the Creative Commons Attribution 4.0 License.

the Creative Commons Attribution 4.0 License.

# Study on the water resources optimal operation based on riverbed wind erosion control in West Liaohe River plain

### Sun Wanguang

### Li Chengzhen

### Fan Baoshan

Rivers are drying up most frequently in West Liaohe River plain and the bare river beds present fine sand belts on land. These sand belts, which yield a dust heavily in windy days, stress the local environment deeply as the riverbeds are eroded by wind. The optimal operation of water resources, thus, is one of the most important methods for preventing the wind erosion of riverbeds. In this paper, optimal operation model for water resources based on riverbed wind erosion control has been established, which contains objective function, constraints, and solution method. The objective function considers factors which include water volume diverted into reservoirs, river length and lower threshold of flow rate, etc. On the basis of ensuring the water requirement of each reservoir, the destruction of the vegetation in the riverbed by the frequent river flow is avoided. The multi core parallel solving method for optimal water resources operation in the West Liaohe River Plain is proposed, which the optimal solution is found by DPSA method under the POA framework and the parallel computing program is designed in Fork/Join mode. Based on the optimal operation results, the basic rules of water resources operation in the West Liaohe River Plain are summarized. Calculation results show that, on the basis of meeting the requirement of water volume of every reservoir, the frequency of reach river flow which from Taihekou to Talagan Water Diversion Project in the Xinkai River is reduced effectively. The speedup and parallel efficiency of parallel algorithm are 1.51 and 0.76 respectively, and the computing time is significantly decreased. The research results show in this paper can provide technical support for the prevention and control of riverbed wind erosion in the West Liaohe River plain.

- Article
(986 KB) - Full-text XML
- BibTeX
- EndNote

With the development and utilization of water resources,
partial rivers were dried up in arid or semi-arid area. After dried up, bare
riverbed becomes the new cradle land of sandstorm. Since 1972, the lower
reaches of the Yellow River has began to dry up (Gao, 1998). The reasons for
dry up in the watershed of the Yellow River can be concluded into three
aspects: (1) natural reasons, the Yellow River basin is located in an arid
and semi-arid region, at the same time, the river dry up may be aggravated by
global warming; (2) human activities, about 30 billion m^{3} runoff were
consumed annually (Fu and Chen, 2006), and large-scale soil and water
conservation measures decreased the streamflow (Shi et al., 2016). The dry up
of river basin has caused many serious problems, such as agricultural and
industrial output reduction, environmental degradation and water quality
deterioration (Fu and Chen, 2006; Wang and Li, 2007).

Wind erosion caused soil loss and then impoverishes lands (Lal, 2001). The blown dust emission from eroding lands impairs air quality (Lee et al., 2003) and poses a threat to human health (Wilson and Sprengler, 1996). The research of wind erosion prevention and control is mainly focused on the following two aspects: (1) the nonerodible grains are placed on the erodible surface to reduce the degree of wind erosion (Bagnold, 1941); (2) the use of vegetation cover or crop residues to prevent wind erosion (Englehorn et al., 1952; Zhang et al., 2003; Jia et al., 2015). The wind erosion control of bare riverbed or lakebed is quite different from sand land. Due to the technical difficulties, it is hard to implement these measures. Bao et al. (2006) estimated the rational water area and inflow of the Ebinur Lake for controlling wind erosion in the dried up lake basin (Bao et al., 2006). For intermittent dried up reach, there is no better erosion control measures so far.

In this paper we analyze the hydrological characteristics West Liaohe River plain and establish the optimal operation model for water resources based on river desertification control.

## 2.1 Water system of the West Liaohe River plain

The West Liaohe River Plain is formed by alluvial deposits of the West Liaohe
River and its main tributaries. West Liaohe River Basin area is 5.2×10^{4} km^{2}. The main rivers of the river system include the lower
reach of the Hongshan Reservoir in the Laoha River, the lower reach of the
Hairisu Water Diversion Project in the Xilamulun River, the Xinkai River, the
main stream of the West Liaohe River, the Jiaolai River and the Wulijimuren
River. From late 1990s, the streamway of West Liaohe River Plain frequent
drying up. After 2000, due to reduced precipitation, water stored by
reservoir, industrial and agricultural water use increased, the trend of
streamway dried up aggravated (Wang and Li, 2007).

This paper collected nearly 8 years (2006 ∼ 2013) of the West Liaohe River plain river hydrological station (including Taihekou, Daxingye, Talagan, Zongban, Sanhetang, Tongliao, Zhengjiatun) daily flow and sediment data (Fig. 1). The statistical results about runoff of each hydrological station in the West Liaohe River Plain are shown in Table 1.

