOLVING A FEASIBLE AND

IMPLEMENTABLE FLOOD CONTROL STRATEGY

Based on the number of flood control strategies being adopted throughout the
world, suitable measures for flood control strategy are listed below which are
generic in nature and are not based on the case study presented alone.
 Reasonably accurate forecast, coordination, flood preparedness, feasible
and practical evacuation plans, and creation of suitable infrastructure.
 Integrated regional and flexible water resources development/watershed
management plan such as dredging of rivers/streams and re-excavation of
abandoned channels based on the ground level situation.
 Interlinking of rivers at a regional level may also reduce the flood
damage.
 Development of more sophisticated and robust mathematical models with
suitable database in close collaboration with academia.
 Research, education and training at all levels such as universities, R&D
organizations (e.g., NIH, CWC, CSIR), water resources departments of
various state governments, and River Valley Development authorities.
 Stakeholder’s survey and involvement of stakeholders at relevant level
for improving participatory flood management situation. This also helps
to get effective feedback to learn lessons from the past events.
 Establishment of National Flood Control Authority which will coordinate
all the relevant agencies.
 Environment impact assessment of each flood.

Page | 29
 CHAPTER 6

CONCLUDING REMARKS
In this chapter, modeling for the flood control and management is described to
suggest suitable management strategies. A case study is presented to
demonstrate the development of an optimal operating policy for a reservoir with
flood control as a major criterion. In addition, the following issues are
addressed: various structural and non-structural approaches for flood control
and management, a step-by-step approach for choosing the most appropriate
flood control strategy and measures for evolving a feasible and implementable
flood control strategy. To conclude, it is essential to develop different modeling
approaches for effective flood control and management so as to ensure
sustainable development of flood-prone regions.

Page | 30
 REFERENCES
Ahmad, S. and Simonovic, S.P. (2006). An Intelligent decision support system for management of floods. Water
Resources Management, 20(3): 391-410.
Akbari , M., Afshar, A. and Sadrabadi, M.R. (2009). Fuzzy rule based models modification by new data: Application
to flood flow forecasting. Water Resources Management, DOI, 10.1007/s11269-008-9392-z.
ASCE Task Committee on Application of Artificial Neural Networks in Hydrology (2000a). Artificial neural
networks in hydrology. I: Preliminary concepts. Journal of Hydrologic Engineering, ASCE, 5(2): 115-123.
ASCE Task Committee on Application of Artificial Neural Networks in Hydrology (2000b). Artificial neural
networks in hydrology. II: Hydrologic applications. Journal of Hydrologic Engineering, ASCE, 5(2): 124-137.
Bahremand, A., Smedt, F.D., Corluy, J., Liu, Y.B., Poorova, J., Velcicka, L. and Kunikova, E. (2007). WetSpa
model application for assessing reforestation impacts on floods in Margecany-Hornad Watershed, Slovakia.
Water Resources Management, 21(8): 1373-1391.
Baliarsingh, F. (2000). Long-Term and Short-Term Optimal Reservoir Operation for Flood Control. Ph.D. Thesis,
Indian Institute of Technology, Kharagpur, West Bengal, India.
Bana e Costa, C.A., Silva, P.A.D. and Correia, F.N. (2004). Multicriteria evaluation of flood control measures: The
case of Ribeira do Livramento. Water Resources Management, 18(3): 263-283.
Bojadziev, G. and Bojadziev, M. (1997). Fuzzy Logic for Business: Finance and Management. World Scientific,
