Differences and variations of drinking water quality between the northern and southern regions of a tropical island and the indicators influenced by meteorological factors

Climate change is a major environmental problem at present and can directly and indirectly affect drinking water. It affects water access for all people in the world, causing severe floods and droughts. Short- or long-term meteorological factors impact the water cycle and precipitation. Moreover, many related factors, including solar radiation and ultraviolet (UV) rays, influence the source and quality of the drinking water. Tropical islands have a unique ecological environment, and the drinking water system has a certain vulnerability. The influence of climate on the quality of source water and drinking water is extensive, with factors, such as rainfall, sunlight, UV, and radiation. Among these factors, rainfall is reported to have a lag effect after 14 days (Murphy et al. 2020). Fang et al. (2018) reported positive correlations among Escherichia coli, Enterococcus, and turbidity with rainfall (r = 0.69–0.95) and negative correlations with total dissolved solids (TDS) (r = −0.58) in the urbanized catchments of Singapore. After Hurricane Harvey, the concentrations of total organic carbon, trihalomethanes, and bacteria in drinking water spiked (Landsman et al. 2019; Skaland et al. 2022). The increase in drinking water pollution has been linked to recent heavy rainfall and high temperatures (Guo et al. 2021).

The northern (19°31′–20°04′N) and southern (18°09′–18°37′N) regions of Hainan Island experience higher temperatures, more rainfall, strong UV rays, and tropical monsoon climate characteristics of tropical storms. In this study, we treated the drinking water of urban and rural areas. The rivers and artificial reservoirs are the primary sources of drinking water in the northern urban area, whereas the drinking water in the southern urban area is sourced from lakes. Reservoir here refers to the artificial excavation of a large impounding reservoir. It is the source of natural groundwater, and the water usually comes from multiple sources.

Meanwhile, in the rural northern area, water is obtained from rivers, reservoirs, and groundwater. At present, reports on the influence of climatic factors on treated water and terminal tap water (TTW) of the water supply system are few. The effects of climatic factors on the quality of treated drinking water from river and lake sources in tropical regions are unknown. Even slight changes in micronutrients or some harmful components in drinking water may impact the health of the population due to long-term drinking. Higher levels of selenium in drinking water and a total hardness (TH) exceeding 2.84 mmol/L can exert a health-promoting effect on the cardiovascular system (Liu et al. 2017; Rapant et al. 2021). Fluorine of less than 1.0 mg/L may exhibit beneficial effects on bone health and calcium metabolism (Allolio & Lehmann 1999). However, increased mercury, arsenic, lead, and disinfection byproducts (DBPs) may have potential adverse effects on human health (Lu et al. 2015; Helte et al. 2023). This study mainly explores the effects of temperature, rainfall, solar radiation, and UV in tropical climate on the general chemical indicators, trace substances, heavy metals, microorganisms, and DBPs of drinking water to provide a basis for its accurate monitoring and early warning as well as offer data for the study of its impact on human health.

Haikou (HK) and Sanya (SY) are coastal regions located in the north and south areas of Hainan Island, respectively. Both regions exhibit a tropical monsoon climate characterized by distinct dry and rainy seasons (Manual 2013). They experience high temperatures (annual average of 24.4–25.5 °C), intense solar radiation (total solar radiation of 4,500–5,800 MJ/m²), and UV rays (8–10 levels). The indicators and units are as follows: turbidity (TUR, NTU), pH (pH value), aluminum (Al, mg/L), ferrum (Fe, mg/L), manganese (Mn, mg/L), zinc (Zn, mg/L), copper (Cu, mg/L), chloride (Cl⁻, mg/L), sulfate (⁠⁠, mg/L), TDS (mg/L), chemical oxygen demand (COD, mg O₂/L), total hardness (TH, mg/L), arsenic (As, mg/L), fluorine (F, mg/L), nitrate (⁠⁠, mg/L), lead (Pb, mg/L), selenium (Se, mg/L), trichloromethane (CHCl₃, mg/L), chlorate (⁠⁠, mg/L), total bacterial count (TBC, CFU/mL), and total coliforms (TC, MPN/100 mL). If any of the indicators did not satisfy the national standard, then the sample is considered unqualified.

