3.1 Temporal and spatial variation characteristics of water quality
Interannual variation of water quality.The annual water quality change trend of North Canal basin from 2015 to 2021, was investigated by the water quality data of 63 water quality stations (Fig. 2). With the implementation of the Action Plan for the Prevention and Control of Water Pollution in China (Meng et al. 2022, Wu et al. 2020), the water quality of the North Canal basin has been improved year by year.
Variation characteristics of water quality within the year. To analyze the rainfall characteristics of North Canal basin, the hydrological data of the Tongxian hydrological station during 1919–2020 was investigated, as shown in Fig. 3. The flood season of North Canal basin was distributed from June to September, and concentrated in July to August, accounting for more than 50% of the annual precipitation. Therefore, June to September was classified as the flood season, while the other months as non-flood season in this study. Monthly monitoring data of 63 water quality stations in the North Canal basin from 2015 to 2021 were investigated to analyze the variation characteristics of water quality within annual year, results as shown in Table 1 and Fig. 4. According to Table 1, the mean concentration of NH3-N, TP,TN and anionic surfactant in non-flood season was significantly higher than that in flood season. The concentration of NH3-N and TP fluctuated greatly in non-flood season,but the fluctuation of TN and anionic surfactant in flood season was slightly higher than that of the non-flood season. Although the mean values of CODMn, COD and BOD5 were close in flood season and non-flood season, the coefficient of variation was small in flood season and large in non-flood season.From the monthly average concentration of water quality indexes from 2015 to 2021, the water quality of the North Canal were better in flood season than in non-flood season, i.e. with lower mean concentration in flood season,which was consistent with the previous conclusion (Chen et al. 2021, Tao et al. 2022, Zhai &Zhang 2018). Zhang Yuhang (2021) found that the water quality in wet season (May to September) was better than that in dry season (January, February, December) and normal season (March, April, October, November) in North Canal basin, which indicated that the runoff in flood season can reduce the concentration of pollutants in the river, and the internal secondary pollution was an important reason for the worst water quality in normal season.
Table 1
Statistical results of water quality indexes of North Canal basin from 2015 to 2021.
Water quality parameters | Period | Min | Max | Mean | S.D. | Skewness | Kurtosis | Coefficient of Variation |
COD(mg/L) | Annual | 8.00 | 194.00 | 30.75 ± 1.75 | 26.77 | 2.73 ± 0.16 | 9.76 ± 0.32 | 0.87 |
Flood | 6.00 | 213.00 | 30.58 ± 1.65 | 25.27 | 3.29 ± 0.16 | 16.40 ± 0.32 | 0.83 |
Non-flood | 8.00 | 185.00 | 30.77 ± 1.90 | 29.11 | 2.74 ± 0.16 | 9.04 ± 0.32 | 0.95 |
CODMn(mg/L) | Annual | 1.97 | 28.50 | 5.99 ± 0.26 | 3.99 | 2.59 ± 0.16 | 9.80 ± 0.31 | 0.67 |
Flood | 1.75 | 25.68 | 5.92 ± 0.21 | 3.26 | 2.13 ± 0.16 | 8.23 ± 0.31 | 0.55 |
Non-flood | 1.83 | 35.85 | 6.03 ± 0.30 | 4.65 | 2.93 ± 0.16 | 12.09 ± 0.31 | 0.77 |
BOD(mg/L) | Annual | 0.50 | 63.22 | 7.35 ± 0.58 | 8.93 | 2.97 ± 0.16 | 10.73 ± 0.32 | 1.22 |
Flood | 0.45 | 67.18 | 7.07 ± 0.53 | 8.08 | 3.49 ± 0.16 | 18.00 ± 0.32 | 1.14 |
Non-flood | 0.50 | 61.24 | 7.50 ± 0.64 | 9.81 | 2.96 ± 0.16 | 9.78 ± 0.32 | 1.31 |
NH3-N(mg/L) | Annual | 0.07 | 41.53 | 3.01 ± 0.37 | 5.77 | 3.15 ± 0.16 | 12.71 ± 0.31 | 1.92 |
Flood | 0.04 | 40.15 | 2.71 ± 0.33 | 5.16 | 3.62 ± 0.16 | 18.10 ± 0.31 | 1.90 |
Non-flood | 0.05 | 42.21 | 3.18 ± 0.40 | 6.29 | 3.05 ± 0.16 | 11.31 ± 0.31 | 1.98 |
