Innovative adaptation of coagulation-sedimentation-filtration process in lightly polluted urban rivers with seasonal high turbidity

Abstract

Urban rivers, as essential components of ecosystems, have endured severe pollution as a consequence of rapid urbanization and economic development. However, the dynamics of pollution and turbidity in urban rivers within residual red clay alluvial plains remain poorly understood, and there is an urgent need for effective engineering strategies focused on pollution control and turbidity management to improve water quality and restore ecosystem functions. This study endeavored to investigate the spatiotemporal mapping of water quality and corresponding treatment strategy for a slow-flowing urban river with lightly-pollution and seasonal high turbidity within the alluvial plain of southern China. Results from 10 sampling events showed the average concentrations of COD (15.13–22.00 mg/L), NH₃-N (1.15–11.70 mg/L), and TP (0.14–0.26 mg/L), corresponding to Class III–V of China’s environmental quality standards for surface water (EQSSW). Annual water transparency ranged from 25 cm to 30.5 cm, although the average value of SS was only in the range of 6.0 mg/L to 24.5 mg/L. Notably, it presented significant spatiotemporal heterogeneity and frequently exceeded the standard. A total of 28 group field pilot scale in-situ tests of the coagulation-sedimentation-filtration process exhibited a highly satisfactory treatment performance on turbidity and TP. The optimal dosages of polyaluminium chloride (PAC) and polyacrylamide (PAM) were determined to be 50 ppm and 1.5 ppm, respectively. The treatment process achieved exceptional removal efficiencies of 99.53% for turbidity and 94.69% for TP, producing effluent with stabilized turbidity < 1 NTU and TP concentrations as low as 0.017 mg/L, fully compliant with Class II for EQSSW. Furthermore, the system was capable of adapting to flow variations during the rainfall events by adjusting the surface load. These findings are of great significance for the in-depth comprehension of urban river pollution dynamics in the plain area and provide effective scientific and technological support for the remediation of such lightly-polluted river with high turbidity.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-09223-4.

Introduction

Urban rivers play a critical role in urban ecosystems, providing essential services such as water supply, ecological balance maintenance, and landscape enhancement^1,2. However, rapid urbanization and economic development have significantly disrupted these ecosystems, leading to increased pollution from point and non-point sources^3,4, as well as artificial channelization^5,6. These anthropogenic activities have resulted in severe contamination of urban rivers, characterized by elevated levels of turbidity and nutrients^7,8. Among these issues, high turbidity has emerged as a particularly pressing challenge, impairing light penetration, reducing primary productivity, and degrading aquatic habitats^9–11. Consequently, many urban rivers currently exhibit mild pollution coupled with seasonal high turbidity, significantly diminishing their ecological, aesthetic, and recreational value.

The spatiotemporal distribution of generic urban river pollution has been widely studied, revealing pronounced temporal variability and regional heterogeneity^12–14. Temporal variations are primarily driven by seasonal changes in industrial and agricultural activities, as well as hydrological dynamics^15,16, while spatial heterogeneity arises from differences in industrial structures, land use patterns, and geological conditions^2,17. For instance, in highly industrialized regions such as China’s Yangtze River Delta and Pearl River Delta, urban river pollution is predominantly industrial in origin^18,19. While previous studies have provided valuable insights, the pollution dynamics in specific geographical regions, such as lowland alluvial plains with residual red clay soils, remain unclear^20,21. In these regions, the clay soils once transported into water bodies via rainfall runoff, exhibit extremely low natural sedimentation rates and form a suspended colloidal state, leading to the coexistence of persistent turbidity and pollutant accumulation. Therefore, systematic investigation of the spatiotemporal pollution dynamics of alluvial plain urban rivers is necessary to support scientific water quality improvement, with particular emphasis on transparency and turbidity—key parameters directly associated with public satisfaction.

Existing river pollution ecological restoration technologies, such as buffer strips^22,23, constructed floodplains²⁴, constructed wetlands^25,26, phytoremediation²⁷, and bioremediation strategies²⁸, are primarily effective for rivers with stable flows and simple pollution profiles. However, they face significant challenges in addressing complex issues, such as high turbidity coupled with multipollutant contamination and substrate blockage in engineered systems, particularly during high-flow rainy seasons. At the core of these challenges is the presence of suspended solids in water, which not only sustain turbidity but also exacerbate clogging in permeable media (e.g., wetlands or floodplains). Nevertheless, as a classic and cost-efficient technology for raw water treatment with exceptional performance, the coagulation-sedimentation-filtration combined process may serve as a suitable pretreatment solution to address these challenges. The combined process operates by adding coagulants to the water to aggregate small particles into large flocs, thereby removing organic matter, turbidity, colour, heavy metals and phosphorus from the water by gravity precipitation^29–31. As the core of coagulation technology, coagulants and flocculants play a significant role in the coagulation-flocculation process to remove contamination³². The most prevalently utilized coagulants are aluminum- or iron-salt polymers, exemplified by polyaluminum chloride (PAC), and long-chain organic polymers like polyacrylamide (PAM)³³. Recently, numerous studies have developed a range of novel coagulants to cater to water treatment requirements, including hybrid coagulants^33,34, bio-flocculants³⁵, plant-based green coagulants³⁶, etc. However, existing research on coagulation and sedimentation has predominantly been limited to raw water purification in waterworks and conventional wastewater treatments, with minimal attention to in-situ river remediation^37,38. Whether this combined coagulation-precipitation-filtration process is applicable to the treatment of rivers with light pollution and high turbidity under continuous flow conditions and whether both traditional and novel coagulants can achieve satisfactory performance, require further investigation.

