Introduction

Constructed wetland (CW), which was composed of substrate, aquatic plants, animals and microorganisms, has been recognized as an ecological technology for wastewater treatment (Vymazal 2010). It has been designed and constructed to purify domestic sewage, urban runoff, and industrial effluents worldwide in the last few decades (Vymazal 2011). Generally, the removal of nutrients mainly depends on sedimentation and adsorption by media, plant uptake, and various microbial process such as nitrification and denitrification (Vymazal 2007). Compared to other conventional sewage treatment processes, CW has its advantages of low energy requirement, stable operation, and simple maintenance.

However, the internal flow pattern and hydrodynamic behavior in different areas of CWs were diverse. There were even some stagnant zones, short-circuiting flow zones, and preferential flow areas leading to the decline of removal efficiencies due to the regional differences of plant species and density, as well as flow passage shape in CW (Sun et al. 2016). Therefore, it was of great importance to investigate the flow pattern and hydraulic characteristics in different areas of CW, to improve its contaminant removal efficiencies (Tang et al. 2017).

At present, many researchers focused on the application of velocity measurement, mathematical models, or tracer to simulate hydraulic flow patterns in CW. Velocity measurement technology based on acoustic Doppler velocimeters (ADVs) has revolutionized the measurement of mean velocities, the flow characteristics, and turbulence parameters in laboratory and field researches in the early 1990s (Muste et al. 2007). This technology was firstly adopted in the field of riverine studies in terms of velocity measurement broadly (Nyantekyi-Kwakye et al. 2017). Das and Mazumdar (2015) used the velocity vector measured by ADVs to deduce the contours and spatial distributions of the time-averaged velocities and turbulence intensities. However, direct velocity measurement in field-scale CW was known as a high-cost and labor-consuming job.

On the other hand, several mathematical models, such as Water Quality Analysis Simulation Program, Surface Water Modeling System, and MIKE (Model-based and Incremental Knowledge), have been introduced as tools to obtain a deep understanding of hydraulic characteristics of CW (Qi et al. 2013). In the previous study, the influence of topographic and plant on the formation of short-circuiting flow zones in CW was estimated by the simulation model MIKE 21 (Min and Wise 2009).

Yet the results obtained by the mathematical model could not reflect the hydraulic flow patterns synchronously because its parameters were mostly empirical. This is why tracer method, which has an advantage of ignoring information of fluid properties before measurement (Lin et al. 2003), was currently widely used to determine the hydraulic flow patterns in CW (Guo et al. 2017; Pugliese et al. 2020; Bodin et al. 2012; Plugatyr and Svishchev 2008). The hydraulic behavior includes residence time distribution and the effective volume ratio in three full-scale surface flow constructed wetlands (SFCWs) were investigated by multi-tracer involving uranine and sulforhodamine B. Residence time distribution analysis shows that the three investigated wetlands displayed very different hydrodynamic properties (Laurent et al. 2015; Liu et al. 2020). However, in large-scale CWs, adding tracers was not a good choice because of the high cost and the potential of secondary pollution induced by external additives. More importantly, the tracer method could not provide information on the location of stagnant zones and short-circuiting flow areas inside CW. Hence, an alternative method should be developed to address these prominent problems (Zheng et al. 2011).

In recent years, more and more attention has been paid to the usage of stable isotope, e.g., hydrogen and oxygen isotopes, to predict the hydraulic characteristics of CW (Bugna et al. 2020). Under the liquid condition, the vapor pressure of isotopic molecule is inversely proportional to its molecular weight, which led to the enrichment of heavy isotopes involving 2H and 18O (Ronkanen and Klove 2007). The abundance of heavy water isotopes (δ2H and δ18O values) therefore would be elevated in stagnant or dead zones with a longer hydraulic retention time (HRT). Compared with conventional methods, the stable hydrogen and oxygen isotopes are more cost-effective and environmental friendly, reflect the on-site conditions of the hydraulic characteristics of CWs. In the previous study, Ronkanen and Klove (2007) indicated that the stable isotope method enables an accurate evaluation of the flow patterns in natural peatlands on the basis of 18O/16O distribution. It is commonly known that vapor pressure increased with temperature; however, to date, annual study on using stable isotope in CW is still lacking.

