Impacts of the Sanmenxia Dam on the Interaction between Surface Water and Groundwater in the Lower Weihe River of Yellow River Watershed

: Sanmenxia Dam, one of the most controversial water conservancy projects in China, has seriously impacted the lower Weihe River of the Yellow River Watershed since its operation. At the Huaxian Station, the dam operation controls the surface water level and leads to the variation of the surface water–groundwater interaction relationship. The river channel switched from a losing reach during the early stage (1959) to a gaining reach in 2010 eventually. The comparison of tracer (Cl − , δ 18 O and δ 2 H) characteristics of surface water in successive reaches with that of ambient groundwater shows that the general interaction condition is obviously a ﬀ ected by the dam operation and the impact area can be tracked back to Weinan City, around 65 km upstream of the estuary of the Weihe River. The anthropogenic inputs (i.e., agricultural fertilizer application, wastewater discharge, and rural industrial sewage) could be responsible for the deterioration of hydro-environment during the investigation periods of 2015 and 2016, as the population and fertilizer consumption escalated in the last 60 years. The use of contaminated river water for irrigation, along with the dissolved fertilizer inputs, can a ﬀ ect the groundwater quality, in particular resulting in the NO 3 − concentrations ranging from 139.4 to 374.1 mg / L. The unregulated industrial inputs in some rural areas may increase the Cl − contents in groundwater ranging from 298.4 to 472.9 mg / L. The ﬁndings are helpful for the improved comprehensive understanding of impacts of the Sanmenxia Dam on the interaction between surface water and groundwater, and for improving local water resources management. There are some river segments along which the concentrations of Cl − increase (decrease) with much higher (lower) Cl − values in the ambient GW (e.g., the segments L2-L1 in May, W2-W1 in August). It may imply that these river segments are gaining reaches. The remaining segments that have not shown such relationships may indicate losing reaches. The Huaxian Station is in the river segment W2-W1. In May, the Cl − values decrease from 25.2 (W2) to 20.6 mg / L (W1) while the average Cl − values in the ambient GW (B41 and B282) is 12.5 mg / L, which means that the segment is a gaining reach. It also occurs in August for the river segment W2-W1. The SW–GW interaction condition is consistent with the results obtained from the water level data before the dam operation.


Introduction
Surface water (SW) and ambient groundwater (GW) are always interacting in the hydrologic cycle [1]. Understanding the interactions between SW and GW is critical for purposes like water resources utilization and management [2,3], water quality assessment [4,5], and ecohydrology analysis [6,7]. One of the main factors controlling the SW-GW interaction is the hydraulic gradient [8], which is variable since levels of SW and GW are persistently affected by various natural and anthropic conditions. Dams provide many benefits while also disturb the river systems and produce some unexpected hydrologic impacts on discharge, sedimentation, and aquatic vegetation, etc. [9][10][11][12]. The SW level can be directly affected by the storing and releasing process of dams, which may change the hydraulic gradients of SW-GW interactions [5,13,14]. SW and GW can also be polluted by many anthropogenic sources, such as agricultural non-point pollutants, domestic sewage, industrial wastewater [15][16][17]. In different river reaches, water quality may respond differently to the SW-GW interaction changes [18,19]. distributed along the Weihe River. The phreatic aquifer in the alluvial plain is composed of Quaternary unconsolidated sediments with the thickness ranging from 10.5 to 63.5 m.

Sanmenxia Dam
The Sanmenxia Dam is located on the middle reach of the Yellow River ( Figure 1b). The catchment area above the dam is about 690,000 km 2 , which accounts for 92% of the whole Yellow River Basin. The construction and water impoundment of the dam were initiated in 1957 and 1960, respectively. The designed normal water level of the dam is 350 m, and due to the estimate on the influence scope of the dam, over 187,000 people in the Weinan City, Dali County, Huayin County, and Huaxian County (Figure 1a) migrated from their original residences. After impoundment of the Sanmenxia Dam, a severe sedimentation problem became evident, especially in the lower Weihe River. The highest operation water level on record is about 333 m and the direct backwater boundary reached upstream to the Huaxian Station. In order to solve the sedimentation problem, the operation scheme of the dam had been changed twice, which also made the dam never meet the initial impounding expectation since its operation. The three operation modes were as follows: (i) storing water mode (Mode A), from September 1960 to March 1962, the reservoir maintained a high water storage level, (ii) detaining flood water and sluicing sediment mode (Mode B), from March 1962 to October 1973, the reservoir kept a low storage level, while detaining floods only during flood seasons (November to June) and sluicing sediment with relatively large discharges, and (iii) storing clear water and releasing muddy water mode (Mode C), from November 1973 to the present, the reservoir was operated at a high level to store relatively clear water in the non-flood seasons and at a low level to release high sediment concentrations in the flood seasons.
Time series data of the pool level of the Sanmenxia Dam during the flood and non-flood seasons and the whole year present a similar change process ( Figure 2). It shows different variation character corresponding to each operation mode. The pool levels in Mode A increased and reached the highest. The pool levels in Mode B show an obvious downward trend. However, the pool levels in Mode C

