Refined Allocation of Water Resources in Pishihang Irrigation Area by Joint Utilization of Multiple Water Sources

Abstract

Refined allocation of water resources is an important means of sustainable water resources utilization. Based on General Water Allocation and Simulation (GWAS), this study uses a Geographic Information System (GIS) to construct spatial topological relationships. A fairness optimal minimum was set as the objective function. Total quantity control, water supply potential, and quality-divided water supply were set as constraint conditions. Considering the dynamic mutual-feedback relationship between the middle-lower-reaches reservoir and the upstream reservoir, this study refines the allocation of water resources combined with the characteristics of “long cane knots melons” in the Pishihang irrigation area. Results showed that at 50%, 80%, and 90% frequencies in the base year, 2025, and 2035, respectively, the water deficient ratio is 0. For continuous drought years at 90% frequency, all water users are faced with different degrees of water shortage. In water source structures, water diversion in the irrigated area is the largest, followed by local surface water; reclaimed water and shallow groundwater are used as supplements. In the case of consecutive drought years, the water shortage degree can be reduced through rational development of local water and additional external water transfer. The model has thus been well applied. This study provides a more accurate method for optimizing water resources allocation.

1. Introduction

Because of global climate change issues, the spatiotemporal distribution of water resources is uneven [1]. Furthermore, the discrepancy between the decreasing amount of water resources and the sharp increase in water consumption has become increasingly prominent. The efficient use and allocation of water resources have become one of the most economical methods for resolving this discrepancy [2].

The optimal allocation of water resources in irrigation areas began with the reservoir optimal scheduling issue proposed by Masee in the 1940s [3,4]. With the development of mathematical programming and simulation technology and its application in the field of water resources, the research results on optimal allocation of water resources at home and abroad are increasing. Romjin and Taminga [5] established a multilevel model of the water resources allocation problem, considered the multifunctionality of water and the relationship between multiple interests, emphasized the cooperation between decision makers and decision analysts, and reflected the multiobjective and hierarchical structure of the water resources allocation problem. Based on the semidistributed water balance model and using a Geographic Information System (GIS), Fortes et al. [6] established the GISAREG model, an irrigation system simulation model meant to improve water use efficiency. Wang and Wang [7] suggested the “binary water cycle” theory. Coupling it with the distributed hydrological model, the rational allocation model of water resources, and the multiobjective decision analysis model, the researchers were able to develop the “natural–artificial” binary water cycle model, which has been applied to water management in many irrigation areas. The “long vines and melons” in the Pishihang Irrigation District is a unique water delivery link method (Section 2.1 for details), and some studies have been carried out to optimize the allocation of water resources. For example, Huang [8] and Wang [9] carried out stochastic simulation optimization on irrigation areas based on the principle of system optimization, and the results showed that this method can better improve the water supply guarantee rate of irrigation areas. Lu et al. [10] used large-scale system decomposition and coordination theory and simulation iteration technology to optimize the water distribution of irrigation areas, and the results showed that the water allocation method recommended can meet multiple objectives of irrigation areas.

In terms of model, although the above allocation research models have completed the optimal allocation of water resources to a certain extent, some model research divisions are too rough, the allocation results are vague, and the allocation accuracy is low, making it difficult to reflect the difference in water supply in a certain period of time. In addition, it is difficult for the conventional model to meet the requirements of refined management of water resources in terms of the mutual feedback of the main body of the water resources system configuration, the complexity of the allocation process, and the dynamics of both supply and demand. From the related research on water resources allocation in the Pishihang Irrigation District, the existing research does not consider the auxiliary regulation effect of small and medium-sized water reservoirs on water resources allocation due to the rough division of units. There are certain deviations between the above research results and the actual situation, and the conclusions obtained also have certain limitations.

Therefore, in order to improve the accuracy of water resources allocation in the irrigation area, we took the Pishihang irrigation area as the research object. The paper used GIS to construct the detailed topological relationship of the area and reflect the “long cane knots melons” irrigation system of the area. Through a long allocation series, the study aimed to understand the water supply and regulation capacity of the six major upper-reaches reservoirs. It was focused on the area’s water source conditions, reservoir characteristics, the water inflow from inside and outside of the irrigation area, the lower-reaches water replenishment, external water transfer, reclaimed water, water division, and drainage. The GWAS model was used to construct a long-sequence water resources allocation model based on the water cycle and water balance of the irrigation area. The model considered the reverse regulatory effect that the middle- and lower-reaches reservoirs may have on the upper-reaches reservoirs. Based on this, the study conducted water resources allocation of multiwater sources, multiobject, and multiusers in irrigation.

