Groundwater level dynamic simulation and soil salinization prediction of the closed hydrogeological unit in the arid irrigation area

Arid and semi-arid areas are rich in land resources and sufficient in light and heat conditions, but water resources are in serious shortage. Irrigated agriculture accounts for about 40% of global crop output (Wang et al. 2019). In the arid irrigation area of northwest China, agricultural production mainly depends on water from the Yellow River for irrigation (Li et al. 2021). However, the transfer of a large number of water resources and the long-term extensive irrigation mode have broken the original law of groundwater movement in irrigation areas (Scanlon et al. 2012). The law of regional water and salt transport has been continuously reorganized and the agricultural water and soil environment has changed slowly. The process of groundwater recharge and movement is different in different hydrogeological units in the arid pumping irrigation area, resulting in different characteristics of groundwater dynamic distribution (Xue et al. 2020). Due to a lack of observational data, special climatic conditions, and geographical environment, it is difficult to accurately assess the future trend of groundwater movement and the dynamic distribution characteristics of groundwater in the closed hydrogeological unit at the regional scale, and it is also difficult to quantitatively predict the degree of soil salinization. The evolution process of groundwater and salinization is unclear and the sustainable development of soil and water environments in the irrigated area is threatened in the future. By constructing a three-dimensional simulation model of groundwater in the closed hydrogeological unit and setting the future scenario model, the purpose of dynamic simulation of groundwater level and prediction of soil salinization can be realized in the future. The groundwater movement characteristics in the closed hydrogeological unit and their relationship with the evolution of soil salinization can be revealed. The results of this study can provide a better understanding of the water and salt transport model of the closed hydrogeological unit, which can provide the basis for formulating scientific and reasonable irrigation and drainage measures in irrigation areas, and have important significance for improving the ecological environment and sustainable development in irrigation areas.

The groundwater level is the key to groundwater circulation in irrigation areas and the study of groundwater dynamic evolution laws and the mechanism is an important direction of agricultural water resources research in irrigation areas (Maheswaran et al. 2016). The combination of the regional hydrological model and groundwater flow model is an important way to study the numerical simulation of groundwater level (Cheng et al. 2014; Wu et al. 2014). Chen et al. (2016) studied the irrigation water-saving potential of the Heihe River Basin in China by the Visual Modular Three-dimensional Finite-difference Ground-water Flow Model (MODFLOW). Mao et al. (2018) established a water–salt balance model for well irrigation and canal irrigation based on the SaltMod and analyzed the evolution law of soil salt under the well-canal combined drip irrigation mode. Wu et al. (2018) simulated the dynamic simulation and prediction of groundwater level in the combined well and canal irrigation area based on the MODFLOW three-dimensional numerical model. Jiang et al. (2020) coupled the Soil and Water Assessment Tool (SWAT) model with the MODFLOW model to obtain the dynamic change of groundwater level in a basin. These research results have confirmed that the dynamic simulation of groundwater level can be carried out through numerical simulation. The groundwater movement model has potential advantages in considering spatial heterogeneity, vertical leakage through a clay layer, irregular site boundaries, steep hydraulic gradient near a drainage ditch, air distribution of recharge and evaporation, etc (Bradford & Acreman 2003; Varouchakis et al. 2015). However, there are few simulation results on groundwater level in closed geological units in water-pumping irrigation areas and there are some problems in groundwater dynamic simulation, such as a short period, the insufficient basis of scenario assumptions, and low accuracy, which cannot better reflect the long-term dynamic characteristics of groundwater in closed hydrogeological units (Xie et al. 2012). It is difficult to realize groundwater prediction under special geological conditions and understand the long-term development trend of groundwater in irrigated areas.

Recent research has studied the relationship between the dynamic change of groundwater level and the ecological environment in arid irrigation areas and the influencing factors. When the groundwater level drops sharply, it may lead to vegetation degradation, land desertification, and wetland drought in irrigation areas (Jiang et al. 2015; Liu et al. 2018). When the recharge is excessive, the groundwater level will rise, which may lead to soil salinization, swamping, and groundwater salinization (Han & Zhou 2018; Çadraku 2021). Affected by natural conditions, the suitable groundwater level in different irrigation areas is the time-space variation (Wang et al. 2022). The dynamic change of groundwater level in irrigation areas is influenced by rainfall, topography, evaporation, surface irrigation, surface vegetation coverage, exploitation and utilization of groundwater, drainage, and other factors (Askri et al. 2010; Syed et al. 2021; Wayangkau et al. 2021). The arid irrigation area lies deep in the hinterland of the mainland, with its special climatic conditions and natural environment. Different hydrogeological units have different groundwater recharge and movement under different irrigation conditions, and the groundwater level has different development trends (Sundararajan & Sankaran 2020). There are many qualitative studies on the relationship between groundwater level and soil salinization (Seeboonruang 2013; Ren et al. 2019), but few quantitative studies on the prediction of soil salinization from groundwater level. In the closed hydrogeological unit, the groundwater movement process, the migration trend of soil salt, and the converging mode of surface water in different hydrologic zones are not clear. It is difficult to predict soil salinization degree in the medium- and long-term quantitatively and the mutual feeding process between soil salinization evolution and groundwater dynamic change is not clear.