In response to the aforementioned challenges, this study aims to: (1) investigate the spatiotemporal characteristics of key water quality parameters in a typical urban river with seasonal high turbidity in residual red clay alluvial plain of southern China; (2) assess the technical feasibility and operational efficiency of the coagulation-sedimentation-filtration process in treating highly turbid and lightly polluted river, particularly its capacity to maintain treatment efficacy during high-flow rainfall events; and (3) compare the performance of traditional and novel coagulants in turbidity reduction and total phosphorus (TP) removal. This study innovatively extends the coagulation-sedimentation-filtration process to in-river treatment, targeting seasonal high turbidity in a red clay alluvial plain urban river, and optimizes operational parameters to achieve high turbidity removal and Class II standards for TP, with results adaptable to rainfall-induced flow variations. Its regional specificity and technological innovation provide a replicable model for turbidity management in similar geologic ecosystems, bridging scientific understanding of turbidity-pollution dynamics and practical water security solutions.

Study region

The Qianhu River (QR), a typical urban lightly-polluted river with high turbidity, is the largest tributary of the Qianhu Lake (QL) basin in Nanchang city, China (Fig. 1A). The Qianhu Lake basin, with a drainage area of 61.8 km², is an important part of the water system pattern of “ten rivers linked one hundred lakes” in Nanchang City, which performs diversified functions of landscape, recreation, entertainment, flood drainage and pollutant transport. With the rapid development of urbanization and economy and the sharply increase of population around the QL, the amount of sewage discharged into the QL has increased from 56,000 t/d in 2012 to more than 157,000 t/d in 2022 (data from https://news.qq.com/rain/a/20220730A00X2U00). The total length of the QR is 7.1 km, with a drainage area of 28.6 km². Its average channel width is 26 m with an extremely small slope of approximately 0.35‰ to 1‰ (Fig. 1A). The rainy season was from March to July while the average annual discharge was 0.741 m³/s, with the maximum runoff of approximately 4.25 million cubic meters/month in June (Fig. 1B). There is a regulating pond (70,000 m²) on the left bank of the QR located at 1.5 km upstream of the QL, which is used to regulate the sediment and water volume (Fig. 1A).

Fig. 1.

Relevant information of study area and river properties: (A) study area location and water system overview, created by QGIS 3.40 L (https://qgis.org/); (B) annual mean hydrologic process curve; and (C) grade curve of particle size of the suspended sediment in the Qianhu River. The map.

The QR basin has two typical features: (1) Water is extremely turbid during rainy season (yellow mud water; Fig. 1A) due to the abundant clay and soil eroded into the river under rapid exploitation and construction around the QR, where landforms are mainly alluvial plain and lake front consist of residual red clay. The suspended clays in the QR are difficult to naturally settle due to the small particle size, ranging from 0.5 μm to 30 μm, and d₉₀ of 6.805 μm (Fig. 1C), which is much smaller than that of the rivers in northern China³⁹; (2) The water pollution appears in QR specially during dry season (Fig. 1A), which was deteriorated by polluted water from the mixed connection and leakage of the urban stormwater and sewer pipe network, as well as by part of the untreated sewage direct discharge. The polluted QR endlessly flows into QL, which greatly destroys the ecological status and landscape function of the QL system and affects the people’s daily life.

Therefore, the local government has carried out a series of engineering measures such as drainage network construction, river dredging, bank ecological protection, water diversion project, and water system connectivity under the China’s major strategy of vigorously promoting comprehensive water environment management and ecological civilization construction in key river basins. Nevertheless, the bottleneck problems of turbidity, low transparency and high nutrient concentration in the QR are still prominent, posing a great risk of eutrophication to the downstream QL. Nanchang Municipal government would ever use the Regulating pond to carry out the treatment project. But the river transparency is still dissatisfactory. Engineering technologies which can directly eliminate water turbidity and pollutants and be at low cost should be explored. Urban rivers situated within alluvial plains were highly susceptible to being polluted by intensive residential and industrial activities, and to being muddied by soil erosion with special physical and chemical properties through direct surface runoff or construction disturbance. The results of this study are of great significance for the treatment of this type of lightly-polluted river with seasonal high turbidity.