In total, 773 treated water and drinking water samples were collected in the north and the south, 666 from the north regions and 107 from the south regions. In total, 192 samples were treated water and collected from the city and rural areas, where 474 samples were TTW. Among these TTW samples, 265 were from the countryside, and 209 samples were from urban areas. The 107 samples from southern China were composed entirely of TTW, with 52 samples collected from urban areas and 55 samples from rural regions.

Urban water samples were collected from all water plants in the southern and northern cities, and rural water samples were collected from the local monitoring sites of the National Rural Drinking Water Safety Project. The sampling sites were designed by stratification of randomization to ensure the samples were collected at key sites and considered the representativeness of all the investigating regions. The northern data were collected during the wet and dry seasons, whereas the southern water was gathered mostly during the wet season.

The water samples in the southern city were from three treatment plants. In total, 52 TTW samples were from 45 sites, and 55 TTW samples were from 37 sites in the south countryside. In the northern urban region, the samples were collected from three treatment plants. Specifically, 209 TTW samples were from 57 sites, 192 were from treated water, and 265 TTW samples were from 75 sites in the countryside.

The drinking water in the north urban areas was sourced from rivers and reservoirs. In the north rural areas, it was from various sources, such as rivers, reservoirs, and the ground. Meanwhile, the drinking water in the southern areas was from lakes and reservoirs.

The methods of transporting and testing were in accordance with the Hygienic Standard for Drinking Water (GB 5749-2006). After 1 April 2023, water samples were collected, preserved, transported, and tested with reference to the new standard: Hygienic Standard for Drinking Water (GB 5749-2022).

From 2022 to 2023, water quality data were collected from the Centers for Disease Control and Prevention in HK and SY. The daily precipitation (mm), daily average temperature (°C), daily UV index, and daily total solar radiation (W/m²) of the two regions in Hainan Island were obtained from the Hainan Provincial Meteorological Bureau. The detection indicators included 12 items of sensory characteristics and general chemical indicators (TUR, pH, Al, Fe, Mn, Zn, Cu, Cl⁻, ⁠, TDS, COD, and TH), 7 items of toxicological indicators (As, F, ⁠, Pb, Se, CHCl₃, and ⁠), and 2 items of bacteriological indicators (TBC and TC).

SPSS 26.0 was employed for analysis, the skewed distribution data were described using the median (M), and differences in water quality among various groups were compared using a non-parametric test (Mann–Whitney U test). Spearman correlation analysis was conducted to calculate the correlation between each variable to determine the water quality indicators related to meteorological factors. p < 0.05 was considered statistically significant.

The binary multifactor logistic regression analysis was further used to explore the major factors that affected water quality. The P50, P75, and P90 values of the TTW indicators in the north and south cities and the countryside (four groups) were compared. The indicators for the regression analysis were explored only in the groups where P90 occurred or the group in which the P50, P75 and P90 values of indicators all occured. The main meteorological factors (i.e., temperature, solar radiation, rain, and source water type), water period, and disinfection were considered independent variables. The assignment details are provided in Supplementary Table S1. OR = 1 meant that the independent variable (X) and the dependent variable (Y) had no correlation. OR > 1 meant that X may have promoted the appearance of Y, and OR < 1 indicated that X hindered the appearance of Y (p < 0.05).

Origin2024 was used for drawing water quality indicators and the correlation of climate index (Figure 1), and Excel2019 was used for drawing high water quality indicators in the frequency of extreme weather (Figure 2).

The qualified rate of all drinking water samples was high according to the Chinese drinking water standard, accounting for 97.41% (753/773). The main failures were TBC and residual disinfectant.

In general, only the F and COD of drinking water in the northern rural areas were lower than those in the southern rural areas, whereas the Mn, pH, TBC, ⁠, Cl⁻, TDS, and TH were more than those in the southern rural areas (p < 0.05). No significant difference in Mn, TH, and F in drinking water was observed between the north and south in urban regions (p > 0.05), but they were inconsistent in different water seasons. CHCl₃, Al, As, and Cu were abundant in the northern urban area during the wet season, whereas pH, TH, TDS, and were more pronounced in the northern urban area during the dry season.

