Factors influencing surface water and groundwater interaction in alluvial fan

Water is a valuable resource; therefore, it is very important to make better use of it (Cantor et al. 2018; Milan et al. 2018; Cuthbert et al. 2019). To improve the scientific management of water resources, it is necessary to study the mechanism of interaction between surface water and groundwater (SGW) and to understand the evolution characteristics of the water cycle in river basins (Epting et al. 2018; Mojarrad et al. 2019). SGW interaction can directly determine the number of local groundwater resources (Tian et al. 2015; Li et al. 2016), especially in arid and semi-arid areas (Lin et al. 2018; Cartwright et al. 2019). Our study area consists of both arid and semi-arid regions; therefore, it is crucial to study the SGW interaction in this area.

Many methods have been used to study the SGW interaction, such as isotopes, SGW coupling model, hydrochemistry method, and the water balance method. Some researchers used stable isotopes to characterize the mechanism of SGW interaction (Wang et al. 2018a; Hao et al. 2019). Although this method can be used to study SGW interaction, it is difficult to quantify the degree of interaction accurately, and sampling and laboratory analysis are required, which are relatively expensive (Zhao et al. 2018; Han et al. 2019). Computer-aided simulation has become a method to explore hydrological interactions; SGW coupling models including SWAT–MODFLOW, HSPF–MODFLOW, and GSFLOW have gradually been accepted. Chunn and Conant used the SWAT–MODFLOW model and a holistic framework model to study the SGW interaction; they identified and discussed the processes of SGW interaction (Chunn et al. 2019; Conant et al. 2019). The GSFLOW model was used to study the damping effects of the hydrological system, which can affect the hydrological cycle across different temporal and spatial scales (Ahn & Kim 2018). Saleem used the principle of water quality conservation to calculate SGW interaction (Saleem & Jeelani 2017). The hydrochemistry method is also widely used to study the SGW interaction. Hillel used the shallow hole hydrochemistry method to identify the temporal and spatial changes of the SGW interaction (Hillel et al. 2019). Guo used hydrochemistry to determine the SGW interaction in coal mines (Guo et al. 2019). Heat transfer can also be used as a research tool for the SGW interaction. Rautio used infrared remote sensing to study the SGW interaction (Rautio et al. 2018), and Zhou used a heat balance method to estimate the amount of surface water leakage (Zhou et al. 2018). Other methods used to study the SGW interaction include the application of new equipment (Cremeans et al. 2018) and the remote sensing and distributed model (Wulf et al. 2016). The water balance method is widely used in the study of water resources, for example, the combined approach of isotope mass balance to constrain the sources of lake water (Elmarami et al. 2017), understanding the integrated surface–water and groundwater processes in prairie wetlands (Hayashi et al. 2016), using water balance to manage water resources in irrigation areas (Ibrakhimov et al. 2018), and using water balance to calculate groundwater supply (Kahsay et al. 2019). The water balance method has also been used to calculate the SGW interaction, and it can develop and accurately calculate the amount of SGW interaction. Machiwal and Tozer used the water balance method to calculate SGW interaction (Machiwal & Jha 2015; Tozer et al. 2018). The calculation process is simple, but the premise is that all inflow and outflow data except SGW interaction are available, which is usually difficult to obtain, so this method is rarely used. However, this method has the advantages of simple calculation process and accurate calculation results of SGW interaction. Our work benefited from a large available data pool sufficient to meet the requirements of the water balance method. Therefore, we chose to use the water balance method to calculate the interaction amount of SGW to ensure the accuracy of the interaction amount calculation results.

Our study area is an alluvial fan in Jilin Province, China. The alluvial fan is rich groundwater resources and can provide significant water resources for the local area (Sakata et al. 2016; Tian et al. 2018; Okakita et al. 2019). In arid and semi-arid areas, the main source of groundwater in alluvial fans is precipitation and SGW interaction (Su et al. 2015; Xu et al. 2017; Gonzales Amaya et al. 2018; Xu & Su 2019). Therefore, it is very important to study the SGW interaction in alluvial fans. There is substantial research on SGW interaction, such as the effects of different alluvial fan strata (Miller et al. 2016; Tsai & Hsu 2018), groundwater sources (Xu et al. 2019), and groundwater pollution sources (Meng et al. 2019). However, there are few studies on the SGW interaction from river runoff or on the factors affecting the SGW interaction from alluvial fans.

