Temporal Variation Characteristics

In the downstream discharge areas of FL1, FL2, and FL3, K⁺ and Na⁺ trajectories displayed certain similarities. K⁺ and Na⁺ concentrations in FL1 remained stable from May to August (Fig. 3). In September 2019, they dropped below the detection threshold, but peaked in October at 0.40 mg/L and 1.99 mg/L, respectively. The concentrations subsequently decreased to 0.35 mg/L and 0.6 mg/L, maintaining these levels in the following months. K⁺ and Na⁺ concentrations in FL2 began to increase in May, peaking in August at 1.06 mg/L and 2.63 mg/L, respectively, before falling to their lowest values in September at 0.18 mg/L and 0.80 mg/L. Thereafter, the levels exhibited a fluctuating upward trend and stabilized. The K⁺ and Na⁺ concentrations in FL3 showed unusual declines in May, September, and April, with K⁺ reaching abnormal levels of 0.51 mg/L, 0.17 mg/L, and 0.35 mg/L, respectively, while other months maintained levels between 0.71 and 0.96 mg/L. The anomalous values for Na⁺ were 1.09 mg/L, 0.57 mg/L, and 0.41 mg/L, with other months maintaining levels between 2.16 and 2.82 mg/L. The minor fluctuations in these ion concentrations were not substantial enough to be definitively attributed to wastewater discharge but could be due to the leaching effects of aluminosilicates in the carbonate region. Between May and August, Ca²⁺ rose from 76.03 mg/L to 100 mg/L, with subsequent months fluctuating between 80 and 100 mg/L in FL1. The Ca²⁺ concentrations at FL1 and FL3 remained at approximately 90 mg/L with minor fluctuations. The Mg²⁺ concentration at FL1 was high at 12 mg/L in May, dropping to 0.6 mg/L in September, rising again to 6 mg/L in October, before steadily decreasing to approximately 3.5 mg/L. Concentrations of Mg²⁺ showed anomalies in September and April 2020, dropping to 3.13 mg/L and 1.54 mg/L in FL3, respectively, with other months fluctuating between 4.6 and 6.8 mg/L.

In the recharge area, Na⁺ and K⁺ concentrations in FL4 showed no significant temporal changes (Fig. 3). The concentration ranges were 0.58-1.66 mg/L for Na⁺ and 0.3-0.94 mg/L for K⁺. The concentration of Ca²⁺ increased in August, peaked at 112.00 mg/L in September, and then sharply decreased to 44.91 mg/L in October, before rising again and stabilizing. At the Overflow Karst Window, FL6, K⁺, and Na⁺ concentrations exhibited a fluctuating downward trend from May 2019 to April 2020, reaching lower values of 0.44 mg/L and 0.84 mg/L, respectively, followed by an upward trend. The Ca²⁺ and Mg²⁺ concentrations displayed opposite trends. The K⁺ and Na⁺ concentrations of FL7 began to increase in September, peaking in April of the following year at 1.25 mg/L and 3.78 mg/L, respectively, and subsequently decreased. The concentration of Ca²⁺ fluctuated downward from May 2019 to July 2020, generally maintaining levels of approximately 95-120 mg/L. The concentration of Mg²⁺ fluctuated and increased from May 2019 to April 2020, peaking at 6.01 mg/L in April.

Fig. 3.

Time series line graph of cation changes.

Fig. 4.

Time series line graph of cation changes at the Nongfa Tunnel.

The discharge outlet of the Nongfa Tunnel drainage (FL5) was influenced by tunnel construction, leading to elevated concentrations of Na⁺ and K⁺ in May and November (Fig. 4). Specifically, the concentration of Na⁺ reached 37.40 mg/L and 27.13 mg/L, respectively, whereas K⁺ peaked at 31.35 mg/L and 28.31 mg/L. The temporal variation curve for Ca²⁺ displayed significant fluctuations; it surged nearly twentyfold to 57.30 mg/L from May to August and remained stable until September. From October 2019 to April 2020, the Ca²⁺ concentration dropped to 20~25 mg/L but recovered to 60 mg/L in July. Fluctuations in Ca²⁺ levels may be associated with the plasma effect caused by the input of other cations (Na⁺ and K⁺). The Mg²⁺ concentration exhibited minimal changes, with notable fluctuations only in May and October. It reached its lowest at 0.78 mg/L in May and peaked at 19.91 mg/L in October; during other months, it fluctuated between 4 and 9 mg/L.