Materials and methods

Sampling and water quality monitoring for QR

Ten sampling events (R1-R10) were conducted in the Qianhu River between November 11, 2021 and July 15, 2022 (Table S1). In each sampling event, six water samples named H1-H6 along the river flow direction (Fig. 1A) were collected by a handheld plexiglass surface water sampler (size: φ150 mm, H 200 mm; volume: 3.5 L). During sampling, a boat was employed to reach the sampling points. Water samples were then taken at 0.5 m below the surface using the plexiglass sampler subsequently while the transparency of the river was measured using the Secchi disc16. Each sample was divided into two parts. One part was used for on-site measurements of dissolved oxygen (DO), and the other part was quickly transferred to a 500-mL polyethylene bottle stored in a 4 ℃ incubator and brought back to the laboratory for chemical oxygen demand (COD), ammonia nitrogen (NH₃-N), suspended solid (SS), and total phosphorus (TP) measurement. DO was measured by an Orion StarTm A326 Portable pH/RDO/DO Meter (relative accuracy: ± 0.2 mg/L). COD, NH₃-N, and TP were analyzed according to the potassium dichromate method, Nessler’s reagent spectrophotometer method, and the ultraviolet spectrophotometric method (UV-2450, Shimadzu), respectively. SS was measured by the gravimetric method (drying and weighing). All the measurements were based on the Water and Wastewater Monitoring and Analysis Methods⁴⁰.

In-situ water treatment experiment set-up and procedure for urban river

Field experimental device

To explore the treatment technology for high-turbidity and lightly-polluted rivers, this study proposed a hybrid process combining coagulation-sedimentation-filtration, biological contact oxidation (BCO), and ecological conservation (EC) in the regulating pond (Fig. 2C). This stems from the consideration that abundant suspended impurities (ranging from small to large sizes) are the key determinant of water transparency, while coagulation-sedimentation-filtration represents a widely recognized and efficient approach for addressing suspended and colloidal particulate matter. To this end, a series of field pilot tests of coagulation-sedimentation-filtration with two coagulants (PAC-PAM combined application and ecological flocculant) were conducted near the regulating pond (H3 sampling site) in QR (Fig. 1A) to explore the removal potential of SS and TP, while N treatment was mainly accomplished by BCO and EC, which will be reported in another study. The experimental system was composed of a submersible pump in QR, pipelines, and a processing-flume located on the shore of QR and equipped with two digital electromagnetic flowmeters and a coagulant dosing system (Fig. 2). The concrete processing-flume had a total length of 11.1 m, a width of 0.6 m, and a height of 1.0 m (Fig. 2A and B). It was divided into seven parts: mixing pool, flocculation basin, sedimentation basin, mud bucket, sludge storage pond, filter tank, and clean-water reservoir. Regarding the filter tank, from bottom to top, it was divided into 10 cm thick pebbles (φ ~ 20 mm), 20 cm thick quartz sand (φ 0.5–1.0 mm), 20 cm thick quartz sand (φ 1.0–2.0 mm) and another 20 cm thick quartz sand (φ 2.0–4.0 mm).

Fig. 2.

Schematic and photographic description of the field experiment: (A) diagram of sedimentation test system; (B) section view of the sedimentation flume; (C) aerial view of the sedimentation tank (left and lower) besides the regulation pond treatment zone (right), where the SZ is the sedimentation zone, the BCO is the bio-contact oxidation zone, and the ECP is the ecological conservation pond.

Experimental materials and instruments

The influent water of experiment was all pumped on site from the QR, leading to the small difference of the water quality of each test. The coagulants used in this study include polyaluminum chloride (PAC) and polyacrylamide (PAM) commonly used in waterworks, and new Ecological Flocculant (EF) composed of igneous rock, natural zeolite, ceramics, hydroaluminite, clay, volcanic soil, gypsum, etc. As a water environment treatment product with great potential, EF was composed of natural mineral materials and had almost no impact on the river ecosystem. PAC and PAM were purchased from Xinqi Polymer Co., LTD., Gongyi City, Henan Province, China, while the EF came from Guohe Biotechnology (Yixing) Co., LTD. The purity of PAC is 28% and the molecular weight of PAM is 11 million. The microstructure and elemental composition of flocculants and sediment sludge samples were analyzed by field emission scanning electron microscope of ZEISS Sigma 300.

Experimental conditions

In order to explore the feasibility of coagulation-sedimentation- filtration process for lightly-polluted and turbidity rivers, a total of 28 sets of on-situ experiments were conducted in this study during Sep. 2023 to Apr. 2024, to study the removal effects on water turbidity and TP for different coagulant type, dosage, and surface load, respectively. A control group was set up to illustrate the natural settlement and self-purification ability of high turbidity lightly-polluted river. The effects of PAC-PAM and EF dosages on coagulation-sedimentation-filtration efficiency were investigated in Exp. 1–12 and Exp. 13–19, respectively, and the effects of surface load using PAC-PAM and EF were studied in Exp. 20–23 and Exp. 24–27, respectively (Table 1). Each experimental group was tested in triplicate. The dosage and design parameters of all pilot-scale tests were determined in advance through indoor beaker tests and in accordance with the outdoor water supply design standard (GB50013-2018).

Table 1.

Field experimental conditions.