Sanmenxia Dam
The Sanmenxia Dam is located on the middle reach of the Yellow River ( Figure 1b). The catchment area above the dam is about 690,000 km 2 , which accounts for 92% of the whole Yellow River Basin. The construction and water impoundment of the dam were initiated in 1957 and 1960, respectively. The designed normal water level of the dam is 350 m, and due to the estimate on the influence scope of the dam, over 187,000 people in the Weinan City, Dali County, Huayin County, and Huaxian County (Figure 1a) migrated from their original residences. After impoundment of the Sanmenxia Dam, a severe sedimentation problem became evident, especially in the lower Weihe River. The highest operation water level on record is about 333 m and the direct backwater boundary reached upstream to the Huaxian Station. In order to solve the sedimentation problem, the operation scheme of the dam had been changed twice, which also made the dam never meet the initial impounding expectation since its operation. The three operation modes were as follows: (i) storing water mode (Mode A), from September 1960 to March 1962, the reservoir maintained a high water storage level, (ii) detaining flood water and sluicing sediment mode (Mode B), from March 1962 to October 1973, the reservoir kept a low storage level, while detaining floods only during flood seasons (November to June) and sluicing sediment with relatively large discharges, and (iii) storing clear water and releasing muddy water mode (Mode C), from November 1973 to the present, the reservoir was operated at a high level to store relatively clear water in the non-flood seasons and at a low level to release high sediment concentrations in the flood seasons.
Time series data of the pool level of the Sanmenxia Dam during the flood and non-flood seasons and the whole year present a similar change process ( Figure 2). It shows different variation character Water 2020, 12, 1671 4 of 18 corresponding to each operation mode. The pool levels in Mode A increased and reached the highest. The pool levels in Mode B show an obvious downward trend. However, the pool levels in Mode C remain relatively stable after a rise that ends in 1980 ( Figure 2). Thus, the impacts from the Sanmenxia Dam that were imposed on the lower Weihe River can be assumed generally unchanged since 1980.
Water 2020, 12, x FOR PEER REVIEW 4 of 18 remain relatively stable after a rise that ends in 1980 ( Figure 2). Thus, the impacts from the Sanmenxia Dam that were imposed on the lower Weihe River can be assumed generally unchanged since 1980.

Data Collection
The hydrochemical data in 1959 (Table S1 in Supplementary Materials) were collected from the results of a hydrogeological survey conducted by the Yellow River Conservancy Commission [36]. SW and GW samples form three campaigns in May, August, and November. A total of 19 SW samples (five sites from the Weihe River and two sites from the Luohe River) and 55 GW samples taken from phreatic wells (19 sites) which were adjacent to the Weihe and Luohe Rivers were chosen for this study (Figure 1a). The gridded meteorological data were from the China Meteorological Data Service Center. The discharge and water level of the Weihe River were measured daily at the Huaxian Station. The ambient GW levels were measured at least every five days at three monitoring wells (B18, B561, and B562). The population, fertilizer consumption, and groundwater exploitation data were gathered from the China's economic and social big data research platform (http://data.cnki.net/) which collected statistical yearbooks of the Weinan City.

Water Sampling
Three field campaigns were conducted in October 2015, June 2016, and September 2016. A total of 22 SW samples (eight sites from the Weihe River, two sites from the Luohe River) and 41 GW samples taken from domestic and agricultural wells (21 sites, phreatic water) adjacent to the Weihe and Luohe Rivers were collected (Figure 1a and Table S1 in Supplementary Materials). The GW samples were taken after purging the wells for a few minutes at each site. The polyethylene bottles (50 and 100 mL) used as sample containers were pre-rinsed three times before sample collection and sealed with adhesive tape. All the water samples were store in the refrigerator at 4 °C after collection.