For temporal matching, the northern rural TTW was included for analysis with the southern rural tap water data in 2022. Significant differences in the 13 indicators of TTW were observed between the northern and southern rural areas (p < 0.05). F and COD in TTW from northern rural areas were lower than those from southern rural areas. However, 11 indicators, such as TBC, TUR, As, Fe, TH, and TDS, were significantly higher in the northern than in the southern rural areas (Table 1, p < 0.05). In particular, AS and Fe showed significant differences.

Table 1

Differences in the quality of terminal tap water between the two rural areas

Indicators .	M (P25, P75) .		p .
	Rural north n = 117 .	Rural south n = 55 .
F (mg/L)	0.13 (0.1, 0.215)	0.16 (0.15, 0.19)	0.001
COD (mg O2/L)	0.48 (0.28, 0.88)	0.97 (0.86, 1.1)	<0.001
TBC (CFU/mL)	19 (15, 34)	0 (0, 2)	<0.001
(mg/L)	6.2 (4, 11)	2.9 (2.7, 3.2)	<0.001
Cl− (mg/L)	10.9 (9.105, 13.4)	8 (6.6, 10.3)	<0.001
Mn (mg/L)	0.0045 (0.00125, 0.018)	0.0031 (0.00064, 0.0069)	0.01
TDS (mg/L)	214 (109, 268.5)	95 (85, 106)	<0.001
As (mg/L)	0.0006 (0.0002, 0.0023)	0.00019 (0.00011, 0.00023)	<0.001
Fe (mg/L)	0.0202 (0.1225, 0.0443)	0.0027 (0.0012, 0.0055)	<0.001
(mg/L)	0.78 (0.5, 1.91)	0.4 (0.3, 0.5)	<0.001
TH (mg/L)	92.8 (40.05, 143.55)	37.4 (23.7, 46)	<0.001
pH	7.69 (7.295, 8.035)	7.17 (7.06, 7.4)	<0.001
TUR (NTU)	0.5 (0.5, 0.53)	0.18 (0.13, 0.24)	<0.001

Indicators .	M (P25, P75) .		p .
	Rural north n = 117 .	Rural south n = 55 .
F (mg/L)	0.13 (0.1, 0.215)	0.16 (0.15, 0.19)	0.001
COD (mg O2/L)	0.48 (0.28, 0.88)	0.97 (0.86, 1.1)	<0.001
TBC (CFU/mL)	19 (15, 34)	0 (0, 2)	<0.001
(mg/L)	6.2 (4, 11)	2.9 (2.7, 3.2)	<0.001
Cl− (mg/L)	10.9 (9.105, 13.4)	8 (6.6, 10.3)	<0.001
Mn (mg/L)	0.0045 (0.00125, 0.018)	0.0031 (0.00064, 0.0069)	0.01
TDS (mg/L)	214 (109, 268.5)	95 (85, 106)	<0.001
As (mg/L)	0.0006 (0.0002, 0.0023)	0.00019 (0.00011, 0.00023)	<0.001
Fe (mg/L)	0.0202 (0.1225, 0.0443)	0.0027 (0.0012, 0.0055)	<0.001
(mg/L)	0.78 (0.5, 1.91)	0.4 (0.3, 0.5)	<0.001
TH (mg/L)	92.8 (40.05, 143.55)	37.4 (23.7, 46)	<0.001
pH	7.69 (7.295, 8.035)	7.17 (7.06, 7.4)	<0.001
TUR (NTU)	0.5 (0.5, 0.53)	0.18 (0.13, 0.24)	<0.001

View Large

A comparison of the water indicators of TTW in the northern and southern urban areas revealed no statistically significant differences in 10 indicators between the two areas (p < 0.05). Only (a byproduct of ClO₂ disinfection) was more abundant in the southern area than in the northern area, whereas the nine other indicators, including CHCl₃ and TBC, were higher in the northern area (p < 0.05, Table 2). In addition, no significant difference in the content of Mn, F, and TH was observed in TTW between the two cities (p > 0.05).