The main purposes of this research are to determine the SGW interactions and study the influence of the selected factors on SGW. The main contribution and novelty of this study lie in the calculation of the SGW interaction in the study area and comprehensive analysis of the effects of various factors on the SGW interaction. This study will guide the theoretical study of the SGW interaction in the Taoer River alluvial fans and has important practical applications in water resource management, planning, and reasonable allocation in the study area.

The Taoer River fan is located in Baicheng City, Jilin Province, China, and covers an area of 2,920 km². The elevation of the study area ranges from 140 to 215 m and gradually decreases from the top to the front of the fan. The study area has a northern, temperate, continental, and monsoon climate. The annual average temperature is typically in the range of −2 to 5 °C, and the annual average precipitation is 395.0 mm. The annual distribution of precipitation is irregular, with 82.59% of the total yearly rainfall occurring from June to September. The average annual evaporation in the study area is 911.2 mm, concentrated from April to September (82.21% of the total annual evaporation).

The main rivers in the study area are the Taoer River, which flows into the study area from the ZhenXi Hydrological Station; the Jiaoliu River, which is the first tributary and flows into the study area from the WuBen Hydrological Station in the northwest of the fan; and the Emutai River, which is the second tributary of the Taoer River. This river generates runoff only in the wet season, and there is no hydrological station. The Taoer and Jiaoliu rivers form a confluence at the bottom of the fan and flow out from the TaoNan Hydrological Station. The three hydrological stations of the Taoer River record most of the runoff data of surface water flow into and out of the study area (Figure 1).

The lithological character of the groundwater aquifer in the study area includes gravel and sandy gravel. The average diameter of the gravel gradually decreases from the top to the front of the fan area, whereas the thickness gradually increases. There is also a fine-grained sandy soil interlayer. The groundwater flows from the top to the front of the fan, and the sand and gravel covering the top of the fan are directly exposed at the ground surface. A weak permeable loess layer covers the front of the fan, while the bottom of the fan is sand and gravel.

The data employed in this study were daily runoff, river water level, evaporation, precipitation, and channel diversion data collected between 1956 and 2014 at ZhenXi, TaoNan, and WuBen hydrological stations. The data were provided by the Songliao Committee of the Ministry of Water Resources of China.

To calculate the SGW interaction with different surface lithology interactions, the river runoff data at the point of change in the surface lithology (Manzhoudai section) were measured from 12 April 2016 to 30 May 2016. Meanwhile, river runoff, precipitation, and evaporation data were collected from the Zhenxi and TaoNan hydrological stations. All of this was processed and analyzed to determine the SGW interaction. The study group conducted hydrogeological surveys within 5 km along both sides of the river bank. Long-term groundwater-level data, covering the period 1995–2014, were collected from 20 observation wells.

The surface water balance method is a reliable technique for calculating the SGW interaction (Rezaei & Mohammadi 2017).

Because there were no measured river runoff or precipitation data for Q_H, this study used the hydrological analogy method to calculate Q_H. To ensure the similarity of data, this study analyzed the runoff producing conditions of the two watersheds. The upper watershed of the Jiaoliu River is composed of semi-arid, hilly land with high terrain and undeveloped vegetation. However, the river runoff conditions in the upper reaches of the Emutai River are very similar to those in the upper reaches of the WuBen Hydrological Station. Moreover, the underlying surface is similar. The precipitation river runoff of the Emutai River was calculated by employing the hydrological analogy method based on the runoff data of the Jiaoliu River (Wulf et al. 2016).

In April 2016, a field survey to collect river runoff, precipitation, evaporation, and other data was conducted in the study area. The latter can be divided into two sections based on different riverbed lithologies: the bare gravel reach (ZhenXi Hydrological Station to Manzhoudai section) and the bare loess reach (Manzhoudai section to the TaoNan Hydrological Station) (Figure 2). In April 2016, the Emutai River and Jiaoliu River were dry; thus, this study only calculated the SGW interaction quantities of two reaches on the Taoer River. The data used for the calculations here were obtained from 12 April 2016 to 30 May 2016.

Automatic linear modeling is an improved linear model method that allows users to input the fewest possible parameters to establish the linear model function automatically. Through parameter selection, automatic data processing, and feature selection, model training is realized. Automatic linear modeling can automatically remove outliers from independent variables during modeling, avoiding the uncertainty of artificial removal of outliers of independent variables. It can automatically find the key-independent variables, and discard non-key-independent variables, enabling the identification of the key-dependent variables affecting the independent variables and achieving the purpose of identifying the dependent variable categories (Wang et al. 2018b). The drawback of this analysis method is that the modeling process is automatically analyzed by computer, focusing on the analysis of the model; hence, the output chart results are too simple, and there is no analysis of the variance results. All of the statistical analyses of the data were completed using the software SPSS 19.0.