Analytical Techniques
Major ion compositions for each sample collected in 2015-2016 were measured at the Center for Physical and Chemical Analysis, the Institute of Geographic Sciences and Natural Resources Research (IGSNRR), Chinese Academy of Sciences (CAS). Major cations (Na + , K + , Ca 2+ , Mg 2+ ) were measured on an inductively coupled plasma optical emission spectrometer (ICP-OES) (Perkin-Elmer Optima 5300DV, Waltham, Massachusetts, USA). Cl − , SO4 2− , and NO3 − were determined by an ion chromatography system (ICS1000). HCO3 − was measured by the diluted vitriol-methylic titration method using 0.016 M H2SO4. The normalized inorganic charge balance (NICB) was applied to ensure

Data Collection
The hydrochemical data in 1959 (Table S1 in Supplementary Materials) were collected from the results of a hydrogeological survey conducted by the Yellow River Conservancy Commission [36]. SW and GW samples form three campaigns in May, August, and November. A total of 19 SW samples (five sites from the Weihe River and two sites from the Luohe River) and 55 GW samples taken from phreatic wells (19 sites) which were adjacent to the Weihe and Luohe Rivers were chosen for this study (Figure 1a). The gridded meteorological data were from the China Meteorological Data Service Center. The discharge and water level of the Weihe River were measured daily at the Huaxian Station. The ambient GW levels were measured at least every five days at three monitoring wells (B18, B561, and B562). The population, fertilizer consumption, and groundwater exploitation data were gathered from the China's economic and social big data research platform (http://data.cnki.net/) which collected statistical yearbooks of the Weinan City.

Water Sampling
Three field campaigns were conducted in October 2015, June 2016, and September 2016. A total of 22 SW samples (eight sites from the Weihe River, two sites from the Luohe River) and 41 GW samples taken from domestic and agricultural wells (21 sites, phreatic water) adjacent to the Weihe and Luohe Rivers were collected ( Figure 1a and Table S1 in Supplementary Materials). The GW samples were taken after purging the wells for a few minutes at each site. The polyethylene bottles (50 and 100 mL) used as sample containers were pre-rinsed three times before sample collection and sealed with adhesive tape. All the water samples were store in the refrigerator at 4 • C after collection.

Analytical Techniques
Major ion compositions for each sample collected in 2015-2016 were measured at the Center for Physical and Chemical Analysis, the Institute of Geographic Sciences and Natural Resources Research (IGSNRR), Chinese Academy of Sciences (CAS). Major cations (Na + , K + , Ca 2+ , Mg 2+ ) were measured on an inductively coupled plasma optical emission spectrometer (ICP-OES) (Perkin-Elmer Optima 5300DV, Waltham, MA, USA). Cl − , SO 4 2− , and NO 3 − were determined by an ion chromatography system (ICS1000). HCO 3 − was measured by the diluted vitriol-methylic titration method using 0.016 M H 2 SO 4 . The normalized inorganic charge balance (NICB) was applied to ensure the charge-balance error of the water sample was in a reasonable scope [37,38]. The results of the major ion compositions analysis were accepted when the charge-balance error was within ±10%.  Table S2 in Supplementary Materials shows the hydrochemical and stable isotopes data.

Statistical Method
The Mann-Kendall test (M-K test) is one of the widely used methods to detect trends of different time series [39]. The statistic S in the Mann-Kendall test is calculated as follow: where x is the variable, n is the length of time series. If n is larger than 8, S will approximately follow a normal distribution [40]. The mean and variance of S can be acquired by Equations (2) and (3) below.
The Z statistics for evaluating the trend is calculated by Equation (4): A positive (negative) Z-value indicates an increasing (decreasing) trend. Z-value also can be used to check if the trend is significant at the selected significance level α. The null hypothesis that there is no significant trend will be rejected when |Z| > Z 1−α/2 .

