## Abstract

Proper management of reservoirs in flood conditions minimizes flood damage and keeps the reservoir in a stable condition. This paper presents a Web-based decision support system of reservoir flood management (WDRFM) for reservoirs with gated spillways. WDRFM is capable of estimating the current situation of the reservoir and before the flood reaches the reservoir, provides the operator with suggestions to have optimal control over gates, using multistage simulation-optimization models to minimize flood damage downstream. Investigating the possibility of changing the dam's discharge gates, carrying out dam pre-release, and announcing relevant flood control warnings can all be performed by WDRFM, while allowing the operator to determine such observations and make final decisions on a time step basis. To assess the performance of WDRFM, 15 scenarios in four groups for flood management were defined on April 14, 2016 at Dez Dam, Iran. A comparison of one scenario with similar initial conditions with the occurrence of flood event has shown that a daily peak discharge shows 997 m3/s decrease from that recorded by the operator. The ability to examine different scenarios based on the conditions at any time in the WDRFM enables the decision-makers and operators to confront various circumstances.

## INTRODUCTION

Nowadays, the use of decision support systems (DSS) has developed dramatically, addressing complex and semi-structured issues of water resource management, and owing much of its increasing success to the speed and accuracy of these systems in calculating and proposing options to users, high-cost savings, ever-increasing hardware, and software advances in computer science, information processing limitations by human resources, theoretical and practical developments in applicable relevant areas such as hydrology and meteorology. DSS is a computerized management advisory system that provides decision-makers with timely management data by drawing upon databases, models, and dialog systems (Grigg 1996). After its confirmation as a practical management approach, DSS were widely used in water resource management systems (Mysiak et al. 2005; Anzaldi et al. 2014). DSS benefit the decision-makers at different levels of management by providing them with a more efficient plan to manage water resources and, at once, forming a better vision for the future. As a leading researcher in this area, Loucks et al. (1985) pursued the development of a DSS for water resources management following an interactive water resources model. More recently, various specific applications of DSS in water resources and commercial software packages have been designed for facilitating water quality management (Camara 1990; Dingfei & Stewart 2004; Paredes et al. 2010), impact assessment of policy scenarios adopted by multiple scales of watershed authorities (Davis et al. 1991; Holmes et al. 2005), reservoir operational management (Simonovic 1992), sustainable water distribution system planning (Freund et al. 2017), regional water resources planning (McKinney et al. 1993), river flow forecasting (Bender & Simonovic 1994; Wang et al. 2017b; Mosavi et al. 2018; Yaseen et al. 2019), drought monitoring and management (Palmer & Tull 1987; Palmer & Holmes 1988; Chang et al. 1996), water allocation (Koutsoyiannis et al. 2003; Yang et al. 2017), flood control management (Ahmad & Simonovic 2001; Cheng & Chau 2004; Mioc et al. 2008; Fotovatikhah et al. 2018), integrated scenario-based multi-criteria decision support system (SMC-DSS) for planning water resources management in a river basin (Weng et al. 2010), and multi-reservoirs' operational management for securing regional water supply (Opricovic 2011).

Among the most efficient types of decision-making systems are Web-based systems with the following advantages: instant connection for collecting data from various sources, portability and availability, no need to install particular software which clogs system memory, the ability to instantly update and repair the system for all users, and finally, the ability to run on basic systems compared to other systems. Hence, local government and local stakeholders can access the Web-based DSS without the need to visit complex organizations. A Web-based DSS can also easily be transferred to other areas with similar features due to the flexible nature. Specific Web-based DSS applications have been designed for urban water resource management (Zeng et al. 2012), flood and fire management in and around urban areas (Kochilakis et al. 2016), irrigation network management considering user management experiences (Wang et al. 2017a), flood forecasting system (Li et al. 2006), single reservoir operation (Jahanpour et al. 2014), reservoir flood control (Yong et al. 2009), smart dam operation (Ahmad & Hossain 2019), real-time flood control (Zhu et al. 2017), and urban flood warning (Li et al. 2011). Considering the user's necessities in designing a DSS is essential (Fernandez & Trolinger 2007). A DSS is developed based on a specific purpose or application in order to provide timely management solutions, and to examine the effects of different decision-making scenarios. Therefore, the development of a Web-based DSS depends on the concept of decision-making that the users of the system need to rely on to be able to solve the problem. It is quite necessary for the DSS to have networking capabilities to log and process simultaneously the collection of independent data coming from sources located in dispersed geographical locations or the sort recorded by the employees of the various departments. The networking capability of the DSS becomes especially relevant when the timing of flood management becomes a critical element in carrying out the necessary actions, especially to reduce flood damage and coordinating the organizations involved in crisis management with the timely information. Moreover, applying a DSS of flood management requires specialized knowledge of hydraulic and hydrological concepts, analyzing the needed data, and computer programming. It is a major challenge for reservoir operators, and it can have a negative impact on their performance, especially under flooding conditions. The capabilities of a Web-based system can overcome this challenge and make the system user-friendly. Therefore, Web-based DSS can be the most appropriate type of DSS for this purpose, as it simultaneously receives various data on flood management and, by imposing different management strategies and analyzing scenarios, provides the most appropriate solution to local and regional managers. Under such circumstances, the system appears to offer optimal functionality as it takes preventive measures such as alerting and evicting residents in downstream areas, blocking roads, warning hospitals and the Red Crescent, and hence providing those in charge of risk-managing organizations with timely information.