In this study, the closed hydrogeological unit in the Jingtaichuan Electric-lifting Irrigation Area in Gansu Province, China was selected as the research area and a three-dimensional model of groundwater movement was established by using Visual MODFLOW software, and the input items of regional climate change and human activity change were analyzed, so that the future groundwater level was simulated and then the development trend of regional soil salinization was predicted. In this study, a groundwater level simulation model suitable for the closed hydrogeological unit in the drought-pumping irrigation area is constructed, and the groundwater dynamic simulation method of the closed hydrogeological unit is further improved. The temporal and spatial changes of groundwater level and salinization degree in the future are predicted, and the transport modes of water and salt at the regional scale under special hydrogeological conditions are clarified. The research results provide a theoretical basis for water and salt control in arid irrigated areas and have important significance for monitoring and prevention of salinization.

Jingtaichuan Electric-lifting Irrigation Area (Phase I) is located in the central part of Jingtai County, Gansu Province, China (between 103°20′–104°04′E and 37°26′–38°41′N). The overall terrain of the irrigation area inclines from west to east, with low mountains and hills slopes. Agricultural production in the area mainly relies on water-lifting irrigation from the Yellow River. Affected by the low-lying and closed terrain and poor irrigation and drainage, the groundwater level shows a continuously increasing trend, which leads to the continuous development of soil salinization in the irrigation area. The total area of the saline–alkali land distributed in the closed hydrogeological units in the irrigation area was 4,500 hm², accounting for 21.75% of the total area of cultivated land.

The study area has a typical temperate continental climate with little rainfall and strong evaporation. The annual average precipitation is 185.7 mm and the annual average evaporation is 2,433.7 mm with most precipitation during June–September. The annual sunshine duration is 2,714 h, the frost-free season lasts for 190 days, and the annual average temperature is 8.5 °C with a large temperature difference (37.3 to −27.3 °C) between summer and winter. The soil surface in the study area is cracked and porous with strong decomposition of organic matter. The soluble salt in the soil matrix is prone to be transported to the surface and accumulate on the surface due to the strong evaporation, which forms surface salt accumulation and saline–alkali soil.

The amount of surface irrigation water in the study area comes from Statistics of Water Consumption in Water Diversion for Jingtaichuan Electric-lifting Irrigation Project Phase I (1972–2016) and the geological parameters come from the Jingtaichuan Electric-lifting Irrigation Area Authority of Gansu Province, including the Technical Design Report for Jingtaichuan Electric-lifting Irrigation Project Phase I (1971), the Land Survey Report for Jingtaichuan Electric-lifting Irrigation Area (1971–2016), and the Hexi Corridor Hydrogeological Survey/Census Report (2015). Groundwater monitoring wells are arranged in the irrigation area by the irrigation area authority and the water level of monitoring wells is recorded on the spot every month. The water levels of seven representative groundwater monitoring wells are selected for simulation and verification. Evaporation and precipitation data come from the monitoring results of the Jingtai County Meteorological Bureau in Gansu Province. The altitude data and cultivated land distribution in the study area were extracted from Digital Elevation Model (DEM) data and remote sensing images downloaded from the website of Geospatial Data Cloud.

This paper studies the simulation and prediction of groundwater levels by using Visual MODFLOW software. The hydrogeological model of the study area is generalized and the numerical simulation model of groundwater in the study area is established. Using the known groundwater level, the geological parameters of the aquifer in the study area are corrected, verified, and calibrated. The future scenarios are assumed, the groundwater changes are simulated, and the soil salinization degree is predicted by combining the previous research results of the research group on the relationship between groundwater depth and soil salt content. The cultivated land distribution in the study area is extracted by coordinate definition, image registration and mosaic cutting, ISO cluster analysis, and image recognition in ArcGIS10.2 Spatial Analyst module, combined with the actual land use situation through maximum likelihood classification.

• (1)

  Generalization of boundary conditions

The main groundwater movement area in the study area is the phreatic layer and the lower confined aquifer is almost unaffected by the outside, so only the phreatic layer is considered in the simulation (Senthilkumar & Gnanasundar 2022). The thickness of the phreatic layer is about 50–100 m. According to the digital elevation data extracted from DEM, the ground elevation is obtained by interpolation with Surfer software, and the top elevation map of the phreatic layer can be obtained by combining the groundwater depth data of each point. According to the well completion depth and geological drilling data in the well completion files of groundwater observation wells at each point, the bottom elevation of the phreatic layer can be known, and finally, the obtained grid file (*.grd) of each layer elevation can be imported into the Visual MODFLOW software.