Water Level Variations
The Huaxian Station is a representative station with relatively continuous data for the study area. Moreover, most of the sediment deposition caused by the Sanmenxia Dam was in the downstream channel of the Huaxian Station [23], which may enhance the variation of SW-GW interaction. The temporal variations of the annual average SW level, ambient GW level, and precipitation can be seen in Figure 3a. The trend of the annual average SW level and runoff were analyzed by the M-K test. Statistic Z of the SW level is 3.99 indicating a significant upward trend, while statistic Z of the runoff is −1.89 showing a downward trend. When the riverbed is stable, the river water level with given discharge is almost the same and the trends of river water level and runoff should be consistent. The inconsistent trends of the SW level and runoff at the Huaxian Station indicate that the riverbed has risen after the operation of the Sanmenxia Dam. The riverbed elevation at the Huaxian Station rose only 3 m during 2500 years from 540 BC to 1960 AD [41]. In 1959 (before the dam operation), the annual average SW level and runoff at the Huaxian Station are 333.4 m and 61.9 × 10 8 m 3 , respectively. However, after 50 years of the dam operation (in 2010), the annual average SW level rose 3 m to 336.4 m with the annual average runoff of 60.2 × 10 8 m 3 . It indicates that the dam operation might be responsible for the elevation of the SW level at the Huaxian Station.
The M-K test was also applied for the annual average GW level series. Statistic Z of the annual average GW at B18 from 1959 to 1985 is 1.56. The statistic Z of the annual average GW at B561 and B561 from 1976 to 2010 is −1.82 and −2.20. The different trends of the GW levels in different periods indicate that the controlling factor of the GW level has changed from 1959 to 2010. In order to seek the reason why the GW levels change, the correlation coefficients among the GW levels, SW level, and the precipitation were calculated and shown in Table 1. The GW levels at B18, B561, and B562 are highly correlated with each other. The GW level at B18 is from 1959 to 1985. The GW levels at B561 and B562 is from 1976 to 2010. The correlation coefficient of the GW level at B18 is significantly correlated to the SW level (0.01 level) and the precipitation (0.05 level). It demonstrates that the SW level and precipitation are the driving factors of the GW level from 1959 to 1985. However, the GW levels at B561 and B562 cannot present a significant correlation with the SW level or the precipitation. The GW exploitation data in the study area increased from 1980 to 2010 ( Figure 4). The correlation coefficient between the GW levels and GW exploitation are −0.72 (B561) and −0.75 (B562), both negatively correlated at a significance level of 0.01, indicating that the driving factor of the GW level is groundwater exploitation after the 1980s across this region. The trend of the annual average SW level and runoff were analyzed by the M-K test. Statistic Z of the SW level is 3.99 indicating a significant upward trend, while statistic Z of the runoff is −1.89 showing a downward trend. When the riverbed is stable, the river water level with given discharge is almost the same and the trends of river water level and runoff should be consistent. The inconsistent trends of the SW level and runoff at the Huaxian Station indicate that the riverbed has risen after the operation of the Sanmenxia Dam. The riverbed elevation at the Huaxian Station rose only 3 m during 2500 years from 540 BC to 1960 AD [41]. In 1959 (before the dam operation), the annual average SW level and runoff at the Huaxian Station are 333.4 m and 61.9 × 10 8 m 3 , respectively. However, after 50 years of the dam operation (in 2010), the annual average SW level rose 3 m to 336.4 m with the annual average runoff of 60.2 × 10 8 m 3 . It indicates that the dam operation might be responsible for the elevation of the SW level at the Huaxian Station.
The M-K test was also applied for the annual average GW level series. Statistic Z of the annual average GW at B18 from 1959 to 1985 is 1.56. The statistic Z of the annual average GW at B561 and B561 from 1976 to 2010 is −1.82 and −2.20. The different trends of the GW levels in different periods indicate that the controlling factor of the GW level has changed from 1959 to 2010. In order to seek the reason why the GW levels change, the correlation coefficients among the GW levels, SW level, and the precipitation were calculated and shown in Table 1. The GW levels at B18, B561, and B562 are highly correlated with each other. The GW level at B18 is from 1959 to 1985. The GW levels at B561 and B562 is from 1976 to 2010. The correlation coefficient of the GW level at B18 is significantly correlated to the SW level (0.01 level) and the precipitation (0.05 level). It demonstrates that the SW level and precipitation are the driving factors of the GW level from 1959 to 1985. However, the GW levels at B561 and B562 cannot present a significant correlation with the SW level or the precipitation. The GW exploitation data in the study area increased from 1980 to 2010 ( Figure 4). The correlation coefficient between the GW levels and GW exploitation are −0.72 (B561) and −0.75 (B562), both negatively correlated at a significance level of 0.01, indicating that the driving factor of the GW level is groundwater exploitation after the 1980s across this region.     