The SO₄^2- and Cl^- concentration curves for the downstream points of Fulong Spring exhibited a high degree of similarity (Fig. 5), suggesting that the influencing factors might have originated from the same source. The concentrations of SO₄^2- and Cl^- in FL3 decreased by more than 5 mg/L from August to October, followed by an increase from October 2019 to April 2020. The NO₃^- concentration decreased by approximately 10 mg/L between September and October 2019, whereas the HCO₃^- concentration curve remained relatively stable. The concentrations of SO₄^2- and Cl^- in FL1 and FL2 increased from November 2019 to April 2020, with the increase at FL2 and FL3 being relatively more significant than that at FL3. There were minimal seasonal variations in HCO₃^- and NO₃^-. Given that Cl^- is a conservative element and does not participate in chemical weathering, variations in this element, compared to other sampling points, might have been caused by factors other than water-rock interactions.

Variations in the Cl^- and HCO₃^- concentrations were not significant in the two upstream karst windows, FL4, FL6, and Danxian Spring FL7, as illustrated in Fig 6. However, from October to April, the sulfate SO₄^2- concentrations increased at FL4 and FL6 and decreased at FL7. From May to September, the NO₃^- concentration at FL4 decreased significantly by over 20 mg/L before increasing, which could be linked to denitrification in an anaerobic environment owing to low dissolved oxygen levels. The concentration returned to its original level by November, and by the rainy season in July, the NO₃^- concentration decreased to near zero owing to dilution effects.

Ion concentrations within the Fulong Springs watershed exhibited substantial variability during the rainy season while remaining relatively stable during the dry season, likely due to the impact of surface runoff on groundwater. Apart from anomalies, K⁺, Na⁺, and NO₃⁻ levels in most samples remained stable throughout the year. Concentrations of Ca²⁺, Mg²⁺, HCO₃⁻, and Cl⁻ were slightly elevated in the dry season and reduced during the rainy season, reflecting the dilution effect of precipitation on groundwater. In contrast, SO₄²⁻ concentrations were marginally higher in the rainy season, possibly due to an increase in sulfate ions from rainfall contributing to groundwater.

Fig. 5.

Time series line graph of anion changes. a) in the Downstream Discharge Area and b) in the Upstream Recharge Area.

Spatial Variation Characteristics

The Na⁺ and K⁺ concentrations exhibited sudden increases only at FL5, as shown in Fig. 6. An increase in Na⁺ concentration above 20 mg/L suggests possible sewage or pollutant intrusion rather than water-rock interactions or precipitation effects. In contrast, the spatial variations in Ca²⁺ and Mg²⁺ concentrations were more complex across the different spring sites, as illustrated in Fig. 6. The differences between the tunnel drainage and karst springs were the most pronounced. Notably, Mg²⁺ experienced an unusual decrease, whereas Ca²⁺ showed significant disparities beginning in October at the Dangui Karst Window compared to other sites, with Ca²⁺ concentrations notably lower than those at other points. In addition, the Mg²⁺ concentration increased in the Overflow Karst Window.

HCO₃^- showed the greatest spatial variation among the four anions, with a significant drop observed in the Tunnel Drainage (FL5), as depicted in Fig. 7. Tunnel construction primarily caused this decline. The other three anions (NO₃^-, Cl^-, and SO₄^2-) exhibited no significant differences at FL4 compared with the other spring sites. Notably, the concentrations of NO₃^- and Cl^- increased more markedly at sites other than FL4. This rise could not be explained by water-rock actions alone, with the NO₃^- rise likely stemming from agricultural activities involving fertilizers. In contrast, spatial variations in SO₄^2- were not significant and appeared random. This variability was mainly due to the catchment area of FL4, partly originating from the Dangui Village Valley, where agricultural activities and fertilizer application likely raised anion concentrations.

Correlation Analysis

To validate and analyze the geochemical characteristics of the Fulong Spring and their interrelations, a statistical correlation analysis was conducted using SPSS 22.0, using 13 parameters collected from May 2019, August to November 2019, and April and July 2020.

Fig. 6.

Line graph of spatial variations in anion concentrations.

Fig. 7.

Gibbs diagram of Fulong Spring.

Fig. 8.

Significance of correlation analysis of geochemical indicators in Fulong Spring. Note: ** Correlation is significant at the 0.01 level. * Correlation is significant at the 0.05 level.