One of the most suitable nonstructural methods for reservoir flood control is the proper operation of dam spillway gates during floods. If a spillway gate is opened more than it should be, a great amount of flow will be released which, in its turn, damages the downstream of the reservoir, reducing the reservoir's role in controlling the flood. Conversely, if the spillway gates are opened less than the required size, the safety of the dam can be damaged, leading to the overtopping of the dam (Che & Mays 2015). In this respect, reservoir flood control has become a complex and semi-structured phenomenon due to the many uncertainties in the flood and multiple decision-makers at different levels of management and thus needs a DSS. Reservoir flood control is also very complicated if the reservoir system lacks a flood forecasting system and volume and peak flow rates are not known. In such a system, the experience and judgment of operators is an important factor in the operation of dam spillway gates during floods. To optimize the operation of spillway gates in systems without a flood forecasting system, Acanal & Haktanir (1999, 2000) and Haktanir et al. (2013) proposed five, six, ten, and fifteen-stage performance policies. In these methods, the volume of reservoir flood control is divided into a small volume of storage, and for each small volume of storage, the reservoir water level and the spillway gate openings are determined on a trial-and-error basis; in this method, a large number of possible answers are given for the operation of spillway gates in times of flood so that an optimal solution can be finalized. Minimizing downstream damage is considered as an objective function to optimize spillway gates' operation (Qin et al. 2010; Valerino et al. 2010; Ahmed & Mays 2013). The objectives functions are the peak release discharge (Malekmohammadi et al. 2010) or the maximum inundation depth at points where downstream damage occurs (Bayat et al. 2011). Zargar et al. (2016) developed a simulation-optimization model based on a multi-stage method with the purpose of responding to floods where there is no prior knowledge of the shape and size of inflow flood hydrographs, and assuming that the reservoir is full. In this method, the decision as to how much of the discharge will be released at each critical level is made only on the basis of the reservoir water level and the objective function of the optimization problem is to minimize the expected annual flood damage. The number of stages and critical discharge values at each stage are determined independently of inflow hydrograph characteristics. It can prevent the sudden drops and jumps in outflow discharge which may cause a major problem in operating the spillway gates. Since all floods from small-to-large floods (a ten-year flood event to PMF) are investigated simultaneously with a single utilization policy by the optimization model, the operator does not need a flood forecasting system but does not use the entire flood control capacity for small floods while using this capacity may create better results in flood management.

### Values of used parameters

Embedding different algorithms in the WDRFM system is only to increase the system's capability. Therefore, based on which algorithm the user of the system is familiar with, the parameters of that algorithm can be entered and used. Due to the collaboration and sharing of information between particles, high convergence speed, better flexibility against local optimal problems of the PSO algorithm, this algorithm was selected for system evaluation in the Dez Dam reservoir management 2016 flood. The values of its parameters are given in accordance with Table 3.

Table 3

The values of common system parameters in all scenarios

Step amount of changing the discharge10 m3/s
Step amount of changing the water level in pre-release 0.1 m
Prioritization of the use of gates 1. powerhouse gates; 2. irrigation gates
Number of stages 10
Permissible error (%) 0.05
Initial spillway discharge
Initial discharge of irrigation gates
Time step 2 hours
Threshold flood discharge 220 m3/s
α (Equation (1)) 0.0123
PSO algorithm Population size 60
Maximum iteration 1,500
W-damp 0.8
C1
C2
Step amount of changing the discharge10 m3/s
Step amount of changing the water level in pre-release 0.1 m
Prioritization of the use of gates 1. powerhouse gates; 2. irrigation gates
Number of stages 10
Permissible error (%) 0.05
Initial spillway discharge
Initial discharge of irrigation gates
Time step 2 hours
Threshold flood discharge 220 m3/s
α (Equation (1)) 0.0123
PSO algorithm Population size 60
Maximum iteration 1,500
W-damp 0.8
C1
C2