Table 2. It is obvious that the rising SW level is the main reason for the change of SW-GW interaction pattern from 1959 to 1981 near the Huaxian Station. The GW levels at B561 and B562 have dropped from 1981 to 2010, while the SW level is still ascending. Therefore, the rising SW level participated in the two shifts of the interaction pattern, which makes the upward SW level the primary cause for the interaction variations at the Huaxian Station.     Table 2. It is obvious that the rising SW level is the main reason for the change of SW-GW interaction pattern from 1959 to 1981 near the Huaxian Station. The GW levels at B561 and B562 have dropped from 1981 to 2010, while the SW level is still ascending. Therefore, the rising SW level participated in the two shifts of the interaction pattern, which makes the upward SW level the primary cause for the interaction variations at the Huaxian Station.  The piper diagram can be used to distinguish different water types of the SW and GW samples in 1959 ( Figure 5). The SW has a narrower range than the GW, plotting within Area 1 and Area 5, while the GW scatters in Areas 1, 4, and 5. Some GW samples plot adjacent to the SW on the diamond diagram, while others do not, indicating that the different patterns of SW-GW interaction may exist in the research area. Cl − behaves relatively conservatively and is commonly used as a tracer [42,43]. In 1959, the rivers in the research can be separated into different segments by the SW sampling sites shown in Figure 6a,b. The tributaries of the lower Weihe River in the research area have little impact on the tracer values since the discharge of them is quite small. The ambient GW samples are also listed in Figure 6a,b. There are some river segments along which the concentrations of Cl − increase (decrease) with much higher (lower) Cl − values in the ambient GW (e.g., the segments L2-L1 in May, W2-W1 in August). It may imply that these river segments are gaining reaches. The remaining segments that have not shown such relationships may indicate losing reaches. The Huaxian Station is in the river segment W2-W1. In May, the Cl − values decrease from 25.2 (W2) to 20.6 mg/L (W1) while the average Cl − values in the ambient GW (B41 and B282) is 12.5 mg/L, which means that the segment is a gaining reach. It also occurs in August for the river segment W2-W1. The SW-GW interaction condition is consistent with the results obtained from the water level data before the dam operation. The piper diagram can be used to distinguish different water types of the SW and GW samples in 1959 ( Figure 5). The SW has a narrower range than the GW, plotting within Area 1 and Area 5, while the GW scatters in Areas 1, 4, and 5. Some GW samples plot adjacent to the SW on the diamond diagram, while others do not, indicating that the different patterns of SW-GW interaction may exist in the research area. Cl − behaves relatively conservatively and is commonly used as a tracer [42,43]. In 1959, the rivers in the research can be separated into different segments by the SW sampling sites shown in Figure 6a,b. The tributaries of the lower Weihe River in the research area have little impact on the tracer values since the discharge of them is quite small. The ambient GW samples are also listed in Figure 6a,b. There are some river segments along which the concentrations of Cl − increase (decrease) with much higher (lower) Cl − values in the ambient GW (e.g., the segments L2-L1 in May, W2-W1 in August). It may imply that these river segments are gaining reaches. The remaining segments that have not shown such relationships may indicate losing reaches. The Huaxian Station is in the river segment W2-W1. In May, the Cl − values decrease from 25.2 (W2) to 20.6 mg/L (W1) while the average Cl − values in the ambient GW (B41 and B282) is 12.5 mg/L, which means that the segment is a gaining reach. It also occurs in August for the river segment W2-W1. The SW-GW interaction condition is consistent with the results obtained from the water level data before the dam operation.