TP(mg/L) | Annual | 0.01 | 3.78 | 0.37 ± 0.04 | 0.59 | 2.54 ± 0.16 | 7.62 ± 0.32 | 1.62 |
Flood | 0.01 | 3.53 | 0.36 ± 0.04 | 0.57 | 2.72 ± 0.16 | 8.87 ± 0.32 | 1.58 |
Non-flood | 0.01 | 3.90 | 0.38 ± 0.04 | 0.63 | 2.45 ± 0.16 | 6.79 ± 0.32 | 1.67 |
TN(mg/L) | Annual | 1.00 | 48.00 | 9.13 ± 0.61 | 8.27 | 1.40 ± 0.18 | 3.09 ± 0.36 | 0.91 |
Flood | 1.00 | 44.00 | 8.1 ± 0.57 | 7.70 | 1.51 ± 0.18 | 3.26 ± 0.36 | 0.95 |
Non-flood | 1.00 | 49.00 | 9.67 ± 0.66 | 8.88 | 1.39 ± 0.18 | 2.93 ± 0.36 | 0.92 |
Anionic surfactant (mg/L) | Annual | 0.02 | 1.93 | 0.14 ± 0.02 | 0.24 | 5.03 ± 0.18 | 30.93 ± 0.35 | 1.68 |
Flood | 0.02 | 1.90 | 0.16 ± 0.02 | 0.25 | 4.16 ± 0.18 | 21.71 ± 0.35 | 1.54 |
Non-flood | 0.02 | 1.91 | 0.18 ± 0.02 | 0.27 | 3.66 ± 0.18 | 16.45 ± 0.35 | 1.52 |
Spatial distribution characteristics of water quality in flood season and non-flood season. As shown in Fig. 4, the water quality of the North Canal basin showed temporal and spatial variations. In general, water quality in flood season was better than that in non-flood season, and water quality in upstream was better than that in downstream, which was consistent with other studies (Wang et al. 2021). In flood season, there were larger water quantity, higher water temperature, longer sunshine duration, which lead to stronger aquatic biological activity and faster pollutant degradation rate. On the other hand, the spatially heterogeneous of water quality characteristics between water quality stations in flood season and non-flood season was investigated. The mean values of COD, CODMn, BOD, NH3-N, TP, TN and anionic surfactants of each water quality station in flood season and non-flood season during 2015–2021 was taken as the evaluation quantity, and the cluster analysis was carried out by hierarchical clustering method (HCA). As shown in Fig. 5, the 63 water quality stations were divided into three types with a distance of 10 as the orientation line. HCA grouped the water quality stations on the basis of similarity and dissimilarities in their characteristics.The cluster 1 was ST49, located in downstream of the North Canal basin, with poorer water quality (COD 101.61mg/l, NH3-N 23.52mg/l), all the seven water quality indexes show that water quality in flood season was significantly better than that in non-flood season, and the difference of COD concentration between flood season and non-flood season > 50mg/l. The second cluster include ST12,ST27,ST28, and ST35, the water quality of COD, CODMn, BOD and NH3-N in flood season was better than that in non-flood season, while the water quality of TP, TN and anionic surfactant in flood season was better than that in non-flood season or there was little difference. The rest of the other stations were included in the third cluster, each water quality index difference between flood season and non-flood season was not significant or characteristics were not uniform.
3.2 Influence of land use type on the water quality
The change of land use in North Canal basin.With the rapid development of society, the demand for land was increasing, the problem of land use was becoming more and more prominent, and the types of regional land use were constantly changing. The analysis results of land use patterns change in the North Canal Basin from 1980 to 2020 were showed in Table 2. It can be seen that the proportion of urban and rural construction land (Artificial surface) had been increasing, from less than 22% in 1980 to more than 45% in 2020, as shown in Fig. 6. Meanwhile, the area and proportion of cultivated land had been declining.Compared to 2015, the area and proportion of cultivated land, forest and Artificial surface declined in 2020, while there was a remarkable increase in the proportion of grassland.