In order to consider the more critical conditions, the value of x is equal to 0.5 and k is equal to 2 hours which is equal to the observed flood travel time between the Talezang station and the reservoir. In fact, the river storage effect will not be considered in the flood routing. The number of stages or the number of decision variables equals 10 which is sufficient to prevent sudden jumps in the water level and the outflow hydrograph from the reservoir. The amount of outflow from the spillway and irrigation gates are considered zero and are assumed to be closed at the start of flood management. The priority of using outlets is to use the gates of the powerhouse at first and after that the irrigation outlets, because opening the irrigation gates discharges the sediment and has environmental limitations. The amounts of changing discharge in these outlets are step by step and 10 m3/s. Decreasing water level for estimating the pre-release is performed with a step of 0.1 m. The lower the discharge and water elevation steps, the lower the processing speed. The permissible 5% error is defined to achieve optimal solutions. The flood start threshold for the Dez River is set at 220 m3/s, which is safe discharge. The α parameter used in estimating the falling limb of the inflow hydrograph is set to 0.0123 using observational hydrograph analysis. Finally, the parameters of the optimization algorithm are determined, with the constants C1 and C2 referring to the learning factors. The PSO algorithm works better when the values of C1, C2 are equal or have a moderate balance. Each time the loop is repeated, it uses a W damping coefficient to reduce the particle velocity so that the algorithm does not have any convergence problems. W is the speed factor at the start of the optimization process. The maximum iteration is assumed as 1,500, which means that the algorithm is not capable of achieving the optimal solution in the next repetitions.

### Scenarios' description

In this system, it is possible to define different scenarios based on several factors such as the objective functions, the initial water levels in the reservoir, outflow from the powerhouse. Also, there are additional limitations on the use of these gates during flood management. In this research, the system capabilities were evaluated for different situations through different scenarios. According to Table 4, for different factors, abbreviations are considered so that different scenarios can be defined accordingly.

Table 4

Abbreviation of factors for scenarios' definition

Factor nameFactor number
1234
Water level (m) (WL340 345 347.32 –
Peak discharge ratio (C1.5 2.5
Initial powerhouse discharge (m3/s) (QH171.8 100 50
Limitations (LimPowerhouse inlets Full use Use of one gate Unchangeable
Irrigation outlets Unchangeable Unchangeable Full use
Factor nameFactor number
1234
Water level (m) (WL340 345 347.32 –
Peak discharge ratio (C1.5 2.5
Initial powerhouse discharge (m3/s) (QH171.8 100 50
Limitations (LimPowerhouse inlets Full use Use of one gate Unchangeable
Irrigation outlets Unchangeable Unchangeable Full use

For example, the defined scenario (WL1-C1-QH1-Lim1) means that the water level at the start of flood control is 340 m, and the ratio of the inflow flood peak discharge to the outflow peak discharge is equal to 1.5, and the powerhouse discharge at the start of flood control is 171.8 m3/s. In this case, the powerhouse inlets can be fully used, and the irrigation gates do not have the ability to change the flow discharge. In fact, to study the effect of the ratio of the inflow flood peak discharge to the outflow peak discharge in the objective (C), different scenarios were defined as Group 1. In this group, the initial water level (W), powerhouse discharge (QH), and the applied limitations on gates (Lim) were considered constant but the objective function was varied. Group 2 scenarios are similar to Group 1 but the initial water level is increased to evaluate its impact. Group 3 scenarios are similar to the reservoir's condition during the April 14, 2016 flood, and in these scenarios, the possibility of flood management under different objective functions and the effect of initial powerhouse discharge are considered. Finally, in Group 4 scenarios, the limitations in using the gates for flood management are evaluated. Generally, the scenarios used in this study and the purpose of defining different scenarios are presented in Table 5. This study is limited to examine only these scenarios. However, other scenarios can also be defined with respect to the capabilities of the Web-based DSS.