There are several advantages and innovations in this study. (1) Firstly, by using the spatial topological relationship, the paper considers all reservoirs, ponds, and dams in the irrigation area in detail, reducing the partition unit scale and enriching storage nodes, thus improving simulation accuracy; (2) secondly, by controlling the filling time of the sluice gate and the amount of water discharged from the reservoir, the small and medium-sized reservoirs in the downstream of the irrigation area can play an auxiliary regulation role in the allocation of water resources, thus making up for the insufficient consideration of the small reservoirs in the Pishihang irrigation area in previous studies; (3) finally, two types of extremely dry years are selected to determine the impact of continuous drought on water distribution in irrigation areas. The results provide theoretical support for optimal water resources allocation in the Pishihang irrigation area.

2. Materials and Methods

2.1. Overview of the Study Area

The irrigation area was established in 1958. It is situated in the central western part of the Anhui Province and the central southern part of Henan Province, China [11]. It envelops a mountainous area, bordering the Dabie Mountains and crossing the irrigation areas of the Yangtze River Basin and the Huaihe River Basin. The area’s exact location is provided in Figure 1. The main rivers in this area are the Pi River, the Shi River, and the Hangbu River. Among the three rivers, the Pi River and the Shi River flow from south to north and into Huaihe River. On the other hand, the Hangbu River flows from west to east and into Chaohu Lake. The control irrigation area is 14,739.7 km², while the designed irrigation area is 733,000 hm². The Pishihang area is one of three super-large irrigation areas in China [12]. It is mainly composed of mountainous geomorphic units, with higher segments in the west and lower ones in the east [13]. The area belongs to the transitional climate zone that is between the subtropical and warm temperate zone and has four distinct seasons, moderate rainfall, and sufficient light. However, the north and south airflow converges here, resulting in large interannual rainfall variation and uneven distribution during the year, making this an area prone to frequent floods and droughts. The average annual rainfall is 1119.0 mm, while the evaporation is 806.9 mm. The temperature reaches its peak between June and August, with the average yearly temperature between 15 and 16 °C [14].

“Long cane knots melons” (Figure 1b) is a water conveyance link method used in the Pishihang irrigation area [15]. The water conservancy system consists of three components, namely the canal head water diversion or water storage project, the water transfer and distribution canal system (cane), and the small reservoirs, ponds, and dams (melons) in the irrigation area [16]. During the non-irrigation season, the canals are used to divert water to irrigate “melons” (small reservoirs and mountain ponds) and thus supplement the water storage deficiency. During the irrigation season, when the water supply is tight, the canal water and the melon water are simultaneously used to irrigate the fields and supply water, thus improving the irrigation and water supply guarantee rate [17]. Reservoirs, ponds, and dams in the middle and lower reaches can collect local precipitation runoff and store the water inflow from the upper-reaches reservoirs during the non-irrigation season. Therefore, these supplies have a vital role in supplying water in an emergency or when precipitation is low, thus assisting reverse regulation under the regulation of upper-reaches reservoirs.

In general, large reservoirs begin storing water from September until the end of October. There are 23 medium reservoirs and several ponds and dams in the middle and lower reaches of our research area. After October, the small and medium reservoirs, ponds, and dams are filled with water. The peak irrigation period is between April and September the following year, in which the planting period is between 1 April and 31 May, and the harvest period is between 5 July and 31 September. The irrigation water is mainly the local reservoir water. The water source structure is divided into the irrigation area water diversion (from the 6 major reservoirs), external water transfer (transferred from other areas through the water transfer projects), local surface water, groundwater, and reclaimed water.

2.2. Data Sources

The data used in this study include the precipitation data from the Pishihang irrigation area, the water inflow data from the 6 upper-reaches reservoirs gathered from 1960 to 2015, and the water diversion from each canal head and water supply in each irrigation area between 2012 and 2018. Detailed data sources are provided in

Table 1. Data sources.

Variable	Data Sources
Precipitation	China Meteorological Data Network, National Meteorological Information Center
Water inflow and water diversion of reservoirs	General Administration of Pishihang irrigation area, Anhui Province; Water Conservancy Census; Statistical Yearbooks
Water supply	Water resources bulletin

. The data cited in this paper were obtained solely from official departments.

