Multi-layer multi-objective cooperative regulation of agricultural water resources in large agricultural irrigation areas based on runoff prediction
Introduction
Globally, agricultural irrigation water for food production is one of the largest users of freshwater resources, accounting for approximately 70% of annual freshwater withdrawals (Saccon, 2018). Continuously growing populations and persistent socioeconomic development require the efficient use of agricultural water resources for agricultural production under the condition of a limited water supply. Water allocation is an effective measure to achieve the rational, efficient and sustainable use of agricultural water resources (Playan and Mateos, 2006). Irrigation districts carry the heavy burden of agricultural production, and the efficient utilization of limited regional water resources to promote sustainable agricultural development is an important issue facing agricultural development at present. With the continuous advancement of research and technology, the connotation of the optimal allocation of agricultural water resources has been increasingly enriched (Ge et al., 2013, Homayounfar et al., 2014, Jiang et al., 2019a, Yu and Shang, 2016). The research scope has been expanded from optimal water allocation at the farm and crop scales to multilevel decision making and joint optimization of different water sources at the regional scale (Li et al., 2022, Li et al., 2020b, Marques et al., 2015, Ren et al., 2019). Research objectives have also been expanded from simply considering the economic benefits of water allocation to simultaneously considering multiobjective problems such as the economic and social effects of water allocation. Mathematical descriptions of optimization problems have begun to consider the trade-offs between different aspects and use multiple objectives to solve allocation optimization problems on different scales (Li et al., 2017a).
The large-system decomposition–coordination method provides the advantage of decomposing complex problems into combinations of relatively simple optimization problems, thereby reducing the optimization difficulty, and it is an effective way to solve the multilevel optimization problem of regional agricultural water allocation. The theory of large-system optimization and coordination with a multilevel recursive structure is an effective way to solve multilevel optimization problems by dividing the whole irrigation system into several interrelated levels according to different irrigation processes, thus producing a recursive structure, and within the levels, optimal allocation of water resources is first separately conducted. The results are then regulated and fed back layer by layer under the overall goal of the multilevel large system, yielding optimal water resource allocation results for the large system. For example, Ahmad constructed a two-level model with upper-level decision makers and lower-level followers to coordinate and optimize water resources (Ahmad et al., 2018), Li et al. developed a two-level planning model to coordinate the interests of decision makers at different levels in the irrigation area (Li et al., 2017b), and Jiang et al. developed a water supply allocation model with a two-level step structure to optimize water allocation involving multistage pumping stations (Jiang et al., 2019b). The large-system recursive model has been widely applied in regional water allocation problems.
With the increase in research depth, in progressive regulation of multilevel agricultural water resources, in addition to the usual consideration of planting benefits, the trade-off between water savings and benefits has been considered within the context of water scarcity, and the economic, social and environmental effects of the entire irrigation area have been accounted for. Coupling multiobjective optimization with a large system model for the optimal allocation of agricultural water resources has become an effective tool to promote the sustainable use of water resources in irrigation areas. For example, Jiang et al. developed a three-level stepwise regional irrigation water optimization model with the objectives of water scarcity and energy consumption minimization(Jiang et al., 2019a). Li et al. established a two-level multiobjective large system model that simultaneously considers farmers' benefits and coordinated economic, social, and environmental development of the irrigation area (Li et al., 2021). Zhang et al. proposed a multiobjective three-level stochastic optimal water allocation model that simultaneously considers water allocation economic benefits, equity, efficiency, and environmental impacts in the development of sustainable water allocation schemes for arid agricultural areas (Zhang et al., 2020). However, the existing studies on multilevel and multiobjective regulation of agricultural water resources insufficiently consider water distribution at different fertility stages of farmland, water transmission through backbone channels in irrigation areas and optimal regional water distribution within a system at the same time, in addition to weighing farmers' income, water distribution fairness, economic benefits of water distribution and scale of the engineering volume. In contrast, considering these objectives could effectively optimize the water use structure in the region and promote the efficient utilization of regional agricultural water resources.
The large number of irrigation areas in large agricultural irrigation districts and the lack of data on certain irrigation areas generate challenges in water resource allocation, low accuracy and data collection difficulties, resulting in a low efficiency and accuracy of water resource allocation in large systems. Therefore, how to efficiently address the problems of a large number of irrigation areas and lack of information in large agricultural irrigation districts is necessary to solve the problem of large-system agricultural water resource regulation. Data mining focuses on correlation analysis without the need for strict logical reasoning and convergence analysis; thus, this approach can avoid the complicated process of objective function modeling and boundary condition analysis to achieve the extraction of noteworthy, implicit and valuable knowledge from massive data (Bao et al., 2020). Irrigation data mining in agricultural water resource large-system regulation can effectively match irrigation districts with similar characteristics and reduce the difficulty of acquiring irrigation district data information. Combining agricultural water resource large-system regulation with data mining can reduce the complexity of large-system problem solving and improve the water resource allocation efficiency.