Table 5

Defined scenarios and purpose of defining each group

Group 1Group 2Group 3Group 4
Abbreviated name for scenarios identification WL1-C1-QH4-Lim1, WL1-C2-QH4-Lim1 WL1-C3-QH4-Lim1 WL1-C4-QH4-Lim1 WL2-C1-QH4-Lim1, WL2-C2-QH4-Lim1, WL2-C3-QH4-Lim1, WL2-C4-QH4-Lim1 WL3-C3-QH1-Lim1, WL3-C3-QH2-Lim1, WL3-C3-QH3-Lim1, WL3-C2-QH3-Lim1 WL3-C3-QH2-Lim2, WL3-C3-QH2-Lim3, WL3-C2-QH3-Lim3
Purpose of each group Effect of peak discharge ratio (c) Effect of initial water level Effect of initial discharge Effect of limiting the use of gates
Group 1Group 2Group 3Group 4
Abbreviated name for scenarios identification WL1-C1-QH4-Lim1, WL1-C2-QH4-Lim1 WL1-C3-QH4-Lim1 WL1-C4-QH4-Lim1 WL2-C1-QH4-Lim1, WL2-C2-QH4-Lim1, WL2-C3-QH4-Lim1, WL2-C4-QH4-Lim1 WL3-C3-QH1-Lim1, WL3-C3-QH2-Lim1, WL3-C3-QH3-Lim1, WL3-C2-QH3-Lim1 WL3-C3-QH2-Lim2, WL3-C3-QH2-Lim3, WL3-C2-QH3-Lim3
Purpose of each group Effect of peak discharge ratio (c) Effect of initial water level Effect of initial discharge Effect of limiting the use of gates

## RESULTS

It should be noted that how the gates open at any interval completely depends on the conditions of the reservoir at the beginning of that period, which itself depends on the proposed management performance of the system in the prior period. Therefore, the condition of opening in stages varies during the flood control process and in different intervals. As shown in Figure 9, by comparing reservoir water level of the scenarios of Group 1, with the increase of the C ratio, it is clear that the outflow discharge is expected to be lower than the spillway, the final water level of the reservoir increases. In any case, this final water level does not exceed the defined value of 352 m, which is the upper level of the spillway. Regarding the reservoir water levels, in these scenarios, it can be seen that during the intervals when the reservoir inflow discharge exceeds the total discharge from the dam, the water level is increasing. Spillway outflow hydrograph shows that in the rising limb of the spillway outflow hydrograph, the spillway outflow discharge can be reduced compared to the previous time frame, and it is not necessary to be strictly increasing; as in a specific time period, the gates' opening pattern is presented in such a way that if the pattern is applied to the spillway operation, the amount of outflow discharge from the spillway will decrease compared to the previous interval. In this case, the chart of the water level, due to an increase in inflow relative to outflow, has a mutation in increasing the water level. Also, any fluctuations in the falling limb of the spillway outflow hydrograph are also found to be oscillations at the reservoir water level.

Figure 9

Release hydrographs and reservoir water level under Group 1 scenarios.

Figure 9

Release hydrographs and reservoir water level under Group 1 scenarios.

According to Figure 10 for Group 2 scenarios at initial water level 345 m, the peak discharge ratio of 3 does not meet the objective function and pre-release is necessary. In this case, the initial water level is high, and the final reservoir water level exceeds the above threshold (352 m), so spillway must be performed at its maximum capacity (6,000 m3/s) to prevent the dam overtopping. However, the objective function can never be satisfied arising from the choice of a large amount for the C ratio in the objective function. Under such circumstances, therefore, the system first considered the possibility of using the full capacity of the powerhouse gates, which again was not possible to achieve the given objective. Therefore, the only possible solution was to pre-release the flood and reduce the initial level of water in the reservoir, which the system proposes.

Figure 10

Release hydrographs and reservoir water level under Group 2 scenarios.

Figure 10

Release hydrographs and reservoir water level under Group 2 scenarios.

Comparison of Group 2 scenarios with initial reservoir water level of 345 m compared to the first group scenarios with initial reservoir water level of 340 m shows that achieving a higher peak discharge ratio (C) would be more difficult by increasing the initial reservoir water level. It is not possible to use the 3-peak discharge ratio in the initial water level of 345 m, unlike the 340-m initial water level. This indicates that the initial reservoir water level has a very influential role in determining the peak discharge ratio. At the beginning of the flood, the higher the peak discharge ratio can be used in the lower the initial reservoir water levels. In addition, due to the amount of powerhouse discharge in these conditions, with the initial water level of 345 m, compared with the scenarios with a 340 m initial water level, it is shown that as the peak discharge ratio increases, so does the powerhouse discharge, while in the scenario WL2-C3-QH4-Lim1 the powerhouse discharge has reached 400 m3/s. In the case of the WL1-C4-QH4-Lim1 scenario, the increase in the powerhouse discharge peaked to 320 m3/s. In general, the desired peak discharge ratio can be entered at each time the program performs t hours. It is generally more acceptable to use low peak discharge ratios in the initial time intervals. In the early moments of the flood, the peak of the inflow flood hydrograph is lower than the safe discharge of the river downstream. Consequently, with a lower ratio of C and even being closer to 1, the water level in the reservoir shows an insignificant increase, and even shows a decrease in some cases; in the future, it can have a critical role in controlling floods by increasing reservoir storage capacity. For example, for the Dez Dam where the peak outflow from the dam is less than 668.25 m3/s, there is no damage to the downstream areas (Malekmohammadi et al. 2010). Therefore, for inflow discharges smaller than this value, a peak discharge ratio (C) can be used in the objective function that the peak discharge of the total outflow hydrograph is smaller or equal to 668.25 m3/s.