No.	Inlet flow (m3/h)	Mixing time (s)	Flocculation time (min)	Settling time (h)	Surface load (m3/m2·h)	Filtering velocity (m/h)	PAC dosage (ppm)	PAM dosage (ppm)	EF dosage (ppm)
Control group	1.50	-	30	1.40	0.60	4.17	0	0	0
Exp. 1	1.50	86.40	30	1.40	0.60	4.17	30	1.00	-
Exp. 2								1.50
Exp. 3								2.00
Exp. 4								2.50
Exp. 5	1.50	86.40	30	1.40	0.60	4.17	50	1.00	-
Exp. 6								1.50
Exp. 7								2.00
Exp. 8								2.50
Exp. 9	1.50	86.40	30	1.40	0.60	4.17	100	1.00	-
Exp. 10								1.50
Exp. 11								2.00
Exp. 12								2.50
Exp. 13	1.50	900	-	1.40	0.60	4.17	-	-	30
Exp. 14									50
Exp. 15									80
Exp. 16									100
Exp. 17									125
Exp. 18									150
Exp. 19									200
Exp. 20	2	64.80	22.50	1.05	0.80	5.56	50	1.50	-
Exp. 21	3	43.20	15	0.70	1.20	8.33
Exp. 22	4	32.40	11.25	0.53	1.60	11.11
Exp. 23	5	25.92	9	0.42	2.00	13.89
Exp. 24	2	11.25	-	1.05	0.80	5.56	-	-	50
Exp. 25	3	7.50		0.70	1.20	8.33
Exp.26	4	5.63		0.53	1.60	11.11
Exp. 27	5	4.50		0.42	2.00	13.89

Water monitoring and data analysis

Phosphorus is recognized as one of the key factors in eutrophication control. Turbidity is the key factor to affect the transparency of river water. Therefore, turbidity (Turb) and TP of the inlet and outlet water of the sedimentation basin and the clean-water reservoir were monitored. Turbidity was measured with a turbidimeter (Shanghai, China, Leizhi WZS-185 A model), to avoid the uncertainties from unsuitable light conditions, water depth, and the surveyor’s subjectivity compared with transparency obtained through the Secchi disk¹¹. The determination of TP was the same as the method in section Sampling and water quality monitoring for QR. The figures were graphed by Origin SR1.

Results and discussion

Spatiotemporal distribution of water quality in QR

Contaminations in the QR

The QR water presented a grayish-green color and turbidity, giving off a slight fishy smell (Fig. 1A). The mean COD ranged from 15.13 to 22.00 mg/L with a large heterogeneity within each point (Table S2), which consistent with the scattered point characteristics shown in Fig. 3. In most of the time, COD was less than 30 mg/L (Table S2; Fig. 3B). It was in the Class IV standard limit of China’s Environmental Quality Standards for Surface Water (GB3838-2002) (EQSSW), indicating that the QR was slightly organic pollution. However, the NH₃-N was extremely serious. The average values of 67% monitoring points (4 sampling points) exceeded the Class V limit of EQSSW (2.0 mg/L), with a least concentration of 1.28 mg/L. Analogously, the average values of TP were between 0.14 and 0.26 mg/L (range from Class II (0.1 mg/L) to Class IV (0.3 mg/L)) with large standard deviations, indicating significant time variability. Such a high TP concentration is almost comparable to that in construction-farmland zone¹³. In addition, there were cases of point H1 and point H3, their water quality values exceeded the Class IV. The DO of QR was between 4.70 and 5.23 mg/L, which met the Class IV (3 mg/L) standard.

Fig. 3.

Pollutant concentration at each sampling site of QR: (A) Physicochemical index; (B) Biochemical index.

The concentrations of SS in six sampling sites were relatively small, with average values between 5.33 and 12.00 mg/L, while the transparencies were only between 23.00 and 30.86 cm (Table S2; Fig. 3A). This phenomenon was primarily attributed to the exceedingly small particle size rather than content of SS in the QR. The particle size distribution was bimodal and non-normal, ranging from 0.2 to 34 μm, with standard particle size d₅₀ of 1.031 μm, d₉₀ of 6.805 μm, and d₉₇ of 16.641 μm (Fig. 1C). The SS was mainly composed of fine clay particles, coarse clay particles, fine silt particles, and a small amount of medium silt particles, but dominated by red clay with high mineral content, which resulted in the water being yellow or even reddish-brown.

In addition, another factor that cannot be overlooked was the high proportion of diatom biomass in the QR, with a maximum value of 54.09%. The biomass of diatoms was considerably greater than that of green algae (Fig. S1). Consequently, diatoms emerged as the dominant algae in the QR, suppressing the growth of green algae and exerting an impact on the transparency of the river.

Temporal distribution of water quality in QR

The various water quality indexes essentially displayed significant temporal heterogeneity during the study period, similar to recent studies on P pollution of the surface sediment in southern China⁴¹ and on COD, NH₃-H, and TP in northern China¹⁶. The temporal distributions of transparency, SS, COD, NH₃-N, TP, and DO were respectively shown in Fig. 4. The transparencies of the QR were essentially low throughout a whole year with a weak fluctuation (average value ranged from 25.0 cm to 30.5 cm), only approaching 30 cm in summer from May 30 to June 9, 2022 (Fig. 4A). However, in the corresponding period, the SS content of QR was relatively low (average value ranged from 6 mg/L and 24.5 mg/L; Fig. 4B). Although the maximum concentration of SS occurred on June 20, 2022, it was also much lower than that in high turbidity rivers with tens to hundreds of milligrams per liter of Yellow River in northern China39. The unrelated trend of transparency and SS seems to indicate that SS content was not the main factor affecting the transparency of QR.

