Agricultural Water Management 303 (2024) 109061

Contents lists available at ScienceDirect

Agricultural Water Management

journal homepage: www.elsevier.com/locate/agwat

Water-agriculture-ecology nexus synergetic management based on

spatiotemporal equilibrium and water transformation: A case study in Aksu
River Basin, China
Taishan Wang a,b , Xiaoling Su a,b,* , Haijiang Wu a,b
a
Key Laboratory of Agricultural Soil and Water Engineering in Arid and Semiarid Areas, Ministry of Education, Northwest A&F University, Yangling 712100, China
b
College of Water Resources and Architectural Engineering, Northwest A&F University, Yangling 712100, China

A R T I C L E I N F O A B S T R A C T

Handling Editor: Dr R Thompson The temporal variability and spatial heterogeneity characteristics of the water-agriculture-ecology (WAE) nexus
system have aggravated the difficulties in its synergetic management. Besides, in the inland river basin, the
Keywords: surface water and groundwater are tightly linked by the combination of canal and well irrigation. To address
Multi-objective programming these issues, a spatiotemporal equilibrium-water transformation based water-agriculture-ecology nexus syner­
Robust optimization
getic management (SEWT-WAE) model was proposed by incorporating a spatio-temporal robust optimization
Conjunctive use
method and linear water transformation model. The SEWT-WAE model was then applied to the Aksu River Basin,
Multiple criteria analysis
Water resource management an inland river basin of Xinjiang, China. The results indicated that the SEWT-WAE model was highly effective in
Irrigation water achieving spatiotemporal equilibrium in groundwater balance and ecological water utilization, as well as in the
integrated management of surface water and groundwater across upstream and downstream regions. The optimal
synergetic management scheme was obtained based on the coordinated development degree. Compared to the
current situation: (i) the irrigation amount provided by the surface water (groundwater) in the Tabei (Tanan)
irrigation district was increased (decreased) by 21.4 % (70.2 %); (ii) the irrigated areas of grain crops and
gardens were increased by 30.4 % and 20.1 %, respectively, while the irrigated area of cotton was decreased by
19.4 %; (iii) the ecological water utilization of the Populus euphratica forest was increased by 17.81 %. Overall,
this study presents a new optimization model for achieving spatiotemporal equilibrium and conjunctive use of
surface water and groundwater and provides decision support for WAE nexus synergetic management in the
inland river basin.

1. Introduction is paramount to support regional sustainable development to engage in

water-agriculture-ecology (WAE) nexus synergetic management (Jia
In agriculture basins, water resources, agriculture, and ecology are et al., 2024). Moreover, to obtain a synergetic management scheme, a
intricately interconnected, forming an indispensable nexus (Fischer holistic method is required to coordinate water resources, agriculture,
et al., 2014). The availability of water resources directly influences the and ecology within a system. Bi-level programming (BP), with a
yield and quality of crops and is crucial for the stability and sustain­ two-level structure, is an effective method for solving multi-objective
ability of agricultural production. Rational utilization and protection of optimization issues (Sakawa et al., 2000). However, traditional BP is
water resources contribute to fostering biodiversity and preserving merely able to solve the optimization problem with two objectives. For
ecological balance (Dalin and Outhwaite, 2019). As one of the primary instance, Lu et al. (2006) established a framework for the bi-level mul­
ways of water consumption, excessive agricultural irrigation can lead to ti-followers decision problem. Jin et al. (2020) developed a stochastic
water stress and negatively affect regional ecology (Gordon et al., 2010). multidivisional hierarchy model by coupling bi-level multi-followers
The ecosystem health, supported by water availability and quality, en­ programming (BMP) and interval chance-constrained programming.
sures the water resources conservation and contributes to the sustain­ Their results revealed that BMP can resolve conflicts and provide useful
ability of agricultural production (Pander and Geist, 2013). Therefore, it information for the integrated management of multiple sectors.

* Correspondence to: College of Water Resources and Architectural Engineering, Northwest A&F University, Yangling, Shaanxi, China.
E-mail address: [email protected] (X. Su).

https://doi.org/10.1016/j.agwat.2024.109061
Received 25 May 2024; Received in revised form 4 September 2024; Accepted 7 September 2024
Available online 16 September 2024
0378-3774/© 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-
nc/4.0/).
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Effectively managing water resources is a huge challenge due to their Section 5 offers the conclusions of the study.
temporal variability and spatial heterogeneity (Li et al., 2024). Tem­
poral variability of water resources is influenced by precipitation pat­ 2. Material and methods
terns, evapotranspiration rates, and climate variability, which can cause
the stochastic inter-annual distribution of available water resources and This study proposed the SEWT-WAE model and achieved WAE nexus
potentially result in water stress, crop failures, and ecological degra­ synergetic management, based on spatiotemporal equilibrium and water
dation (Abbas et al., 2022). Spatial heterogeneity of water resources transformation, the detailed steps are shown as follows (see Fig. 1):
influenced by geographic features, climatic conditions, and human ac­
tivities, may lead to localized groundwater imbalances and ecological (1) setting the topology of water resources allocation and the model’s
degradation within basins (Dalin et al., 2014). Stochastic robust opti­ parameters according to the data and the policies of the study
mization (SRO) is a type of uncertainty programming method that in­ area;
corporates risk aversion into optimization models, and it can be used to (2) constructing a linear water transformation model by combining
find robust solutions to resource management problems by considering groundwater balance calculation and streamflow component
the variability of resources (Mulvey and Ruszczyński, 1995). Based on investigation method;
SRO, multiple uncertainty optimization models were developed and (3) developing spatio-temporal robust optimization (STRO) method
applied to the water resources management system (Xu et al., 2009). through analogizing with the SRO method and introducing the
Results indicated that SRO can generate stable water resources alloca­ concept of spatiotemporal equilibrium;
tion patterns and decrease water deficit under low available water re­ (4) proposing the SEWT-WAE model by incorporating the linear
sources by reducing water demand to achieve temporal equilibrium water transformation model constructed in (2) with the STRO
(Ortiz-Partida et al., 2019). However, traditional SRO cannot be used to method developed in (3) into the water resource allocation
address spatial heterogeneity. Coupling coordination degree (CCD) can model.
clearly describe the level of coordinated development between systems, (5) solving the SEWT-WAE model using BMP under different
Li et al. (2022) constrained the level of the spatial equilibrium of water spatiotemporal equilibrium tendency scenarios (in total 36 sce­
allocation (SEWA) through CCD, increasing the level of SEWA. How­ narios) to obtain multiple decision schemes.
ever, the trade-off between spatial equilibrium and overall development (6) analyzing the influence of spatiotemporal equilibrium tendency
cannot be realized by only constraining the level of SEWA. and considering water transformation; obtaining the optimal
Surface water and groundwater are integral components of the water scheme including canal-well water utilization structure, water-
cycle (Karatzas et al., 2019). In the inland river basin of China, surface adapting irrigated area, and ecological water utilization
water and groundwater are tightly linked by the combination of canal allocation.
and well irrigation (Zhao and Chang, 2014). Inadequate consideration of
groundwater alongside surface water management may lead to under­
estimation of local development potential or the emergence of severe 2.1. Study area
environmental issues (Jain et al., 2023). Thus, the conjunctive use of
surface water-groundwater is paramount for water resources manage­ The Aksu River is a transboundary river originating in Kyrgyzstan,
ment, and many scholars have developed optimization models to approximately located between 75◦ 35’ to 82◦ 00’ E and 40◦ 00’ to 42◦ 27’
address the problem (Zeinali et al., 2020). For example, Wang et al. N (Fig. 2). The Aksu River Basin lies on the northwest edge of the Tarim
(2016) developed a type-2 fuzzy interval programming (T2FIP) Basin, far from the ocean, characterized by a warm temperate arid
approach and proposed a new optimization model, in which the surface climate with significant continental features. The dry climate is
water and groundwater supply of each irrigation subarea was con­ accompanied by high evapotranspiration rates, scarce precipitation, and
strained, to obtain optimal conjunctive water allocation patterns. Pre­ large annual and seasonal variations (Ouyang et al., 2007). Aksu River is
vious studies on the conjunctive use of surface water-groundwater have one of the main tributaries of the Tahe River, formed by the confluence
mainly focused on coupling simulation and optimization models (Song of the Kumalak River and the Toshkan River, and joins the Tahe River in
et al., 2020). For example, Dehghanipour et al. (2019) developed a Xiaojiake hydrological station. Numerous diversion gates are distributed
coupled simulation model integrating surface water (WEAP) and in the Kumalak and Toshkan Rivers and the Aksu River’s main channel,
groundwater (MODFLOW). Dehghanipour et al. (2020) then applied this forming a complex water conservancy infrastructure (Rumbaur et al.,
model in a simulation-optimization (SO) approach to identify effective 2015).
water management strategies for irrigated endorheic river basins, In terms of agriculture, the Aksu River Basin is responsible for
demonstrating its utility in addressing complex water management providing irrigation water to the First Region irrigation district. The
challenges. However, since water transformation is not considered in the irrigation district can be further divided into five smaller irrigation
optimization part, it is hard to systematically regulate the allocation of districts: Shajingzi irrigation district, Tuo irrigation district, Ku irriga­
surface water and groundwater across the whole basin. tion district, Tabei irrigation district, and Tanan irrigation district, as
Herein, this study aims to propose a novel optimization model, i.e., illustrated in Fig. 2. In terms of ecology, the Aksu River Basin plays a
the spatiotemporal equilibrium-water transformation based water- crucial role in ecological water transfer between the Eichmann Lake
agriculture-ecology nexus synergetic management (SEWT-WAE) wetland and Populus euphratica forest. The release of streamflow into
model. The model can achieve spatiotemporal equilibrium in ground­ the river channel is necessary to ensure the ecological balance both
water balance and ecological water utilization, systematically coordi­ within and outside the Tahe River basin (Nie et al., 2022).
nating the use of surface water and groundwater within a basin. The
SEWT-WAE model would be applied to the WAE nexus synergetic 2.2. Data description
management and solved by using BMP in Aksu River Basin, to disclose
the optimal WAE nexus synergetic management scheme in Aksu River In this study, the irrigation data, such as annual irrigated area, irri­
Basin. The remainder of this paper is organized as follows: Section 2 gation volumes, crop yield, and annual surface water and groundwater
describes the study area, data source, and methodology, including amounts in each irrigation district, were derived from the water man­
developing and solving of SEWT-WAE model; Section 3 presents the agement comprehensive statistical report (2004–2019) and Aksu water
main results, covering the influence of spatiotemporal equilibrium ten­ resources bulletin (2004–2019) that was provided by the Aksu Water
dency and the optimal decision scheme; Section 4 provides the discus­ Resources Bureau. The benefits and costs of agricultural products were
sion, highlighting the advantage and limitation of SEWT-WAE model; derived from the “National Agricultural Product Cost and Benefits Data