As shown in Table 1, the annual average runoff of Taihekou (Xi) station is
14 657.3 × 10^{4} m^{3} and the annual average runoff of Taihekou
(Xin) is 3019.9 × 10^{4} m^{3}. The result indicate that only
about 17.1 % of total runoff of the Xilamulun River were diverted into the
Xinkai River. The annual average runoff of Talagan station on the Xinkai
River is 879.7 × 10^{4} m^{3}. Because of seepage loss along the
streamway, only about 29.1 % of runoff of Taihekou (Xin) can reach Talagan
station. The reach below Daxingye station of the Xinkai River were drying up
all the time. From 1998, Hongshan Reservoir, which located in the Laoha
River has been in a low water level running, so it did not discharge to
downstream and the streamway below Hongsha Reservoir has dried up all the
time. The runoff of main stream of the West Liaohe River mainly comes from
Taihekou (Xi) station. This runoff diverted into Mengjiaduan Reservoir
partly, deducting the seepage along the streamway, the remaining runoff were
diverted into Molimiao Reservoir and Yinliaojiqing channel. So the reach
below Molimiao Water Diversion Project was dried up all the time. The annual
average runoff of Zhengjiatun station is 4901.5 × 10^{4} m^{3}
and the runoff is mainly from Wulijimuren River.

According to the analysis results of hydrological characteristics, the reach of West Liaohe River Plain is divided into intermittent river reach and long dry reach. This paper focuses on the prevention and control of riverbed wind erosion in intermittent river reach.

The river of West Liaohe River Plain has a large amount of sediment.
According to the Taihekou (Xi) station monitoring data, the average sediment
concentration is 9.07 kg m^{−3}. The sediment concentration increases with
the increasing of flow rate. At the same time, the average of median size of
riverbed is 0.13 mm, so the change of riverbed erosion and deposition is very
severe (Fig. 2).

In the summer flood season in 2011, the maximum flow rate is 585 m^{3} s^{−1} at
the Taihekou (Xi) hydrological station. After flood season, the riverbed
were cut seriously. In the summer flood season in 2012, the maximum flow
rate is 151 m^{3} s^{−1} at the same cross-section, but the riverbed were
deposited seriously. We can come to the conclusion that the longitudinal
stability of the riverbed is poor. According to the results of the field
investigation, vegetation cover is an effective measure to resist wind
erosion in the bare river bed. Due to the longitudinal variation of
riverbed, it is difficult to form effective vegetation cover on the riverbed
surface. At the drying up time, bare riverbed has become the main source of
wind erosion. The optimal water resource operation measures can ensure that
the vegetation cover of the river bed is not destroyed frequently by the
sediment laden flow.

## 2.2 Modeling analysis

There are three large beside reservoirs (the total capacity of more than
1 × 10^{8} m^{3}) and four large water diversion project in the
West Liaohe River plain (Fig. 3). The runoff from upstream is controlled by
the water diversion project and diverted into beside reservoir partly. The
water resources stored in beside reservoir are used for local industrial and
agricultural water supply. Under the premise of ensuring the local
industrial and agricultural water supply, whether the water diversion
project can be optimized to reduce the water diversion frequency on the
riverbed, and then to achieve the purpose of wind erosion control.

## 2.3 Analysis of water flow evolution considering seepage loss

In the intermittent river reach, a thick unsaturated zone formed beneath the riverbed. When the flow of water through the riverbed, seepage loss is very strong. There are two stages of seepage loss in the intermittent river reach, one is initial infiltration stage and the other is stable infiltration stage. Under the initial infiltration stage, the infiltration rate is large, but the decay rate is faster, and the cumulative infiltration of this stage accounted for a large proportion (Cheng et al., 2015).

Due to the difficulty of accurately dividing the seepage stage, the two stages of seepage loss are considered as a whole. From 2006 to 2013, only 3 years (summer flood season in 2011 and 2012, spring flood season in 2013) that the runoff diverted into Xinkai River by Taihekou Water Diversion Project can reached Talagan Hydrological Station. The daily average flow rate of Taihekou (Xin) station and Talagan station is compared (Fig. 4).

By statistics, the runoff of Talagan station accounted for only 20.38 % of the Taihekou (Xin) station in summer flood season in 2011, and the same ratio is 20.07 % in summer flood season in 2012. This indicates that the water seepage loss from the Taihekou (Xin) station to Talagan station reaches about 80 % of the total runoff. In spring flood season in 2013, the runoff of Talagan station accounted for 49.34 % of the Taihekou (Xin) station, and the seepage loss is relatively small, mainly because of the seasonal frozen soil in the riverbed has not been fully melted.