Fig. 4.

Temporal distribution of contaminants in QR: (A) Transparency; (B) SS; (C) COD; (D) DO; (E) NH₃-N; and (F) TP. The error bar represents the standard deviation of all the samples in the 6 sampling sites. The horizontal dot dashed lines indicate corresponding values of water quality limits for Class III, Class IV and Class V in EQSSW, respectively.

The COD displayed a high concentration (26–36 mg/L) in autumn and winter (R1 and R2, dry season; Fig. 4C), exceeding the limit of Class IV in EQSSW, which was closely related to the little discharge (0.2–0.3 m³/s) of the QR in winter (Fig. 1B). The total amount of organic contaminant carried by sewage discharged into the QR from sewage plants and combined drainage networks changed little during a whole year, while minimal amount of water in winter led to higher concentrations of COD. This phenomenon is relatively common in China^15,16,42. Subsequently, the COD gradually decreased from May to August in summer ranging from 7.4 mg/L to 24.75 mg/L, basically meeting the requirements of Class III or Class IV in EQSSW (Fig. 4C). This variation pattern is comparable to that of water quality dynamics in northern Fen River Basin15 and southern Yangtze River Basin19 in China. Indeed, the high standard deviation indicated that there are sub-standard points along the river in study reach. On the contrary, the annual variation of DO was relatively small, low in winter (4.5 mg/L) and high in summer (5.4 mg/L; Fig. 4D). Although saturated DO increases as water temperatures becomes low in winter, high algal biomass produces more oxygen in summer.

The concentration of NH₃-N in the QR showed a trend of high levels in winter and low levels in summer, with the highest concentration in late November (11.7 mg/L) and the lowest concentration in mid-July (1.15 mg/L; Fig. 4E). In winter, the NH₃-N was much higher than the Class V of EQSSW (2.0 mg/L), with a maximum value of 5.85 times than the limit of Class V. In summer, it was basically maintained at Class IV standard, but still existed Class V or even inferior Class V. The annual TP changed similar to NH₃-N, with a maximum of 0.40 mg/L reached the limit of Class V at the end of November (Fig. 4F), while it can basically meet Class III in summer except for May 19 and July 11, 2022.

Spatial distribution of contaminants in QR

The sensory quality of the QR was extremely poor, with a relatively milder variability of transparency from upstream to downstream, all less than 30.86 cm (Fig. 5A). The transparency showed a trend of decreasing firstly along the flow, reaching the lowest point at H3 (average of 23 cm), and then gradually increasing. The highest transparency located at the Point H6 principally owing to the low velocity at estuary jacked by the Qianhu Lake, facilitating to the settlement of suspended particles. The SS of QR displayed a trend of firstly increasing and then gradually decreasing along the channel flow, ranging from 5.33 mg/L to 12 mg/L (Fig. 5B), which may be related to the re-suspension of bed sediment generated by the relatively high velocity in large longitudinal slope of the upper reaches than downstream channel^16,43. Nevertheless, the downstream channel of H3 section had gentle slope and small velocity.

Fig. 5.

Spatial distributions and variations of water quality indexes in the QR. The horizontal dashed lines indicate corresponding values of water quality limits Class III, Class IV and Class V in EQSSW.

Analogously, the little changes in average COD of QR along the flow were basically maintained at the Class III of EQSSW, except for the poor water quality of point H5 with a mean concentration of 22.00 mg/L. However, the large standard deviation in time scale indicated that the COD not only fluctuated greatly over a whole year but also exceeded the Class III standard (Fig. 5C). The relatively small spatial variability of COD reflected two problems: Firstly, there are persistent sewage discharged into the QR (especially near H5 reach); Secondly, the water self-purification ability was poor with low degradation coefficient, owing to the extremely low longitudinal slope of the QR. On the contrary, the DO along the flow of QR presented substantially stable tendency, which maintained at the Class III of EQSSW, with average values ranging from 4.70 mg/L to 5.48 mg/L (Fig. 5D). While COD provides a comprehensive measure of organic pollutant load, future studies could incorporate BOD analysis to differentiate biodegradable vs. refractory organic matter.

Quite differently, NH₃-N exceeded the limit value of Class V of EQSSW (2.0 mg/L), and belonged to the category of poor Class V. The lowest NH₃-N in Site H2 belonged to Class IV of EQSSW (1.5 mg/L), with an average concentration of 1.28 mg/L and a low standard deviation of 0.44 mg/L (Fig. 5E). From Site H2 to Site H3, the NH₃-N increased significantly as the effluent from the sewage treatment plant entered the Site H3 of QR, which was also reflected in the spatial distribution characteristics of TP with a mean concentration of 0.26 mg/L (Fig. 5F). This phenomenon may be attributed to the discharge from a sewage treatment plant, where substantial volumes of treated effluent continuously enter the water system, significantly impacting local water quality parameters. This finding is consistent with what was presented in a previous study regarding the water quality distribution downstream of a sewage treatment plant⁴⁴. Fortunately, the other sampling sites were polluted relatively moderate and were remained within the limit of Class III of EQSSW (0.2 mg/L).