Herein, the main objectives of this study were as follows: (1) to assess the long-term feasibility of the stable isotope method for determining hydraulic flow patterns in CW; (2) to evaluate how important role the inflow and outflow types and aquatic plant played in hydraulic flow patterns of CW; (3) to explore the correlation between hydraulic characteristics and contaminants removal of CW.

Materials and methods

Study area description

In this study, two surface flow CWs, namely Xiaomei river CW and Tuhai river CW, were selected as the research objects. Effluent from local wastewater treatment plants (WWTPs) dominated pollutant loads (N, P) discharged into both CWs.

Xiaomei river CW, with an area of 1.4 ha and the water treatment capacity of 20,000 m3/day, was constructed along the watercourse of Xiaomei river located in Liaocheng (115° 16′ ~ 116° 32′ E, 35° 47′ ~ 37° 02′ N), Shandong Province. Geographically, Liaocheng is dominated by the warm temperate continental monsoon climate, which has an annual mean precipitation of 602.5 mm, an annual mean temperature of 13.1 °C, and an annual mean relative humidity of 62%. Wastewater flows into the CW through water distribution ditch and then discharged into Xiaomei river with the outflow type of unilateral overflow. The surface flow area of Xiaomei river CW, which is planted with Phragmites, Lotos, and Crispus, has an irregular pentagon shape and the designed water depth is 0.5 m.

Tuhai river CW is the combination of subsurface flow CW and surface flow CW, with the water treatment capacity of 30,000 m3/day. Its total area is 2.1 ha and is located in Yucheng of Dezhou (116° 22′ ~ 116° 45′ E, 35° 41′ ~ 37° 12′ N), Shandong Province. The climate of Yucheng is similar to that in Liaocheng, with an annual mean precipitation of 555.5 mm, annual mean temperature of 13.3 °C, and an annual mean relative humidity of 65%. Inlet and outflow of this CW are equipped with over-fall weirs. The surface flow area of Tuhai river CW are planted with emergent vegetation (Zizania aquatica); the average design water depth is 0.7 m and has an irregular shape. The detailed information on hydraulic parameters of both CWs is shown in Table 1.

Table 1 The information on hydraulic parameters in experimental CWs
Full size table

Sampling

We set up three vertical lines and divided each vertical line equally. Then, five sampling points are established uniformly in each vertical line. Each point is spaced approximately 20 m apart (SEPA 2002). Water samples in Xiaomei river CW were collected seasonally in April, July, October, and December of 2017, while the samples of Tuhai river CW were collected in identical months of 2017, respectively. For each sampling point, three parallel water samples were collected 10 cm below water level, under the specific condition that is windless and no precipitation for ten consecutive days or more. Collected water samples were stored in an icebox using 50-mL polyethylene bottles, subsequently transported to the laboratory and kept in a freezer with a controlled temperature of − 4 °C for the following analyses. The software of Google Earth and GPS (629sc, Garmin, Taiwan, China) were employed in this study to determine the detailed geographic location of all sampling points, such as latitude and longitude. The sampling points and wetland shapes are shown in Fig. 1.

Fig. 1
figure 1

Geographical location and sampling map of Xiaomei river CW and Tuhai river CW (arrows represent the direction of water flow; orange dots represent sampling points; red pentagram represents artificial island)

Full size image

Principles and analysis

Among the various elements in nature, the number of atoms is the determining factor in the chemical properties of the various elements. Therefore, there is almost no difference in chemical properties between various isotopes with the same atomic number. However, due to the different number of neutrons, the mass of each isotope is different. The difference in mass causes the isotope fractionation phenomenon (Lécuyer et al. 2013). Phase change is the main factor that causes isotope sharing, such as evaporation of water, condensation of rain, snow, and melting. Stable isotopes in different regions and different water forms have regular changes. The isotope analyses were conducted by Picarro Isotopic Water Analyzer L2140-I, based on wavelength-scanned cavity ring-down spectroscopy (WS-CRDS) technology (Gupta et al. 2009). 2H/1H and 18O/16O ratios of samples were measured with pre-treatment of 0.45-μm high capacity filter (Gooddy et al. 2016). Results of stable isotope analyses were expressed in delta notation (δ) as part per thousand (‰) deviations versus Vienna Standard Mean Ocean Water (VSMOW). Analytical precision was ± 0.100‰ and ± 0.025‰ for δ2H and δ18O, respectively (Sun et al. 2016). Water samples for ammonia nitrogen and total phosphorus concentration analyses were filtered by a 0.22-μm filter following the standard photometric method (APHA 2017).