Table 2

Differences in water quality in urban terminal tap water

Indicators .	M (P25, P75) .		p .
	Northern city (n = 113) .	Southern city (n = 52) .
CHCl3 (mg/L)	0.0156 (0.00745, 0.01995)	0.00165 (0.00765, 0.02825)	0.037
Cu (mg/L)	0.0012 (0.0008, 0.00205)	0.0007 (0.0003725, 0.003375)	0.04
COD (mg O2/L)	1.35 (1.115, 1.515)	1.0 (0.8625, 1.275)	<0.001
TUR (NTU)	0.5 (0.5, 0.5)	0.15 (0.13, 0.18)	<0.001
TBC (CFU/mL)	18 (14.5, 30)	0.000 (0.0, 5.75)	<0.001
(mg/L)	7.67 (4.88, 9.72)	3.3 (3, 3.575)	<0.001
Cl− (mg/L)	13.4 (11.2, 15.7)	10.900 (7.375, 11.875)	<0.001
(mg/L)	0.106 (0.009, 0.1415)	0.365 (0.3025, 0.42)	<0.001
TDS (mg/L)	131 (96, 195.5)	94 (88.25, 98)	<0.001
pH	7.55 (7.265, 7.835)	7.195 (7.02, 7.31)	<0.001

Indicators .	M (P25, P75) .		p .
	Northern city (n = 113) .	Southern city (n = 52) .
CHCl3 (mg/L)	0.0156 (0.00745, 0.01995)	0.00165 (0.00765, 0.02825)	0.037
Cu (mg/L)	0.0012 (0.0008, 0.00205)	0.0007 (0.0003725, 0.003375)	0.04
COD (mg O2/L)	1.35 (1.115, 1.515)	1.0 (0.8625, 1.275)	<0.001
TUR (NTU)	0.5 (0.5, 0.5)	0.15 (0.13, 0.18)	<0.001
TBC (CFU/mL)	18 (14.5, 30)	0.000 (0.0, 5.75)	<0.001
(mg/L)	7.67 (4.88, 9.72)	3.3 (3, 3.575)	<0.001
Cl− (mg/L)	13.4 (11.2, 15.7)	10.900 (7.375, 11.875)	<0.001
(mg/L)	0.106 (0.009, 0.1415)	0.365 (0.3025, 0.42)	<0.001
TDS (mg/L)	131 (96, 195.5)	94 (88.25, 98)	<0.001
pH	7.55 (7.265, 7.835)	7.195 (7.02, 7.31)	<0.001

View Large

The comparison of the water quality in two rural areas during the wet season showed that the nine indicators of Mn, As, pH, Fe, TBC, ⁠, Cl⁻, TDS, and TH were lower, and F was higher in the southern rural area than that in the northern area (p < 0.05, Supplementary Table S2). In the northern rural region, CHCl₃, Cl⁻, ⁠, and of TTW were more abundant during the wet season than the dry season (p < 0.05, Supplementary Table S3). The concentration of in the dry season was found to be only 0.1 times that observed in the wet season.

A comparison of the drinking water in the northern and southern urban areas during the wet season revealed that none of the indicators was greater in the south than in the north, similar to what is shown in Table 2 (p < 0.05).

The treated water and TTW were compared with determine the changes in the pipe network. The results showed differences only in Al, Pb, Cu, Zn, and CHCl₃ between the TTW and treated water in the northern rural areas (p < 0.05). Among these indicators, Al increased by 189.2%, Zn rose by 121.62% (Table 3), Pb increased by 42.86%, Cu increased by 50%, and CHCl₃ enhanced by 200%. Correlation analysis showed that all indicators were positively correlated (r = 0.242–0.508). The weakest correlation was found for Al, and the highest correlation was achieved by F, suggesting that Al changed remarkably in the pipe network. This finding is consistent with the results of the difference analysis of these water samples. The water in the south was not analyzed given that the treated water samples were few.

Mn, As, Fe, pH, F, TBC, ⁠, Cl⁻, TDS, TH, and in the rural northern treated water and TTW was not statistically significant (p > 0.05, Table 3).