• (3)

  Generalization of hydraulic characteristics

The aquifer in the study area is widely distributed and the groundwater movement conforms to Darcy's law. The recharge and discharge items of the whole groundwater system change with time and space, which is an unstable flow. The geological parameters of the aquifer change with the spatial position, which is heterogeneous; after comprehensive analysis, the groundwater flow system in the study area is generalized as a three-dimensional, heterogeneous, and isotropic unsteady flow.

• (1)

  Establishment of the numerical simulation model.

The simulation area of the study area is about 320 km². The simulation area is divided into 200 m × 300 m unit grids by Visual MODFLOW.

• (3)

  Parameter partition

The recharge items in the study area mainly include rainfall infiltration recharge, channel leakage recharge, field irrigation recharge, lateral ground recharge, and irrigation water inflow recharge. The discharge items include diving evaporation and drainage ditch outflow (Liu et al. 2018).

• (a)

  Precipitation infiltration recharge

Irrigation water that has not penetrated the ground and some groundwater with higher water levels will be discharged out of the study area through the drainage ditch, which will be sorted out according to the collected monitoring data of each water outlet in the irrigation area.

The correction process of the model was constructed using manual correction in this study. According to the data of the study area, the permeability coefficient, water supply degree, and other major parameters of the model were preliminarily formulated, and the groundwater level was simulated. The preliminary simulation results were compared with the observed values. According to the errors, the preliminary parameters were adjusted manually by subregion until the simulated values of the model is consistent with the observed values.

It can be seen from Table 4 that the maximum MAE of the numerical simulation results of each monitoring well is 0.175 m, the maximum MRE is 0.016 m, the maximum RMSE is 0.237 m, and the minimum value of R² is 0.8399. The simulation error is small and the precision is high, which can meet the needs of irrigation district management. It can be seen from Figure 6 that the predicted water level of each monitoring well matches the actual water level well in the identification period. Although there are errors in some times, the changing trend of the simulated value is consistent with the observed value. The model can accurately predict the end-of-year value of groundwater depth in the identification period and only needs to fine-tune the parameter zoning and geological parameters.

It can be seen from Table 5 that in the numerical simulation results of representative monitoring wells in the verification period, the maximum MAE of groundwater level is 0.184 m, the maximum MRE is 0.104 m, and the maximum RMSE is 0.221 m. There are certain differences in R² values, but they are all above 0.6. The simulation accuracy can meet the prediction requirements of the actual groundwater level in the study area. As can be seen from Figure 7, the predicted water level of each monitoring well is different from the actual water level during the verification period, and the predicted water level is higher than the observed value in many cases. The water level of the Shicheng Village observation point had no obvious change at the end of the year, and the water level of other observation points increased to some extent at the end of the year. After verification, the prediction accuracy of this model can meet the research needs, and it can predict the dynamic change of groundwater in the future study area. The final parameter values are shown in Table 6.

It can be seen from Table 7 and Figure 8 that, under the above climate scenario model, the MAE of the simulated water level of each observation well is 0.051–0.385 m, and the MRE is 0.002–0.129 m. Only the large value of 1.598 m appears in the total of the two monitoring wells and the RMSE is 0.061–0.417 m. The prediction accuracy meets the prediction requirements and the multi-year average value can be used as the input of future climate conditions for groundwater prediction.

It can be seen from Figure 10 that the spatial distribution pattern of groundwater depth in the study area has not changed greatly in the future and that the groundwater depth is still gradually decreasing from west to east. The groundwater depth of 0–5 m in the study area will continue to expand and will exceed 50% of the study area by 2031. Groundwater will overflow the surface in the northern basin area, the overflow area will continue to increase, and the maximum overflow depth will reach 0.26 m by 2031. The area where the groundwater depth is more than 25 m will decrease continuously and will disappear in the study area by 2031. As can be seen from Table 9, the minimum groundwater depth began to appear negative after 2021, with the maximum value decreasing by 1.79 m and the average value decreasing by 2.51 m. In the study area, the buried depth of groundwater will not change much in the future, but it will continue to decrease.

In spatial distribution, the total salt content of the soil in the study area increased gradually from southwest to northeast. From 2016 to 2031, the area of non-alkali soil in the southwest will change little, the area of saline–alkali soil and mild saline–alkali soil will decrease by 18.78%, and the area of medium saline–alkali soil will increase continuously, reaching 16.73% by 2031. Combined with the groundwater depth distribution in Figure 10, after 2021, a small area of salt marsh will appear near the boundary between mild saline–alkali soil and medium saline–alkali soil, and the area of salt marsh will keep increasing. In the study area, only the minimum salt content of non-alkali soil has a decreasing trend, which will decrease to 0.008% by 2031. The average regional soil salt content will continue to increase, reaching 1.59% by 2031. The maximum salt content of medium saline–alkali soil can reach 4.72%. In the future, the salt content of regional soil will increase rapidly and the degree of salinization will be aggravated as a whole.