As shown in Figure 11, in Group 3 scenarios, at the initial time intervals, the reservoir's water level has been steadily decreasing, which also allows the system to control the flood. In these scenarios, the initial reservoir water level is 347.32 m, a ratio of 2.5 is considered as the objective function, in which the gates of the powerhouse can release, but the irrigation outlets cannot. The high initial water level of the reservoir has become a major challenge in this case, and the system can only manage the flood if the initial water level is reduced to a level of 344.5 m in the early steps. In the WL3-C3-QH1-Lim1 scenario the powerhouse discharge flow is 171.8 m3/s. The system manages the flood by using the spillway outflow without changing the outflow from the powerhouse. In this scenario, in the early moments of the flood the sum of outflow discharge from powerhouse and spillway, exceeded the reservoir outflow discharge, so that the water level of the reservoir was reduced, which is why the system was able to manage the flood. But in the WL3-C3-QH2-Lim1 scenario, the discharge flow from the gates of the powerhouse is 100 m3/s. The system initially tries to manage the flood through the outflow from the spillway without changing the outflow from the powerhouse, but the decreased water level is not what it has to be. Therefore, it has increased the discharge rate of the powerhouse until the water level reaches a level that can be managed. The results show the effectiveness of the initial outflow discharge from the reservoir in flood management, so more powerhouse discharge at the start of the flood makes flood control easier. It can be concluded that any safe release discharge from the above-mentioned gates and other available dam gates at the beginning of the flood, could be very helpful.

Figure 11

Release hydrographs and reservoir water level under Group 3 scenarios.

Figure 11

Release hydrographs and reservoir water level under Group 3 scenarios.

Group 4 scenarios were defined to examine possible constraints on the use of dam gates in flood control. For example, due to dispatching center orders and repairs or overhaul of the equipment or any other reason there are limitations in using the powerhouse gates. Irrigation gates are also facing another form of limitation, as their functionality in being fully open has been marred due to environmental concerns, although sometimes the opening of these gates will be a priority during flood events due to the possibility of flushing operation. Opening the irrigation gates causes large volumes of sediment to flow downstream and creates problems for aquatic life, agricultural land, and irrigation and water supply networks (Moridi & Yazdi 2017). Therefore, the opening of irrigation gates is subject to limitations and, in this assessment, it is assumed that sedimentation from the reservoir will not occur during the flood. Only in the fourth group scenario (Lim3) has the use of irrigation gates been investigated, where the aim was to investigate the limitation on the maximum capacity utilization of powerhouse gates. Figure 12 shows that limiting the operation of gates on the selection of the peak discharge ratio in the objective function and flood control has a significant effect, and the system is not able to manage the flood for some constraints based on the objective function. In these two scenarios, by selecting a ratio of 2.5 as the objective function, the initial reservoir water level (347.32 m) and initial discharge flow from the powerhouse (100 m3/s), there are two different limitations of Lim2 and Lim3 in using gates. In Lim2, powerhouse gates can discharge up to 240 m3/s and irrigation outlets cannot be changed. In Lim3, powerhouse discharge cannot be changed, and irrigation outlets can operate with any three drain outlets (maximum discharge is 180 m3/s). The WL3-C3-QH2-Lim2 scenario allows flood management to be done for the desired objective function by increasing the discharge to 240 m3/s, but if the powerhouse discharge cannot be changed in this scenario, flood management is not possible for the given objective function due to impassability of using the irrigation outlets. In the WL3-C3-QH2-Lim3 scenario it is not possible to change 140 m3/s discharge by the powerhouse gates with 100 m3/s discharge ability, but instead it is possible to use irrigation gates. Thus, by using the irrigation outlets, it is possible to manage the flood for the given objective function.

Figure 12

Release hydrographs and reservoir water level under Group 4 scenarios.

Figure 12

Release hydrographs and reservoir water level under Group 4 scenarios.