2
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 1. The research framework of water-agriculture-ecology (WAE) collaborative management based on spatiotemporal equilibrium and water transformation.

Compilation-2021” (National Development and Reform Commission derived from Dong (2018), Nie et al., (2022), and Zhao (2023). Besides,
and Price Cost Survey Center, 2021) and were obtained from Wang the water utilizations of ecology from the inside and outside river
(2022) and Gao (2023). The situation of the water conservancy project channels were derived from the streamflow of the Alar hydrological
was available from the “Comprehensive Water Resources Planning station, “2019 Ecological Water Dispatch Implementation in Aksu Pre­
Report for Alar City” which was provided by the Aksu Water Resources fecture” and “2018 Ecological Water Transfer Plan for Aksu Area” that
Bureau and was partly obtained from Rousidan (2021) and Zhao (2023). was provided by the Aksu Water Resources Bureau. The specific
The streamflow data of each hydrological station was obtained from the description for these data resources can be found in Appendix A.
Aksu Water Resources Bureau, and the meteorological data was ob­ Identifying and addressing the uncertainty intervals within the
tained from the China Meteorological Data Service Center (China model’s input data helps us to elucidate parameter inaccuracies, model
Meteorological Data Service Center, 2023). Additionally, the water de­ structure uncertainties, and the inherent randomness of water resources.
mands of ecology from the inside and outside river channels were By acknowledging these uncertainties, the study can make more

3
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 2. Location and digital elevation model (DEM, its spatial resolution is 30 m×30 m) of Aksu River Basin.

informed decisions regarding water resource management, ultimately

xj ≥ 0, j = 1, 2, …J (1e)
enhancing the credibility and applicability of the results in real-world
scenarios. The handling of uncertainties in this study is discussed in
ytj ≥ 0, j = 1, 2, … J; t = 1, 2, … T (1f)
Section 4 “Discussion”.
Vtj ≥ 0, j = 1, 2, … J; t = 1, 2, …T (1g)
2.3. spatial-temporal robust optimization method
Utj ≥ 0, j = 1, 2, … J; t = 1, 2, …T (1h)
Two-stage stochastic programming (TSP), an uncertain optimization
method, is proposed to address the randomness issues (Huang, 2000). where, xj and ytj represent the first-stage and second-stage decision
The decision-maker relies on experience to make decisions before sto­ variables, respectively; pt denotes the probability of occurrence for
∑
chastic events and adjusts them afterward to minimize losses (Li and random event t, and st=1 pt = 1; aj, bj, and wtj indicate model parame­
Zhang, 2020). Although TSP addresses uncertainties expressed as ters in constraints; cj and dtj are the coefficients of xj and ytj in the
randomness, it overlooks system risk. The robust optimization (RO) objective function, respectively; xj and ytj are constrained with param­
method integrates risk-aversion strategies to address risk issues (Mulvey eters and variables in constraints (Eq. 1b), respectively; TE and SE
et al., 1995). By combining TSP and RO, stochastic robust programming indicate the temporal and spatial equilibrium coefficients, respectively;
(SRP) handles both randomness and risk issues (Li et al., 2014). Based on Vtj and Utj are temporal recourse cost and spatial recourse cost, respec­
the SRP method, Zhang et al. (2022) defined the positive deviation from tively, representing the positive deviation from the expected level across
the expected environmental penalty as recourse costs. In this study, we time and space; AIj is the weight of unit j in space; constraints (1e)–(1 h)
used recourse costs to reflect the positive deviation from the expected define xj, ytj, Vtj, and Utj being always positive.
level across both time and space, thereby formulating a spatiotemporal
robust optimization (STRO) method, which can be represented as 2.4. Linear water transformation model
follows:
Objective function: The streamflow component investigation method, based on water
( ) ( ) balance, eliminates the impacts of various human activities on stream­
∑J ∑T ∑n ∑T ∑J flow to restore measured streamflow to natural streamflow (Loboda
Maxf = cj xj − pt dtj ytj − TE t=1 pt Vtj
j=1 t=1 j=1 j=1 et al., 2005). The variation in groundwater storage was calculated using
( ) the groundwater balance calculation method. The main recharge and
∑T ∑J
− SE pt Utj discharge of groundwater included canal irrigation infiltration, canal
t=1 j=1
system seepage, well irrigation return, river seepage, rainfall infiltra­
(1a) tion, artificial exploitation, and other terms (Jing et al., 2023). In this
study, streamflow component investigation and groundwater balance
Constraints: calculation method were combined through streamflow infiltration,
aj x + bj ytj ≤ wtj , j = 1, 2, …, J; t = 1, 2, …T (1b) canal system infiltration, canal field infiltration, and irrigation drainage,
and then a linear water transformation (LWT) model was established
∑s [ ( )] ( )
(see Equations (2o)–(2af) in Section 2.5). In the LWT model, natural
Vtj ≥ pt cj xj − dtj ytj − cj xj − dtj ytj ,
t=1 (1c) streamflow was used as the water yield to minimize the difference be­
j = 1, 2, 3, …, J; t = 1, 2, 3, …, T tween the simulated streamflow values and their corresponding
/ / / observed values. The Nash-Sutcliffe efficiency value for streamflow at
J ( ) ∑J ( )
∑ the Alar hydrologic station from 2004 to 2019 is 0.94 (see Fig. 3). The
Ujh AIj ≥ cj xj − dtj ytj AIj − cj xj − dtj ytj AIj
j=1 j=1
(1d) conceptual model is depicted in Fig. 4.
j = 1, 2, 3, …, J; t = 1, 2, 3, …, T

4
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 3. The observed and simulated streamflow in the Alar hydrologic station.

Fig. 4. Water transformation model.

2.5. Spatiotemporal equilibrium-water transformation based water- the linear water transformation model was incorporated into the model
agriculture-ecology nexus synergetic management model as a constraint (See constraints (B15) and (B29) in Appendix B). This
allowed for the representation of key water transformation processes,
We integrated the constructed LWT model and STRO method to such as streamflow infiltration, irrigation seepage, and river channel
establish the SEWT-WAE model (See Fig. 1). Practical details were confluence, within the model. By including these processes in the opti­
considered in the case study as follows. Firstly, to achieve coordinated mization, the model ensured that water resources were managed more
development of the WAE nexus, the maximization of economic benefits efficiently and effectively, enhancing the overall optimization outcomes.
was set as the upper-level objective, with the maximization of ground­ Lastly, based on local policies and actual conditions, constraints such as
water balance, crop yield, and ecological water volume set as the lower- irrigation water utilization, irrigated area, crop yield, ecological water
level objectives for the water, agriculture, and ecology subsystems, demand, discharge of the Tahe River, and spatiotemporal balance were
respectively (See Eqs. (B1)–(B4) in Appendix B). Secondly, the STRO all considered in the model (See constraints (B5)–(B14) and (B30)
method was incorporated into the model to avoid uneven spatiotem­ –(B32) in Appendix B). The detailed descriptions with respect to the
poral distribution of groundwater balance and ecological water utili­ SEWT-WAE model can be found in Appendix B.
zation, which may lead to groundwater over-extraction and ecological
damage. This defined the spatial and temporal recourse costs of
groundwater balance and ecological water in the constraints of the 2.6. Bi-level multi-follower optimization method
model (See constraints (B33)–(B40) in Appendix B). These costs were
considered in the lower-level objective function of the water resources Bi-level multi-follower programming (BMP) was proposed to balance
and ecology subsystems (See Eqs. (B2) and (B4) in Appendix B). Thirdly, objectives between decision-makers of two hierarchical levels (Esmaeili
et al., 2019). In the BMP method, multiple decision-makers were