In-situ treatment of QR water via flocculation-sedimentation-filtration

Effect of coagulants dosage on turbidity removal

The “flocculation-sedimentation-filtration” process exhibited outstanding effects on the treatment of high turbidity water. The sedimentation stage with the promotion of flocculant can remove majority SS from the water, reducing the pollution load for the ensuing quartz sand filter. Subsequently, the effluent from the filter was crystal clear (Fig. S2).

Under natural sedimentation and filtration condition (Control group), the sedimentation efficiency (ƞ_Turb−Se) and filtration efficiency (ƞ_Turb−Fi) were merely 46.98% and 60.82%, respectively (Fig. 6D). The addition of PAC-PAM and EF significantly enhanced the turbidity removal (Fig. 6), which was mainly due to their high positive surface charge density, long molecular chain, or unique natural-based components, promoting the aggregation of particles^29,34. As the dosage of PAM increased, the ƞ_Turb−Se and ƞ_Turb−Fi gradually increased, except for PAC = 30 ppm (parts per million; Fig. 6A, B, and C). When the dosage of PAC was 50 and 100 ppm (Exp. 5–12), the turbidity of the sedimentation effluent (Turb_Se) was remarkably low, ranging from 4.23 to 8.43 NTU, and the corresponding ƞ_Turb−Se was in the range of 64.07 to 77.04%. Moreover, the turbidity of filter effluent (Turb_Fi) could be as low as 0.06–0.49 NTU, while the ƞ_Turb−Fi reached 93.32-98.77%. The total turbidity removal efficiency (ƞ_Turb−To) was in the range of 97.85–99.69% (Fig. 6B and C). Even at the smallest dose of 30 ppm, after the sedimentation and filtration processes, the ƞ_Turb−To ranged from 96.64 to 98.84%, while the effluent turbidity could be reduced to 0.2–0.35 NTU. However, considering the removal efficiency and economic benefit, the best combination was 1.5 ppm PAM to jointly match 50 ppm PAC, whose ƞ_Turb−to and Turb_Fi reached 99.53% and 0.15 NTU, respectively, which was better than the constructed wetlands for polluted river treatment^45,46.

Fig. 6.

Removal effects of different doses of two flocculants on turbidity: (A) PAC = 30 ppm; (B) PAC = 50 ppm; (C) PAC = 100 ppm; (D) The use of EF. The Turb represents turbidity, and the subscripts In, Se and Fi represent influent water, sedimentation effluent and filter effluent, respectively. The ƞ represents treatment efficiency, and subscripts Turb-Se, Turb-Fi, and Turb-To represent sedimentation, filtration, and total efficiency. The error bar represents the standard deviation.

Compared with the PAC and PAM, although the ecological flocculants (EF) was composed of natural mineral materials and had almost no impact on the river ecosystem, EF had a relatively weak effect on the removal of turbidity, mainly reflected in the filtration process (Fig. 6). When the EF dosage increased from 30 ppm to 200 ppm, the removal efficiency of the sedimentation increased significantly (maximum up to 83.7%) and the residual turbidity gradually decreased (the lowest was 1.82 NTU) (Fig. 6D). This turbidity removal efficiency is slightly lower than that of composite inorganic polymer flocculants with removal efficiencies ranging from 85 to 96% (such as polysilicate-ferroaluminum-sulfate⁴⁷ and polyaluminum ferric chloride⁴⁸). This discrepancy may be related to the differences in coagulant dosage, the experimental scale, and influent water quality parameters, etc. However, some of the natural minerals in EF were feebly soluble in water under low temperature, resulting in viscous substances that slightly blocked the filter material in the filter. Therefore, the turbidity removal of the filtration was not satisfactory, ranging from 44 to 72% (Fig. 6D). In general, when EF was added to the water, the sedimentation and filtration had a certain effect on the removal of turbidity in the water, with the ƞ_Turb−To ranging from 84.31 to 95.31%, the Turb_Fi ranging from 0.57 NTU to 1.64 NTU. It was worth mentioning that when EF was used, the backwashing frequency of the filter increased significantly. Based on a comparison of the results from Exp. 1–19, joint use of PAC and PAM with dosage of PAC = 50 ppm and PAM = 1.5 ppm seems a reasonable option. EF is not suitable due to its inherent nature in spite of effectiveness as a new type of flocculant. In addition, the addition of the two flocculants did not introduce other harmful elements or substances, thereby avoiding potential impacts on the aquatic ecosystem (Fig. S3 and S4).

Effect of coagulants dosage on TP removal

Phosphorus is a limiting factor in the process of controlling eutrophication in water bodies. Therefore, P removal in the flocculation-sedimentation-filtration process is vital to verify whether this process can produce water quality that is good enough to alleviate the eutrophication problem of downstream lake. Figure 7 illustrates the details of each water sample under conditions of Exp. 1–19 and control group from Table 1.

Fig. 7.

Removal effects of different doses of two flocculants on TP: (A) PAC = 30 ppm; (B) PAC = 50 ppm; (C) PAC = 100 ppm; (D) The use of EF. The red, blue, rose, and blank dashed lines respectively represent the TP limits of Class II (0.1 mg/L), Class III (0.2 mg/L), Class IV (0.3 mg/L), and Class V (0.4 mg/L). The FB represents filter backwash.