Data analysis

Based on the Kriging method and Surfer 11.0 software, all data, including isotopic composition and pollutant concentration, were processed by the interpolation method to generate distribution maps visually (Woodard 2000).

Results and discussions

Annual feasibility study of stable isotopes

The annual hydrogen and oxygen isotopic compositions of Xiaomei river CW

The annual variation of stable isotope composition in Xiaomei River CW is shown in Table 2. In each season, the difference in stable isotope composition caused by evaporation was observed, indicating the regional difference of hydraulic flow patterns. The correlation relationship between δD and δ18O in Xiaomei river CW was δD = 6.67 δ18O-5.87 (r = 0.92), with the correlation coefficient of 0.86, as shown in Fig. 2. All stable isotopes dropped to the right side of the eastern China meteoric water line (ECMWL), with the slope and intercept of the fitting line less than that of ECMWL (Yu et al. 1987). This was caused by the strong evaporation in the area where Xiaomei river CW located.

Table 2 Annual stable isotope composition of Xiaomei river CW
Full size table
Fig. 2
figure 2

Relationship between surface water signatures of δD and δ18O for CW relative to the eastern China meteoric water line (ECMWL)

Full size image

In the winter, δD fluctuated between − 61.88 and − 58.21‰ with the amplitude of 3.67‰, while δ18O was ranged from − 8.34 to − 7.89‰ with the amplitude of 0.45‰ in Xiaomei river CW. The average values and amplitude variation were all weaker than other seasons, which was caused by the decrease of air temperature and evaporation in winter. This also explained the decreased amplitude variation of isotopic composition (Kele et al. 2011). And interestingly, the mean δD and δ18O values in summer were − 56.56‰ and − 7.69‰, respectively, similar to that in spring and autumn. Relatively strong evaporation did not lead to a significant increase in average water isotopic compositions, mainly due to the supplement of precipitation during the process of water flow. Dayem et al. (2010) demonstrated that precipitation had a great influence on the isotopic composition of the surface water, and δ18O values even expressed a fatal decline in rainy season.

Water isotope distribution

The oxygen isotope distribution in Xiaomei river CW during different seasons is shown in Fig. 3. Throughout the year, the region with minimum δ18O value was located at the inlet of CWs. As the wastewater flowed through CW, δ18O values raised with the increasing cumulative evaporation. However, the peak value of δ18O appeared inside the CW, rather than the outlet of CW. To better describe the water flow patterns, dark area with high δ18O value was regarded as a stagnant area (SA), while the area with sparse contours was defined as preferential flow area (PFA). The stagnant area is the area where the water flow is stagnant for some reason and the water flow speed is slow. The preferential flow is the area where the water flows preferentially for some reason and the area where the water flow is faster.

Fig. 3
figure 3

The δ18O distribution of Xiaomei river CW in spring (a), summer (b), autumn (c), and winter (d). (Arrows represent the direction of water flow)

Full size image

Figure 3 a showed that SAs (δ18O > − 7.05‰) were located at E 116.05849° N 36.46389° and E 116.05858° N 36.46339° in spring, where heavy isotope has a large degree of enrichment. The west side of CWs was considered as PFAs, in which δ18O values ranged from − 7.82 to − 7.73‰, and the 18O content was closed to the inflow of CW, indicating that evaporation of surface water body in CW was not sufficient and heavy isotope 18O failed to enrich.

In summer, the location of SAs (E 116.05850° N 36.46388° and E 116.05856° N 36.46337°) were well-marked; however, the size of SAs was slightly larger than that in spring. Likewise, the size of PFAs was expanded as well despite weak geographic change. In autumn, the number of SA increased and δ18O values were higher than − 7.35‰. The location of SAs was E 116.05870° N 36.46425°, E 116.05838° N 36.46386°, and E 116.05857° N 36.46339°, which were in agreement with those in spring and summer. This was possibly caused by the withered aquatic plants in autumn. In addition, the contours of δ18O values were relatively uniform in autumn and no distinct PFA was determined.

Compared to other seasons, δ18O values required longer hydraulic retention time to change in winter because of low evaporation. Therefore, it is unqualified to judge whether PFAs present or not in terms of contour distribution of δ18O values, although one SA (E 116.05855° N 36.46339°) could be determined in Fig. 3d. In conclusion, stable oxygen isotope technology could identify the location of SA and PFAs, and determine the hydraulic flow patterns in CWs annually, except for winter.