5
T. Wang et al. Agricultural Water Management 303 (2024) 109061

typically divided into an upper-level decision-maker (leader) and mul­ can be introduced as the “coordination degree” among different systems
tiple lower-level decision-makers (followers) (Beykal et al., 2022). Their to measure the coordination status. The coordinated development de­
objectives often conflicted and mutually influenced each other. BMP can gree was used to measure the level of coordinated development between
be represented as follows: (Jin et al., 2020)： systems (Ji et al., 2021). The calculation formula for the degree of co­
∑n ordinated development among subsystems was listed as follows:
maxFU = j=1
cU,j xj (upperlevel) (2a)

∑n (1) Data standardization: For n evaluation indicators and m evalua­

maxFL1 = cL1,j xj (lowerlevel1) (2b) tion objects, the original information matrix, X = (xij)m×n, was
j=1
established, where xij represents the evaluation value of the i-th
∑n evaluation object under the j-th evaluation indicator (Zang and
minFL2 = cL2,j xj (lowerlevel2) (2c)
j=1 Su, 2019). The original matrix was standardized to obtain the
∑n standardized information matrix Y = (yij) m×n, as shown in Eq. 4a.
minFLk = j=1
cLk,j xj (lowerlevelk) (2d) ⎧
⎪ xij − min(xi )
⎨ max(xi ) − min(xi ), positive indexes
⎪
⎪
subject to:
yij = (4a)
( )
G xj ≤ 0 (2e) ⎪ max(xi ) − xij
⎪
⎪
⎩ , negative indexes
max(xi ) − min(xi )
FU is the objective of the upper-level decision maker, FL1, FL2, … FLk is
the objectives of a specific lower-level decision maker, respectively; xj is
the decision variable; cU, j, cL1, j, cL2, j, …, cLk are the coefficients of xj in (2) Coordination calculation: In the quantitative evaluation of coor­
FU, FL1, FL2, …, FLk; and G(xj) is constraint relaxed to xj. To obtain an dinated development systems, the discrepancy coefficient was
overall satisfactory solution, an interactive fuzzy approach was intro­ used to measure the closeness of different indices comprehen­
duced to solve the BMP model. Compared to other algorithms, the sively. Based on the principle of the minimum analogy of the
interactive fuzzy approach can simplify the complexity of the WAE discrepancy coefficient of the comprehensive evaluation index, a
nexus issue and solve a two-level problem in a single iteration, and then formula for the coordination of multiple systems can be derived:
one optimal solution is obtained (Shih et al., 1996). The detailed solu­ [ ]/( / )2
m
tion process es of BMP can be shown as follows: 2 ∑( ) ∑
Ci = yij ⋅yij yij m (4b)
step 1: solving multiple models by separately setting the objective of m(m − 1) j∕
=jʹ i=1
the upper-level or lower-level decision maker as the objective function,
and obtaining the maximum values FH H H H
U, FL1, FL2, …, FLk, and minimum
L L L L
values FU, FL1, FL2, …, FLk for each objective function. (3) Coordinated development degree calculation: To measure the
step 2: letting λ less than the membership functions of each objective, level of coordinated development between systems, combining
with λ being maximized as the objective (see constraints 3a–3e), then the coordination calculation with the development degree
letting the membership function of the upper-level objective greater calculation, the coordinated development degree model can be
than its acceptable minimum satisfaction level σ (see constraint 3 f), to expressed as (Zhu et al., 2024):
transform the multi-objective model (2) into a single-objective model
m ( ) m
(3). ∑ ∑
Di = Cηi × T1− η
, Ti = λj yij , λj = 1, 0 ≤ η ≤ 1 (4c)
step 3: solving model (3), to obtain the optimal solution xopt j and the
i
j=1 j=1
optimal values of each objective function FU(xopt opt opt
j ), FL1(xj ), FL2(xj ), …,
FLk(xopt
j ). 3. Results
maxλ (3a)
3.1. Analysis of the influence of spatiotemporal equilibrium tendency on
subject to: schemes
( )
FU xj − FLU
≥λ (3b) Fig. 5 presents the influence of spatiotemporal equilibrium tendency
FUU − FLU on the temporal and spatial distribution of groundwater balance in
( ) diverse irrigation districts. High equilibrium tendency (both TE and SE
FL1 xj − FLL1 were set as 1) improved groundwater balance in the Tanan irrigation
≥λ (3c)
FUL1 − FLL1 district (the area with the lowest groundwater balance value) and hin­
( ) dered groundwater balance in the Tabei irrigation district. Under low
FUL2 − FL2 xj spatiotemporal equilibrium tendency (both TE and SE were set as 0),
≥λ (3d)
FUL2 − FLL2 groundwater balance was achieved in the Tuo, Ku, and Tabei irrigation
( ) districts, with the groundwater in the Shajingzi irrigation district
FULk − FLk xj approaching balance and the Tanan irrigation district exhibiting lower
≥λ (3e)
FULk − FLLk groundwater balance. Under high spatiotemporal equilibrium tendency,
the expected groundwater balance value in the Tabei (Tanan) irrigation
( )
FU xj − FLU district decreased from 0 to –522.8×104 m3 (increased from
≥σ (3f) –2220.4×104 m3 to –1767.7×104 m3). This was mainly because (1) the
FUU − FLU
areas of Shajingzi, Tabei, and Tanan irrigation districts are similar, while
( )
G xj ≤ 0 (3g) the Tanan irrigation district has a lower groundwater balance value than
the other irrigation districts; (2) under high spatiotemporal equilibrium
tendency, the increased surface water irrigation in Tanan irrigation
2.7. Coordinated development degree district enhances canal system seepage and canal irrigation infiltration,
while the decreased irrigation groundwater reduces discharge, which
“Coordinated development” refers to the mutual coordination among leads to a decrease in surface water availability in other irrigation
systems during the development process. The discrepancy coefficient

6
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 5. The influence of spatiotemporal equilibrium tendency on groundwater balance value. (a) Spatial distribution of groundwater balance value, (b) Temporal
distribution of groundwater balance value. Note: N means that spatiotemporal equilibrium is not considered; Y means that spatiotemporal equilibrium is considered.

districts; (3) the decreased surface water availability in the Tabei irri­ while the ecological water utilization in Eichmann Lake wetland and
gation district brings the reduced surface water irrigation and increased Populus euphratica forest was increased by 1581.4×104 m3 and
irrigation groundwater, decreasing groundwater balance value. More­ 2167.7×104 m3, respectively. This was primarily because (1) the
over, a high spatiotemporal equilibrium tendency can reduce inter- streamflow must be maintained at a relatively high level under the
annual groundwater balance differences in the Tanan irrigation dis­ constraint for discharge to the mainstream of the Tahe River, resulting in
trict and slightly enhance inter-annual groundwater balance differences a much higher ratio of ecological water utilization to ecological water
in the Tabei irrigation district. Compared with the low spatiotemporal utilization demand in the river compared to other ecological zones; (2)
equilibrium tendency, the differences between actual and expected annual ecological water utilization in the mainstream of Tahe River is
groundwater balance values in the Tanan irrigation district during dry, reduced to the lowest discharge requirement, and the decreased
normal, and wet years were decreased by 754.5×104 m3, 452.7×104 m3, ecological water utilization is transferred to ecological water utilization
and 452.7×104 m3 under high spatiotemporal equilibrium tendency, outside the river. Additionally, a high spatiotemporal equilibrium ten­
respectively. However, the differences in the Tabei irrigation district dency can reduce the differences in ecological water utilization among
were increased by 581.9×104 m3, 335.3×104 m3, and 355.1×104 m3 for different hydrological years in each ecological zone. The results also
dry, normal, and wet years, respectively. This is mainly because the indicated that high spatiotemporal equilibrium tendency brought re­
irrigation water, under high spatiotemporal equilibrium tendency, was ductions in deviation between actual and expected ecological water
decreased during dry and normal years and increased during wet years. utilization levels, contrasting with low spatiotemporal equilibrium
Fig. 6 presents the influence of spatiotemporal equilibrium tendency tendency scenarios. In the Eichmann Lake wetland, the deviation of
on the temporal and spatial distribution of ecological water utilization in ecological water utilization was decreased by 4146.2×104 m3 and
ecological zones. The results indicated that a high spatiotemporal 1631.4×104 m3 during normal and wet years, respectively, while that
equilibrium tendency reduced ecological water utilization in the main­ was slightly increased by 169.8×104 m3 during dry years. Similarly, in
stream of Tahe River, and increased ecological water utilization in other the Populus euphratica forest, the reductions of ecological water utili­
ecological zones. Compared with the low spatiotemporal equilibrium zation were 8955.8×104 m3 and 1980.2×104 m3 during the normal and
tendency, at the high spatiotemporal equilibrium, ecological streamflow wet years, respectively, alongside slight increases of 2167.7×104 m3
in the mainstream of Tahe River was decreased by 3832.5×104 m3, during dry years. Moreover, in the mainstream of the Tahe River, the

7
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 6. The influence of spatiotemporal equilibrium tendency on ecological water utilization. (a) Spatial distribution of ecological water utilization, (b) Temporal
distribution of ecological water utilization.