In general, the flocculation-sedimentation-filtration process exhibited a high level of efficiency in P removal similar to turbidity. The TP content in the final effluent using PAC-PAM was capable of fulfilling the level of Class II in EQSSW, except for three conditions of Exp. 1, Exp. 9, and control group (Fig. 7). Under the condition of Exp. 1, the flocs probably cannot fully adsorb and entrap P in suspended particles and dissolved P in water due to the small dosage of flocculant. In the case of Exp. 9, the main reason was that the small doses of PAM did not match well with PAC. In most cases, the sedimentation and filtration efficiency of TP increased with the rise of PAM at a certain PAC dosage (Fig. 7A, B, and C), consistent with recent studies⁴⁹⁵⁰,. Nevertheless, excessive PAM dosage may not match well with PAC, such as the reducing ƞ_TP−Se in Exp. 4 (Fig. 7A). The sedimentation and filtration processes exhibited a favorable complementary effect. In cases where the sedimentation efficiency was relatively low, the filtration process can effectively guarantee the water quality of the effluent. Consequently, the overall quality of the effluent remained stable, and the best TP removal efficiency (ƞ_TP−To) can attain a value of 95%, with an extremely low concentration of 0.014 mg/L (Fig. 7A, B, and C), whose performance was much higher than a hybrid constructed wetlands in China⁵¹. Of course, the treatment efficiency did not improve significantly when the PAC dosage was increased from 50 ppm to 100 ppm (Fig. 7B and C), which may indicate the optimal dosage of PAC = 50 ppm and PAM = 1.5 ppm.

The sedimentation efficiency of EF can be comparable to that of PAC (Fig. 7). But it exerted a more substantial impact on the subsequent filtration process. The ƞ_TP−Se experienced an increment from 20.12 to 45.66% upon the addition of 30 ppm of EF compared with the Control Group (Fig. 7D). Subsequently, it gradually ascended to 75.24% with the continuous augmentation of the EF dosage. However, the ƞ_TP−Fi merely fluctuated within the range of 17.36–43.75%. A notable feature is that the filtration efficiency is significantly improved after three filter backwashes. After the two sequential stages of flocculation, sedimentation, and filtration, the TP concentration in the effluent from the clean-water reservoir predominantly attained the quality of Class II, with TP values spanned from 0.045 mg/L to 0.124 mg/L, and the ƞ_TP−To varied from 55.09 to 86.06%.

Performance evaluation of the treatment under various surface load

To assess the ability of coagulation-sedimentation-filtration process to cope with large discharge during rainy season or rainfall event, detailed performance of this process under various surface load (SL) is illustrated in Fig. 8. In the experiment of PAC-PAM, the ƞ_Turb−Se decreased significantly as the SL increased, with a lowest value of 18.55% under SL = 2 m³/m²·h (Fig. 8A). Nevertheless, the turbidity of filter effluent could still be as low as 0.15 ~ 0.28 NTU, and the corresponding removal efficiency could be stable at about 98%. Accordingly, during the implementation of the experiment, the greater the SL was, the faster the filtration rate was, and the more frequently the backwashing of the filter tank was required. In contrast, when EF was employed as a coagulant, an increase in SL did not exert a significant impact on the sedimentation effect. This can be attributed to the relatively large dead weight of natural mineral flocculants and the faster blooming and settling rate of SS formed in water. Especially when the influent turbidity was 101 NTU, the turbidity of the effluent from the sedimentation tank could be reduced to 7.65 NTU, with the corresponding sedimentation efficiency reaching 90% (Fig. 8B). In general, the coagulation-sedimentation-filtration system can still cope with high turbidity of incoming water at large SL, achieving satisfying treatment performance. During the rainfall event, appropriately increasing the surface load of sedimentation and filter tanks can accommodate higher influent flow rates; following the rainy season, intensifying backwashing is sufficient to ensure normal system operation (Fig. 7D).

Fig. 8.

Removal effects of different surface load: (A) Turbidity using PAC and PAM; (B) Turbidity using EF; (C) TP using PAC and PAM; (D) TP using by EF.

Similar to turbidity, the increase of SL also reduced the removal efficiency of TP. When PAC-PAM was used as coagulant and SL = 0.6 m³/m²·h, the sedimentation efficiency (ƞ_TP−Se), filtration efficiency (ƞ_TP−Fi), and the total efficiency (ƞ_TP−To) of TP were 70.31%, 82.11% and 94.69%, respectively (Fig. 8C). At this time, TP of sedimentation effluent (0.095 mg/L) can reach Class III water in EQSSW, while that of filtration effluent is tremendously low (0.017 mg/L). With the increase of SL, the removal efficiencies of sedimentation and filtration decreased significantly, and the lowest ƞ_TP−Se and ƞ_TP−Fi were 35.52% and⁵⁴0.19%, respectively. Fortunately, the TP of the final effluent consistently remained close to the Class III water limit. In contrast, when EF was used as coagulant, the increase of SL had relatively little effect on the removal of TP. The ƞ_TP−Se, ƞ_TP−Fi and ƞ_TP−To were 51.90%, 50.88% and 76.37% in the condition of SL = 0.6 m³/m²·h, respectively (Fig. 8D). When SL increased from 0.6 to 1.6 m³/m²·h, the ƞ_TP−Se did not decrease but slightly increased, which may be related to the increase of influent concentration. In general, PAC-PAM was more sensitive to SL of sedimentation process, while EF was more sensitive to SL of filtration process. Incorporating the two treatment processes, the overall ƞ_TP−To can consistently reach 70–75%.