## SUMMARY AND CONCLUSION

In this study, a user-friendly, Web-based DSS was developed using ASP.NET software based on the C# programming language. The system is linked to the user through graphical interfaces and tools such as textboxes, checkboxes, checklists, and charts. It takes the necessary information and scenarios from the user, and according to the user choosing one of the algorithms of GA, PSO, or SA, and using the multistage simulation in the opening spillway gates, determines the decision parameters that are the same as the opening values in each critical level, so that the objective function is satisfied. In this system, a reliable approximate method is used to estimate the reservoir inflow flood hydrograph. The objective optimization function can identify the release hydrograph from the reservoir with a specific peak ratio compared to the peak discharge of the inflow flood hydrograph (C) or to reach the water level of the reservoir to the desired water level so that the type of objective function and the accuracy of the answer for it can be verified by the operator of the system. In this system, the possibility of engaging the powerhouse and irrigation gates has been considered by dedicating priorities and related constraints on use in flood management. In addition, if a pre-prelease is required to reach the intended target, the system calculates the required pre-prelease volume, and if the pre-release process is not feasible, the system alerts the user and provides a timetable for preventive actions. Conversely, in this system, it is possible to apply different scenarios according to the changing conditions of the reservoir, so that the user can decide on their scenario with greater accuracy and confidence. This WRDFM system is capable of executing automatically at certain times and also has the ability to develop and deploy other reservoirs.

In order to evaluate the efficiency of the WRDFM system, various scenarios were defined for the type of objective function, the initial water level of the reservoir, the initial outflow discharge, and the limitation of the use of irrigation gates and powerhouse gates. Then, according to each scenario, the flood management on April 14, 2016 was simulated by the system. The difference in results for different scenarios showed that the user and the system administrator are able to make different choices for each of the flood conditions. It is also able to see the result of applying a decision to manage a flood and, if necessary, make any decision at any time to modify its decision. The comparison of the WL3-C3-QH1-Lim1 scenario similar to that of the reservoir during April 14 was indicative of the daily peak inflow of 5,841.75 m3/s, the spillway peak discharge reached 2,437.24 m3/s while the powerhouse discharge during the whole duration of the flood reached only a maximum of 191.8 m3/s in the 6-hour period and remained unchanged at other discharge periods at the powerhouse staying at 171.8 m3/s. No irrigation gate was used. Eventually, the peak discharge of the outflow hydrograph from the reservoir reached 2,548.04 m3/s. Nevertheless, management by dam operators had led to the release hydrograph with a daily peak discharge of 3,545 m3/s, which is 997 m3/s more than the peak discharge of the outflow hydrograph from the dam that it is proposed by the system. By comparing these two values, according to the study of Malekmohammadi et al. (2010), the damage was reduced only on the basis of the peak discharge outflow from the dam, showing a reduction, namely, from 299.55 billion rials ($2.5 million) to about 215.65 billion rials ($1.8 million). In order to increase the efficiency and accuracy of the system in management of the flood, the development of the system in the following cases is highly recommended: the use of more precise methods for river routing; considering the runoff in the middle basin; improving optimization methods to form the best solution in the shortest time; linking the system to software such as GIS to determine flood plains and flood damage; linking the system with real-time flood forecasting systems instead of using an estimated flood; auto-determining of the C ratio in the objective function in terms of peak flood inflow during flood control.

## ACKNOWLEDGEMENTS

This work has been supported by the Jundi-Shapur University of Technology (JSU) in the context of the Graduate Study Program (GSP). The authors would like to thank the Khuzestan Water and Power Authority, Iran for providing the observed data used in this study.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this paper is available online at https://dx.doi.org/10.2166/hydro.2020.185.