ecological water utilization was decreased by 3976.9×104 m3 and groundwater balance in the current year and minor increase in surface
12958.4×104 m3 during dry and normal years, respectively, with a water irrigation, after optimization, the groundwater consumption in
minor increase by 2144.9×104 m3 during wet years. the Tanan irrigation district during dry years was significantly decreased
by 70.2 %, decreasing groundwater discharge to achieve groundwater
balance. Moreover, with the increase in surface water resources during
3.2. Decision scheme under optimal spatiotemporal equilibrium tendency normal and wet years, the proportions of groundwater irrigation
scenario decreased in all irrigation districts compared to dry years, leading to an
overall decrease in groundwater irrigation proportion from 3.8 % to
Based on the coordinated development degree, the performances 2.9 % and 2.5 %. Lastly, compared to the current year, surface water
under different scenarios were evaluated by considering system benefit, consumption (irrigation water consumption) increased by 2.1 % (3.4 %)
groundwater balance value, total grain yield, ecological water utiliza­ and decreased by 21.6 % after optimization during dry years in the study
tion, and spatial equilibrium cost relaxed to groundwater balance value area, respectively.
and ecological water utilization. Results indicated that the scenario with Fig. 9 illustrates the optimized crop planting structure under
temporal and spatial equilibrium coefficients of 0.2 and 1.0 was deter­ different hydrological years compared to the current situation. Fig. 10
mined as the optimal spatiotemporal equilibrium tendency scenario. displays the difference between the optimized scheme in the dry year
Fig. 7 displays the optimized irrigation channels and wells structure and the current scheme in irrigated areas of different irrigation districts
under different hydrological years against the current situation, while and crops. Results indicated that irrigated areas and proportions of
Fig. 8 displays the optimized irrigation water and groundwater balance cereal crops and gardens have significantly increased, while the irri­
values under different hydrological years compared to the current sit­ gated area and proportion of cotton have notably decreased. In com­
uation. The results demonstrated that the irrigation surface water in the parison with the current situation, under the dry years of the optimized
Tabei irrigation district was increased by 21.4 % after optimization. This scheme, the irrigated area of grain crops (wheat, rice, and coarse ce­
is primarily due to the substantial groundwater discharge in the current reals) has increased by 30.4 %, and the garden has increased by 20.1 %,
year and the high yield of rice in the Tabei irrigation district, necessi­ while the irrigated area of cotton has decreased by 19.4 %, with an
tating an increase in surface water irrigation to increase the canal irri­ overall reduction in the total irrigated area has increased by 4.2 %. This
gation infiltration and canal system seepage. In addition, due to the low

8
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 7. Irrigation water structure in irrigation channels and wells. (a) Current structure, (b) Optimized structure during dry years, (c) Optimized structure during
normal years, (d) Optimized structure during wet years. Note: SW stands for irrigation surface water, and GW stands for irrigation groundwater.

Fig. 8. Irrigation water consumption of surface water and groundwater and groundwater balance value in irrigation districts.

was primarily because (1) maximizing system benefits and grain yield unit-water benefit of the cotton was lower and its current planting area
was set as objective; (2) the unit-water benefit of the garden was higher, was wide. Additionally, under the optimization, cotton remained the
and its suitable planting area and current planting area were wide; the largest irrigated crop, accounting for 50.8 %, 49.9 %, and 50.3 % of the

9
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 9. Crop planting structures in different irrigation districts. (a) Current structure, (b) Optimized structure during dry years, (c) Optimized structure during normal
years, (d) Optimized structure during wet years.

Fig. 10. The difference between the optimized scheme in dry years and the current scheme in irrigated areas. Note: Gray means no planting under the cur­
rent situation.

total irrigated area under dry, normal, and wet years, respectively. to ecological zones. Furthermore, with the increase in water resources,
Following cotton, garden accounted for 41.1 %, 40.4 %, and 40.8 % of the ecological water utilization of the outside river channel (the main­
the total irrigated areas under dry, normal, and wet years, respectively, stream of the Tahe River), compared to dry years, was increased by
and rice, accounted for 6.5 %, 8.1 %, and 7.2 %, under dry, normal, and 87.6 % (28.9 %) under normal years and 136.6 % (33.4 %) under wet
wet years, respectively. years. Moreover, the ecological water utilization in each ecological zone
Fig. 11 illustrates ecological water utilization in each ecological zone apologies the ecological water utilization demand during the wet years.
under different periods. Compared to the current situation, under dry Fig. 12 displays the changes in the groundwater balance of different
years, the ecological water utilization of the Eichmann Lake wetland irrigation districts. The results indicated that incorporating water
remained unchanged, and that of the mainstream of the Tahe River was transformation into the optimization model can reduce surface water
increased slightly, while the ecological water utilization of the Populus irrigation in upstream Shajingzi, Tuo, and Ku irrigation districts (areas
euphratica forest was increased by 17.8 %. This is mainly because the with higher groundwater balance difference), and increase surface
ratio of current ecological water utilization to ecological water utiliza­ water irrigation in downstream Tabei and Tanan irrigation districts
tion demand was the lowest for the Populus euphratica forest, compared (areas with lower groundwater balance difference). The transferred

10
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Fig. 11. ecological water utilization in each ecological zone.

Fig. 12. The changes in groundwater balance terms of different irrigation districts. Note: RS stands for river seepage, CS stands for canal system seepage, CI stands
for canal irrigation infiltration, OR stands for other recharge terms, DT stands for discharge terms, C stands for under current situation, and O stands for under
optimization scheme.

irrigation water among diverse irrigation districts was enhanced canal Tanan irrigation districts, respectively.
system seepage by 1629.5×104 m3 and canal irrigation infiltration by
2281.2×104 m3 in the Tabei irrigation district and canal irrigation 4. Discussion
infiltration by 377.3×104 m3 and 175.4×104 m3 in Tanan irrigation
district. Additionally, increasing the streamflow at the Aral hydrologic The uncertainties in this study were primarily reflected in parameter
station (the streamflow of the mainstream of Tahe River) can augment inaccuracies, model structure uncertainties, and the randomness of
river seepage for the Tabei and Tanan districts. However, due to the water resources. Regarding parameter inaccuracies, the study employs
need to sustain economic development and grain yield, the increase in statistical data from 2019 or close to 2019 as input for the model due to
streamflow during dry years was limited, resulting in only a slight in­ the limited data availability. Concerning the model structure, the pro­
crease of 0.3×104 m3 and 0.6×104 m3 in river seepage for Tabei and posed linear water transformation model in this study showed

11
T. Wang et al. Agricultural Water Management 303 (2024) 109061

satisfactory simulation accuracy for streamflow. As for the randomness Table 1

of water resources, the P-III distribution function was employed to fit the The upstream inflow with a probability distribution.
uncertainty of total upstream inflow. As shown in Fig. 13, the fitted Hydrological year type Dry Normal year Wet
results were compared to actual inflow distributions, indicating a suit­ year year
able fit with the P-III distribution (R2 = 0.94). Based on the stochastic Inflow in Xiehela hydrologic station (106 4105.1 4955.8 5280.6
simulation of the total upstream inflow and the classification standards m3)
of hydrological years, this study used the multi-year average inflow of Inflow in Shaliguilanke hydrologic station 2305.6 2874.8 3287.6
each hydrological year type as the inflow for each hydrological year in (106 m3)
Total upstream inflow (106 m3) 6410.7 7830.5 8568.2
the SEWT-WAE model (see Table 1). Frequency (%) P≤ 37.5 < P ≤ P > 62.5
In this study, the advantages of the proposed SEWT-WAE model lie in 37.5 62.5
three aspects: Probability (%) 37.5 25 37.5

(1) In terms of spatiotemporal equilibrium, some researchers miti­

method, a linear programming approach, to handle spatiotem­
gated the risk of water scarcity in different hydrological years by
poral equilibrium issues with higher efficiency. Additionally, the
using the SRO method, to obtain a robust decision scheme and
linear water transformation model, introduced as constraints in
achieve temporal equilibrium (Hassas et al., 2021). Li et al.
the optimization model, minimally impacts the model’s solution
(2022) quantitatively reflected the level of the spatial equilib­
efficiency compared to constructing a simulation-optimization
rium of water allocation (SEWA) through coupling coordination
model. Furthermore, an interactive fuzzy approach is intro­
degree (CCD). These previous studies only focused on temporal or
duced to solve the SEWT-WAE model, simplifying the complexity
spatial equilibrium. In this study, to achieve a spatiotemporal
of the WAE nexus problem and solving the model in a single
equilibrium distribution of groundwater balance and ecological
iteration, resulting in short computation times and high effi­
water utilization, the STRO method was introduced into the
ciency (Shih et al., 1996).
SEWT-WAE model. The developed STRO method not only
considered the spatial equilibrium but also reflected the spatial
Additionally, based on the comprehensive evaluation, the optimal
equilibrium. The results also indicated that the SEWT-WAE
decision scheme is obtained. In the scheme, irrigation surface water in
model was effective in reducing spatial and temporal disparities
Tabei irrigation district is increased, and irrigation groundwater in
in groundwater balance and ecological water utilization (Figs. 4
Tanan irrigation district is decreased, leading to a better groundwater
and 5).
balance in each irrigation district and relieving the spatial inequilibrium
(2) In terms of conjunctive utilization between surface water and
of groundwater in different irrigation districts (Figs. 6 and 7). Compared
groundwater, a coupled simulation-optimization approach was
to the current situation, the irrigated areas of garden and grain crops are
utilized to obtain the optimal water allocation (Song et al., 2020).
increased under dry years, which is in line with the demands of local
However, in the optimization part, the available water resources
agricultural development and increased total grain yield (Figs. 8 and 9).
in each region were constrained, and it is difficult to optimize the
The ecological water utilization of the Populus euphratica forest is
allocation among diverse regions. Based on the water balance
increased, which is conducive to spatial equilibrium in the ecological
method, we constructed an LWT model and integrated it as a
water utilization of different ecological zones (Fig. 10). This indicates
constraint condition into the SEWT-WAE model to achieve
that a scientific and reasonable scheme can be obtained based on this
optimal water resource allocation. Compared to previous
study.
research, this model can systematically coordinate surface and
Due to varying local conditions, the obtained decision scheme cannot
groundwater resources between upstream and downstream areas,
be directly applied to other regions. However, by adjusting model pa­
and can further achieve more efficient water resource allocation
rameters and input data according to the specific conditions of different
(Figs. 7 and 10).
areas, the SEWT-WAE model can be applied to other regions. In addi­
(3) Compared to employing constraints like Nash equilibrium or
tion, due to the limited data, the SEWT-WAE model achieved temporal
coordination development degree, this model uses the STRO