In Equations (5)–(7), n is the number of samples; X_i and Y_i are the values of two variables, ̀X and ̀Y are the average values of two variables, and i is the ordinal number of variables.

In addition, the following general rules were applied to analyze the correlation coefficient results: r ≤ 0.39 represented weak correlations, r between 0.40 and 0.69 moderate correlations, r between 0.70 and 1 strong or high correlations, and r ≥ 0.9 very high correlations (Fu et al. 2020).

In our study, the runoff, precipitation, and evaporation data of the three hydrological stations were collected (Table 1), and the water diversion data of the canal were used to establish the surface water balance equation of the study area and calculate the SGW interaction.

From 1956 to 2014, the amount of SGW interaction in the study area varied significantly. Based on the linear trend analysis, the amount of SGW interaction increased from 1956 to 2014. The maximum value of SGW interaction appeared in 1998, and the minimum value appeared in 2002. The annual average value was 27,772.7 × 10⁴ m³ (Figure 3).

To express the distribution of the SGW interaction within a year, the calculated results were converted into monthly averages. Within a year, the total number of interactions initially increased and then decreased, with the largest interaction occurring in July. The interaction in the second half of the year was significantly greater than that in the first half (Figure 4).

To study the influence of different lithologies on the degree of SGW interaction, this study calculated the interaction amount per kilometer of the two river sections. The calculation results show that the interaction in each stream segment in the bare gravel reach is 49.29 × 10⁴ m³/km, while that in the bare loess reach is 36.44 × 10⁴ m³/km. The amount of SGW interaction in the bare gravel reach is more than that in the bare loess reach (49.29 × 10⁴ > 36.44 × 10⁴ m³/km), indicating that gravel reach is more conducive to the SGW interaction (Table 2).

Five hydrological variables were tested to determine the key and non-key-dependent variables such as river runoff, groundwater level, precipitation, evaporation, and groundwater abstraction. Based on the analyses, it is clearly shown that the river runoff and groundwater level are the key-dependent variables that significantly affected the SGW interaction, with levels of importance of 0.92 and 0.08, respectively (Figure 5). Using the correlation analysis method, the influence of groundwater extraction, precipitation, and evaporation on the SGW interaction was further analyzed.

According to the results of key-dependent variables analysis, groundwater exploitation, precipitation, and evaporation are the non-key factors affecting the SGW interaction. However, this statement does not suggest that these factors have no influence on the SGW interaction. This study first assumed that the non-key factors indirectly affected the SGW interaction by influencing the key factors and then verified the validity of this hypothesis through correlation analysis of the key and non-key factors.

The correlation analysis results show that groundwater exploitation is strongly positively correlated with the groundwater level, with a significance (Sig) = 0.000 < 0.05 and a correlation coefficient of 0.718. Precipitation is moderate and positively correlated with river runoff, with Sig = 0.001 < 0.05 and a correlation coefficient of 0.665. For the river runoff and groundwater level, Sig = 0.439 > 0.05, indicating that there is no correlation. For evaporation, river runoff, and groundwater level, Sig = 0.093 > 0.05 and 0.322 > 0.05, respectively, which shows that there are no correlations; however, these results do not mean that evaporation has no effect on the SGW interaction. That inference needs to be based on the actual situation, and hence, further analysis of the situation is required. It is worth noting that Sig = 0.025 < 0.05 for groundwater exploitation and river runoff, and the correlation coefficient is −0.500, which is a moderate negative correlation (Table 3). The reason for this result is that it is extensive on both sides of the river in the study area. When the river water is high enough, irrigation mainly by channel diversion leads to a decrease in groundwater exploitation.

Both sides of Taoer River are farmland, without construction activity, and there is no significant change in riverbed lithology, which is called the stability factor in this study. According to the key-dependent variable analysis and correlation analysis results, the river runoff and the groundwater level are the key factors affecting the SGW interaction, and groundwater exploitation, precipitation, and evaporation are the non-key factors.