Table 2
The change of land use in North Canal basin (Beijing) from 1980 to 2020.
Year | Proportion of Cultivated Land | Proportion of Forest | Proportion of Grassland | Proportion of Water Body | Proportion of Artificial Surfaces | Proportion of Other |
1980 | 55.13% | 21.44% | 0.51% | 1.85% | 21.07% | 0.00% |
1990 | 55.14% | 21.46% | 0.53% | 1.84% | 21.03% | 0.00% |
1995 | 43.22% | 22.39% | 0.44% | 2.55% | 31.40% | 0.00% |
2000 | 42.31% | 22.38% | 0.44% | 2.57% | 32.29% | 0.00% |
2005 | 35.99% | 22.04% | 0.44% | 2.24% | 39.29% | 0.00% |
2010 | 33.09% | 18.97% | 0.58% | 0.89% | 46.45% | 0.02% |
2015 | 29.74% | 18.95% | 0.51% | 0.84% | 49.94% | 0.02% |
2020 | 27.05% | 17.86% | 5.11% | 1.04% | 48.88% | 0.06% |
Influence of land use type on river water quality. Spearman correlation analysis was conducted between the water quality indexes of different water quality stations and the land use structure (area proportion) of the corresponding region from 2015 to 2020. As shown in Table 3, all the water quality indexes showed significant correlation with cultivated land and forest land, among which, there was a significant negative correlation with forest land, and a positive correlation with cultivated land in flood season or non-flood season. In addition, the concentration of CODMn and COD were negatively correlated with water, the concentration of NH3 was negatively correlated with artificial surfaces, in non-flood season. The concentration of TP was negatively correlated with artificial surfaces, and the concentration of anionic surfactants were negatively correlated with other land in flood season or non-flood season. Except TP, the correlation between other indexes and non-flood season was more significant.
Table 3
Spearman rank correlation analysis of water quality and proportion of land use type (2015–2020)
| Period | Cropland | Forest | Grassland | Water | Impervious_ surface | Others |
COD | Non-flooda | .302** | − .353** | 0.037 | − .215* | -0.108 | -0.065 |
Floodb | .226* | − .289** | 0.152 | -0.081 | -0.068 | 0.032 |
Annualc | .292** | − .329** | 0.085 | -0.163 | -0.109 | -0.029 |
CODMn | Non-flood | .339** | − .307** | 0.021 | − .209* | -0.141 | -0.11 |
Flood | .270** | − .273** | 0.095 | -0.096 | -0.124 | -0.042 |
Annual | .325** | − .307** | 0.046 | -0.168 | -0.14 | -0.088 |
BOD | Non-flood | .354** | − .335** | 0.112 | -0.166 | -0.163 | 0.015 |
Flood | .298** | − .272** | 0.177 | 0.007 | -0.123 | 0.032 |
Annual | .321** | − .335** | 0.123 | -0.105 | -0.132 | 0 |
NH3 | Non-flood | .427** | − .202* | 0.132 | -0.124 | − .298** | 0.056 |
Flood | .347** | − .218* | 0.076 | -0.074 | -0.171 | -0.044 |
Annual | .408** | − .209* | 0.135 | -0.081 | − .262** | 0.029 |
TP | Non-flood | .409** | − .238* | 0.108 | -0.094 | − .221* | 0.027 |
Flood | .427** | -0.129 | 0.186 | 0.014 | − .270** | 0.102 |
Annual | .417** | − .211* | 0.141 | -0.058 | − .244* | 0.063 |
TN | Non-flood | .287** | -0.201 | 0.062 | -0.067 | -0.171 | -0.147 |
Flood | 0.202 | − .213* | 0.057 | -0.081 | -0.078 | -0.159 |
Annual | .259* | − .223* | 0.04 | -0.072 | -0.131 | -0.166 |
Anionic surfactant | Non-flood | .248* | -0.19 | -0.069 | -0.071 | -0.113 | − .277** |
Flood | .204* | − .251* | − .266** | -0.182 | -0.052 | − .368** |
Annual | .241* | − .233* | -0.143 | -0.139 | -0.087 | − .345** |
a,bn=112; cn=110. |
**.The correlation was significant when the confidence (double measure) was 0.01. |
*. The correlation was significant when the confidence (double measure) was 0.05. |