2.3. Development of GWAS Model

2.3.1. Model Introduction

The two basis notions behind the GWAS model are the refined water allocation and simulation model (WAS) and the river water demand-oriented water allocation and calculation framework (EWAS) [18]. The WAS model comprises a water cycle simulation module, a water resources allocation module, a reclaimed water module, an industrial theoretical water consumption module, and a statistical analysis module. The WAS model’s operation process is composed of several steps. The water cycle module is firstly used to simulate the regional water resources changes. The industrial water consumption module is then used to analyze the regional theoretical water consumption, while the reclaimed water module stimulates the amount of recycling. After, the WAS model uses the allocation strategy to calculate the rational allocation of water resources. Simultaneously, data on artificial intake, utilization, consumption, and drainage are sent back to the water cycle module to simulate the regional water resources changes for the upcoming period in real time. This process is repeated until the final optimal scheme is proposed [19,20,21,22]. The EWAS represents a simulation calculation framework and is an extension of the water resources allocation model. In general, it does not change the internal calculation structure of the model. Rather, the framework incorporates the river section into the traditional allocation with the help of virtual units and water sources. This enables a more refined overview of the section ecological unit and allows for a comprehensive allocation of water resources centered around the ecological flow of the river [23].

Based on the WAS model, the GWAS model also incorporates the EWAS component into its structure. The GWAS model also improves the process of dealing with the water quantity in the river channel. This modification enables the GWAS model to tackle both the conventional socioeconomic water allocation and the ecological scheduling and power generation scheduling issues. To form a rule-based scheduling scheme, the reservoir operation rules are incorporated into the model for simulation calculation. On the other hand, optimal scheduling objectives are set and optimal simulation is conducted through algorithms to form new optimal schemes.

2.3.2. Allocation Principles

The optimal allocation of Pishihang’s water resources adopts the following rules:

(1)

The priority of water resources utilization schemes:

• The use of surface water should be prioritized for the agriculture and ecological environment;
• Taking into account the landscape environment, water diversion should be rationally used for agricultural production, urban and rural living, and production;
• Boundary water pumping should be developed for agricultural irrigation water along rivers, lakes, and canal ends;
• For emergency water, it should be possible to transfer water from outside the irrigation area and use the water diverted from the Yangtze River to the Huaihe River, and the Yangtze River to the west to replace the agricultural irrigation water as an urban water supply or as ecological water.

(2)

Water use schemes for water users should be prioritized:

• Taking into account industrial, agricultural, and other ecological water uses, priority should be given to ensuring domestic water use;
• Allocation should fully consider the types of local water sources and the overall balance between supply and demand. Based on this, a reasonable priority of water source supply should be established;
• The water resources of the urban living water supply include the Pishihang irrigation area water and the local reservoir storage. The source of the ecological environment water supply comprises local surface water and reclaimed water. On the other hand, the agricultural water supply sources include the Pishihang irrigation area water, local surface water, and the appropriate use of reclaimed water.

2.3.3. Objective Function and Constraint Conditions

The core objectives of water resources allocation are regional water supply security and water supply fairness. This paper used optimal fairness and the minimum water deficient ratio as objective functions for determining optimal allocation. The constraint conditions were water supply nodes at all levels, water balance and total quantity control of water supply units at all levels, water supply potential, quality-divided water supply, and non-negative variable.

(1)

Objective function

The optimal fairness objective:

$$M\mathrm{in}F(x)={\displaystyle {\sum }_{y=1}^{myr}{\displaystyle {\sum }_{n=1}^{12}{\displaystyle {\sum }_{h=1}^{mh}{q}_h\times GP(X_h)}}}$$

(1)

$$GP(X_h)=\sqrt{\frac{1}{mu-1}\times {\displaystyle {\sum }_{u=1}^{mu}{\left|x_h^u-{\overline{x}}_h\right|}^2}}$$

(2)

In the equation above, $F\left(x\right)$ represents the fairness objective, while GP(X_h) denotes the fairness function. Next, q_h represents the penalty function of industrial users, while $X_h^u$ denotes the water deficient ratio of industrial user h in the regional unit u. Furthermore, $\overline{x}$_h stands for the average water deficient ratio of industrial user h in regional unit u, while myr represents the number of years in the calculation period. The symbol n denotes the month, while mh represents the number of $GP$ industrial water uses in the region. Finally, mu denotes the number of units in the region.