Stable Isotope Characteristic in Water
The values of δ 18 O and δ 2 H in the Weihe River range from −8.6‰ to −7.8‰ and −64‰ to −51‰, respectively. The values of the Luohe River vary from −8.2‰ to −6.5‰ for δ 18 O and from −61‰ to −52‰ for δ 2 H, respectively. The compositions of groundwater range from −11.3‰ to −7.4 for δ 18 O and −80‰ to 53‰ for δ 2 H. The δ 18 O and δ 2 H values of SW and GW are plots in Figure 6. The global

Stable Isotope Characteristic in Water
The values of δ 18 O and δ 2 H in the Weihe River range from −8.6% to −7.8% and −64% to −51% , respectively. The values of the Luohe River vary from −8.2% to −6.5% for δ 18 O and from −61% to −52% for δ 2 H, respectively. The compositions of groundwater range from −11.3% to −7.4 for δ 18 O and −80% to 53% for δ 2 H. The δ 18 O and δ 2 H values of SW and GW are plots in Figure 6. The global meteoric water line (GMWL, δ 2 H = 8δ 18 O + 10) [30], and the local meteoric water line (LMWL, δ 2 H = 7.49δ 18 O + 6.13, precipitation in the Xi'an Station, IAEA) are also shown in Figure 7. The water samples are scattered along the GWML and the LWML, indicating that both the SW and GW most likely come from modern precipitation. meteoric water line (GMWL, δ 2 H = 8δ 18 O + 10) [30], and the local meteoric water line (LMWL, δ 2 H = 7.49δ 18 O + 6.13, precipitation in the Xi'an Station, IAEA) are also shown in Figure 7. The water samples are scattered along the GWML and the LWML, indicating that both the SW and GW most likely come from modern precipitation. The SW samples show a narrow range in Figure 7. The Luohe River water samples are more enriched in heavier isotopes than the Weihe River, probably resulting from the different evaporation rate of these two rivers. The GW samples show a broader range of stable isotopes than SW. Some GW samples are adjacent to the SW, possibly indicating the close interaction of the SW and GW. The average deuterium excess of the Weihe River (5.6‰) is close to the intercept of the local meteoric water line (6.13‰), so it can be assumed that the evaporation is not obvious. The average deuterium excess of the Luo River is only 0.1‰, which may imply that the evaporation in the Luohe River is significant. As Figure 6c,d shows, in 2016, there are also some river segments along which δ 18 O or δ 2 H enrich (deplete) with much higher (lower) tracer values in the ambient GW. It suggests these river segments are gaining reach, e.g., the segments LP2-LP1 in June, WP4-WP3 in August. The remaining segments that have not shown such relationships may indicate losing reaches. In June, the Huaxian Station is in the river segment WP5-WP3. δ 18 O enriches from −8.6‰ (WP5) to −7.8‰ (WP3) while the average δ 18 O value of the ambient GW (B41 and B282) is −7.8‰, indicating the segment is not a gaining reach. In September, the Huaxian Station is in the river segment WP6-WP4. δ 18 O enriches from −8.5‰ (WP6) to −8.1‰ (WP4) while the average δ 18 O value of the ambient GW (B41 and B282) is −8.8‰, showing that the segment is not a gaining reach. The SW-GW interaction is also consistent with the results deduced from the water level data in current status.

Surface Water and Groundwater Interaction Variations
The Cl − concentrations of the SW and GW alongside the rivers from two campaigns in May 1959 and August 1959 were selected to obtain the condition of the SW-GW interaction before the dam operation (Figure 6a  The SW samples show a narrow range in Figure 7. The Luohe River water samples are more enriched in heavier isotopes than the Weihe River, probably resulting from the different evaporation rate of these two rivers. The GW samples show a broader range of stable isotopes than SW. Some GW samples are adjacent to the SW, possibly indicating the close interaction of the SW and GW. The average deuterium excess of the Weihe River (5.6% ) is close to the intercept of the local meteoric water line (6.13% ), so it can be assumed that the evaporation is not obvious. The average deuterium excess of the Luo River is only 0.1% , which may imply that the evaporation in the Luohe River is significant.
As Figure 6c,d shows, in 2016, there are also some river segments along which δ 18 O or δ 2 H enrich (deplete) with much higher (lower) tracer values in the ambient GW. It suggests these river segments are gaining reach, e.g., the segments LP2-LP1 in June, WP4-WP3 in August. The remaining segments that have not shown such relationships may indicate losing reaches. In June, the Huaxian Station is in the river segment WP5-WP3. δ 18 O enriches from −8.6% (WP5) to −7.8% (WP3) while the average δ 18 O value of the ambient GW (B41 and B282) is −7.8% , indicating the segment is not a gaining reach. In September, the Huaxian Station is in the river segment WP6-WP4. δ 18 O enriches from −8.5% (WP6) to −8.1% (WP4) while the average δ 18 O value of the ambient GW (B41 and B282) is −8.8% , showing that the segment is not a gaining reach. The SW-GW interaction is also consistent with the results deduced from the water level data in current status.

Surface Water and Groundwater Interaction Variations
The Cl − concentrations of the SW and GW alongside the rivers from two campaigns in May 1959 and August 1959 were selected to obtain the condition of the SW-GW interaction before the dam operation (Figure 6a,b). δ 18 O and δ 2 H values of surface water and groundwater alongside the rivers from two campaigns in June 2016 and September 2016 were selected to obtain the condition of the SW-GW interaction in the current status, as shown in Figure 6c,d. The variations of the SW-GW interaction along the Weihe and Luohe Rivers can be seen from Figure 6.
By comparing the interaction conditions before the dam operation and in current status (comparing May 1959 with June 2016; August 1959 with September 2016), several variations of the SW-GW interaction can be found. Before the dam operation, the channels upstream W3 are losing reaches, whereas the channels downstream W3 are gaining reaches. In current status, some channels downstream W3 are losing reaches, which may result from the rising SW level. As indicated by the relationship between SW level and ambient GW level at the Huaxian Station, an assumption can be made that the interaction pattern of SW-GW under the predominant impacts of the dam may shift from gaining reaches before the dam operation to losing condition after the dam operation. The shifting direction consistent with the assumption can be found along the reach from near WP5 to WP3 and the reach from near WP6 to WP5, indicating that the impact of Sanmenxia Dam on the interaction of SW-GW can affect as far as to WP6. The interaction pattern shifting from losing condition to gaining condition along the reach from WP8 to WP6 in August-September and the reach from LP2 to LP1 in May-June may result from the obvious decline of runoff in the Weihe and Luohe Rivers with the minor influence of the Sanmenxia Dam. In the current status, the reach WP2-WP1 nearest the Sanmenxia Dam is a losing reach in two campaigns, indicating the reach is under stronger influence of the dam than other river segments.