The influence of design and operation parameters on flow pattern of CWs

To help to optimize the hydraulic characteristics of CW, correlation analysis was conducted between the δ18O distribution and design parameter of different CW.

Effect of inflow and outflow types on flow pattern

Water distribution channel was applied in Xiaomei rive CW to guarantee the good distribution of influent. And it can be seen from Fig. 4, there was no obvious SA or PFA around the inlet area throughout the year, except for autumn. In autumn, SA was observed in the east of the inflow channel, which was attributed to the withered reeds. Differed from the Xiaomei river CW, over-fall weirs were adopted in Tuhai river CW. In spring and summer, SAs were gathered on the west side of the inflow channel, while PFAs occurred on the east side near inflow channel (Fig. 4). In autumn and winter, there was no significant difference in δ18O values that appeared around the over-fall weirs. This phenomenon was attributed to the manual removal of macrophytes in Tuhai river CW after summer. Thus, both inflow types were competent to achieve a uniform and stable water flow, and the SA around the inlet area was likely caused by other factors, i.e., aquatic plants.

Fig. 4
figure 4

The isotopes δ18O distribution of Tuhai river CW in spring (a), summer (b), autumn (c), and winter (d). (Arrows represent the direction of water flow)

Full size image

Outlet ditch which means unilateral overflow was adopted in Xiaomei river CW and was constructed in its southwest corner. The lowest water level on the southwest corner led to a problem that the water flow on the west side had a stronger preferential pathway than that on the east side, resulting in a clear flow stratification. Consequently, the apparent boundary between SAs and PFAs was yielded, and SAs were located on the east side, while PFAs on the west side. The difference in topography was considered an alternative factor for this trend superficially. However, it was no doubt that unilateral overflow aggravated the occurrence of this phenomenon. In Tuhai river CW, over-fall weirs was adopted by the construction of braiding channel. The equal length of flow paths from each cross-section to outlet, due to the lowest water level in the whole outflow cross-section, was beneficial to ensure the stability of outflow in Tuhai river CWs. However, compared with Xiaomei river CW, relatively serious stagnant flow phenomenon was obtained, mainly attributed to the construction of the artificial island in Tuhai river CWs. Thus, over-fall weirs seem to be a better choice than unilateral overflow.

Effect of the plant on flow pattern

Large stretches of macrophytes, either submerged or emergent plants, could increase the flow resistance and change the flow characteristic in CWs. The previous study demonstrated that the flow resistance relied not only on the species of vegetation but also on the density of plant beds (Pham et al. 2011). In the present study, the mixture of submerged plants and emergent plants was introduced into Xiaomei river CW, while Tuhai river CW was dominated by emergent plants (Zizania aquatica).

In Xiaomei river CW, SAs were located at E 116° 3′ 30.6″ W 36° 27′ 50.00″, where reed and Vallisneria asiatica were planted with the density of 277 plants m−2, while the adjacent area was only covered by reed with the density of 74 plants m−2. Either submerged plant or emergent plant could produce flow resistance, although dissimilar resistance intensity appeared with the different plant species in the same flow direction (Aberle and Jarvela 2013). As a consequence, the water flow would have priority to pass the areas with smaller flow resistance; meanwhile, extensive SAs were produced in large resistance areas (Ferguson 2012). Stagnant flow phenomenon faded along with the harvest of reed, as well as the withered of Vallisneria asiatica in winter, which also explained the disappearance of SAs in Fig. 4d.

In addition, plant density was also a major factor influencing flow patterns in CW (Cui et al. 2012; Hua et al. 2014). In Tuhai river CW, stagnant flow appeared in the west of the over-flow weir and the central location of wetland in spring and summer, which were mainly attributed to the difference of plant density. In SAs of Tuhai river CW, plant density of 211 plants m−2 and 174 plants m−2 were observed, separately, which was higher than other areas. The resistance toward water flow increased with the plant density, due to the energy loss caused by the interaction between water flow and plant (Ruggiero et al. 2003). Stagnant flow disappeared as a consequence of Zizania aquatic harvest in autumn. In general, plants played a very important role in hydraulic flow patterns of CW and need careful consideration during its design to prevent negative hydraulic flow patterns.