In the process of groundwater level change prediction in a closed hydrogeological unit based on the Visual MODFLOW model, although the assumptions of the climate scenario and human activities change are not completely consistent with the actual situation, the prediction accuracy can meet the application requirements for the regional scale groundwater development trend research under the present conditions. The increase in prediction error in the validation period of the model may be related to the substantial increase in irrigation water quantity after the completion of the supplementary project in 2002. Although the irrigation water quota has not changed, the irrigation area and the total amount of irrigation have increased. Under special terrain conditions, the groundwater level rises due to irrigation infiltration, so both the measured value and the simulated value in the verification period show an increasing trend (Nian et al. 2014; Liu et al. 2015). With the influence of global climate change in the future, the climate scenario assumed by this model may increase its influence on the simulation results. With the completion of the first-phase project and the second-phase project of the irrigation area, the amount of water in the irrigation area increases, the irrigation area increases, and some areas of the first-phase project become the groundwater storage areas of the second-phase project (Xu et al. 2019). The rise of groundwater level, the increase of groundwater salinity and the increase of soil salinization in closed hydrogeological units will affect the parameters such as soil permeability coefficient and specific yield, and then affect the prediction accuracy of the model. This process is a complex process of mutual feedback coupling and its influence on prediction accuracy needs further study.

The prediction results of groundwater level in the study area show that there is no significant change of groundwater level in the future, but the overall trend is gradually decreasing. The western region is at a high altitude and after the irrigation water is infiltrated, it moves to the eastern region, and the groundwater depth remains at a large depth. With the continuous convergence of groundwater in Caowotan Basin in the north and Luyang Basin in the east, the buried depth of groundwater increases slowly, groundwater overflows in some areas, and the area is expanding to the surrounding areas.

According to the groundwater migration situation of the closed hydrogeological units, the rise of the groundwater level is objectively affected by topography and climate conditions. Under the action of long-term surface irrigation, surface water infiltrates into groundwater, and the groundwater in the irrigation area gradually moves and concentrates under the action of basin convergence. In this process, soil salinity varies with the movement of groundwater, and the basin becomes the natural storage area of groundwater and salinity in the closed hydrogeological unit of the irrigation area (Wang et al. 2020). Under the action of strong evaporation, salt in soil and groundwater accumulates with the evaporation of water, which leads to salinization in irrigation areas. The rise of groundwater levels is subjectively influenced by the long-term unreasonable irrigation behavior of human beings. Long-term water-lifting irrigation and unreasonable irrigation methods cause excess water resources to overflow, and cause groundwater to rise as a whole (Gao et al. 2015; Wen et al. 2020). Especially in closed hydrogeological units, the groundwater lacks drainage channels, and the drainage in irrigation areas is not smooth, which leads to the continuous rise of groundwater level and aggravation of soil salinization.

In this study, by constructing a three-dimensional groundwater simulation model, the future groundwater level of closed hydrogeological units in arid irrigation areas was simulated and predicted. Under the existing meteorological and human activities, by 2031, the groundwater level in the study area will rise as a whole, especially in the eastern and northern basins. The development trend of groundwater level in the next 10 years is consistent with the overall pattern of the past 20 years. In the study area, the degree of soil salinization is increasing, the soil is changing from mild saline–alkali soil to medium saline–alkali soil, and salt marshes may appear. Three key hydrological zones, namely, ‘irrigation infiltration’, ‘solute transport’ and ‘water collection and salt accumulation’, are formed in the closed hydrogeological unit. Under long-term irrigation, groundwater moves slowly in three hydrological zones, which leads to the continuous rise of groundwater level in the ‘water-collecting and salt-accumulating zone’, and leads to soil and water environmental problems such as soil salinization. The movement trend of groundwater is objectively influenced by topographic and climatic conditions, and subjectively by long-term unreasonable irrigation and drainage activities. Developing new water-saving irrigation methods, rationally formulating irrigation systems, and unblocking groundwater discharge channels are effective ways to control the sustained growth of groundwater in closed hydrogeological units in arid irrigation areas.

The study was supported by the National Key Research and Development Program of China (2021YFC3201205, 2019YFC1904303, 2021YFC3201202), the National Natural Science Foundation of China (51579102), the Zhongyuan Science and Technology Innovation Leading Talent Support Program of Henan, China (204200510048), and the Key Technologies R&D and Promotion Program of Henan Province (212102310273).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.