Anthropogenic Sources and Impacts on Surface Water and Groundwater
Human activity is one important factor influencing water quality. The general increase of the Cl − and NO 3 − ions may result from the rapid growth of the population and the fertilizer consumption in this area from 1959 to 2020 [44,45]. The population and the fertilizer consumption increase from 158.4 to 325.9 million tons and from 0.1 to 12.4 million tons, respectively (shown in Figure 8). There are serval kinds of anthropogenic inputs including agricultural activities, domestic wastewater, and untreated industrial or urban sewage that may affect the SW and GW. The threshold values of Cl − and NO 3 − that can be used to distinguish whether water samples are potentially affected by anthropogenic inputs can be estimated by cumulative probability plots [46,47] (Figure 9). By referring to the national standard of Cl − (250 mg/L) and NO 3 − (10 mg/L), the background levels of Cl − and NO 3 − (shown in Figure 10) are determined as 203.6 and 24.5 mg/L, respectively. Based on the background levels, four water types are identified among the water samples taken from sampling sites shown in Figure 1a for the research area. A detailed water type classification is provided as Figure 10a, the spatial distribution of the dominant anthropogenic input types for each sampling site are shown in Figure 10b,c. The percentages of each water type in different water bodies and periods are estimated and shown in Table 3.
Water 2020, 12, x FOR PEER REVIEW 12 of 18 made that the interaction pattern of SW-GW under the predominant impacts of the dam may shift from gaining reaches before the dam operation to losing condition after the dam operation. The shifting direction consistent with the assumption can be found along the reach from near WP5 to WP3 and the reach from near WP6 to WP5, indicating that the impact of Sanmenxia Dam on the interaction of SW-GW can affect as far as to WP6. The interaction pattern shifting from losing condition to gaining condition along the reach from WP8 to WP6 in August-September and the reach from LP2 to LP1 in May-June may result from the obvious decline of runoff in the Weihe and Luohe Rivers with the minor influence of the Sanmenxia Dam. In the current status, the reach WP2-WP1 nearest the Sanmenxia Dam is a losing reach in two campaigns, indicating the reach is under stronger influence of the dam than other river segments.

Anthropogenic Sources and Impacts on Surface Water and Groundwater
Human activity is one important factor influencing water quality. The general increase of the Cl − and NO3 − ions may result from the rapid growth of the population and the fertilizer consumption in this area from 1959 to 2020 [44,45]. The population and the fertilizer consumption increase from 158.4 to 325.9 million tons and from 0.1 to 12.4 million tons, respectively (shown in Figure 8). There are serval kinds of anthropogenic inputs including agricultural activities, domestic wastewater, and untreated industrial or urban sewage that may affect the SW and GW. The threshold values of Cl − and NO3 − that can be used to distinguish whether water samples are potentially affected by anthropogenic inputs can be estimated by cumulative probability plots [46,47] (Figure 9). By referring to the national standard of Cl − (250 mg/L) and NO3 − (10 mg/L), the background levels of Cl − and NO3 − (shown in Figure 10) are determined as 203.6 and 24.5 mg/L, respectively. Based on the background levels, four water types are identified among the water samples taken from sampling sites shown in Figure 1a for the research area. A detailed water type classification is provided as Figure 10a, the spatial distribution of the dominant anthropogenic input types for each sampling site are shown in Figure 10b,c. The percentages of each water type in different water bodies and periods are estimated and shown in Table 3.