The minimum water deficient ratio objective:

$$M\mathrm{in}Y(x)={\displaystyle {\sum }_{y=1}^{myr}{\displaystyle {\sum }_{n=1}^{12}{\displaystyle {\sum }_{h=1}^{mh}{q}_h\times SW(X_h)}}}$$

(3)

$$SW\left(X_h\right)=\frac{1}{mu}\times {{\displaystyle \sum }}_{u=1}^{mu} \left|{\left(x_h^u-So{b}_h^n\right)}^2\right|$$

(4)

In the equation above, Y(x) represents the water supply stress objective, while SW(Xh) denotes the water supply stress function. Next, q_h represents the penalty function of the industry user, while $X_h^u$ denotes the water deficient ratio of industrial user h in regional unit u. The symbol $so{b}_h^n$ represents the ideal water supply stress objective of industrial user h for each month in region u. Moreover, myr represents the number of years during the calculation period, while n denotes the month. Finally, mh represents the number of industrial water use types in the region, while mu denotes the number of units within the region.

(2)

Constraint conditions

Total water quantity constraint:

$${\displaystyle \sum }_{j=1}^J{x}_{iu}\le C_i$$

(5)

In the equation above, Ci represents the control quantity of water source i in the total water consumption control index of the research area. In the above equation, i means water source (excluding unconventional water), u represents water user, and x_ij denotes the water supply amount of i water source to u water user.

Water supply potential constraint:

$$W_{m,t}\le Q_{m,t}$$

(6)

In the above equation, W_m,t denotes the water supply (m³) of water source m during period t, while Q_m,t denotes the quantity of available water resources (m³) of the same water source for period t.

User water quality constraint:

$$q_{m,u}\le {\overline{q}}_u$$

(7)

In the equation above, q_m,u represent the water quality of water source m to user u, while q_u denotes the minimum water quality standard for users.

Non-negative variable constraint:

All variables ≥ 0

(8)

2.3.4. Model Solution

The model is solved by a non-dominated genetic improvement algorithm that is based on an elite strategy. The water distributed to each user from each water source represents the decision variable, coded into a feasible solution set. By judging the satisfaction degree of each individual, the survival of the fittest is carried out, so as to generate a new generation of feasible set of solutions, in order to complete the optimal allocation of water resources through repeated iterations.

2.3.5. Calculation Unit as well as Adjustment and Storage Node

Six large reservoirs situated in the upper reaches of the Dabie Mountains function as the main water sources. Water resources are diverted through the Hengpaitou Junction, Hongshizui Canal Head Junction, Niujiaochong Inlet Gate, Meiling Inlet Gate, Liji Junction, and the Nanganqu Inlet Gate. The water is then transferred from the canal into the irrigation area. There are a total of 23 medium reservoirs, 1200 small reservoirs, and several ponds and dams that make up the middle and lower reaches. In addition, there is also a large number of water-pumping stations built within the irrigation area.

Table 2. Major hydrojunctions in irrigation areas.

Reservoir Type	Reservoir Name	Reservoir Type	Reservoir Name
Large irrigation area Large reservoir	Foziling	Medium reservoir	Guanwan
	Mozitan		Hexi
	Bailianya		Zhongxing
	Xianghongdian		Dajing
	Meishan		Dongfanghong
	Longhekou		Taolaoba
Medium reservoir	Huaguo		Hongqi
	Xiezishan		Duji
	Longtan		Weilaohe
	Shuimentang		Yongdou
	Modun		Shuanghe
	Zhangqiao		Longmensi
	Daguantang		Anfengtang
	Caitang		Yongfeng

shows the main hydrojunctions. Small and medium reservoirs and ponds and dams were treated as one adjustment node, with the ponds and dams being calculated by the re-storage coefficient, which was 1.8. The calculation unit division of the irrigation area and the distribution of main adjustment and storage calculation nodes are provided in Figure 2.