The quality of the urban TTW from northern river and reservoir sources was analyzed, CHCl₃ of the river-sourced TTW accounted for more than 3.55 times that found in TTW from reservoir resources. Moreover, Cu was 0.4 times higher, was 0.83 times higher, and Se and COD were slightly elevated in water from the river source. However, Mn, TBC, ⁠, TDS, and TH were 0.8, 0.5, 0.31, 0.42, and 0.44 times lower, respectively. In addition, no significant difference in Al, Pb, As, Fe, Zn, F, TUB, and was found between the two groups (p > 0.05).

The comparison in water quality of urban TTW between reservoirs in the north and south, considering the lack of water samples from rivers in the south. was 0.72 times lower in TTW from reservoir sources in the north than that in the south. Other indexes with high concentrations in the north reservoirs were as follows: CHCl₃, TBC, F, TUR, Mn, Cl⁻, ⁠, TDS, and TH at 1.3, 34, 0.49, 2.33, 1.07, 0.4, 1.68, 1.27, and 0.55, respectively. Cu and COD in the urban TTW sourced from the two regions did not show statistically significant differences (p > 0.05).

The positive correlations with drinking water and meteorological factors in the urban areas from rivers (p < 0.05) were mainly as follows: CHCl₃, As, TBC, Mn, Al, COD, and temperature; CHCl₃, Al, As, TBC, and solar radiation; ⁠, Cl⁻, ⁠, TDS, Cu, F, COD, and rainfall. In particular, CHCl₃ was found to be more strongly correlated with temperature (r = 0.633–0.669) and solar radiation (r = 0.444–0.532).

The correlation analyses of meteorological factors on drinking water from urban areas in the northern reservoirs exhibited the following results: a strong positive correlation existed between Fe (r = 0.408–0.478) and rainfall; Pb (r = 0.380–0.458) and (r = 0.505–0.579, not the main DBP_S) had positive correlations with solar radiation; Pb, Fe, and Zn had positive correlations with temperature; Cu was correlated with rainfall; was partially positively correlated with temperature and solar radiation (p < 0.05) but not correlated with rainfall (p > 0.05), as shown in Figure 1(b).

The stronger negative correlations were found among F, ⁠, and rainfall, followed by TBC and rainfall with partial correlations. Nevertheless, some indicators had no statistically significant correlation with meteorological factors (p > 0.05).

The results of the correlation analysis of meteorological factors on TTW in urban areas sourced from southern lakes are shown in Figure 1(c).

The strongest negative correlations were found for (r = −0.673 to −0.340) and CHCl₃ (r = −0.634 to −0.483) with temperature (not the main DBPs), and (r = −0.560 to −0.296) and pH (r = −0.709 to −0.278) with total solar radiation. Other indicators with negative correlations were as follows: COD, ⁠, and partial Cl⁻ with solar radiation.

Strong positive correlations were also observed for COD (r = 0.360–0.415), TH (r = 0.404–0.447), and TBC (r = 0.570–0.737) with temperature; pH, Cl⁻, ⁠, COD, and TDS with rainfall; partial TDS with temperature and total solar radiation. Those with moderately positive correlations were as follows: ⁠, partial TH, and rainfall; TC, F, ⁠, ⁠, and temperature; TBC and total solar radiation.

Analysis of rural TTW in the north showed that the slightly stronger positive correlations were for (r = 0.150–0.315) and CHCl₃ (r = 0.248–0.432), which were positively correlated with rainfall and temperature in the previous 1–7 days. Temperature and solar radiation can promote an increase in TBC (r > 0.147, p < 0.05). The aluminum concentration was 0.0094 mg/L, and the mean level was positively related to the rainfall among the previous 2–4 days (r = 0.144–0.148). Surprisingly, few significant negative correlations were found between the indicators and meteorological factors. In addition, no positive correlations were observed between other water quality indicators and meteorological factors.

TTW in the rural south was more strongly influenced by meteorological factors, with stronger positive correlations for COD, AS, ⁠, and rainfall (0.412 > r > 0.684). Temperature and solar radiation were more strongly positively correlated with TUR (r = 0.512–0.672), and pH was more strongly positively correlated with solar radiation. Solar radiation was negatively correlated with ⁠, ⁠, and COD (−0.811 to −0.414).