Annual evaluation of pollutants removal and distribution

Ammonia nitrogen removal and distribution

In Xiaomei river CW, the ammonia nitrogen removal efficiencies were 57.1%, 54.9%, 47.6%, and 18.3% with the influent concentration of 0.28, 0.91, 0.21, and 1.74 mg/L in spring, summer, autumn, and winter, respectively. The high concentration of ammonia nitrogen in CW influent (in other words, effluent of local WWTP) in summer was caused by increasing water consumption and discharge of residents. And in winter, decreased treatment efficiency of WWTP under low temperature was the main reason for the high ammonia nitrogen concertation of its effluent. Low temperature also led to the low ammonia nitrogen removal efficiency (18.3%) of Xiaomei river CW in winter.

Distribution of ammonia nitrogen of Xiaomei river CW in different seasons is illustrated in Fig. 5. It can be seen that the ammonia nitrogen concentration did not decrease along with the flow direction. Instead, some distinct differences at the same transect existed, which was explained by different hydraulic retention time, as well as the differences of plant and microorganism effect in different flow paths (Stottmeister et al. 2003; Toet et al. 2005). In spring and autumn, the ammonia nitrogen concentration in SAs was even less than that in the CW effluent. This distribution property was compatible with those of peatlands in Northern Finland (Ronkanen and Klove 2008). Hydraulic retention time had a significant impact on ammonia removal in CWs, and longer HRT gave arise to maximize nitrification and plant adsorption (Chen et al. 2006). In addition, it was difficult to exchange water between SAs and other areas, which prevented the flow in water with higher ammonia nitrogen concentration (Fan et al. 2016; Stein and Hook 2005).

Fig. 5
figure 5

The ammonia nitrogen distribution of Xiaomei river CW in spring (a), summer (b), autumn (c), and winter (d). (Arrows represent the direction of water flow)

Full size image

Total phosphorus removal and distribution

The Xiaomei river CW showed significant seasonal variations in total phosphorus removal efficiency. Total phosphorus removal efficiencies were 57.1%, 40.2%, and 48.6% in spring, summer, and autumn, respectively, while no phosphorus removal was observed in winter. In the present study, the ability of phosphorus adsorption by substrate reaches saturated since Xiaomei river constructed wetland had been operated for six years, and plant adsorption was considered to be the main pathway of phosphorus removal.

As can be seen from Fig. 6, the peak of TP concentration occurred in the inlet area, similar to that of the ammonia nitrogen. In spring and summer, the minimum phosphorus concentration was found in SAs, due to the optimization of efficiency (vigorous macrophyte and active microorganism) in CWs. In autumn, phosphorus concentration decreased along with flow direction. The SAs located in the eastern of water distribution drain showed highest phosphorus concentration, which was attributed to the phosphorus release by lodging plants and decayed leaves and stems (Cheesman et al. 2010; Doig et al. 2017). The distribution of total phosphorus in winter was quite irregular, compared with the other three seasons. And more interestingly, TP concentration in most areas of CW was higher than that in the influent, which was 0.90 mg/L. And the maximum phosphorus concentration in CW could even reach 1.23 mg/L. This could be explained by phosphate desorption from the substrate of CW (Xiang and Zhou 2011). And the relatively low TP concentration in the SAs was mainly relying on the existence of Potamogenton crispus, which maintains alive in winter and has been reported to have a considerable effect on phosphorus removal (Wang et al. 2010; Zhang et al. 2015). In general, hydraulic characteristics have a different impact on nitrogen and phosphorus in CW, and the optimization of hydraulic flow patterns in CW should consider its impact on contaminant removal.

Fig. 6
figure 6

The TP distribution of Xiaomei river CW in spring (a), summer (b), autumn (c), and winter (d). (Arrows represent the direction of water flow)

Full size image

Conclusions

In the present study, stable hydrogen and oxygen isotope were introduced to explore the hydraulic flow patterns of field-scale CW. The results of the whole year’s measurements showed that stable isotopes of oxygen and hydrogen could be used to determine the hydraulic flow patterns of CW, except for winter when the evaporation was low. Both ammonia nitrogen and total phosphorus distribution were associated with flow patterns. SAs showed better ammonia nitrogen removal performance, while its effect on total phosphorus removal was highly dependent on seasons. Phosphorus release, rather than removal, was observed in winter. Finally, aquatic plants played a significant role in regulating the hydraulic flow pattern of CW and should be well considered during the design of CW.