The Muskingum method is used to calculate river water flow evolution in the West Liaohe River plain (Eq. 1). The water flow evolution parameters are the results (Table 2) of the “West Liaohe River flood control planning report” (Northeast investigation and Design Institute of Ministry of water resources, 2005).

where *Q*_{up, 1}, *Q*_{up, 2}, *Q*_{down, 1},
*Q*_{down, 2} are the inflow of beginning and end period of the
upper section, and the outflow of beginning and end period of the down
section respectively, m^{3} s^{−1}. *C*_{0}, *C*_{1}, *C*_{2} are
parameters. *K* is propagation time of the reach under steady flow, hour. *x*
is the flow proportion factor, in addition to reflect on the role of wedge
storage flow, also reflect the storage capacity of the river. Δ*t* is
the calculation period, hour.

Considering the optimization calculation is too large, and the degree of concern for water quantity is higher than the flow process, simplified flow evolution and seepage loss coupling and used the comprehensive seepage loss coefficient. However, it is necessary to distinguish between spring flood season and summer flood season, expression is as flow:

where *S*_{f, r} is the comprehensive seepage loss coefficient of
reach *r*. The comprehensive loss coefficient can be approximately considered
as linear correlation with the length of the river. Because of the lack of
measured hydrological data, the comprehensive loss coefficient of seepage of
other reach can be calculated by the correlation between the loss
coefficient and the length of the reach which is from Taihekou to Talagan
Water Diversion Project:

where *L*_{r} is the length of reach *r*, km.

To verify the comprehensive river seepage loss coefficient, the measured data of Daxingye station and Sanhetang station on the Xinkai River were selected from 1999 to 2000. The reach length is 36.7 km between the two hydrological stations. The comprehensive loss coefficients of spring and summer flood season were 0.78 and 0.65 respectively using Eq. (3). However, the measured values were 0.72 and 0.69 respectively, so the calculated result is close to the measured value. It is proved that the assumption of linear correlation between the loss coefficient and the length of the reach is reasonable. It is needed to be explained that the comprehensive seepage loss coefficient of reach was obtained under the condition that the river dried up for a long period previously. So the river water conditions will have a certain impact on the coefficient.

## 2.4 Hydraulic connection

The distribution of river flow is controlled by water diversion project. Through the water diversion project, river flow can be diverted into beside reservoir or lower reach. The hydraulic connections which take water diversion project as a node are described below.

where *Q*_{Taihekou_in}, *Q*_{Taihekou_out},
*Q*_{Taihekou_diversion} are inflow, outflow and diversion flow of
the Taihekou Water Diversion Project respectively, m^{3} s^{−1}.

where *Q*_{Talagan_in}, *Q*_{Talagan_out},
*Q*_{Talagan_diversion} are inflow, outflow and diversion flow of the
Talagan Water Diversion Project respectively, m^{3} s^{−1}.
*Q*_{Taihekou_diversion_evolution} is the diversion flow rate that
evolved from Taihekou to Talagan Water Diversion Project, m^{3} s^{−1}.

where *Q*_{Sujiapu_in}, *Q*_{Sujiapu_out},
*Q*_{Sujiapu_diversion} are inflow, out flow and diversion flow of
Sujiapu Water Diversion Project respectively, ,m^{3} s^{−1}.
*Q*_{Taihekou_out_evolution} is the outflow rate that evolved from
Taihekou to Sujiapu Water Diversion Project, m^{3} s^{−1}.

where *Q*_{Zongban_in}, *Q*_{Zongban_out}, *Q*_{Zongban_diversion} are inflow,
outflow and diversion flow of Zongban Water Diversion Project, m^{3} s^{−1}.
*Q*_{Sujiapu_out_evolution}
is the outflow rate that evolved from Sujiapu to Zongban Water Diversion
Project, m^{3} s^{−1}.

According to the analysis results of hydraulic system, the system is a mixed type system and the hydraulic connection between upstream and downstream is closely. The distribution of water resources were controlled by four water diversion projects, so the dimension of the system space is four dimensional. Based on the results of hydrological statistical analysis, the Water Diversion Project of Talagan and Zongban had no outflow in recent 8 years (from 2006 to 2013). So we use regular operation mode for the Water Diversion Projects of Talagan and Zongban, and the two Water Diversion Projects no longer participate in optimal operation. Specific to Zongban Water Diversion Project as an example:

- 1.
When

*Q*_{Zongban_in}>*Q*_{Zongban_diversion_max}, then*Q*_{Zongban_diversion}=*Q*_{Zongban_diversion_max}, and then ${Q}_{\mathrm{Zongban}\mathrm{\_}\mathrm{out}}={Q}_{\mathrm{Zongban}\mathrm{\_}\mathrm{in}}-{Q}_{\mathrm{Zongban}\mathrm{\_}\mathrm{diversion}\mathrm{\_}\mathrm{max}}$. - 2.
When

*Q*_{Zongban_in}≤*Q*_{Zongban_diversion_max}, then*Q*_{Zongban_diversion}=*Q*_{Zongban_in}, and*Q*_{Zongban_out}=0. Where*Q*_{Zongban_diversion_max}is the maximum capacity of diversion flow of Zongban Water Diversion Project, m^{3}s^{−1}.

After using regular operation mode for the Water Diversion Projects of Talagan and Zongban, the four dimensional optimization problem can be reduced to two dimensional optimization problem, and only the Water Diversion Projects of Taihekou and Sujiapu can be seen as targets of the optimal operation. The outflow and diversion flow are optional state variables for every water diversion project. In order to meet the threshold setting and the realization of the target, the diversion flow is chosen as the state variable in Taihekou Water Diversion Project and the outflow is chosen as state variable in Sujiapu Water Diversion Project.

## 3.1 Objective function

The inflow of Taihekou Water Diversion Project is mainly concentrated in the
spring flood season (from 10 March to 20 April) and summer flood season
(from 25 June to 20 August). Other times the riverbed is in the drying
up state. The maximum inflow were 147 and 1074 m^{3} s^{−1} of the
Taihekou Water Diversion Project in spring flood season and summer flood
season respectively.

Because the water diversion project has no capacity for runoff regulation, the inflow either diverted into beside reservoir or discharged into lower reach. Due to the extreme shortage of local water resources, the water eventually was diverted into beside reservoir regardless of diversion flow or outflow. Therefore, it is a prerequisite to optimize the operation of water resources to meet the requirements of the local industry and agriculture. Based on the prevention and control of wind erosion of riverbed, it is beneficial that the river discharge or dry up for a long time. If the upstream flow is less, the flow is firstly diverted into nearest beside reservoir and reduce interference on the downstream of the river. If the upstream flow is larger, the flow is diverted into long distance beside reservoir preferentially, concentrated flow and reduce the frequency of water through of riverbed. Therefore, as the objective function, it is needed to consider diversion flow or outflow and river length. Specific expressions are as follows:

where obj is objective function; *j* is the Number of reservoir; obj_{j} is
objective function value of *j* reservoir water storage. *L*_{j} is the
distance from Taihekou Water Diversion Project to *j* reservoir, km;
*V*_{j, cap} is utilizable capacity of *j* reservoir,
10^{4} m^{3}; *I*_{j} is the volume of water that had diverted into *j*
reservoir in calculation period, 10^{4} m^{3}; *P* is penalty
coefficient; *Q*_{j, t} is the flow rate that diverted into *j* reservoir.
${Q}_{j,\phantom{\rule{0.125em}{0ex}}s,\phantom{\rule{0.125em}{0ex}}\mathrm{max}}$ is the maximum flow rate of flood,
m^{3} s^{−1}; *Q*_{j, thr} is the lower threshold of flow rate
that diverted into *j* reservoir, m^{3} s^{−1}; sign is indicator of
status.

The objective function includes two aspects: (1) the objection of water
volume that diverted into beside reservoir and (2) the objection of wind
erosion prevention and control of riverbed. The objective function is
essentially a function of the reach length (*L*_{j}) and the amount of water
diverted into beside reservoir (*I*_{j}). ${V}_{j,\phantom{\rule{0.125em}{0ex}}\mathrm{cap}}/{I}_{j}$ is a
dimensionless variable coefficient, when the amount of water diverted into
*j* reservoir increases, the coefficient becomes smaller. Through the
objective function can ensure the amount of water diverted into beside
reservoir, and thus guarantee the local industrial and agricultural water
supply.

The lower threshold flow is one of the key variables of the objective function. Through this variable, it can avoid the interference of high-frequency of small flow rate to the lower reach. To obtain the lower threshold flow, we can use two methods of optimization and simulation. If the optimization method is used to solve the problem, it is necessary to add two dimension on the basis of the original two dimensional optimization problem; and the use of simulation method to solve, we can learn from the operation experience, given a reasonable lower threshold flow. This paper uses simulation method to obtain the lower threshold of diversion flow of Taihekou Water Diversion Project and outflow of Sujiapu Water Diversion Project respectively. The lower threshold flow setting is directly related to the volume of diversion flow of Talagan and Zongban Water Diversion Project. The optimal value of the lower threshold of the diversion flow is based on the measured water diversion of the Talagan and Zongban Water Diversion Project.