Singapore, 232 pp.
Carsell, K.M., Pingel, N.D. and Ford, D.T. (2004). Quantifying the benefit of a flood warning system. Natural
Hazards Review, 5(3): 131-140.
Chau, K.W., Wu, C.L. and Li, Y.S. (2005). Comparison of several flood forecasting models in Yangtze River.
Journal of Hydrologic Engineering, ASCE, 10(6): 485-491.
Clark, M.J. (1998). Putting water in its place: A perspective on GIS in hydrology and water management.
Hydrological Processes, 12: 823-834.
Clarke, K.C. (1999). Getting started with geographic information systems. Prentice Hall, New Jersey, 338 pp.
De Vantier, B.A. and Feldman, A.D. (1993). Review of GIS applications in hydrologic modeling. Journal of Water
Resources Planning and Management, ASCE, 119(2): 246-261.
Deb, K. (2001). Multiobjective Optimization using Evolutionary Algorithms. Wiley, Singapore, 515 pp.
Deka, P. and Chandramouli, V. (2005). Fuzzy neural network model for hydrologic flow routing. Journal of
Hydrologic Engineering, ASCE, 10(4): 302-314.
DeMers, M.N. (1997). Fundamentals of Geographic Information Systems. John Wiley & Sons, New York, 480 pp.
Dhar, O.N. and Nandagiri, S. (1998). Floods in Indian Rivers and their Meteorological Aspects. In: V.S. Kale
(editor), Flood Studies in India, Memoir 41, Geological Society of India, Bangalore, pp. 1-25.
Engman, E.T. and Gurney, R.J. (1991). Remote Sensing in Hydrology. Chapman and Hall, London, 225 pp.
Garde, R.J. (1998). Floods and flood control: Engineering approach. In: V.S. Kale (editor), Flood Studies in India,
Memoir 41, Geological Society of India, Bangalore, pp. 173-193.
Hajkowicz, S. and Collins, K. (2007). A review of multiple criteria analysis for water resource planning and
management. Water Resources Management, 21(9): 1553-1566.
Hall, J.W., Meadowcroft, I.C., Sayers, P.B. and Bramley, M.E. (2003). Integrated flood risk management in England
and Wales. Natural Hazards Review, 4(3): 126-135.
Heywood, I., Cornelius, S. and Carver, S. (1998). An Introduction to Geographical Information Systems. Pearson
Education, 279 pp.
Holz, K.P., Hildebrandt, G. and Weber, L. (2006). Concept for a web-based information system for flood
management. Natural Hazards, 38(1-2): 121-140.
Ignizio, J P. (1991). An Introduction to Expert Systems: Development and Implementation of Rule-Based Expert
Systems. McGraw-Hill Inc, New York, 384 pp.
Islam, M.M. and Sado, K. (2002). Development priority map for flood countermeasures by remote sensing data
with geographical information system. Journal of Hydrologic Engineering, ASCE, 7(5): 346-355.
Jain, S.K. and Singh, V.P. (2006). Water Resources Systems Planning and Management: Developments in Water
Science. Elsevier, New Delhi, 858 pp.
Jain, S.K., Saraf, A.K., Goswami, A. and Ahmad, T. (2006). Flood inundation mapping using NOAA AVHRR data.
Water Resources Management, 20(6): 949-959.
Jain, S.K., Singh, R.D., Jain, M.K. and Lohani, A.K. (2005). Delineation of flood-prone areas using remote sensing
techniques. Water Resources Management, 19(4): 333-347.
Kale, V.S. (1998), Flood studies in India. Edited volume, Memoir 41, Geological Society of India, Bangalore, 256.
Karamouz , M., Abesi, O., Moridi, A. and Ahmadi, A. (2009). Development of optimization schemes for floodplain
management: A case study. Water Resources Management, 10.1007/s11269-008-9350-9.
Khan, S., Ahmad, A. and Wang, B. (2007). Quantifying rainfall and flooding impacts on groundwater levels in
irrigation areas: GIS approach. Journal of Irrigation and Drainage Engineering, ASCE, 133(4): 359-367.
Kumar, R. and Chatterjee, C. (2005). Regional flood frequency analysis using L-moments for North Brahmaputra
Region of India. Journal of Hydrologic Engineering, ASCE, 10(1): 1-7.