Fig. 13. Hydrological frequency curve.

12
T. Wang et al. Agricultural Water Management 303 (2024) 109061

equilibrium on an annual scale and spatial equilibrium on irrigation conditions across different areas, the obtained decision scheme cannot
district and ecological zone scales but did not achieve temporal and be directly applied elsewhere. However, by adjusting model parameters
spatial equilibrium on a monthly or grid scale. Furthermore, the linear and input data according to local conditions, the model can be applied in
water transformation model was simplistic. In this model, for surface other similar study areas. This study offers valuable insights into the
water, the natural streamflow was used to substitute water yield, which spatiotemporal equilibrium of groundwater balance and ecological
failed to detail the water yield process; for groundwater, the ground­ water utilization as well as conjunctive use between surface water and
water balance difference for irrigation areas was only calculated, but the groundwater. Based on the STRO model, future research can further
change in groundwater level was not described. explore how to achieve temporal equilibrium on a monthly scale and
spatial equilibrium on a grid scale. Besides, the water transformation
5. Conclusions model can be further enriched (e.g., coupling water quality processes
with the water transformation model) to obtain more efficient and
In this study, the SEWT-WAE model was proposed to support scientifically sound decision-making schemes.
collaborative management of the WAE nexus in the Aksu River Basin.
The SEWT-WAE model has advantages in supporting spatiotemporal CRediT authorship contribution statement
equilibrium of groundwater balance and ecological water utilization, as
well as coordinating water transformation in upstream and downstream Taishan Wang: Writing – original draft, Software, Methodology,
basins. Then 36 scenarios were designed considering different spatio­ Formal analysis, Data curation, Conceptualization. Xiaoling Su: Writing
temporal equilibrium tendencies. Several findings were revealed: (1) – review & editing, Supervision, Project administration, Methodology.
High spatiotemporal equilibrium tendency increased the expected Haijiang Wu: Writing – review & editing.
groundwater balance of Tanan irrigation district and the expected
ecological water utilization of Eichmann Lake wetland and Populus Declaration of Competing Interest
euphratica forest while reducing their inter-annual variations through
introducing the STRO method. (2) Based on the water transformation The authors declare that they have no known competing financial
process between upstream and downstream, the water resources allo­ interests or personal relationships that could have appeared to influence
cation scheme of Aksu was optimized, increasing the groundwater the work reported in this paper.
recharge of Tabei and Tanan irritation districts, by incorporating the
water transformation model into the optimization model. (3) In the Data availability
optimal scenario during dry years, compared to the current situation,
irrigation amount of surface water (groundwater) in the Tabei (Tanan) The data that has been used is confidential.
irrigation district was increased (decreased) by 21.4 % (70.2 %); the
irrigated area of grain crops, garden and cotton was increased by Acknowledgment
30.4 %, increased by 20.1 %, and decreased by 19.4 %, respectively;
ecological water utilization of the Populus euphratica forest was This work was supported by the National Key Research and Devel­
increased by 17.8 %. Additionally, due to varying natural and social opment Program of China (2022YFD1900501).

Appendix A. The specific data description and sources

Data description Time Data source

interval

Irrigated area, irrigation volumes, crop yield, surface water, and 2004–2019 First division water management comprehensive statistical report that was provided by Aksu
groundwater utilization in each irrigated area Water Resources Bulletin
The benefits and costs of wheat, rice, cotton, coarse, coarse cereals, 2020 National Agricultural Product Cost and Benefits Data Compilation− 2021 (National
pepper, melons and vegetables, and potato Development and Reform Commission and Price Cost Survey Center, 2021)
The benefits and costs of garden and alfalfa 2019 Analysis of the production efficiency and effectiveness for different irrigation patterns of
Alfalfa in Hexi Corridor (Gao, 2023)
Analysis and Research on Comprehensive Benefit of Jujube Industry in Fourteenth Division
(Wang, 2022)
The situation of the water conservancy project \ The comprehensive water resources planning report for Alar City
Research on joint operation and spatial equilibrium allocation of water resources in the three
source reservoir group of the Tarim River (Zhao, 2023)
Supply and Demand Analysis of Water Resources in Tailan River Basin (Rousidan, 2021)
Streamflow of Xiehela, Shaliguilanke, Xiaota and Aral 2004–2019 Aksu Water Resources Bureau
Meteorological data 2004–2019 China Meteorological Dataset (China Meteorological Data Service Center, 2023)
Water demand of ecology from the inside river channel \ Study on water quality evolution and ecological water requirement in Tarim River Basin (Dong,
2018)
Research on joint operation and spatial equilibrium allocation of water resources in the three
source reservoir group of the Tarim River (Zhao, 2023)
Water demand of ecology from the outside river channel \ Operation optimization of ecological water conveyance in Aksu River Basin based on ecological
restoration (Nie et al., 2022)
Research on joint operation and spatial equilibrium allocation of water resources in the three
source reservoir group of the Tarim River (Zhao, 2023)
Water utilization of ecology from the inside river channel \ Streamflow of Alar hydrological station Aksu provided by Water Resources Bureau
Water utilization of ecology from the outside river channel 2018–2019 The 2019 ecological water dispatch implementation in Aksu Prefecture
The 2018 ecological water transfer plan for the Aksu area

13
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Appendix B. Modeling formulation and Nomenclature of the SEWT-WAE model

Modeling formulation
Objective functions
Upper-level objective
To ensure local economic development, maximizing system benefit is set as an upper-level objective, in which target irrigation benefit and costs for
water deficit of each irrigation district and crop are considered (the description of subscripts, coefficients, and decision variables in the SEWT-WAE
models can be seen in Nomenclature).
[ ]
∑ I ∑ J T I ∑ J
( ) ∑ ∑ ( / )
max fU = Bj × FXtij × Xij − pt × CWDj × WDtij WXtij (B1)
i=1 j=1 t = i=1 j=1

Lower-level objective (water system)

Water resources are one of the important factors in agricultural basin, maintaining groundwater balance is key to ensuring the sustainable
development of agriculture and ecosystems. GWEti is introduced as the groundwater balance value. When maximizing GWEti is set as objective if the
groundwater balance difference is less than 0, GWEti equals it, otherwise, GWEti equals 0 (see constraints (B18) and (B19)). Maximizing the difference
between groundwater balance value and the associated spatiotemporal disequilibrium costs is set as a lower-level objective relaxed to the water
system. Adjusting the TE and SE can reflect the spatiotemporal equilibrium tendency of decision-makers.
( ) ( )
∑T ∑I ∑ T ∑I
max fLW = pt × GWWti − TE × pt × VWti
t=1 i=1 t=1 i=1
T
(
I
) (B2)
∑ ∑
− SE × pt × UWti
t=1 i=1

Lower-level objective (Agriculture system)

To ensure food security is the main purpose of agricultural production, maximizing total grain yield is set as a lower-level objective relaxed to the
agriculture system, in which the yields of wheat, rice, and coarse grains are considered.
[ ( )]
∑T ∑I ∑2
max fLA = pt × SFtij + SFti4 (B3)
t=1 i=1 j=1

Lower-level objective (ecology system)

A healthy ecosystem plays a crucial role in preventing soil erosion and farmland encroachment, thus ensuring the sustainable development of the
basin. Therefore, maximizing the difference between the ecological water utilization and the associated spatiotemporal disequilibrium costs is set as a
lower-level objective relaxed to the ecology system, in which Populus euphratica forest, the Eichmann Lake wetland, and Tahe River are considered.
( ) ( )
∑ T ∑E ∑T E
∑
max fE = pt × OEWte − TE × pt × VEte
t=1 e=1 t=1 e=1
T
(
E
) (B4)
∑ ∑
− SE × pt × UEte
t=1 e=1