In this paper, the problem of optimal flood control is not studied, and the flood discharge of each river is given in the form of constraint.

## 3.2 Constraints

### 3.2.1 Water balance of reservoir

where *V*_{j, t} and ${V}_{j,\phantom{\rule{0.125em}{0ex}}t+\mathrm{1}}$ are water storage at the beginning and
end of the period *t* respectively in reservoir *j*; *I*_{j, t} is runoff of
reservoir *j* in period *t*; *W*_{j, t} is water supply of reservoir *j* in
period *t*; *E*_{j, t} is the loss of evaporation and seepage of *j*
reservoir in period *t*; *A*_{j, t} is surplus water of *j* reservoir in
period *t*.

### 3.2.2 Constraint of discharge capacity

where *Q*_{r, t} is the flow rate of reach *r* in period *t*,
m^{3} s^{−1}; *Q*_{r, max} is the discharge capacity of reach
*r*, m^{3} s^{−1};

### 3.2.3 Constraint of upper bound of reservoir water storage

where *V*_{j} is water storage of reservoir *j* in period *t*,
10^{4} m^{3}; *V*_{j, nor} is upper bound of water storage of
reservoir*j*, 10^{4} m^{3}.

## 3.3 Search method

### 3.3.1 Constraint handling method

In the process of water resources optimal operation, the constrained optimization problem is transformed into unconstrained optimization problem by penalty function generally. In this paper, the discharge capacity and the upper bound of reservoir water storage are all mandatory constraint. When the variables violate the constraints, the penalty term is added on the basis of the results of the original objective function. Specific expressions are as follows:

where *F*_{1} and *F*_{2} are the penalty coefficient of the flow rate and
water storage respectively; obj^{′} is the objective function value that is
added by the penalty value.

### 3.3.2 Solving process

Water diversion project itself does not have the ability to regulate runoff, only transform the inflow into diversion flow or outflow. Importantly there is a time effect of water flow evolution, that is, there is post-efficiency, which is essentially different from reservoirs. Therefore, the solution method of the optimal operation of the water diversion project group is quite different from the reservoir group.

In order to simplify the problem solving, multi-dimensional optimization problems are usually transformed into one-dimensional optimization problems, such as DPSA (Dynamic Programming Successive Approximation) in dynamic planning (Opan, 2010; Yi et al., 2003). Multi-stage optimization can also be translated into a single two-stage optimization problem, such as POA (Progressive Optimality Algorithm) in dynamic plan (Liu et al., 2011; Guo et al., 2011). If DPSA and POA are combined (DPSA-POA), the multi-dimensional and multi-stage optimization problems can be transformed into one-dimensional and two-stage optimization problems (Bai et al., 2015; Zhang et al., 2016), and the difficulty of solving optimization problems reduced greatly, at the same time, the problem of post-efficiency can be solved effectively.

To solve the problem of the water resources optimal operation based on riverbed wind erosion control in West Liaohe River plain, and improve computational efficiency, this paper presents a “multi-core parallel solution algorithm for optimal operation of water diversion project group”.

From the above analysis, we can see that the upper and lower reaches are
closely linked to each other of the optimal operation system in this paper,
and two independent state variables find the optimal combination of states
through the DPSA method under the framework of POA to get the optimal
operation scheme. According to the relationship between upstream and
downstream, Taihekou Water Diversion Project is the first level, Sujiapu
Water Diversion Project is the final level. Under the framework of POA, we
only change the state variable of the last stage *i* at a time, and the first
state variable is fixed. The full range of two complete series are brought
into the calculation method of objective function. According to the
calculation method of water flow evolution, the diversion flow and outflow of
each water diversion project are calculated, and then the objective function
value is calculated. When the traversal of the final level at time *i* is
completed, the state variable of its parent is changed. When all traversal
completed at time *i*, then $i=i+\mathrm{1}$, until all the time were completed.
The whole process is repeated *n* times, until the difference between the two
calculation results is less than the set threshold, you can stop the
calculation. Calculation process is shown in Fig. 5. In the figure, *M* and
*K* are the number of diverting states of Tahekou and Sujiapu Water Diversion
Project, respectively, and their values can be selected comprehensively
according to the state change field and the calculation precision.