Selenium was only detected in the north region, and the TTW concentration from the northern river sources was high at 0.001 mg/L. This indicator was negatively correlated (p < 0.05) with temperature (r = −0.419 to −0.248) and solar radiation (r = −0.476 to −0.306). By contrast, no statistically significant correlation was observed for TTW from the reservoirs.

The significant results of regression with the correlation analysis were seldom. The water period was influenced by the complex state of several meteorological factors. The analysis results for the high value groups are shown in Table 4. The findings suggest that solar radiation will decrease the concentration of (the main DBPs) in TTW in rural areas in the south. Additionally, an increase in temperature is anticipated to raise the concentration of CHCl₃ in the northern city. Furthermore, high temperatures have led to an uptick in both TBC and AS in TTW within north rural areas, with the dry season similarly contributing to this escalation.

Table 4

Multivariate regression analysis of factors influencing water quality in special groups of the terminal tap water

Region .	Indicators .		B .	SE .	Wald χ2 .	Odds ratio (95CI%) .	p-value .
South rural		Solar radiation	−0.179	0.09	4.004	0.836 (0.701–0.996)	0.045**
North city	CHCl3	Temperature	0.454	0.128	12.62	1.574 (1.226–2.022)	<0.001***
North rural	TBC	Temperature	0.162	0.05	10.467	1.176 (1.066–1.298)	0.001***
		Water period	2.159	0.827	6.81	8.665 (1.712–43.860)	0.009***
North rural	AS	Temperature	0.101	0.049	4.19	1.106 (1.004–1.218)	0.041**
		Water period	1.637	0.819	3.99	5.138 (1.031–25.599)	0.046**

Region .	Indicators .		B .	SE .	Wald χ2 .	Odds ratio (95CI%) .	p-value .
South rural		Solar radiation	−0.179	0.09	4.004	0.836 (0.701–0.996)	0.045**
North city	CHCl3	Temperature	0.454	0.128	12.62	1.574 (1.226–2.022)	<0.001***
North rural	TBC	Temperature	0.162	0.05	10.467	1.176 (1.066–1.298)	0.001***
		Water period	2.159	0.827	6.81	8.665 (1.712–43.860)	0.009***
North rural	AS	Temperature	0.101	0.049	4.19	1.106 (1.004–1.218)	0.041**
		Water period	1.637	0.819	3.99	5.138 (1.031–25.599)	0.046**

**p< 0.05, ***p < 0.01, the assignment of the independent and dependent variable tables in the Supplementary.

View Large

The TTW from the north river and the south lakes in cities also underwent complementary analysis to investigate more factors. The north reservoir was not adopted because of the unusual artificial storage of the original water. The regression analysis also revealed that COD changes with temperature in urban TTW (OR = 1.234) and source types (OR = 0.084); for TBC and radiation, OR = 1.074 and for source type, OR = 0.146 (p< 0.05). These findings reveal that solar radiation, temperature, water period, and source types were the main factors that increased the water quality values. The results were highly consistent with those of the correlation analysis.

This study illustrates the occurrence of extreme weather when water quality is ≥P90. The high value of indicators for drinking water from the northern urban river was mostly affected by extremely high temperatures and solar radiation, especially CHCl₃ and Al (Figure 2).

Temperature and solar radiation had the greatest and fastest effect on the high values of the and Cu of TTW from the northern reservoirs, especially for ⁠, with corresponding extremes of 100, 200, and 300%. Temperature or solar radiation alone had no elevating effect on Cu. Another factor was CHCl₃, with corresponding high temperatures and storms of 75, 150, and 225% for the first 3 days. Extreme weather also affected the elevation of TDS and Pb of high value in the northern reservoirs. The high value of F corresponded to a composition ratio of 50% of strong solar radiation on the previous day, but slightly affected the first 2 and 3 days.

The high values for TBC in TTW from lakes in the southern urban area occurred after extreme temperatures and total solar radiation, especially for temperature and TBC, with temperature extremes of 60, 120, and 160% for the three previous days. Meanwhile, high values of ⁠, associated with extremely high temperatures over the 3 days, ranging from 60 to 100%. This finding corresponded to heavy rainfall of 60% from the 3 days prior to the previous day.