2.3.6. Joint Scheduling of Multiple Water Sources

Water sources in the Pishihang irrigation area mainly come from the six upper-reaches reservoirs, ponds, and dams within the irrigation area, and the pumping stations at Wabuhu Lake, Chaohu Lake, Chengdonghu Lake, and Chengxihu Lake, situated on the outskirts of the irrigation area. In addition, water diversion projects, such as the one diverting resources from the Yangtze River to the Huaihe River and to the west, serve as emergency supplementary water sources that will be necessary in the future. Therefore, when calculating the water resources balance, it was necessary to consider the scheduling and operation of various water source projects to conduct joint adjustments. As shown in Figure 3, the network diagram of the water resources system in the Pishihang irrigation area was generalized in accordance with the current water supply system and the supplementary water sources that will be needed in the future.

2.3.7. Auxiliary Regulation Setting of Small and Medium Reservoirs

To simulate the regulatory effect of the lower-reaches reservoir nodes on water use and the joint mutual feeding effect on the upper-reaches reservoirs, the paper set the following conditions in the model:

• Between September and October, large reservoirs in the upper reaches should be filled with water. During the period between October and March, the water should be released to the lower-reaches reservoirs, filling ponds and reservoirs. The sluice gate is set to control the reservoirs’ filling time. When the lower-reaches reservoir requires water for operation, it should be discharged to fill the reservoir.
• During the main water use period between April and September, when crop planting and the harvest guarantee period are taking place, priority should be given to using water from small and medium reservoirs.

Within the control operation principle, the reserved water volume of the reservoir ensures the safety of the urban water supply. The control operation principle is as follows:

Medium reservoirs: 20% may be used at the end of irrigation.

Small reservoirs: 60% may be used by the end of June, while another 20% may be used after irrigation.

Ponds and dams: at least 25% may be used by the end of June, while the rest may be used up after irrigation.

2.3.8. Model Data Input and Related Parameters

(1)

Data Input

• The engineering characteristic data of each reservoir, sluice, pumping station, and channel, i.e., the relationship curve of the reservoir capacity and water level, the flood limit water level at different time periods, the overflow capacity of the sluice, the water lifting capacity of the pumping station, and the overflow capacity of the channel.
• By sharing or collecting the forecast data of water inflow of each reservoir, each unit and river in each time period, the water diversion plan issued from each water diversion outlet overseas, and the time scale of these data have been interpolated or divided into a minimum scale smaller than or equal to the model configuration.
• The water demand of each water distribution unit is calculated according to the quota method.
• Initial water volume of each reservoir.

(2)

Related Parameters

In the process of water distribution, the characteristics of water supply of water distribution units and water users are mainly considered. According to the principle and formula of the model, the sensitive parameters of the configuration module are as follows:

• For the water distribution ratio of various users S_hp (the ratio of each industry water demand and water source supply, 0 < S_hp < 1), we give the initial value based on the historical experience of the long series of years from 1960 to 2015, while allowing the model to seek the optimal value of the parameters.
• For the water user weight coefficient H_qz (living water volume: industrial water volume: agricultural water volume: ecological water volume), in general, the domestic water volume is greater than the agricultural water volume, but we consider this an irrigation area, and ensuring agricultural irrigation is as important as ensuring urban water supply, so all user weight coefficients are equal.

3. Results

According to the rainfall data of 56 years in the irrigation area from 1960 to 2015, the hydrological frequency analysis was carried out. Finally, 1971 is selected as the typical year at 50% frequency (median water year), and 1999 is the closest typical year at 80% frequency (dry year). Two typical years were selected under the frequency (extremely dry year), the typical year 1 was 2004 (the previous year was a wet year), and the typical year 2 was 1966 (the previous year was a continuous drought year). The water inflow in the above years is used as the basis for the allocation of water resources in the base year (2018), 2025, and 2035.

3.1. Water Resources Allocation Results for Each Water User

According to the allocation results (Figure 4), the water users’ supply and demand in the base year, 2025, and 2035 would be balanced at 50% and 80% frequencies. Furthermore, there would be no water deficiency in the irrigation area. Overall, the highest water supply and demand came from agriculture, followed by urban living, industry, ecology, and finally rural living needs. At 50% frequency, the total water demand in the base year, 2025, and 2035 was 4.461 billion m³, 4.650 billion m³, and 4.863 billion m³, respectively. Agricultural water demand accounted for 3.205 billion m³, 2.937 billion m³, and 2.690 billion m³, respectively. Following it was the urban living water demand which accounted for 503 million m³, 721 million m³, and 1.095 billion m³, respectively. The rural areas’ water demand accounted for 149 million m³, 148 million m³, and 171 million m³, respectively. Finally, the ecological water demand amounted to a total of 230 million m³. On the other hand, at 80% frequency, in the base year, 2025, and 2035, the total water demand in the irrigation area was 5.512 billion m³, 5.601 billion m³, and 5.730 billion m³. In this case, agricultural water demand accounted for 4.256 billion m³, 3.888 billion m³, and 3.557 billion m³, respectively. Next, the water demand in urban areas was 503 million m³, 721 million m³, and 1.095 billion m³. Rural area water demand accounted for 149 million m³, 148 million m³, and 171 million m³, respectively. Lastly, the ecological water demand amounted to 230 million m³.