Constraints
Constraints for irrigation water:
The target irrigation water for each irrigation district and crop in the growth period equals the product of its target irrigated area and the irrigation
volumes in the growth period.
IRWtij = WXtij × Xij (B5)

The net irrigation water in the growth period for each irrigation district and crop is defined as the difference between the target irrigation water
and the water deficit. It comprises the product of surface water extraction and the effective utilization coefficient of surface water, along with the
product of groundwater extraction and the effective utilization coefficient of groundwater.
IRWtij − WDtij = UIti × UOti × IRSWtij + UIti × IRGWtij (B6)

As there is no well irrigation in the Shajingzi irrigation district, the irrigation groundwater in the Shajingzi irrigation district is set to 0.
IRGWtij = 0, i = 1 (B7)

Constraints for irrigated area:

The actual irrigated area of each irrigation district and crop is defined as the difference between the target irrigated area and the area experiencing
yield reduction due to water deficit.
/
AXtij = Xij − WDtij WXtij (B8)

The total target irrigated area should not exceed the maximum irrigated area for each irrigation district.
J
∑
Xij ≤ MAXXi (B9)
j=1

The target irrigated area should not exceed the maximum suitable irrigated area for each irrigation district and crop.
Xij ≤ MAXIij (B10)

14
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Constraints for crop yield:

The yield of each crop and irrigation district equals the product of the actual irrigated area and the yield per unit irrigated area.
SFtij = FXtij × AXtij (B11)

The total yield of each crop in each hydrological year should not be lower than the minimum yield requirement.
I
∑
SFtij ≥ DFj (B12)
i=1

The total grain yield in each hydrological type should not be lower than the current level.
I ∑
∑ 2 I
∑
SFtij + SFti4 ≥ FAN (B13)
i=1 j=1 i=1

Net system benefits in each hydrological type should not be lower than the current level.
I
∑ J
∑ I ∑
J
( ) ∑ ( / )
Bj × FXtij × Xij − CWDj × WDtij WXtij ≥ FBN (B14)
i=1 j=1 i=1 j=1

Water transformation model:

Based on the water balance, the outflow at each intake point is set as the difference between the water inflow at the intake point and the water
supply, where the water supply includes basin water diversion, ecological water utilization, other non-irrigation water in the basin, and irrigation
surface water in of the First Region irrigation district.
Note: RWEwe is a known parameter, RWEwe = 1 (RWEwe = 0) means that water intake point w can (cannot) supply water to ecological zone e; RWIwi
is also a known parameter, RWIwi = 1 (RWIwi = 0) means that water intake point w can (cannot) supply water to irrigation district i.
∑E
WIRtw = IURtw − OIWw − (RWEwe × OEWte ) − OSWw
e=1
I
{ [ / /
J
]} (B15)
∑ ∑
− RWIwi × ODSWi + (THZWi + BIRWi ) UIi UOi + IRSWtij
i=1 j=1

The upstream inflow at each water intake point includes the outflow from its upstream water intake points, natural runoff between this water
intake point and its upstream water intake points, inflow from Xiehela, Shaliguilanke, and Xiaota upstream, irrigation return flow from irrigation
districts that do not belong to the First Region irrigation district, and the irrigation return flow from the First Region irrigation districts.
Note: IOVWvw and OVWvw are known parameters, IOVWvw = 1 (IOVWvw = 0) means that the water intake point v is (not) an adjacent upstream
water intake point of the water intake point w; OVWvw = 1 (OVWvw = 0) means that other water intake points v are (not) water intake point w, and the
discharge from water intake point v does (not) equal the discharge from water intake point w.
V
∑
IURtw = (IOVWvw × OWVtv ) + NRtw + ORWtw + BRWw × OIRWw
v=1
I
[ (
J J
)] (B16)
∑ ∑ ∑
+ IOIWiw × BRi × IRWtij − WDtij + BISWi + BIGWi + THZWi
i=1 j=1 j=1

W
∑
OWVtv = (OVWvw × WIRtw ) (B17)
w=1

GWWti is defined as the groundwater balance value of irrigation district i in hydrological type t, which does not exceed the groundwater balance
difference and maximum at 0.
GWWti ≤ GWti (B18)

GWWti ≤ 0 (B19)
The groundwater balance difference should not be less than the current situation in each irrigation district and hydrological type.
GWti ≥ NGWti (B20)
Groundwater balance difference equals the difference between groundwater recharge and discharge in each irrigation district and hydrological
type.
GWti = GWBti − GWPti (B21)
The groundwater recharge includes rainfall infiltration, lateral inflow recharge, river seepage, canal system seepage, reservoir recharge, canal
irrigation infiltration, and well irrigation return.
GWBti = GWJti + LIFi + GWHti + GWQti + RLRi + GWQTti + GWJTti (B22)
The groundwater discharge includes groundwater evaporation, lateral outflow, well irrigation extraction, other extractions, spring overflow, and
drainage ditch discharge.
GWPti = EVDi + LOFi + GWJKti + OGWi + SOFi + DCDWi (B23)
Rainfall infiltration equals the product of rainfall infiltration coefficient and rainfall.

15
T. Wang et al. Agricultural Water Management 303 (2024) 109061

GWJti = PRIi × PREti (B24)

River seepage equals the product of the river seepage coefficient and the runoff of the adjacent river.
W
∑
GWHti = (CSWIwi × WIRtw ) (B25)
w=1

Canal system seepage equals the product of canal system seepage coefficient and surface water extraction.
[ / / ]
J
∑
GWQti = CLCi × IRSWtij + (BISWti + THZWti ) UIti UOti + ODSWi (B26)
j=1

Canal irrigation infiltration equals the product of canal irrigation infiltration coefficient and effective surface water utilization for irrigation.
[ ]
∑J
GWQTti = CICi × IRSWtij × UIti × UOti + (BISWti + THZWti ) (B27)
j=1

Well irrigation return equals the product of well irrigation return coefficient and effective groundwater utilization for irrigation.
( )
∑J
GWJTti = WICi × BIGWti + IRGWtij × UIti (B.28)
j=1

Groundwater extraction from wells includes groundwater extraction for crop irrigation during the growing period, pre-sowing groundwater
extraction for irrigation, and extraction of other groundwater.
/
∑ J
GWJKti = IRGWtij + BIGWti UIti + OGWIti (B29)
j=1

Constraints for ecological water utilization:

The ecological water utilization of the third ecological zone is defined as the discharge of the Tahe River, namely the runoff at the Alar hydrological
station.
OEWte = WIRtw , e = 3, w = 8 (B30)
The annual average discharge of the Tahe River must not fall below the minimum discharge.
T
∑
(pt × WIRtw ) ≥ AAWSw (B31)
t=1

The water supply to primary ecological zones should not fall below the minimum ecological water utilization requirement and should not exceed
the suitable ecological water utilization requirement.
BOERe ≤ OEWte ≤ MOERe (B32)
Constraints for spatiotemporal equilibrium:
When spatial (temporal) equilibrium is considered in an objective, if the groundwater balance value of each irrigation district or ecological water
utilization of each ecological zone in different hydrological years is less than the weighted average with area of the irrigation district (occurrence
frequency of hydrological year) as weights, its spatial (temporal) recourse cost about groundwater balance value or ecological water utilization is
defined as the difference between it and its weighted average; otherwise, the cost equals 0.
UWti /AIi ≥ EUWt − GWWti /AIi (B33)
/
I
∑ I
∑
EUWt = GWWti AIi (B34)
i=1 i=1

UEte /MOERe ≥ EUEt − OEWte /MOERe (B35)

/
E
∑ E
∑
EUWt = OEWti MOERe (B36)
e=1 e=1

VWti ≥ EVWi − GWWti (B37)

T
∑
EVWi = (pt × GWWti ) (B38)
t=1

VEte ≥ EVEe − OEWte (B39)

T
∑
EVEe = (pt × OEWte ) (B40)
t=1

Constraints for non-negativity variables:

Except for GWWti, GWti, EVWti, and EUWti, all variables are non-negative.

