Construction and application of the cloud model service platform for water cycle simulation and water resources configuration

A cloud model of water resources has been explored in the early stages in recent years. However, it is still lacking on the cloud model service platforms for the water cycle simulation and water resources allocation. This study aims to construct the water cycle models for water resource allocation using multi-objective optimization. A modular modeling approach was also adopted with the model library as the core. The B/S architecture pattern was utilized to configure the basic network structure. The cloud model service platform was designed using the classic cloud computing architecture. Four layers were divided into: IaaS (Infrastructure as a Service), PaaS (Platform as a Service), SaaS (Software as a Service), and Visualization Service Layer. The calculation services were provided by the water cycle model and water resources optimization model (SaaS layer) on the server side. The core of the entire system included the solving methods, numerical models, file systems, and database components. The backend database selected MySQL to support data for the entire system, including graphical, modeling, calculation, and hydrological data. Two parts were on the browser side: 1) Modeling and flow field display module included the numerical model building, parameter setting, simulation statistics and querying, and flow field display; 2) Data transmission and calls between the browser and Web server was achieved via the Python interface in HTML, enabling the seamless data exchange. Once an external request was initiated from the browser, it was sent to the server via the Python interface. Subsequently, the Python logic on the server side invoked Python models to perform analysis and calculations using the received request. The computed data from the models was then presented within the HTML framework through the server, indicating the seamless integration and display of the calculation on the user interface. Taking the Shaanxi section of the Weihe River Basin as an example, the new platform was used to simulate the water cycle evolution after multi-objective optimization for the water resources allocation in the long-term planning. The results show that: 1) The relative errors during the calibration and validation periods above Zhuangtou were −4.32% and −1.62%, respectively. The monthly average runoff Nash-Sutcliffe efficiency coefficients were 0.80 and 0.82, respectively. For the Linjia Village to Xianyang reach, the relative errors during the calibration and validation periods were −4.96% and −1.71%, respectively. The monthly average runoff Nash-Sutcliffe efficiency coefficients were 0.79 and 0.81, respectively, indicating that the water cycle model of the cloud model service platform effectively characterizes the variation process of the runoff during the periodic and verification period. The total rainfall and the total evaporation values were 21.78, and 17.97 billion cubic meters, respectively, under the multiple-year average inflow frequency in 2025. The average depth of annual runoff was 121.80 mm with a total runoff volume of 5.63 billion cubic meters. The inflow and outflow water volume values were 3.39, and 7.20 billion cubic meters, respectively. The total water withdrawal was 5.10 billion cubic meters for economic and social purposes, where 2.767, 2.369, and 0.40 billion cubic meters were from the surface water, shallow groundwater, and deep groundwater, respectively. The water outputs in the Shaanxi Province of the Wei River Basin were evaporation and socio-economic water withdrawal, accounting for 77.89% and 22.11%, respectively. The spatial distribution of runoff depth depended mainly on the distribution of rainfall. A significant decreasing trend was also found in the runoff volume at various hydrological stations. 2) The Pareto optimal frontier was obtained using the NSGA-III (Non-dominated Sorting Genetic Algorithm III). An optimal plan of water resources allocation was obtained using a population multi-attribute decision-making model, where five decision-makers were assumed to participate in the decision-making. As such, the water shortage rate, grain yield reduction rate, and GDP reduction were 7.97%, 1.23%, and 0.15%, respectively, in the entire region in 2025. The agricultural water deficit reached 429 million cubic meters, accounting for 84.84% of the total water deficit. Water shortages were concentrated mainly in the areas from Baoji Gorge to Xianyang North, Xianyang to Tongguan South, and Xianyang to Tongguan North. The findings can provide technical references and insights for the smart water conservancy system and the refined allocation of water resources in the Weihe River Basin.