During the extremely dry year, when the frequency was 90%, the supply and demand of each water user for a typical year 1 could reach a balance, without the occurrence of water deficiency. For a typical year 2, the total water demand would not change significantly. However, compared with typical year 1, the water supply would drop sharply. In the case of typical year 2, the total water demand in the base year, 2025, and 2035 was 5.973 billion m³, 6.040 billion m³, and 6.125 billion m³, respectively. On the other hand, the total water supply was 4.425 billion m³, 4.851 billion m³, and 5.033 billion m³, respectively. This will lead to varying degrees of water shortage for each user.

3.2. Results of Water Source Structure Allocation

Under 50% and 80% of each level year, the largest water supply source in the Pishihang irrigation area would be the water diversion from the irrigation area. At 50% frequency, the water diversion in the base year, 2025, and 2035 was 2.904 billion m³, 2.946 billion m³, and 3.091 billion m³, respectively. Furthermore, at 50% frequency, the available supply of surface water in each level year was 1.417 billion m³, 1.497 billion m³, and 1.507 billion m³, respectively. On the other hand, the available supply of surface water at 80% frequency in each level year was 1.456 billion m³, 1.458 billion m³, and 1.448 billion m³, respectively. The external water transfer was almost 0, while the supply of groundwater and reclaimed water was quite small and mostly used as supplementary water sources. The allocation results show that the water supply of reclaimed water increased yearly. For example, at 80% frequency, the water supply of reclaimed water in the base year, 2025, and 2035 was 87 million m³, 147 million m³, and 203 million m³, respectively. During the 90% frequency year, the water diversion of Typical year 1 had a relatively small attenuation. The total water division of the irrigation area in the base year, 2025, and 2035 was 4578 million m³, 4406 million m³, and 4440 million m³, respectively. In 2025, there was a small amount of external water transfer, namely 150 million m³. Compared with the typical year, the water diversion of Typical year 2 experienced a sharp drop. The total water division of the irrigation area in the base year, 2025, and 2035 was 2872 million m³, 2853 million m³, and 2957 million m³, respectively. However, the amount of surface water, groundwater, and reclaimed water did not notably differ compared to Typical year 1 (Figure 5).

3.3. Water Deficient Ratio

Considering the impact of small and medium reservoirs, ponds, and dams, the paper draws the water deficient ratio of each water user from the model in each level year as shown in Figure 6.

At 50% frequency, the water deficient ratio in the base year, 2015, and 2035 was 0, with the water supply capable of meeting the water demand of each user. At a frequency of 80%, agriculture would experience a slight water deficiency during the base year, with a water deficient ratio of 0.26%. However, other uses would not experience water deficiency, with a water deficient ratio of 0.2%. On the other hand, in 2025 and 2035, no users would experience water deficiency. During the extremely dry years which have a 90% frequency, water users would be satisfied with the water supply during Typical year 1. Moreover, in this case, the water deficient ratio would be 0, indicating that the irrigation area could maintain a strong water supply potential without encountering continuous drought years. In Typical year 2, water users in the base year, 2025, and 2035 would all experience water deficiency to varying degrees. Among them, urban living has the lowest water shortage rates, 0.23%, 1.25%, and 2.79%, respectively, followed by rural living, with 1.44%, 3.02%, and 6.71%, and industrial water shortage rates are 1.58%, 2.93%, and 5.75%. The highest water shortage rate in the base year is agricultural water, which is 31.16%, followed by ecological water, which is 29.81%. For 2025 and 2035, the highest water shortage rate is the ecological water use of 33.07% and 34.98%, respectively, followed by the agricultural water use of 25% and 23.56%. The agricultural water shortage rate will decrease to a certain extent, and the ecology will increase. Furthermore, in the base year and 2025, the comprehensive water deficient ratio of the Shi River irrigation area would be higher than that in other irrigation areas. However, in 2035, the highest water deficient ratio would be that of the Hangbu River irrigation area. These observations indicate that the water supply in the irrigation area could be seriously damaged if it were to encounter consecutive drought years, especially in the Shi River area. The regulatory capacity of water resources in this irrigation area in the base year and 2025 is relatively lower than that in other irrigation areas.