## REFERENCES

REFERENCES
Acanal
N.
Haktanir
T.
2000
Six-stage flood routing for dams having gated spillways
.
Turkish Journal of Engineering and Environmental Sciences
23
(
6
),
411
422
.
Afshar
M. H.
2012
Large scale reservoir operation by constrained particle swarm optimization algorithms
.
Journal of Hydro-Environment Research
6
,
75
87
.
S. K.
Hossain
F.
2019
A Web-based decision support system for smart dam operations using weather forecasts
.
Journal of Hydroinformatics
21
(
5
),
687
707
.
Ahmed
S. M. S.
Mays
L. W.
2013
Model for determining real-time optimal dam releases during flooding conditions
.
Journal of Natural Hazards
65
(
3
),
1849
1861
.
S.
Simonovic
S. P.
2001
A decision support tool for evaluation of impacts of flood management policies
.
Journal of Hydrological Science and Technology
17
(
1–4
),
11
22
.
Anzaldi
G.
Rubion
E.
Corchero
A.
Sanfeliu
R.
Domingo
X.
Pijuan
J.
Tersa
F.
2014
Towards an enhanced knowledge-based decision support system (DSS) for integrated water resource management (IWRM)
.
Procedia Engineering
89
,
1097
1104
.
Banach
W.
2011
Determination of synthetic flood hydrograph in ungauged catchments
.
Infrastruktura I Ekologia Terenów Wiejskich
12
,
147
156
.
Bayat
B.
Mousavi
S. J.
Montazeri Nami
M.
2011
Optimization–simulation for short-term reservoir operation under flooding conditions
.
Journal of Water Supply: Research and Technology-AQUA
60
(
7
),
434
447
.
Bender
M. J.
Simonovic
S. P.
1994
Time Series Modelling for long-range stream flow forecasting
.
Journal of Water Resources Planning and Management, ASCE
120
(
6
),
857
870
.
Camara
A. S.
1990
Decision support system for estuarine water-quality management
.
Journal of Water Resources Planning and Management, ASCE
116
(
3
),
417
432
.
Chang
T. J.
Kleopa
X. A.
Teoh
C. B.
1996
Development of an Expert System for daily drought monitoring
.
Journal of Computing in Civil Engineering, ASCE
10
(
1
),
20
24
.
Cheng
C. T.
Chau
K. W.
2004
Flood control management system for reservoirs
.
Environmental Modeling & Software
19
(
12
),
1141
1150
.
Davis
J. R.
Nanninga
P. M.
Biggins
J.
Laut
P.
1991
Prototyped Decision Support System for analyzing impact of catchment policies
.
Journal of Water Resources Planning and Management, ASCE
117
(
4
),
399
414
.
Dingfei
L.
Stewart
T. J.
2004
Object-oriented decision support system modeling for multi-criteria decision making in natural resource management
.
Computers and Operations Research
31
,
985
999
.
Fernandez
C. J.
Trolinger
T. N.
2007
Development of a Web-based decision support system for crop managers
.
Agronomy Journal
99
(
3
),
730
737
.
Fotovatikhah
F.
Herrera
M.
Shamshirband
S.
Chau
K. W.
S.
Piran
M. J.
2018
Survey of computational intelligence as basis to big flood management: challenges, research directions and future work
.
Engineering Applications of Computational Fluid Mechanics
12
(
1
),
411
437
.
Freund
A.
Aydin
N. Y.
Zeckzer
D.
Hagen
H.
2017
A decision-support system for sustainable water distribution system planning
.
IEEE Computer Graphics and Applications
37
(
1
),
44
55
.
Grigg
N. S.
1996
Water Resources Management
.
McGraw Hill
,
New York
,
USA
.
Haktanir
T.
Citakoglu
H.
Acanal
N.
2013
Fifteen-stage operation of gated spillways for flood routing management through artificial reservoirs
.
Hydrological Sciences Journal
58
(
5
),
1013
1031
.
He
Y.
Xu
Q.
Yang
S.
Liao
L.
2014
Reservoir flood control operation based on chaotic particle swarm optimization algorithm
.
Applied Mathematical Modelling
38
(
17–18
).
doi:10.1016/j.apm.2014.02.030
.
Holmes
M. G. R.
Young
A. R.
Goodwin
T. H.
Grew
R.
2005
A catchment-based water resource decision-support tool for the United Kingdom
.
Environmental Modelling and Software
20
,
197
202
.
Jahanpour
M.
Sandoval-Solis
S.
Afshar
A.
2014
A Web-based application for optimization of single reservoir operation
.
Journal of the American Water Works Association
106
(
11
),
E509
E517
.
Kochilakis
G.
Poursanidis
D.
Chrysoulakis
N.
Varella
V.
Kotroni
V.
Eftychidis
G.
Makropoulos
C.
2016
A web based DSS for the management of floods and wildfires (FLIRE) in urban and periurban areas
.
Environmental Modelling & Software
86
,
111
115
.
Koutsoyiannis
D.
Karavokiros
G.
A.
Mamassis
N.
Koukouvinos
A.
Christofides
A.
2003
A decision support system for the management of the water resource system of Athens
.
Physics and Chemistry of the Earth
28
,
599
609
.
Li
X. Y.
Chau
K. W.
Cheng
C. T.
Li
Y. S.