As shown in Fig. 8, K⁺ exhibited a significant positive correlation with Na⁺, SO₄^2-, and pH but a slightly less significant correlation with sulfate. K⁺ showed a significant negative correlation with Ca²⁺, HCO₃^-, electrical conductivity (EC), and total dissolved solids (TDS). Na⁺ was significantly positively correlated with pH values and significantly negatively correlated with Ca²⁺, HCO₃^-, EC, and TDS. Ca²⁺ displayed a significant negative correlation with SO₄^2- and had significant negative correlations with pH and DO values; it was significantly positively correlated with HCO₃^-, EC, and TDS. HCO₃^- had significant positive correlations with EC and TDS and significant negative correlations with pH, SO₄^2-, and DO values. Cl^- showed a positive correlation with sulfate and a negative correlation with DO values.

In analyzing the hydrochemical indicators of the Fulong Spring, K⁺ and Na⁺ showed remarkably similar direct correlations with other chemical indicators, whereas the correlations between Ca²⁺ and HCO₃^- with other ions significantly overlapped. This indicates that the source mechanisms or influencing factors of these two pairs of ions are consistent, further confirming that the Fulong Spring tunnel is not only influenced by water-rock interactions, but also by external mechanisms.

Fig. 8 shows each geochemical indicator has direct and distinct correlations with the other indicators, complicating the analysis of the influencing factors. Analyzing individual indicators directly results in the loss of information and inaccurate results [23, 24]. Therefore, following the correlation analysis of the hydrochemical indicators of Fulong Spring, dimensionality reduction was performed. Principal component analysis [25, 26] was subsequently employed to further investigate the factors influencing the hydrochemical characteristics.

Principle Component Analysis (PCA)

Using SPSS 22.0, analyses were conducted on Fulong Spring’s geochemical indicators. Before principal component analysis, KMO and Bartlett tests were performed, yielding a KMO value >0.6 and a significance level of 0.00, indicating significant correlations and mutual independence suitable for PCA [27, 28]. PCA yielded four components with eigenvalues greater than 1, along with their variance and cumulative contribution rates. The first principal component accounted for 47.148% of the variance, while the remaining three principal components accounted for approximately 10%. The cumulative variance contribution rate was 79.704% (Table 2).

The following presents a detailed analysis of the four principal components. Fig. 8 and Table 3 show the first principal components, TDS, EC, HCO₃^-, and Ca²⁺, all have loadings greater than 0.8. These four indicators are closely related to carbonate dissolution, indicating that the first principal component primarily reflects the impact of water-rock interactions on Fulong Spring. In the second principal component, the loadings of Cl^- and SO₄^2- are 0.7779 and 0.6675, respectively. These indicators are closely related to human activities, including tunnel construction and agricultural activities. In PC3 and PC4, only NO₃^- has a significant load (0.5866). Based on previous analysis, this is attributed to pollution from agricultural fertilization activities in the valley affecting the Dangui Karst Window, as well as the nitrification and denitrification processes resulting from active microbial states in the water under anaerobic conditions.

Factor Analysis of the Nognfa Tunnel

Due to significant spatial differences between the sampling sites in the Nongfa Tunnel and the other sampling sites in Fulong Spring, a separate factor analysis was conducted for the Nongfa Tunnel.

Based on the PCA, four main factors with eigenvalues greater than one were identified, with a cumulative variance contribution rate of 97.899% (Table 4). This indicated that these four factors reflected 97.899% of the information in the overall sample. The component matrix was rotated using the varimax rotation method, resulting in a rotated factor-loading table (Table 5).

Based on the factor analysis, as shown in the table, the loads of EC, TDS, Ca²⁺, and HCO₃^- in RF1 are 0.959, 0.961, 0.577, and 0.921, respectively. The primary minerals in the carbonate region are calcite and dolomite. The concentrations of Ca²⁺ and HCO₃^- in water are closely related to karst processes. Therefore, RF1 can be considered controlled by water-rock interactions in the carbonate region. In RF2, the loads of Na⁺ and K⁺ are 0.828 and 0.793, respectively, whereas that of SO₄^2- was 0.973. Previous analyses showed that the concentrations of these three ions do not exhibit significant seasonal variations in response to precipitation, evaporation, or karst processes. The negative loading of Ca²⁺ further excluded the influence of karst processes. Thus, RF2 can be interpreted as reflecting the impact of the Na⁺ and K⁺ components from tunnel discharge on Nongfa Tunnel sampling. In RF3, only Ca²⁺ and Mg²⁺ have significant loads; therefore, RF3 can be considered to represent the influence of the tunnel discharge containing Ca²⁺ and Mg²⁺ on the Nongfa Tunnel. In RF4, only NO₃^- and Cl^- have significant loads. Based on previous analyses, this factor can be attributed to pollution from agricultural fertilization in the surrounding valley bottom, which affects runoff.