4. Discussion

4.1. Discussion of Model Advantages, Applications and Result Reliability

At present, a number of studies have dealt with the simulation and regulation of water resources systems using hydrological methods and intelligent algorithms. A lot of research has been conducted and many models have been developed, such as Xin’an Jiang Model, SIMHYD, MIKE, TOPMODEL, ROWAS, VIC, etc. [24,25,26,27,28,29,30,31,32]. Both the conventional allocation model and the GWAS model are capable of meeting the allocation characteristics of multiple water sources and water users. However, the number of nodes on both the supply and demand side of the GWAS model is about 1000 times that of the conventional allocation model. In terms of water quality and quantity, the GWAS model upgrades the static characteristics to dynamic characteristics, and the water system is examined as a whole. The difference between the GWAS model and the conventional model is provided in

Table 3. The comparison of characteristics between the GWAS model and the conventional model.

Category	Item	General Allocation Model	GWAS Model
Allocation object	Multiple water supply sources	Yes	Yes
	Multiple users	Yes	Yes
Number of units	Number of nodes on both supply and demand sides	≤200 × 200	≤150,000 × 150,000
	Model development method	Conventional array method	(Conventional + improved sparse matrix COO-RC) smart matching
Allocation method	Rule allocation	Yes	Yes
	Optimal allocation	Yes	Yes
	All-time optimization	Weak	Strong
Water balance simulation	Node water quantity simulation	Yes	Yes
	Basin water cycle simulation	No or loosely coupled	Yes and tightly aggregated
Allocation objective	Supply and demand balance	Yes	Yes
	Overall consideration of the water system	No	Yes

.

The GWAS model is capable of dynamically considering both the aspects of supply and demand. Furthermore, this model satisfies the dynamic mutual-feedback relationship of various water sources and allocation subjects, thus completing the refined allocation of water resources. For example, Yan [33] applied the GWAS to the Fu River Basin to complete the multiobjective water resources allocation and meet water supply, power generation, and river ecology demands of the area. The GWAS model is capable of meeting the requirements of water demand calculation, water and sediment scheduling, ecological scheduling, and optimal operation of reservoirs. For example, Yan Ziqi [19] used the GWAS model to comprehensively consider the characteristics of the ecological flow constraint model and the ecological flow target model to coordinate the socioeconomic and ecological water demand and took Pingshan River Basin as an example to study the river ecological water supply, and achieved feasible effects.

To test the reliability of the model, we compared the actual water supply data in 2018 with the simulated water supply results of the model, as shown in Figure 7. In 2018, the actual total water supply of the irrigated area was 4.32 billion m³, and the total water supply was 4.461 billion m³, with a total error of 3.2%. The correlation analysis between the actual water consumption data and the allocated water quantity of each industry shows that the correlation coefficient reaches 0.995, and the correlation coefficient between the water diversion of each irrigation area and the actual data reaches 0.993, indicating that the established model results are good.

According to the allocation results, the urban living water demand will gradually increase between the base year, 2025, and 2035. This expansion will be directly related to the expansion of the city and the resulting increase in water demand for economic development. As a result of economic development, the irrigation area would be further controlled, and water-saving technologies would be improved, leading to a decline in agricultural water demand. These results are in line with local planning and water allocation objectives. The water supply of reclaimed water gradually increased as a result of water source structure allocation and thus met the rigid requirements of local planning.

4.2. Influence of Continuous Drought Conditions on Water Resources Allocation in Irrigated Areas

In the research results, we found that for 90% of the special dry year, the irrigation area in each level year can maintain a strong water supply capacity under the premise of not encountering continuous drought years, and the water shortage rate is basically 0. When the irrigated area experienced successive drought years in the early stage, the water supply would be seriously damaged, especially in the Shihe irrigated area, while the available water quantity decreased sharply, indicating that the multiyear regulation ability of water resources in this area was poor. Water users have different degrees of water shortages: the ecological and agricultural water shortage rate is the largest, and the urban water shortage rate is the smallest. In 2025 and 2035, the water shortage rate of the whole basin would decrease to a certain extent compared with the base year.