From the entire calculation process, the calculation is the largest in the final cycle. It is necessary to calculate the evolution of water flow under fixed stages of two water diversion project. It is necessary to calculate the amount of diversion flow and out flow of each water diversion project and calculate the entire objective function value. In this paper, the last stage is changed into parallel computing mode to balance the load of each thread, so as to avoid the cost of thread synchronization at the end of the cycle. The Fork/Join framework is a framework for performing parallel tasks. In the final cycle, each subtask is independent of each other, so the task can be distributed to multiple threads, and the results of multiple subtasks can be combined into a total calculation result.

The performance of parallel algorithm is evaluated by speedup ratio and parallel efficiency (Peng et al., 2014), and the expression is as follows:

where *S*_{p} is speedup ratio; *T*_{s} is serial computing
time consuming; min; *T*_{p} is parallel computing time consuming;
min; *E*_{p} is parallel efficiency; *P* is processor number.

Taking Taihekou Water Diversion Project daily measured flow from 2006 to 2013 as input condition, on the basis of guaranteeing the total water separation of various reservoirs, avoid the destruction of riverbed vegetation by the frequent overflow of the river through the optimal operation of Taihekou and Sujiapu Water Diversion Project, and achieve the goals of riverbed wind erosion control. The combination of the different thresholds of water diversion at the Taihekou and the Sujiapu Water Diversion Project are analyzed. Take Talagan and Zongban Water Diversion Project measured water diversion volume value as the verification. The results are shown in Table 3.

It can be seen from the calculation results in the table that the lower
threshold of water diversion for Taihekou Water Diversion Project is set to
150 m^{3} s^{−1} and the release threshold of Sujiapu Water Diversion
Project terminal set to 15 m^{3} s^{−1} is more appropriate (Scheme 2
in Table 3). The simulated diversion water volume of the Talagan Water
Diversion Project is 6636×10^{4} m^{3}, and the measured value is
7038×10^{4} m^{3}. The simulated diversion water volume of the
Zongban Water Diversion Project is 27 919×10^{4} m^{3}, and the
measured value is 27 001×10^{4} m^{3}. The simulation results and
the measured values are relatively close.

Under scheme 2, Taihekou Water Diversion Project has diverted into Xinkai
River for 3 years, summer flood in 2008, summer flood in 2011 and summer
flood in 2012 respectively. The number of draining days at Talagan Water
Diversion Project is 29 days. In the measured value, Taihekou Water Diversion
Project has diverted into Xinkai River for 4 years, spring flood in 2006,
summer flood in 2011, summer flood in 2012 and summer flood in 2013
respectively. The number of draining days at Talagan Water Diversion Project
is 44 days. The results of the comparison between the calculated and the
measured are shown in Fig. 6. In the spring flood season of 2006, the maximum
inflow of Taihekou Water Diversion Project was 132 m^{3} s^{−1}. In the
actual operation, the maximum diversion flow into Xinkai River is
19.4 m^{3} s^{−1}, and the amount of water diverted into Xinkai River
is 503×10^{4} m^{3} in the whole spring flood season. Leakage
along the river consumes the vast majority of water. While not meeting the
purpose of diversion water storage, the large amount of sediment carried by
the water current also undermined the growth environment of the bed
vegetation. After the optimal operation, only when the inflow of Tahekou
Water Diversion Project exceeds the set threshold, it will be diverted into
Xinkai River. Taking the summer flood season in 2011 as an example, the
maximum of inflow of the Taihekou Water Diversion Project is
1074 m^{3} s^{−1}, and the total flood runoff in this session is
25 320×10^{4} m^{3}. After optimization calculation, the maximum
flow diverted into Xinkai River is 698 m^{3} s^{−1} (the measured value
is 489 m^{3} s^{−1}), and the amount of water diverted into Xinkai
River is 11 900×10^{4} m^{3} (measured value is 8655×10^{4} m^{3}). The maximum diversion flow and total amount of diversion
water have been greatly increased compared with the measured values.

Under scheme 2, Zongban Water Diversion Project has inflow water for 7 years and the total number of inflow days is 176 days. In the measured values, Zongban Water Diversion Project also has inflow water for 7 years and the total number of inflow days is 200 days.

The results of comparison between the optimized operation and the measured data show that the number of years of water passing through the Zongban Water Diversion Project is the same with the measured value and the number of years that the Taihekou Water Diversion Project diverted into Xinkai River decreased by 1 year after the optimal operation is implemented. For the number of draining days, the number of draining days at Talagan and Zongban Water Diversion Project decreased to some extent contrast with the measured value, but the reduction in the number of draining days at Talagan Water Diversion Project was relatively large.