Large agricultural irrigation districts usually carry important responsibilities in regard to food production and regional agricultural development, and regional decision makers are concerned not only about the current situation but also about future changes when allocating agricultural water resources. Water is a key factor in agricultural production, and the water supply in irrigation districts mainly originates from runoff. There are multiple rivers in large irrigation districts, and accurate runoff forecasting therefore becomes a key constraint in regional agricultural water allocation. Whereas studies have reported that a single model suffers the disadvantage of difficult adaptation to different watersheds, a combination of multiple models can represent a suitable approach to improve the prediction accuracy by adjusting parameters according to watershed runoff characteristics (Zhao et al., 2021). There have been comparative preferences of single runoff prediction models and their use for runoff allocation among multiple farms and irrigation districts. (Khan and Harris, 2020, Khan and Noor, 2019; ). Therefore, to achieve efficient water resource utilization in large agricultural irrigation areas, it is necessary to combine comprehensive runoff forecast models with large-system water allocation (Huang et al., 2019, Li et al., 2020a, Zhang et al., 2022; ).
To this end, this paper established a three-layer system water distribution model of farmland-irrigation district-region, which considers the mutual feedbacks of the water supply and demand in each layer, thereby weighing the interests of multiple subjects such as farmers' income, water distribution equity, regional economy and water diversion projects. This system could yield efficient and stable decisions on the irrigation water amount during different crop growth periods, water flow in backbone channels and surface and groundwater allocation to irrigation districts at different water supply levels within a large system. To explore future water allocation plans, a combined prediction model for runoff originating from different rivers was developed, and the cluster matching method was integrated with irrigation district matching to improve the decision efficiency. This study aimed to solve the problem of the efficient and sustainable allocation of agricultural water resources in complex systems comprising large agricultural irrigation areas under fluctuations in the water supply through multilayer and multiobjective cooperative regulation of water resources among irrigation systems.
Section snippets
Runoff prediction model
To improve the prediction accuracy of the runoff prediction model, a combined runoff prediction model could be used, and the concept of a combined prediction model was first proposed by Bates (Bates and Granger, 2001). The basic principle indicates that a number of distinct single deterministic models are used to predict basin hydrometeorological factors at a given prediction moment, and finally, the single deterministic models are combined in a linear weighted form to obtain combined
Study area
Located in the eastern part of Heilongjiang Province, China (43°49′55″N-48°27′20″, 129°11′20″E-135°5′10″E), the Sanjiang Plain is a low plain formed via alluvial accumulation of the Heilongjiang, Ussuri and Songhua Rivers. The total area of the Three Rivers Plain reaches approximately 108,900 km2, with an annual production of 15 million tons of grain. The per capita arable land area and per capita grain production are more than four times the national average levels, and the scale of
Runoff simulation results and quantification of the water supply
In this paper, runoff simulations were performed using long-term runoff monitoring data from 1985 to 2015 recorded at eight hydrological stations, including HL, BQ, and BQL, thereby considering three surface water source runoff levels and four runoff simulation models, i.e., the ARIMA, SVM, BP neural network, and combined ARIMA-SVM-BP (A-S-B) models, and adopting 1985–2015 as the rate period and 2015–2020 as the test period. The runoff prediction results are shown in Fig. 5.
Among them, the
Conclusion
To address the problem of the optimal allocation of water resources within the regional-irrigation district-agricultural water distribution problem, in this study, a large system for optimal allocation of water resources in large-area agricultural irrigation districts was constructed with the synergistic objectives of maximizing the income of farmers, maximizing the economic output and fairness of water distribution in irrigation districts, maximizing the benefits of the regional water supply
CRediT authorship contribution statement
Mo Li: Funding acquisition, Methodology, Formal analysis, Writing – review & editing. Wuyuan Liu: Methodology, Software, Writing – original draft. Qiang Fu: Methodology, Supervision. Dong Liu: Formal analysis, Investigation. Tianxiao Li: Software, Validation. Yaowen Xu: Visualization. Ruochen Shang: Data curation.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This research was supported by the National Natural Science Foundation of China (No. 52222902, No. 52079029 and No. 42277492), Natural Science Foundation of Heilongjiang Province (YQ2022E006), and Key Laboratory Fund of Efficient Use of Agricultural Water Resources, Ministry of Agriculture and Rural Affairs of the People′s Republic of China (AWR2021001)
References (26)
- et al.
A decision support system for irrigation water allocation along the middle reaches of the Heihe River Basin, Northwest China
Environ. Model. Softw.
(2013) - et al.
Optimizing regional irrigation water allocation for multi-stage pumping-water irrigation system based on multi-level optimization-coordination model
J. Hydrol. X
(2019) - et al.
Mid- and long-term runoff prediction based on time-varying weight combination and Bayesian correction
Prog. Geogr.
(2020) - et al.
Sustainable management of agricultural water and land resources under changing climate and socio-economic conditions: a multi-dimensional optimization approach
Agric. Water Manag.
(2022) - et al.
Managing agricultural water and land resources with tradeoff between economic, environmental, and social considerations: a multi-objective non-linear optimization model under uncertainty
Agric. Syst.
(2020) - et al.
An intuitionistic fuzzy multi-objective non-linear programming model for sustainable irrigation water allocation under the combination of dry and wet conditions
J. Hydrol.
(2017) - et al.
Multi-scale modeling for irrigation water and cropland resources allocation considering uncertainties in water supply and demand
Agric. Water Manag.
(2021) - et al.
Multi-objective optimization of water distribution systems based on a real options approach
Environ. Model. Softw.
(2015) - et al.
Modernization and optimization of irrigation systems to increase water productivity
Agric. Water Manag.
(2006) - et al.
Integrated multi-objective stochastic fuzzy programming and AHP method for agricultural water and land optimization allocation under multiple uncertainties
J. Clean. Prod.
(2019)