Page | 31
Lekuthai, A. and Vongvisessomjai, S. (2001). Intangible flood damage quantification. Water Resources Management,.
Levy, J.K. (2005). Multiple criteria decision making and decision support systems for flood risk management.
Stochastic Environmental Research and Risk Assessment, 19(6): 438-447.
Li, Z.J. and Zhang, K. (2008). Comparison of three GIS-based hydrological models. Journal of Hydrologic
Engineering, ASCE, 13(5): 364-370.
Liu, Y.B. and Smedt, F.D. (2005). Flood modeling for complex terrain using GIS and remote sensed information.
Water Resources Management, 19(5): 605-624.
Lund, J.R. (2002). Floodplain planning with risk-based optimization. Journal of Water Resources Planning and
Management, ASCE, 128(3): 202-207.
Machado, S.M. and Ahmad, S. (2007). Flood hazard assessment of Atrato River in Colombia. Water Resources
Management, 21(3): 591-609.
Mays, L.W. (1996). Water Resources Hand Book. McGraw-Hill, New York, 1568 pp.
Mohapatra, P.K. and Singh, R.D. (2003). Flood management in India. Natural Hazards, 28: 131-143.
Nagesh Kumar, D. (2002). Remote sensing applications to water resources. In: Rama Prasad and S. Vedula (editors),
Research Perspectives in Hydromechanics and Water Resources Engineering. World Scientific, Singapore.
Nagesh Kumar, D. (2006). Lecture Notes on Soft Computing. Indian Institute of Science, Bangalore, India, 120 pp.
Nagesh Kumar, D. and Baliarsingh, F. (2003). Folded dynamic programming for optimal operation of multireservoir
system. Water Resources Management, 17: 337-353.
Nakicenovic, N., Alcamo, J., Davis, G., de Vries, B., Fenhann, J., Gaffin, S., Gregory, K., Grübler, A., Jung, T.Y.,
Kram, T., La Rovere, E.L., Michaelis, L., Mori, S., Morita, T., Pepper, W., Pitcher, H., Price, L., Raihi, K.,
Roehrl, A., Rogner, H.H., Sankovski, A., Schlesinger, M., Shukla, P., Smith, S., Swart, R., van Rooijen, S.,
Victor, N., Dadi, Z. (2000). IPCC Special Report on Emissions Scenarios. Cambridge University Press,
Cambridge, United Kingdom and New York, NY, 599 pp.
Needham, J.T., Watkins, D.W., Lund, J.R. and Nanda, S.K. (2000). Linear programming for flood control on the
Iowa and Des Moines Rivers. Journal of Water Resources Planning & Management, ASCE, 126(3): 118-127.
Noman, N.S., Nelson, E.J. and Zundel, A.K. (2003). Improved process for floodplain delineation from digital
terrain models. Journal of Water Resources Planning and Management, ASCE, 129(5): 427-436.
Patri, S. (1993). Data on Flood Control Operation of Hirakud Dam. Irrigation Department, Government of Orissa,
Bhubaneswar, India, 200 pp.
Ranjithan, S.R. (2005). Role of evolutionary computation in environmental and water resources systems analysis.
Journal of Water Resources Planning and Management, ASCE, 131(1): 1-2.
Rao, D.P., Bhanumurthy, V., Rao, G.S. and Manjusri, P. (1998). Remote sensing and GIS in flood management in
India. V.S. Kale (editor), Flood Studies in India. Memoir 41, Geological Society of India, Bangalore, 195-218.
Rumelhart, D.E., Hinton, G.E. and Williams, R.J. (1986). Learning internal representation by back-propagating
errors. Nature, 32(3): 533-536.
Sanders, R., Shaw, F., Galy, M.H. and Foote, M. (2005). National flood modeling for insurance purposes using
IFSAR for flood risk estimation in Europe. Hydrology and Earth Systems Sciences, 9(4): 449-456.
Seth, S.M. (1998). Flood hydrology and flood management in India. In: V.S. Kale (editor), Flood Studies in India.
Memoir 41, Geological Society of India, Bangalore, pp. 155-172.
Simonovic, S.P. (1991). Knowledge-based systems and operational Hydrology. Canadian Journal of Civil
Engineering, 18(1): 1-11.
Simonovic, S.P. and Li, L. (2004). Sensitivity of the Red River Basin flood protection system to climate variability
and change. Water Resources Management, 18(2): 89-110.
Singh, V.P. and Fiorentino, M. (1996). Geographical Information Systems in Hydrology. Kluwer Academic
Publishers, Dordrecht, Germany, 468 pp.
Slutzman, J.E. and Smith, J.A. (2006). Effects of flood control structures on flood response for hurricane floyd in
the Brandywine Creek Watershed, Pennsylvania. Journal of Hydrologic Engineering, ASCE, 11(5): 432-441.
Solomatine, D.P. and Xue, Y. (2004). M5 model trees and neural networks: Application to flood forecasting in the
upper reach of the Huai River in China. Journal of Hydrologic Engineering, ASCE, 9(6): 491-501.
Subramanya, K. (1994). Engineering Hydrology. Tata McGraw-Hill Publishing, New Delhi, 392 pp.
Travis, Q.B. and Mays, L.W. (2008). Optimizing retention basin networks. Journal of Water Resources Planning
and Management, ASCE, 134(5): 432-439.
Varas, E.A. and Von Chrismar, M. (1995). Expert system for the selection of methods to calculate design flood
flows. Hydrological Sciences Journal, 40(6): 739-749.
Venkatesh, M., Olivera, F., Arabi, M. and Edleman, S. (2008). Uncertainty in flood inundation mapping: current
issues and future directions. Journal of Hydrologic Engineering, ASCE, 13(7): 608-620.
Verbyala, D.L. (2002). Practical GIS analysis. Taylor and Francis, London, 294 pp.
Vieux, B.E. (2001). Distributed Hydrologic Modeling using GIS. Kluwer Academic Publishers, Dordrecht, Germany.
Wei, C.C. and Hsu, N.S. (2008). Multireservoir flood-control optimization with neural-based linear channel level
routing under tidal effects. Water Resources Management, 22(11): 1625-1647.
Wu, J.S., Han, J., Annambhotla, S. and Bryant, S. (2005). Artificial neural networks for forecasting watershed
runoff and stream flows. Journal of Hydrologic Engineering, ASCE, 10(3): 216-222.
Yu , Y.B., Wang, B.D., Wang, G.L. and Li, W. (2004). Multi-person multiobjective fuzzy decision-making model.

Page | 32