Previous studies observed that the similar relationship between land use pattern and water quality as well. Jian et al.(2011)found that the river water quality was negatively correlated with the relative area proportion of farmland all year round and positively correlated with the proportion of woodland area in non-flood season. Bahara and Yamamuro (2008) reported that forest land was negatively correlated with all ions chemistry of the Shimousa Upland in Japan, farmland coverage and residential areas were positively correlated with the major ion chemistry.Forest land was rich in vegetation and the process of vegetation root absorption and soil interception can effectively reduce pollutants (Guo et al. 2020), meanwhile, human activities in this region often weak. Water quality index was positively correlated with cultivated land due to the not fully utilized fertilizer and insecticides together with runoff along with the erosion of rainfall(Wang et al. 2021). On the other hand, a high proportion of agricultural and residential land always lead to more non-point pollutants (Du et al. 2016), which was driven by surface runoff and discharged into the river network (Deng 2019).
3.3 Influence of rainfall on the water quality
Non-point source pollution caused by rainwater has become one of the important reasons for the deterioration of urban water environment. In order to investigate the influence of rainfall on water quality in Beijing, the correlation analysis between monthly mean water quality and rainfall during 2015–2021 and the correlation analysis between daily mean water quality and rainfall in 2021 of ST19 were conducted. Spearman correlation analysis showed that the concentration of CODMn, NH3 and TP were significantly positively correlated with rainfall in flood season, and the concentration of TN was significantly negatively related to rainfall in non-flood season (Table 4). Spearman correlation analysis between rainfall and runoff in Tongxian hydrological station from 1919 to 2020 was analyzed as well, results showed that rainfall and runoff were positively correlated with a correlation coefficient of 0.4300, indicating a significant correlation when the confidence (double measure) was 0.01 (Fig. 7).
Table 4
Spearman correlation analysis between water quality and rainfall of ST19 from 2015 to 2021
| Period | Group | CODMn | NH3 | TP | TN |
Rainfall | Flood | monthly water quality and rainfall in 2015-2021a | 0.016 | 0.495* | 0.444* | 0.188 |
daily water quality and rainfall in 2021b | 0.358** | 0.686** | 0.500** | 0.172 |
Non-flood | monthly water quality and rainfall in 2015-2021c | -0.133 | -0.074 | -0.123 | -0.039 |
daily water quality and rainfall in 2021d | 0.005 | 0.086 | -0.038 | -0.200** |
a n = 26; b n = 54; C n = 121; dn=241. |
**.The correlation was significant when the confidence (double measure) was 0.01. |
*. The correlation was significant when the confidence (double measure) was 0.05. |
In flood season, there were larger rainfall, runoff and water quantity.Increased precipitation and runoff during flood season could result in a greater contribution of sewage-derived organic matter (Xuan et al. 2020) and non-point source pollution loads (Du et al. 2016, Ma et al. 2021, Zeng et al. 2021). In a short period of time, with the increase of precipitation and urban runoff, the concentration of CODMn, NH3, TP increased (as shown in Table 4) in flood seasons, except TN. Previous studies indicated that nitrogen loading was more sensitive to changes in temperature than precipitation (Zhu et al. 2022). Particulate-P and NH4-N were delivered primarily from overland sources and transported by runoff (Mihiranga et al. 2021). On a long time series scale (seasonal), combining other factors such as higher water temperature, longer sunshine duration, stronger aquatic biological activity and faster pollutant degradation rate, the overall water quality in flood season was better than that in non-flood season in the North Canal basin in Beijing (as shown in Table 1).