Miller studied the influence of different lithologies on the groundwater level of fan-shaped land but did not directly note the influence of different lithologies on the SGW interaction (Miller et al. 2016). This study calculated the SGW interaction in different areas to determine the influence of different lithologies. According to the SGW interaction results from April to May 2016 for different river sections, the SGW interaction in the study area is more prominent in the bare gravel reach than in the bare loess reach (Table 1). The sand and gravel atop the fan are directly exposed at the surface, thereby facilitating flow between SGW. Furthermore, the aquifer located at the front of the alluvial fan is covered with loess soil and loam (the main component of which is silt with a diameter of 0.05–0.005 mm); this weakens the hydraulic connection between the SGW, significantly reducing the interaction (Figure 6). In summary, the lithological character of the landform significantly influences the SGW interaction.

The key-dependent variable analysis shows that the river runoff and the groundwater level are the key factors affecting the SGW interaction. Thus, quantitative analyses of river runoff and groundwater level were performed.

According to an analysis of the key-dependent variables, the river runoff is one of the key factors affecting the SGW interaction. The changes in the amount of river runoff over time were found to be similar to the changes in the amount of the SGW interaction, indicating that there is a significant positive correlation between river runoff and SGW interaction.

It is noted that the water level and runoff of the river will have a certain impact on the SGW interaction, but the amount of SGW interaction was not calculated (Scibek et al. 2007; Vrzel et al. 2019). Our study also found that when the river runoff was large, such as from 1956 to 1999 and 2011 to 2014, the amount of SGW interaction remained between 10% and 30% of the river runoff. When the amount of river runoff was relatively small, such as from 2000 to 2010, the proportion was high, about 90% (Figure 7). For the monthly SGW interaction, when the river runoff was relatively large (April–November), the proportion was also between 10% and 30%. When the river runoff was relatively small, the proportion increased correspondingly to between 40% and 50% (Figure 8). These results show that the amount of SGW interaction changes with the amount of river runoff and is less disturbed by other factors. These findings further indicate that river runoff is a key factor affecting the SGW interaction.

Observation well 26600014 is located at 122°33′49.37″ N and 45°34′00.64″ E (Figure 1) and is located 2,360 m from the Taoer River. Data from the well obtained between 1995 and 2014 were analyzed.

According to the analysis of key-dependent variables, the groundwater level is one of the key factors affecting the SGW interaction. The key factors controlling the groundwater level are groundwater recharge and exploitation, which directly determine the groundwater level and subsequently affect the difference between the groundwater and surface water levels. This water level difference determines the SGW interaction.

From 1995 to 2000 and 2012 to 2014, the river water flow was continuous, and the change trends of the surface water and groundwater were similar (Figure 9). From 2001 to 2012, little river runoff entered the study area; therefore, when it reached the observation well, the river had dried up, and the result was a sharp drop in the groundwater level. In some months from 2005 to 2009, the river runoff was slightly higher, the surface water level rose, and the groundwater level rose accordingly (Figure 9). The surface water and groundwater levels are positively correlated. During the year, the surface water and groundwater levels showed similar trends from January to July. The surface water level reached its lowest point in April, and the groundwater level of the observation well reached its lowest point in May (Figure 10). However, agriculture is well developed in the study area. A large rice crop was planted near the observation well. In May, during the initial stage of rice growth, there was little precipitation; consequently, the water requirements of the rice could not be met. Therefore, it was necessary to extract a large amount of groundwater for irrigation, and the groundwater level fell sharply. From May to August, although water was still needed for rice growth, precipitation levels increased rapidly; therefore, it was not necessary to extract a large amount of groundwater for irrigation. Thus, after May, the SGW interaction increased as the surface water level rose (Figure 4) and the groundwater level started to rise. After July, the surface water level began to decline and the rising trend of the groundwater level slowed down. After October, the surface water level was low, and there was little river runoff, which led to a reduction in the amount of SGW interaction. It is speculated that the quantity of groundwater runoff discharged was larger than the amount of surface water converted into groundwater, which led to a decrease in the groundwater level. In summary, the groundwater and surface water levels exhibited a certain correlation but were affected by groundwater exploitation and groundwater runoff. In addition, Ke indicates that the change of groundwater level is related to the SGW interaction (Ke 2014). Therefore, this study concludes that the groundwater level affects the SGW interaction by affecting the water level difference between surface water and groundwater.