Performance comparison of two flocculants

To provide dependable support for engineering applications, the treatment performances of the two coagulants on turbidity and TP in sedimentation and filtration stage were comprehensively analyzed in Fig. 9. The data of turbidity removal efficiency were divided into 4 Groups (Fig. 9A). Group 1 and Group 2 exhibited extremely high ƞ_Turb−Fi (90-99%), while Group 3 and Group 4 demonstrated high ƞ_Turb−Se (70-90%). The optimal ƞ_Turb−Se and ƞ_Turb−Fi were 77% and 99% (Exp. 12, Table 1), respectively, comparable to those of electroflocculation⁵² and hybrid coagulants²⁹. Significantly, the coagulation-sedimentation-filtration technology employed in this study offered simplicity in operation and stable treatment efficiency. This made it well-suited for large-scale river treatment projects. Even under conditions with larger SL, Group 2 and Group 4 exhibited higher filtration and sedimentation efficiencies compared to the control group (ƞ_Turb−Se = 47%, ƞ_Turb−Fi = 61%). Thus, it was evident that the turbidity processing performance was enhanced via the addition of coagulants, particularly in Group 1 and Group 3. Overall, the turbidity removal efficiency of PAC-PAM was superior to that of EF.

Fig. 9.

Comparison of the water treatment effects of the two coagulants: (A) turbidity; (B) TP. CG represented the Control Group. Above the diagonal, filtration exhibited a better effect than sedimentation; below the diagonal, sedimentation exhibited a better effect than filtration.

Similarly, coagulation-sedimentation-filtration displayed satisfactory treatment performance for TP. The treatment efficiencies can be divided into five groups, namely Group 5 - Group 9 (Fig. 9B). Group 5 exhibited excellent sedimentation and filtration efficiencies on TP, with values around 70% and 80% respectively. In this study, the treatment efficiencies of Group 5 were consistent with those of biofilters packed with ceramsite⁵³. Moreover, they were significantly higher than the floating biofilm reactors (9.38%)⁵⁴ and hybrid constructed wetlands⁵¹. Compared with EF, PAC-PAM demonstrated superior filtration performance. To ensure optimal performance of EF, it is crucial to timely backwash the filter material and address any potential blockage issues.

Conclusion

This study investigated the water quality spatiotemporal distribution of urban lightly-polluted river with high turbidity in alluvial plain with residual red clay. Furthermore, it comprehensively evaluated the performance of the most reliable and economical coagulation-sedimentation-filtration process in treating turbidity and TP in such river, especially under the variable influent loads. The QR has been contaminated due to intensive residential and industrial activities. Notably, the pollution status was more severe in autumn and early winter compared to summer, presented significant spatial heterogeneity. The average water quality approximately ranged from Class III to IV of EQSSW, belonging to the state of lightly-pollution while extremely high turbidity especially in the rainy season. The annual water transparency varied from 25 cm to 30.5 cm, although the average value of SS was only in the range of 6.0 mg/L to 24.5 mg/L. The factors affecting transparency were not only SS, but also the high content of diatoms in water. The pilot-scale in-situ coagulation-sedimentation-filtration process has demonstrated excellent treatment effects on turbidity and TP in this type of river. After treatment, the turbidity of the water can be stably reduced to below 1 NTU, and the TP can steadily meet the target level of Class II of EQSSW. The optimal dosages of PAC and PAM were determined to be 50 ppm and 1.5 ppm, respectively. As a novel coagulant composed of natural minerals, EF exhibited considerable application potential, despite its performance was marginally inferior to that of PAC combined with PAM. Elevating the surface load during the rainy period enables the system to handle higher influent flow rates, while enhanced backwashing post-rainy maintains operational stability. These results are of great significance for understanding the pollution dynamics of urban river network in plain area, and provide important support for the application of coagulation-sedimentation-filtration technology in river pollution control projects.

Future research should investigate chemical modification of natural coagulants (e.g., EF) and eco-friendly composite blends to enhance flocculation efficiency for high-turbidity waters, and prioritize long-term ecological monitoring of coagulant residues and integration of microbial dynamics (e.g., diatoms) with water quality to optimize sustainable operational parameters in alluvial plain river systems.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Y.Y. conceptualized the study, performed validation and investigation, and wrote the original draft; T.Z. conducted investigation and methodology development, reviewed/edited the manuscript, and acquired funding; Y.Z. developed conceptual framework, supervised research, and reviewed/edited the manuscript; Y.L. supervised the project and coordinated administration; S.H. conducted experimental investigations and secured funding; Y.C. and YB.L. performed formal data analysis and manuscript editing; S.W. curated datasets and contributed to manuscript review/editing.

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.