16
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Nomenclature
Subscripts
e Ecological zones with e = 1, 2, 3 for Eichmann Lake wetland, Populus euphratica forest, mainstream of Tahe River
i Irrigation districts with i = 1, 2, …, 5 for Shajingzi irrigation district, Tuo irrigation district, Ku irrigation district, Tabei
irrigation district, Tanan irrigation district
j Crops with j = 1, 2, …, 11 for wheat, rice, cotton, coarse cereals, garden, pepper, alfalfa, melons and vegetables, potato, other
crops, Tuohaixiang (comprehensive)。
t hydrologic years with t = 1, 2, 3 for dry years, normal years, wet years.
w Water intake points/hydrometric stations with w = 1, 2, …, 8 for yuejin-qiugeer-lianhe headwork, hongxing trunk canal,
xiehela diversion junction, qiagelake main canal, geming-duolang headwork, xidaqiao diversion junction, yimapaxia diversion
junction, aral hydrographic station
v Other diversion junctions/hydrometric stations with v = 1, 2, …, 8 for yuejin-qiugeer-lianhe headwork, hongxing trunk canal,
xiehela diversion junction, qiagelake main canal, geming-duolang headwork, xidaqiao diversion junction, yimapaxia diversion
junction, aral hydrographic station
Coefficients
FBN, FNA System benefit and total grain yield under the current situation (104 yuan, 104 kg)
BOERe, MOERe The current and suitable ecological water utilization of ecological zone e (104 m3)
BRi Recession coefficient of irrigation district i
BRWw Recession coefficient of irrigation district between water intake point w and adjacent upstream water intake point.
CICi, CLCi, DCDWi, PRIi, WICi Canal irrigation infiltration, canal system seepage, drainage ditch discharge, rainfall infiltration, well irrigation return
coefficient of irrigation district i
CSWIwi River seepage coefficient of water intake point w in irrigation district i.
EVDi, LIFi, LOFi, RLRi, SOFi, OGWi, Phreatic evaporation, lateral recharge, lateral discharge, reservoirs recharge, spring overflow other artificial exploitation of
irrigation district i (104 m3)
MAXXi The biggest irrigated area of the irrigation district i (ha)
UIi, UOi Utilization rate of canal system outside and inside irrigation district i
Bij Benefit per unit yield of crop j in irrigated area i (yuan/kg)
FXtij, WXtij Yield and water consumption per irrigated area of crop j in irrigation district i under hydrologic year t (104 kg/ha, 104 m3/ha)
IOIWiw If IOIWiw = 1, irrigation district i discharges water to the river between the water intake point w and the upstream adjacent
water intake points; if IOIWiw = 0, it is not.
DFj Yield demand for crop j (104 kg)
pt Probability of occurrence of hydrological year t
PRE ti Rain of irrigation district i during hydrological year t (mm)
OIWw Water transferred to Tailan River basin from water intake point w (104 m3)
OIRWw Other irrigation water from adjacent water intake points upstream of water intake point w (104 m3)
OSWw Other water consumption of the irrigation district that does not belong to the First Region from water intake point w (104 m3)
NRtw Natural runoff between water intake point w and upstream water intake points during hydrological year t (104 m3)
ORWtw Water inflow from upstream outside the study area to water intake point w during hydrological year t (104 m3)
IOVWvw If IOVWvw = 1, water intake point v is adjacent to water intake point w and upstream of water intake point w, the outflow of
water intake point v is included in the inflow of water intake point w; if IOVWvw = 0, it is not.
OVWvw If OVWvw = 1, the water intake point v is the water intake point w; if OVWvw = 0, it is not.
AAWSw When AAWSw = 8, the parameter is the annual average discharge demand of the mainstream of the Tahe River; otherwise, the
parameter equals 0 (104 m3)
RWEwe If RWEwe = 1, water intake point w supplies water to ecological zone e; if RWEwe = 0, it does not.
RWIwi If RWIwi = 0, water intake point w supplies water to irrigation district i; if RWIwi = 1, it does not.
Decision variables
Xij Objective irrigated area of crop j in irrigation district i (ha)
OEWte ecological water utilization of ecological zone e during hydrological year t (104 m3)
AXtij, IRWtij, IRGWtij, IRSWtij, SFtij, WDtij Actual irrigated area, irrigation water, irrigation groundwater, irrigation surface water, total grain yield, and Water deficit of
crop j in irrigation district i during hydrological year t (ha, 104 m3, 104 m3, 104 m3, 104 kg)
OWVtv The outflow of other water intake v during hydrological year t (104 m3)
IURtw Outflow of water intake w during hydrological year t (104 m3)
WIRtw Runoff after water intake at water intake point w during hydrological year t (104 m3)
GWti, GWBti, GWPti, GWJti, GWHti, GWQti, GWQTti, Groundwater balance difference, Groundwater recharge, Groundwater discharge, Rainfall infiltration, River seepage, Canal
GWJTti, GWJKti, GWWti system seepage, Canal irrigation infiltration, Well irrigation return, Well irrigation exploitation, and Groundwater balance
value of irrigation district i during hydrological year t (104 m3)
VWti, VEte Temporal recourse cost relaxed to groundwater balance value of irrigation district i ecological water utilization of ecological
zone i and during hydrological year t (104 m3)
UWti, UEte Spatial recourse cost relaxed to groundwater balance value of ecological zone i and ecological water utilization of ecological
zone i during hydrological year t (104 m3)
EVWi, EVEe Using the probabilities of occurrence for each hydrological year as weights, the weighted average groundwater balance value of
irrigation district i and ecological zone e (104 m3)
EUWt Using the current irrigated area of each irrigation district as weights, the weighted average groundwater balance value during
hydrological year t (104 m3)
EUEt Using the ecological water utilization demand of each ecological zone as weights, the weighted average ecological water
utilization during hydrological year t (104 m3)

References Abbas, M., Zhao, L.S., Wang, Y.N., 2022. Perspective impact on water environment and
hydrological regime owing to climate change: a review. Hydrology 9 (11). https://
doi.org/10.3390/hydrology9110203.
China Meteorological Data Service Center, 2023. China Meteorological Dataset (2004-
Beykal, B., Avraamidou, S., Pistikopoulos, E.N., 2022. Data-driven optimization of
2019). https://data.cma.cn/en)%E3%80%82.
mixed-integer bi-level multi-follower integrated planning and scheduling problems
National Development and Reform Commission and Price Cost Survey Center, 2021.
under demand uncertainty. Comput. Chem. Eng. 156, 107551 https://doi.org/
National Agricultural Product Cost and Benefit Data Compilation-2021.
10.1016/j.compchemeng.2021.107551.