2006
A Web-based flood forecasting system for Shuangpai region
.
37
(
3
),
146
158
.
Li
C. Q.
Jia
C.
Xu
B. S.
2011
Development of a decision support system for flood warning in Jinan city
.
219
,
1267
1270
.
Loucks
D. P.
French
P. N.
Taylor
M. N.
1985
Developing and implementing decision support systems
.
Water Resources Bulletin
31
,
571
582
.
B.
Zahraie
B.
Kerachian
R.
2010
A real-time operation optimization model for flood management in river-reservoir systems
.
Natural Hazards
53
(
3
),
459
482
.
McKinney
D. C.
Maidment
D. R.
Tanriverdi
M.
1993
Expert geographic information system for texas water planning
.
Journal of Water Resources Planning and Management, ASCE
119
(
2
),
170
183
.
Mioc
D.
Nickerson
B. G.
Anton
F.
Fraser
D.
MacGillivray
E.
Morton
A.
Liang
G.
2008
Web-GIS application for flood prediction and monitoring
.
WIT Transactions on Ecology and the Environment
118
,
145
154
.
Moridi
A.
Yazdi
J.
2017
Sediment flushing of reservoirs under environmental considerations
.
Water Resources Management
31
(
6
),
1899
1914
.
Mosavi
A.
Ozturk
P.
Chau
K. W.
2018
Flood prediction using machine learning models: literature review
.
Water
10
(
11
),
1536
.
Mysiak
J.
Giupponi
C.
Rosato
P.
2005
Towards the development of a decision support system for water resource management
.
Environmental Modelling and Software
20
,
203
214
.
Olukanni
D. O.
T. A.
Salami
A. W.
A. A.
2018
Optimization-based reliability of a multipurpose reservoir by genetic algorithms: Jebba Hydropower Dam, Nigeria
.
Cogent Engineering
5
(
1
),
1438740
.
Opricovic
S.
2011
Fuzzy VIKOR with an application to water resources planning
.
Expert Systems with Applications
38
,
12983
12990
.
O'Sullivan
J. J.
Ahilan
S.
Bruen
M.
2012
A modiﬁed Muskingum routing approach for ﬂoodplain ﬂows: theory and practice
.
Journal of Hydrology
470–471
,
239
254
.
Palmer
R. N.
Holmes
K. J.
1988
Operational guidance during droughts: expert system approach
.
Journal of Water Resources Planning and Management Division, ASCE
114
(
6
),
647
666
.
Palmer
R. N.
Tull
R. M.
1987
Expert system for drought management planning
.
Journal of Computing in Civil Engineering, ASCE
1
(
4
),
284
297
.
Paredes
J.
Andreu
J.
Solera
A.
2010
A decision support system for water quality issues in the Manzanares River (Madrid, Spain)
.
Science of the Total Environment
408
,
2576
2589
.
Simonovic
S. P.
1992
Reservoir systems analysis: closing gap between theory and practice
.
Journal of Water Resources Planning and Management, ASCE
118
(
3
),
262
280
.
Teegavarapu
R. S. V.
Simonovic
S. P.
2002
Optimal operation of reservoir systems using simulated annealing
.
Water Resources Management
16
(
5
),
401
428
.
Valerino
O. C. S.
Koike
T.
Yang
K.
Yang
D.
2010
Optimal dam operation during flood season using a distributed hydrological model and a heuristic algorithm
.
Journal of Hydrologic Engineering, ASCE
15
(
7
),
580
586
.
Wang
W.
Cui
Y.
Luo
Y.
Li
Z.
Tan
J.
2017a
Web-based decision support system for canal irrigation management
.
Computers and Electronics in Agriculture
.
https://doi.org/10.1016/j.compag.2017.11.018
Wang
W. C.
Chau
K. W.
Xu
D. M.
Qiu
L.
Liu
C. C.
2017b
The annual maximum flood peak discharge forecasting using Hermite projection pursuit regression with SSO and LS method
.
Water Resources Management
31
(
1
),
461
477
.
Yang
G.
Liu
L.
Guo
P.
Li
M.
2017
A flexible decision support system for irrigation scheduling in an irrigation district in China
.
Agricultural Water Management
179
,
378
389
.
Yong
P.
Yongri
H.
Wenlong
L.
Yongfeng
W.
2009
Development of Fengman Reservoir Flood Control Decision Support System Based on Web Service
.
Water Resources and Power
4
,
46
49
.
Yoo
C.
Lee
J.
Lee
M.
2017
Parameter estimation of the Muskingum channel flood-routing model in ungauged channel reaches
.
Journal of Hydrologic Engineering
22
(
7
),
05017005
.
Zargar
M.
Samani
H. M.
Haghighi
A.
2016
Optimization of gated spillways operation for flood risk management in multi-reservoir systems
.
Natural Hazards
82
(
1
),
299
320
.
Zeng
Y.
Cai
Y.
Jia
P.
Jee
H.
2012
Development of a Web-based decision support system for supporting integrated water resources management in Daegu city, South Korea
.
Expert Systems with Applications
39
(
11
),
10091
10102
.
Zhu
F.
Zhong
P. A.
Sun
Y.
Yeh
W. W. G.
2017
Real-time optimal flood control decision making and risk propagation under multiple uncertainties
.
Water Resources Research
53
(
12
),
10635
10654
.