The correlation between water quality and extreme weather in northern rural areas was lower than that in urban areas. Moreover, the frequency of rainstorms corresponding to high concentrations of CHCl₃ and Cu was higher at 38.46–107.69% and 26.92–73.08% on the first 3 days.

High values of F in the southern rural areas corresponded to the frequency of strong solar radiation ranging from 100 to 300%. Moreover, high values of TBC, ⁠, TDS, and PH corresponded to high temperatures and solar radiation in the range of 100–300%. Furthermore, high values of COD corresponded to heavy rainfall occurrences at 100%.

This study shows that temperature, rainfall, and solar radiation affect the quality of TTW popularly and complexly. The application of various statistical analyses and comparisons is conducive to discovering important influencing factors, new problems, and conducting supervision from the view of public health. These factors must be further explored in terms of health promotion and risk prevention. Notably, in tropical areas, high temperatures and strong UV and solar radiation usually occur. By contrast, strong solar radiation typically corresponds to extremely low rainfall. As a result, indicators highly influenced by rainfall are prone to low concentrations of water quality during strong solar radiation. Thus, they must be identified further. Usually, the factors affecting water quality include water purification and disinfection, varying sources (rivers, reservoirs, and underground), storage, regional geological elements, variations in surface contamination, and intersection of climatic factors. Although the correlation differs from the results of the regression analysis or the frequency analysis of extreme weather in some cases, the conclusion could be distinguished by the adaptive region and types of water. The common results in different analyses are worthwhile studying in similar regions.

Urban water quality from rivers in the north and lakes in the south of Hainan Island is more strongly influenced by meteorology. Meanwhile, TTW seems less influenced by meteorology in rural regions. We believe that the diversity of drinking water sources in the countryside mainly influences the analysis results. In general, urban drinking water quality from the northern reservoir is less affected by meteorology. This result is probably because the original water would be settled naturally for 1–2 days prior to treatment. The northern region is downstream of the largest river on the island, explaining the relatively high levels of most substances in its drinking water.

Indicators include microorganisms, COD, ⁠, TDS, DBPs, geological elements (e.g., TH, F, AS, and Se), and common toxicological indicators (e.g., Pb), which greatly impact human health. Notably, Zn, F, Se, and TH with higher concentrations may be more beneficial to human health while still adhering to standards of drinking water. This study revealed that Al, Pb, and Zn may be released in the pipe network from the northern countryside (Li et al. 2020; Zhang et al. 2022). Similarly, Lytle reported an increase in Pb in the water supply networks of Columbus, USA (Lytle et al. 2004). Meanwhile, the increase in CHCl₃ in the north may be related to the increase in temperature from the correlation and regression analyses. This increment is similar to the increase in the CHCl₃ content of drinking water during the hot season in Tunisia (Rabhi et al. 2016).

Particularly, F is positively correlated with temperature in the south and is opposite in the north. We believe that the difference may be due to the source of F in the north and south. In the south, F is mainly sourced from the geological environment. At high temperatures, geological elements enter the source water as a result of evaporative traction in groundwater. Rainfall is effective in reducing F in the north reservoir, which is presumably related to water dilution by rain. TDS was positively correlated with rainfall (TTW from rivers in the north and reservoirs in the south) probably because surface washout occurred in the source water, but further investigation is needed. Different studies have obtained inconsistent results on the changes in TDS in source water (Fang et al. 2018; Woldeab et al. 2018).

COD is more pronounced in the rural south, rainfall, temperature, and water source may all be its important influences, especially for its high values from the heavy rainfall.

The microbial indicators are positively correlated with temperature and solar radiation. Most of them are negatively correlated with rainfall, indicating that sufficient treatment by water supply plants in the two areas occurs prior to rainfall. The temperature and wet period were further found to be risky for microorganisms in rural areas by several analyses in this study. Thus, during extremely high temperatures, the risks to microorganisms must be considered. Similar to other studies, the main DBPs (CHCl₃ or ⁠) in water plants are positively correlated with the type of disinfectant and temperature or solar radiation (Zhang et al. 2013; Suh & Mitch 2021).