Through the above analysis we can see that the setting of the lower threshold of diversion flow and the amount of water diverted into the reservoir are conflicting. On the basis of guaranteeing the amount of water diverted into the reservoir, setting a reasonable lower threshold of flow diverted into the new Xinkai River of Taihekou Water Diversion Project and lower threshold of outflow of the Sujiapu Water Diversion Project can make the Xinkai River from Taihekou to Talagan Water Diversion Project (river length of 85 km) and the West Liaohe River from Sujiapu to Zongban Water Diversion Project (length of 65.7 km) through a larger water flow and control the lower flow in the reach. Reduce the frequency of river flow as much as possible, and reduce the frequent damage of water flow to the vegetation cover of the riverbed. The goal of water resources optimization of West Liaohe River Plain based on the prevention and control wind erosion of riverbed has been achieved.

By optimizing the operation, we can summarize the following rules: (1) When
the inflow of Tahekou Water Diversion Project is greater than or equal to
150 m^{3} s^{−1}, the Xinkai River floodgate is opened to
preferentially divert the water into Xinkai River, and more water from the
same flood can be diverted into the Xinkai River. (2) When the inflow of
Tahekou Water Diversion Project is less than 150 m^{3} s^{−1}, and the
flow rate is less than 15 m^{3} s^{−1} after it evolves to Sujiapu
Water Diversion Project, it should be diverted into Mengjiaduan reservoir as
far as possible. When the inflow of Sujiapu Water Diversion Project is
greater than or equal to 15 m^{3} s^{−1}, give priority to the
downstream drainage.

The optimized operation solution test environment is a notebook computer. CPU type is CORE i7, dual-core, 4G memory. After calculation, the serial algorithm takes 14.61 min and the parallel algorithm takes 9.66 min. The speed ratio of the parallel computation is 1.51, and the parallel efficiency is 0.76. It can be seen that the parallel computation of multi core shortens the time consuming and improves the computational efficiency.

- 1.
According to the measured data and actual operation, the West Liaohe River Plain water conservancy project system is simplified to be operation system that is composed of the four water diversion project and three reservoirs. On the basis of this, through the rules and regulations of Talagan and Zongban Water Diversion Projection, the four-dimensional optimization problem is further reduced to a two-dimensional optimization problem.

- 2.
An optimal water resources operation model of West Liaohe River Plain based on the prevention and control of wind erosion of riverbed was established. The model objective function involves the factors such as the amount of water diverted into the reservoir, the length of the river reach and the lower threshold of the flow rate, and on the basis of ensuring the requirements of water diversion in each reservoir, make the river flow through the reach in a short period of time and avoid the destruction of vegetation above the riverbed by the frequent overflow of the river.

- 3.
Proposed a multi-core parallel solution method of water resources optimal operation in West Liaohe River Plain. Find the optimal combination of states by DPSA method under the framework of POA and adopting the Fork/Join mode design parallel computing algorithm in OpenPM programming model, we proposed the calculation flow.

- 4.
The calculation results show that on the basis of ensuring the demand of water diversion into Talagan and Molimiao reservoirs, the threshold of diversion flow from Taihekou Water Diversion Project to Xinkai River is 150 m

^{3}s^{−1}, while the threshold of outflow of the Sujiapu Water Diversion Project is 15 m^{3}s^{−1}. The optimal operation effectively reduced the water diversion frequency in the Xinkai River reach from Taihekou Water Diversion Project to Talagan Water Diversion Project. Using parallel algorithms, speedup and parallel efficiencies were 1.51 and 0.76, respectively, and computational efficiency improved significantly. - 5.
The research results of this paper show that it is feasible to achieve the prevention and control of wind erosion of riverbed in the intermittent over-water reach of West Liaohe River Plain through the optimized water resources operation.

The hydrological data used in this study all come from the hydrologic Yearbook (Hydrology data of Liaohe River Basin). According to the People's Republic of China hydrological regulations, hydrological data are not allowed to be made public.

The authors declare that they have no conflict of interest.

This article is part of the special issue “Innovative water resources management – understanding and balancing interactions between humankind and nature”. It is a result of the 8th International Water Resources Management Conference of ICWRS, Beijing, China, 13–15 June 2018.

The study is financially supported by the National Non-Profit Research
Program of China (No.201401015).

Edited by: Depeng Zuo

Reviewed by: two anonymous referee

Bagnold, R. A.: The physics of blown sand and desert dunes, Methuen, London, 265 pp., 1941.

Bai, T., Chang, J., Chang, F., Huang, Q., Wang, Y., and Chen, G.: Synergistic gains from the multi-objective optimal operation of cascade reservoirs in the Upper Yellow River basin, J. Hydrol., 523, 758–767, https://doi.org/10.1016/j.jhydrol.2015.02.007, 2015.