17
T. Wang et al. Agricultural Water Management 303 (2024) 109061

Dalin, C., Hanasaki, N., Qiu, H.G., Mauzerall, D.L., Rodriguez-Iturbe, I., 2014. Water Lu, J., Shi, C., Zhang, G., 2006. On bilevel multi-follower decision making: General
resources transfers through Chinese interprovincial and foreign food trade. Proc. framework and solutions. Inf. Sci. 176 (11), 1607–1627. https://doi.org/10.1016/j.
Natl. Acad. Sci. 111 (27), 9774–9779. https://doi.org/10.1073/pnas.1404749111. ins.2005.04.010.
Dalin, C., Outhwaite, C.L., 2019. Impacts of global food systems on biodiversity and Mulvey, J.M., Ruszczyński, A., 1995. A new scenario decomposition method for large-
water: the vision of two reports and future aims. One Earth 1 (3), 298–302. https:// scale stochastic optimization. Oper. Res. 43 (3), 477–490. https://doi.org/10.1287/
doi.org/10.1016/j.oneear.2019.10.016. opre.43.3.477.
Dehghanipour, A.H., Schoups, G., Zahabiyoun, B., Babazadeh, H., 2020. Meeting Mulvey, J.M., Vanderbei, R.J., Zenios, S.A., 1995. Robust optimization of large-scale
agricultural and environmental water demand in endorheic irrigated river basins: A systems. Oper. Res. 43 (2), 264–281. https://doi.org/10.1287/opre.43.2.264.
simulation-optimization approach applied to the Urmia Lake basin in Iran. Agric. Nie, Y., Guo, Y.R., Tan, Y., Huang, W.D., Liu, X.H., 2022. Operation optimization of
Water Manag. 241 https://doi.org/10.1016/j.agwat.2020.106353. ecological water conveyance in Aksu River Basin based on ecological restoration.
Dehghanipour, A.H., Zahabiyoun, B., Schoups, G., Babazadeh, H., 2019. A WEAP- Arid Land Geogr. 45 (2), 325–337. https://doi.org/10.12118/j.
MODFLOW surface water-groundwater model for the irrigated Miyandoab plain, issn.1000–6060.2021.294.
Urmia lake basin, Iran: Multi-objective calibration and quantification of historical Nie, Y., Yin, C., Wang, P., He, X.Y., Cao, J.J., Yu, J., 2022. Temporal and spatial evolution
drought impacts. Agric. Water Manag. 223 https://doi.org/10.1016/j. of Eichmann Lake Wetland in Aksu River basin and its response to ecological water
agwat.2019.105704. supply. Int. J. Environ. Res. Public Health 20 (1). https://doi.org/10.3390/
Dong, N., 2018. Study on water quality evolution and ecological water requirement in ijerph20010351.
Tarim River Basin (D). Xi’an University of Technology, Xi’an. https://doi.org/. Ortiz-Partida, J.P., Kahil, T., Ermolieva, T., Ermoliev, Y., Lane, B., Sandoval-Solis, S.,
Esmaeili, S., Anvari-Moghaddam, A., Jadid, S., 2019. Optimal operational scheduling of Wada, Y., 2019. A two-stage stochastic optimization for robust operation of
reconfigurable multi-microgrids considering energy storage systems. Energies 12 (9), multipurpose reservoirs. Water Resour. Manag. 33 (11), 3815–3830. https://doi.
1766. https://doi.org/10.3390/en12091766. org/10.1007/s11269-019-02337-1.
Fischer, S., Pluntke, T., Pavlik, D., Bernhofer, C., 2014. Hydrologic effects of climate Ouyang, R.L., Cheng, W.M., Wang, W.S., Jiang, Y., Zhang, Y.C., Wang, Y.Q., 2007.
change in a sub-basin of the Western Bug River, Western Ukraine. Environ. Earth Sci. Research on runoff forecast approaches to the Aksu River basin. Sci. China Ser. D.
72 (12), 4727–4744. https://doi.org/10.1007/s12665-014-3256-z. -Earth Sci. 50 (S1), 16–25. https://doi.org/10.1007/s11430-007-5008-3.
Gao, L.T., 2023. Analysis of the production efficiency and effectiveness for different Pander, J., Geist, J., 2013. Ecological indicators for stream restoration success. Ecol.
irrigation patterns of Alfalfa in Hexi Corridor (D). Lanzhou University, Lanzhou. Indic. 30, 106–118. https://doi.org/10.1016/j.ecolind.2013.01.039.
https://doi.org/10.27204/d.cnki.glzhu.2023.003740. Rousidan, T., 2021. Supply and demand analysis of water resources in Tailan River basin.
Gordon, L.J., Finlayson, C.M., Falkenmark, M., 2010. Managing water in agriculture for Water Sci. Technol. Econ. 27 (3), 65–71. https://doi.org/10.3969/j.issn.1006-
food production and other ecosystem services. Agric. Water Manag. 97 (4), 512–519. 7175.2021.03.014.
https://doi.org/10.1016/j.agwat.2009.03.017. Rumbaur, C., Thevs, N., Disse, M., Ahlheim, M., Brieden, A., Cyffka, B., Duethmann, D.,
Hassas, M.A., Kalantari, N.T., Mohammadi-Ivatloo, B., Safari, A., 2021. Sustainable Feike, T., Frör, O., Gärtner, P., Halik, Ü., Hill, J., Hinnenthal, M., Keilholz, P.,
management of the electrical-energy–water–food nexus using robust optimization. Kleinschmit, B., Krysanova, V., Kuba, M., Mader, S., Menz, C., Othmanli, H., Pelz, S.,
Sustainability 14 (1), 172. https://doi.org/10.3390/su14010172. Schroeder, M., Siew, T.F., Stender, V., Stahr, K., Thomas, F.M., Welp, M.,
Huang, G.H., 2000. An inexact two-stage stochastic programming model for water Wortmann, M., Zhao, X., Chen, X., Jiang, T., Luo, J., Yimit, H., Yu, R., Zhang, X.,
resources management under uncertainty. Civ. Eng. Environ. Syst. 17 (2), 95–118. Zhao, C., 2015. Sustainable management of river oases along the Tarim River
https://doi.org/10.1016/j.jenvman.2013.04.027. (SuMaRiO) in Northwest China under conditions of climate change. Earth Syst.
Jain, S., Ramesh, D., Trivedi, M.C., Edla, D.R., 2023. Evaluation of metaheuristic Dynam. 6 (1), 83–107. https://doi.org/10.5194/esd-6-83-2015.
optimization algorithms for optimal allocation of surface water and groundwater Sakawa, M., Nishizaki, I., Uemura, Y., 2000. Interactive fuzzy programming for multi-
resources for crop production. Agric. Water Manag. 279, 108181 https://doi.org/ level linear programming problems with fuzzy parameters. Fuzzy Sets Syst. 109 (1),
10.1016/j.agwat.2023.108181. 3–19. https://doi.org/10.1016/S0165-0114(98)00130-4.
Ji, J.W., Wang, S.X., Zhou, Y., Liu, W.L., Wang, L.T., 2021. Spatiotemporal change and Shih, H.S., Lai, Y.J., Lee, E.S., 1996. Fuzzy approach for multi-level programming
coordinated development analysis of “population-society-economy-resource- problems. Comput. Oper. Res. 23 (1), 73–91. https://doi.org/10.1016/0305-0548
ecology-environment” in the Jing-Jin-Ji Urban Agglomeration from 2000 to 2015. (95)00007-9.
Sustainability 13 (7), 4075. https://doi.org/10.3390/su13074075. Song, J., Yang, Y., Sun, X.M., Lin, J., Wu, M., Wu, J.F., Wu, J.C., 2020. Basin-scale multi-
Jia, D.N., Zhang, T., Wu, L.Z., Su, X.L., Bai, T., Huang, Q., 2024. Multi-objective objective simulation-optimization modeling for conjunctive use of surface water and
cooperative optimization of reservoir scheduling and water resources allocation for groundwater in northwest China. Hydrol. Earth Syst. Sci. 24 (5), 2323–2341.
inter-basin water transfer project based on multi-stage coupling model. J. Hydrol. https://doi.org/10.5194/hess-24-2323-2020.
630, 130673 https://doi.org/10.1016/j.jhydrol.2024.130673. Wang, C.X., Li, Y.P., Huang, G.H., Zhang, J.L., 2016. A type-2 fuzzy interval
Jin, S.W., Li, Y.P., Yu, L., Suo, C., Zhang, K., 2020. Multidivisional planning model for programming approach for conjunctive use of surface water and groundwater under
energy, water and environment considering synergies, trade-offs and uncertainty. uncertainty. Inf. Sci. 340-341, 209–227. https://doi.org/10.1016/j.ins.2016.01.026.
J. Clean. Prod. 259, 121070 https://doi.org/10.1016/j.jclepro.2020.121070. Wang, H., 2022. Analysis and Research on Comprehensive benefit of Jujube Industry in
Jing, J.N., Wang, W.K., Duan, L., Ma, J.J., Ma, Z.T., Shi, H.Y., 2023. An analysis of the Fourteenth Division (D). Tarim University, Alear. https://doi.org/10.27708/d.cnki.
changes in groundwater balance elements of the Baoding Plain. Hydrogeol. Eng. gtlmd.2022.000219.
Geol. 50 (4), 115–126. https://doi.org/10.16030/j.cnki.issn.1000-3665.202208083. Xu, Y., Huang, G.H., Qin, X.S., 2009. Inexact two-stage stochastic robust optimization
Karatzas, G.P., Yetilmezsoy, K., Varouchakis, E.A., 2019. A decision-making framework model for water resources management under uncertainty. Environ. Eng. Sci. 26
for sustainable management of groundwater resources under uncertainty: (12), 1765–1776. https://doi.org/10.1089/ees.2009.0212.
combination of Bayesian risk approach and statistical tools. Water Policy 21 (3), Zang, L.Z., Su, Y.Q., 2019. Internal coordinated development of China’s Urbanization
602–622. https://doi.org/10.2166/wp.2019.128. and its spatiotemporal evolution. Sustainability 11 (3), 626. https://doi.org/
Li, M., Zhang, C., 2020. Two-stage stochastic variational inequality arising from 10.3390/su11030626.
stochastic programming. J. Optim. Theory Appl. 186 (1), 324–343. https://doi.org/ Zeinali, M., Azari, A., Heidari, M.M., 2020. Multiobjective optimization for water
10.1007/s10957-020-01686-x. resource management in low-flow areas based on a coupled surface
Li, M.S., Yang, X.H., Di, C.L., Wang, K.W., Xiang, W.Q., Bian, D.H., Babuna, P., 2024. water–groundwater model. J. Water Resour. Plan. Manag 146 (5), 04020020.
A novel framework for integrative assessment of water balance health in China. https://doi.org/10.1061/(ASCE).
J. Clean. Prod. 443, 141199 https://doi.org/10.1016/j.jclepro.2024.141199. Zhang, J.L., Li, Y.P., You, L., Huang, G.H., Xu, X.M., Wang, X.Y., 2022. Optimizing
Li, M.S., Yang, X.H., Wu, F.F., Babuna, P., 2022. Spatial equilibrium-based multi- effluent trading and risk management schemes considering dual risk aversion for an
objective optimal allocation of regional water resources. J. Hydrol. Reg. Stud. 44, agricultural watershed. Agric. Water Manag. 269 https://doi.org/10.1016/j.
101219 https://doi.org/10.1016/j.ejrh.2022.101219. agwat.2022.107716.
Li, W., Liu, M., Wu, S.Z., Xu, Y., 2014. An inexact optimization model associated with Zhao, W.Z., Chang, X.L., 2014. The effect of hydrologic process changes on NDVI in the
two robust programming approaches for water resources management. Int. J. desert-oasis ecotone of the Hexi Corridor. Sci. China-Earth Sci. 57 (12), 3107–3117.
Environ. Sci. Technol. 12 (7), 2401–2414. https://doi.org/10.1007/s13762-014- https://doi.org/10.1007/s11430-014-4927-z.
0639-3. Zhao, Y.X., 2023. Research on joint operation and spatial equilibrium allocation of water
Loboda, N.S., Glushkov, A.V., Khokhlov, V.N., 2005. Using meteorological data for resources in the three source reservoir group of the Tarim River (D). Xi’an University
reconstruction of annual runoff series over an ungauged area: Empirical orthogonal of Technology, Xi’an. https://doi.org/10.27398/d.cnki.gxalu.2023.000165.
function approach to Moldova–Southwest Ukraine region. Atmos. Res. 77 (1-4), Zhu, X.Y., Su, X.L., Hu, X.X., Chu, J.D., Jia, D.N., Wu, H.J., Zhang, T., 2024. Optimal
100–113. https://doi.org/10.1016/j.atmosres.2004.10.020. allocation of multiple water sources in arid areas based on water-agriculture-ecology
coordinated regulation. Trans. Chin. Soc. Agric. Eng. 40 (2), 239–248. https://doi.
org/10.11975/j.issn.1002-6819.202307012.

18