Conceptual Model and Environmental Implications
Previous studies have not paid much attention to the SW-GW interaction under the impact of the Sanmenxia Dam. However, the SW-GW interaction has changed since the Sanmenxia Dam operation. A conceptual model of the SW-GW interaction at a representative section (the Huaxian Station) is presented as Figure 11. In order to give some environmental implications, several components like unregulated rural industrial input and domestic input that are not found around the Huaxian Station at current status but appear in the whole research area are also integrated into this model. Thus, this model helps to draw a more comprehensive picture of the SW-GW interaction and the potential aquatic environment problems. The uplifted riverbed caused by the Sanmenxia Dam makes the river water level higher in the current status and eventually induces the variation of SW-GW interaction. Anthropogenic sources like industrial wastewater, domestic sewage, irrigation return flow, and surface runoff both affected by agricultural activities are also influencing the water quality.
The SW and GW are not two isolated parts but an interconnected system. The pollutants discharging into SW may migrate to GW, of which the inverse process can also happen in the lower Weihe River Basin, through the SW-GW interactions. Since the dam has an obvious impact on the interaction, it is essential to take the dam operation into account when coping with the serious water environment in this area. Figure 11. Conceptual diagram of the SW-GW interaction integrated with potential anthropogenic inputs: (a) before the dam operation, (b) at current status.

Conclusions
The Sanmenxia Dam has dramatically increased the water level in the lower Weihe River since 1959. The SW level rise has inevitably affected the SW-GW interaction that plays a vital role in hydroenvironment. Meanwhile, the lower Weihe River area is being faced with hydro-environment pressures. Therefore, it is crucial to assess the impact of the dam on the SW-GW interaction. The main conclusions of the study are as follows: (1) At the Huaxian Station, the dam operation plays a key role in raising the SW level. After the 1980s, the GW level decreased because of the increased groundwater exploitation. The variations of SW and GW levels induced the shift of SW-GW interaction. Three different patterns of the SW-GW interaction can be identified in the past 60 years including (i) gaining reach during the early stage (i.e., in 1959), (ii) gaining reach in non-flood season and losing reach in flood season (i.e., in 1981), and (iii) losing reach in 2002 and 2010. (2) Detailed SW-GW interaction patterns along the lower Weihe River are first examined from the multi-environmental tracers (Cl − , δ 18 O, and δ 2 H) of the water samples. Before the dam operation, the channels upstream W3 are losing reaches, while downstream W3 are gaining reaches. In the current status, the distribution of SW-GW interaction along the Weihe River is more complex. Some channels downstream W3 are losing reaches, which may result from the rising SW level induced by the dam operation. Gaining reaches also appeared upstream W3, indicating the impact of the dam lessens as the distance increase. The impact of the dam on the SW-GW interaction can reach to W3 (65 km from the estuary of the Weihe River).

Conclusions
The Sanmenxia Dam has dramatically increased the water level in the lower Weihe River since 1959. The SW level rise has inevitably affected the SW-GW interaction that plays a vital role in hydro-environment. Meanwhile, the lower Weihe River area is being faced with hydro-environment pressures. Therefore, it is crucial to assess the impact of the dam on the SW-GW interaction. The main conclusions of the study are as follows: (1) At the Huaxian Station, the dam operation plays a key role in raising the SW level. After the 1980s, the GW level decreased because of the increased groundwater exploitation. The variations of SW and GW levels induced the shift of SW-GW interaction. Three different patterns of the SW-GW interaction can be identified in the past 60 years including (i) gaining reach during the early stage (i.e., in 1959), (ii) gaining reach in non-flood season and losing reach in flood season (i.e., in 1981), and (iii) losing reach in 2002 and 2010. (2) Detailed SW-GW interaction patterns along the lower Weihe River are first examined from the multi-environmental tracers (Cl − , δ 18 O, and δ 2 H) of the water samples. Before the dam operation, the channels upstream W3 are losing reaches, while downstream W3 are gaining reaches. In the current status, the distribution of SW-GW interaction along the Weihe River is more complex. Some channels downstream W3 are losing reaches, which may result from the rising SW level induced by the dam operation. Gaining reaches also appeared upstream W3, indicating the impact of the dam lessens as the distance increase. The impact of the dam on the SW-GW interaction can reach to W3 (65 km from the estuary of the Weihe River). GW sampling sites of BP2, BP11, and BP17 (mean Cl − concentration of 372.2 mg/L) might be affected by industrial input in the rural area. (4) Evidence has also shown that different SW-GW interaction patterns may produce different influences on the water system. The impacts of the dam which change the SW-GW interaction patterns and make the water system more variable present a challenge to water resource management in this area. Further research may focus on how the different SW-GW interaction patterns affect the pollutant migration and make a suitable operation mode of the dam to improve the local water resources management.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4441/12/6/1671/s1, Table S1: Major ions and stable isotope concentrations in surface water and groundwater in 1959, Table S2: Major ions and stable isotope concentrations in surface water and groundwater in 2015-2016.