We believe that the main water sources of the Pishihang irrigation area come from the six upstream reservoirs. Under a reasonable water source allocation, climate changes such as short-term drought will not have a fatal impact on the water diversion in the irrigation area. However, the damage degree of continuous drought is much higher than that of short-term drought [34], and the impact of continuous dry climate on the reduction in river and other water volume is very significant [35,36]. Under continuous arid climate conditions, the water supply of upstream reservoirs is greatly reduced or cannot be supplied in time. At the same time, the imbalance between precipitation and water evaporation leads to a serious reduction in surface runoff [37], and the shortage of total available water will result in a serious water shortage for all water users.

In the allocation principle, domestic water is given priority to ensure users, so it is less affected by insufficient water supply. With the improvement of agricultural water-saving level, agricultural water consumption decreases year by year, and the water is transferred from outside to replenish water for agriculture and urban life. Therefore, the water scarcity rate of the whole basin increased to a certain extent in 2025 and 2035, which indicates that the regulatory capacity of water resources in 2025 and 2035 has further decreased. However, there are not many detailed studies on water resources allocation under continuous drought, which need further analysis and discussion.

4.3. Policy Proposal

To prevent water shortage in irrigated areas during continuous drought years, the paper makes the following suggestions:

(1)

The first suggestion concerns implementing a scientific allocation of water supply. Studies should consider the demand of water for life, production, and ecology within the irrigated area. Water supply management should also be strengthened in accordance with the principle of “total quantity control, quota management, supply based on demand, water right to county and county boundaries”. Furthermore, the annual water supply (use) plan should be prepared in accordance with the law of demand and the trend of water consumption in the irrigated area. The water supply scope, water supply period, and water quantity should all be specified by the plan [38]. Lastly, the General Administration should conduct unified water supply operation in the irrigated area.

(2)

Secondly, this paper makes suggestions related to water conservation and the improvement of water use in agriculture. Namely, agricultural water accounts for most of the water used in irrigated areas. Water use efficiency needs to be further improved, while the agricultural water-saving irrigation technology should be expanded.

(3)

Based on existing engineering facilities and the plans to renew supporting facilities, engineering and non-engineering measures should be combined to support the reconstruction of existing projects, enhance the water supply capacity of channels, and improve the efficiency of the existing projects. Government establishes and improves water resources management institutions, formulates feasible regulations on water resources management, and strengthens publicity and education on water law, water resources management, and water-related economic policies.

(4)

For successful dual allocation and drought prevention and control, the amount of water in dry years and consecutive dry years needs to be properly reserved [39]. In the case of consecutive drought years, the water shortage at the 90% guaranteed rate is relatively serious. In this case, the water shortage rate under the 90% guaranteed rate can be reduced by 6.22% and 8.08% in 2025 and 2035 compared with the base year. This reduction should occur through the rational development of local reservoir water and additional water transfer from outside the country.

5. Conclusions

This study uses the GIS and GWAS models in accordance with the Pishihang “long cane knots melons” characteristics and builds the fine water resources allocation model of the irrigation area. The model considers the regulating role of some small reservoirs in the allocation of water resources and has been well applied. The research results provide a basis for a water resources allocation scheme of the Pishihang Irrigation District. The research conclusions are as follows:

(1)

For each water user, the water supply at 50%, 80%, and 90%-1 frequency can meet the water demand without water shortage. Additionally, the highest water supply and demand came from agriculture, followed by urban living, industry, ecology, and finally rural living needs. At the frequency of 90%-2, the supply and demand of each water user cannot be balanced.

(2)

For water sources, at 50%, 80%, and 90%-1 frequency, the annual water diversion of irrigation area at each level is the largest, followed by the local surface water and boundary water lifting station, and the middle water and shallow groundwater as supplements.

(3)

The results of water shortage rate show that in the base year, 2025, and 2035, the water shortage rate of each water user is 0 at frequencies of 50%, 80%, and 90%-1. Under the frequency of 90%-2, water shortage of different degrees occurs in all water users in the base year, 2025, and 2035, with the lowest water shortage rate in urban life, followed by rural life, industrial, ecological, and agricultural water.

(4)

Extreme continuous arid climate conditions will cause serious damage to water supply in irrigated areas. The water shortage rate can be reduced by improving the level of agricultural water use, adding external water sources, optimizing water resources allocation, and other measures.