Bao, A., Mu, G., Zhang, Y., Feng, X., Chang, C., and Yin, X.: Estimation of the rational water area for controlling wind erosion in the dried-up basin of the Ebinur Lake and its effect detection, Chinese Sci. Bull., 51, 68–74, https://doi.org/10.1007/s11434-006-8209-9, 2006.

Cheng, L., Wang, Z., Hu, S., Wang, Y., Jin, J., and Zhou, Y.: Flood routing model incorporating intensive streambed infiltration, Sci. China Earth Sci., 58, 718–726, https://doi.org/10.1007/s11430-014-5018-x, 2015.

Englehorn, C. L., Zingg, A. W., and Woodruff, N. P.: The effects of plant residue cover and cold structure on soil losses by wind, Soil Sci. Soc. Am. J., 16, 29–33, https://doi.org/10.2136/sssaj1952.03615995001600010010x, 1952.

Fu, G. and Chen, S.: Water crisis in the Yellow River: facts, reasons, impacts, and countermeasures, Water Practice and Technology, 1, 1–8, https://doi.org/10.2166/WPT.2006028, 2006.

Gao, Y.: Analysis of reasons for the Yellow River's dry-up and its eco-environmental impacts, J. Environ. Sci., 10, 357–364, 1998.

Guo, S., Chen, J., Li, Y., Liu, P., and Li, T.: Joint operation of the multi-reservoir system of the Three Gorges and the Qingjiang cascade reservoirs, Energies, 4, 1036–1050, https://doi.org/10.3390/en4071036, 2011.

Jia, H., Wang, G., Guo, L., Zhuang, J., and Tang, L.: Wind erosion control utilizing standing corn residue in Northeast China, Soil Till. Res., 153, 112–119, https://doi.org/10.1016/j.still.2015.05.009, 2015.

Lal, R.: Soil degradation by erosion, Land Degrad. Dev., 12, 519–539, https://doi.org/10.1002/ldr.472, 2001.

Lee, H. N., Tanaka, T., Chiba, M., and Igarashi, Y.: Long range transport of Asian dust from dust storm and its impact on Japan, Water, Air, & Soil Pollution: Focus, 3, 231–243, https://doi.org/10.1023/A:1023254910362, 2003.

Liu, X., Guo, S., Liu, P., Chen, L., and Li, X.: Deriving Optimal Refill Rules for Multi-Purpose Reservoir Operation, Water Resour. Manag., 25, 431–448, https://doi.org/10.1007/s11269-010-9707-8, 2011.

Northeast investigation and Design Institute of Ministry of water resources: West Liaohe River flood control planning report, Changchun, Report No.: 900-CH7-13, 50 pp., 2005.

Opan, M.: Irrigation-energy management using a DPSA-based optimization model in the Ceyhan Basin of Turkey, J. Hydrol., 385, 353–360, 2010.

Peng, A.-B., Peng, Y., and Zhou, H.,: Multi-core parallel computation for deriving joint operating rule curves in multi-reservoir system under the condition of inter-basin water transfer, J. Hydraul. Eng., 45, 1284–1292, https://doi.org/10.13243/j.cnki.slxb.2014.11.003, 2014.

Shi, H., Li, T., Wang, K., Zhang, A., Wang, G., and Fu, X.: Physically based simulation of the streamflow decrease caused by sediment-trapping dams in the middle Yellow River, Hydrol. Process., 30, 783–794, https://doi.org/10.1002/hyp.10649, 2016.

Wang, X. and Li, L.: The drying-up of the West Liaohe River and the countermeasures to the problem, Journal of Arid Land Resources and Environment, 21, 79–83, 2007.

Wilson, R. and Sprengler, J. D.: Particles in our air: concentrations and health effects, Harvard University Press, Cambridge, Mass, 1996.

Yi, J., Labadie, J. W., and Stitt, S.: Dynamic Optimal Unit Commitment and Loading in Hydropower Systems, J. Water Res. Pl., 129, 388–398, https://doi.org/10.1061/(ASCE)0733-9496(2003)129:5(388), 2003.

Zhang, C. L., Zou, X. Y., Dong, G. R., and Lu, Y. Z.: Wind tunnel studies on influences of vegetation on soil wind erosion, J. Soil Water Conserv., 17, 31–33, 2003.

Zhang, W., Liu, P., Chen, X., Wang, L., Ai, X., Feng, M., Liu, D., and Liu, Y.: Optimal operation of multi-reservoir systems considering time-lags of flood Routing, Water Resour. Manag., 30, 523–540, https://doi.org/10.1007/s11269-015-1175-8, 2016.