The difference in DBPs is that ClO₂ is used in most treatments in the south, whereas NaClO is popularly used in the north. The value is lower than those in cities in northern Chile and in Egypt (Radwan et al. 2021; Muñoz-Arango et al. 2023). in southern rural areas is positively correlated with rainfall, possibly due to increased disinfectants in response to contamination.

In addition, high temperatures increased CHCl₃ possibly because high temperatures cause algal growth and increase organic precursors. The level of CHCl₃ in drinking water in northern rural areas is higher during the wet season than during the dry season, consistent with the results of studies reported (Mazhar et al. 2024).

Arsenic is probably one of the health risks associated with drinking water worldwide. It is positively correlated with temperature and solar radiation in drinking water in the urban areas of the Northern Rivers. Moreover, regression analysis indicated that As increased at TTW during the dry season and with increasing temperature in the north rural region. This finding reminds the public to reduce the health risk in this special state of the enhanced groundwater evapotranspiration, given that the local As is of geological origin (Bonte et al. 2014). Pb is positively correlated with solar radiation in the source water of northern reservoirs. Some studies have reported that certain water distribution pipes contain Pb (Zhang et al. 2022), which cannot be ruled out as a possible cause of the increase in Pb at high temperatures. In addition, the occurrence of Al and Pb with extreme solar radiation and high temperatures suggests that water may release Pb at high temperatures.

A lag is observed among Cl⁻, TC, and Mn extremes in the southern urban area and during extreme weather. Some distance may exist between this contamination and the area of the lakes from which it originated. The source of the strong correlation between Cu and climate in the northern reservoirs has not been clarified thus far. In addition, a positive correlation is observed between COD and temperature in the water from the southern urban lakes during the previous 7 days. This finding may be related to the growth of bacteria and other organisms and is contrary to the results of the study in the countryside and those of Zhong et al. (2021). pH is more complex in its association with these factors. It reveals other differences between rural and urban areas.

Meteorological change, including extreme weather events, may exhibit a wide range of impacts on various water quality indicators of drinking water. However, the impact patterns are inconsistent in different locations. It is difficult to find accurate rules for preventing the risk and improving the healthy function of public water supply. The monitoring of TTW combined with meteorological and related influencing factors would be more effective for health supervision. The concentration of CHCl₃ increases with the increase in sunlight and temperature, requiring effective management. Rain slightly impacts the microbial indicators of drinking water in both areas, suggesting that the water supply system has excellent measures in place to handle tropical storms. Some water samples exhibit a limited increase in F, Zn, and TH concentrations with rising temperatures or precipitation. This condition may have beneficial effects on health within the standard range. However, the concentration of As and Pb in drinking water increases with the increase in temperature, which may have potential adverse effects on human health. However, the patterns are inconsistent, especially the differences between non-extreme weather and extreme weather. This finding suggests the need for enhanced emergency response measures for drinking water using actual conditions. During high temperatures and the wet season, supervision should be strengthened for microorganisms, DBPs, oxygen-consuming substances, and even Pb and Al in tropical areas. The decrease in DBPs of with solar radiation probably reveals the advantage of ClO₂ disinfection in rural South China and similar regions. This study provides valuable methods and information for studying and understanding the health impacts of climate change. It may also serve as a reference for ensuring the safe and stable supply of drinking water in tropical regions and similar subtropical areas.

The authors particularly express thanks to the members of the Climate Center of Hainan Province for their invaluable support, as well as the Key Laboratory of Tropical Translational Medicine of the Ministry of Education, School of Public Health, Hainan Medical University. The College Student Innovation team X202211810086, and the public research center of Hainan Medical University.

W.L. conceptualized the study, reviewed and revised the article, and administered the project. Y.G., Y.ZH., and X.P. sampled and wrote the primary article. H.K., M.H., and H.Y. reviewed the results. S.X. and H.H. reviewed and revised the article.

This study was supported in part by grant 821RC558 (W. Long) and 820MS156 (H. Ye) from the Science and Technology Department of Hainan Province.

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

The authors declare there is no conflict.