Since 2001, farmland has been developed extensively in the study area, and policies for increasing the rice planting area have been implemented, which has led to the extraction of a large amount of groundwater. According to the correlation analysis results, the correlation coefficient between groundwater exploitation and groundwater level is 0.718, which is a strong positive correlation. Hence, groundwater exploitation is closely related to the groundwater level as also found in the study (Chu et al. 2017). Additionally, groundwater-level fluctuations play an important role in the SGW interaction (Stefania et al. 2018). In the study area, the main factors affecting the groundwater level are groundwater exploitation and surface water supply to the groundwater. From 2001 to 2005, the river water volume was very small, resulting in little surface water recharging the groundwater (Figure 7). During this period, the amount of groundwater exploitation increased to a large extent compared with that between 1995 and 2000, resulting in the continuous decline in the groundwater level during this period. From 2005 to 2014, the change in the amount of groundwater exploitation was small, but the amounts of river runoff and surface water supplying groundwater increased (Figure 7), causing the groundwater level to rise (Figure 11). In conclusion, there is a strong correlation between the groundwater exploitation and the groundwater level, but it is also affected by the SGW interaction. Based on the observed effects of the groundwater level on the SGW interaction, this study draws the conclusion that groundwater exploitation indirectly affects the SGW interaction by influencing the groundwater level.

According to the correlation analysis results, the correlation coefficient between precipitation and river runoff is 0.665, which shows that precipitation has a direct impact on river runoff. Studies have shown that precipitation is the main source of river runoff (Deb et al. 2019; Farsi & Mahjouri 2019), but the artificial construction of reservoirs has a greater impact on river runoff. From 2000 to 2009, the reservoirs in the upstream region of the study area were in a state of water storage. Only a small quantity of river water could enter the study area, which resulted in the weakening of the correlation between precipitation and river runoff. However, precipitation generally directly affects the river runoff and indirectly affects the SGW interaction (Figure 12). In addition, the change trend of precipitation is similar to those of surface water and groundwater. Precipitation is one of the main components of climate, which shows that climate change will have certain impacts on river runoff and the SGW interaction.

According to the correlation analysis results, evaporation is not directly related to the runoff and the groundwater level. However, practically, evaporation still has certain impacts on surface water and groundwater. The average river surface evaporation in the study area is 538.7 × 10⁴ m³ annually, which accounts for 0.3% of the river runoff, indicating that evaporation has negligible influence on river runoff. The influence of evaporation on phreatic water is affected by the soil properties and the groundwater level (Xing et al. 2019). When the groundwater level is shallow, evaporation has some influence on it. Research indicates that evaporation has affected the groundwater quantity and the groundwater level (Maxwell & Condon 2016). Thus, it can be concluded that evaporation indirectly affects the SGW interaction by affecting the runoff and the groundwater level.

Based on the results of a 59-year time series of SGW interaction data and the interaction analysis with river runoff, groundwater level, lithology, precipitation, evaporation, and groundwater abstraction, the following conclusions can be drawn.

The exposed surfaces of sand and gravel are more conducive to the SGW interaction than loess. According to the automatic linear model analysis, the key factors affecting the SGW interaction are river runoff and groundwater level; the former is predominant. The SGW interaction changes with changes in river runoff. There is a certain correlation between the groundwater and surface water levels, but these are affected by groundwater exploitation and groundwater runoff. According to the results of the correlation analysis, groundwater exploitation, precipitation, and evaporation affect the SGW interaction indirectly by affecting the groundwater level and river runoff. Groundwater exploitation and evaporation affect the change of the groundwater level and precipitation mainly affects the river runoff.

This study calculated the SGW interaction using time series data in the study area, investigated the factors that affected the SGW interaction, and preliminarily determined the mechanism through which each influencing factor affected the SGW interaction. However, the factors influencing the SGW interaction are not independent, but also interact with each other. Future research with the objective of analyzing the influence of each factor on the SGW interaction separately, potentially by performing studies in the alluvial fan with little human impact or conducting experimental laboratory research, is suggested.

This work was financially supported by the National Natural Science Foundation of China (Project Approval No. 41572216); the Shenyang Geological Survey Center Project of China Geological Survey ‘SongNen Plain Hydrogeology Survey’ Special Topic (DD20190340); Project of Provincial-School Co-construction Plan: Frontier Science and Technology Guidance Class (SXGJQY2017-6); the Key Projects of Geological Exploration Fund of Jilin Province (2018-13); the Natural Science Foundation of Jilin Province, China (20140101164JC); and the Construction of Big Data Platform for Geotechnical Engineering (2017ZDCXL-SF-03-01-01).

We thank the Songliao Water Resources Commission of the Ministry of Water Resources for providing us with the data needed for the calculation.

The authors declare no conflicts of interest.

All data used during the study are proprietary or confidential in nature and may only be provided with restrictions.