Introduction

A healthy landscape lake always plays an important role not only in landscape environment but also in the flood control, water supply, and an increase in conservation value by wildlife and wetland systems (Brack 2005). Water quality and ecosystem of landscape lakes are being increasingly affected by nutrient input from anthropogenic sources (Carpenter et al. 1998; Havens et al. 2001). In China, the water quality problems have increased due to rapid growth of industry, population concentration in large cities, transport, and a much greater per capita use of water (Lei et al. 2008). Although most of the major industrial plants installed primary treatment systems after 1990s, the existing treatments are not sufficient to solve the eutrophication problems in lakes. Deprivation of landscape lakes may trigger a negative influence on a city's environment (Yan and Wang 2005). The phenomenon of an algae outbreak, pellucidity decline, water darkening, and emission of foul odors will not only influence the beauty of landscape water but also lead to waste of water resources and breakdown of the ecological environment.

In China, considerable research has been performed on nutrient dynamics as a first step toward pollution control of the landscape lakes. The value of freshwater resources necessitates that they should be managed ecologically for meeting the water quality standards (Karakoç et al. 2003). Throughout the past two decades, the problems of eutrophication and deterioration of water have gained increasing concern, especially in regions where eutrophication is significant (Morkoç et al. 2008). Resources and natural ecological properties become damaged when the load of wastewater exceeds the self-purification capacity of the receiving water environment (Jiang and Shen 2006; Gulati and Van Donk 2002). Eutrophication problems usually happen in polluted lakes and coastal sea (Justić et al. 1987).

In recent years, eutrophication management method has been researched and developed in Chinese lakes and reservoirs. Many studies have been conducted on water quality, trophic state assessment, and ecological risk assessment in Daya Bay, Songhua Lake, Chaohu Lake, and reservoirs in Fujian in China (Sun et al. 2011; Wang et al. 2008; Yu et al. 2011; Yang et al. 2012). Especially in the study of the reservoirs in Fujian, the shallow lakes or reservoirs appear to be more susceptible to anthropogenic disturbance than the deeper ones (Yang et al. 2012). Fulfilling the requirement of the landscape lake's health is difficult not only because of multiple occurring stressors but also because traditional approaches for managing aquatic ecosystems often fail to account for the potential effects of disturbances on the biota. Therefore, combining modeling and eutrophication assessment would be more effective and useful to assess and manage the ecological status. An ecological model, AQUATOX, which was developed by the United States Environmental Protection Agency has been used to simulate and analyze a number of lakes in China, America, and Turkey (Lei et al. 2008; Carleton et al. 2009; Morkoç et al. 2008). Computational fluid dynamics (CFD) models have been constructed for a number of lakes and reservoirs in the UK, Norway, Finland, and North America to study circulation ways (Falconer et al.1991; Shen et al. 1995; Beletsky 1996; Ali and Othman 1997; Güting and Hutter 1998; Olsen and Lysne 2000). These models have also been extended to study phytoplankton and algae (Olsen et al. 2000) and chlorophyll-a (Hedger et al. 2002). Trophic index (TRIX) was proposed by Vollenweider et al. (1998), which has been widely used over the last 10 years.

This paper uses ANSYS Fluent CFD software (Fluent), AQUATOX, and TRIX to model, assess, and manage the landscape lake's water quality and eutrophication state. The aims of this study are to: (1) reflect and predict water system characteristics, (2) assess the water environment state of the landscape lake, and (3) determine the suitable management method by modeling of different scenarios. The result of this study was to realize well reusing and circulating of the water resources. Model results are important to analyze the actual situation and determine remediation requirements in the future.

Materials and methods

Site description

The lake named HM Lake is located at 39°10′7″ N, 117°22′33″ E, in the center of HM Town in Tianjin, and is composed of two landscape lakes which are named HMLA and HMLB in this paper. The relationship between water resources and lakes can be seen in Fig. 1. In the wet season from April to October, HMLA received the effluence of constructed wetlands and landscape wetlands as the point source (PS) pollution, and the runoff water when raining is also an important non-point source (NPS) pollution. In the dry season, there are no direct pollution sources. The main output of the two lakes is evaporation in wet season and dry season.

Fig. 1
figure 1

Location and water system of the study area

Full size image

In this paper, HMLA was used as the study object for modeling. Water quality characteristics of HMLA monitored from January 2010 to October 2010 enabled us to identify the effect of waste loads on water quality. The general features of HMLA are given in Table 1.

Table 1 General features of HMLA
Full size table

Fluent

In the present studies, the CFD modeling applied to surface water reservoirs involves solving the Reynolds-averaged Navier–Stokes equations describing the flux of mass and momentum within a fixed domain being subject to specified boundary conditions (Kennedy et al. 2006). The CFD code (Fluent Incorporated 2006), with the same turbulence model (k–ε) as used by Güting and Hutter (1998), is used in the present study with focus on the fluid and turbulence modeling capabilities.

In this study, the flow field and the locations with low flow velocity could be modeled by the use of Fluent, and for changing the flow status of HMLA, according to the modeling result, some water replenishing locations were confirmed to optimize the initial flow fields. Finally, the optimal water replenishing location would be determined.

AQUATOX

The results of physical, chemical, and biological analyses were used to run AQUATOX. Modeling with AQUATOX is a deterministic process (USEPA 2004). The model uses differential equations to simulate the changing concentrations of the different state variables (Richard et al. 2008; Rashleigh 2003).

In this study, mean weekly data in 2010 were used to run and adjust the model developed by AQUATOX, and then the adjusted model is used to predict the effect of the pollution result in eutrophication in 2011 and 2012. According to the geographical position and spot investigation of the study region, the microorganism community and main aquicolous plants of HMLA's ecosystem were achieved. For determining the optimal management method, three management scenarios were developed to reduce the pollution from point and non-point sources.

The TRIX assessment method

A water quality index provides a convenient means of summarizing complex water quality data and facilitating its communication to a general person. In order to identify the trophic condition of the HMLA, the water quality TRIX index (Vollenweider et al. 1998) was calculated. It is a linear combination of four state variables related to primary production (chlorophyll-a and oxygen) and nutritional condition (dissolved inorganic nitrogen, inorganic phosphorus) (Melaku et al. 2003). This index was calculated as follows:

$$ \mathrm{TRIX}=\frac{{\left[ {\log \left( {\mathrm{chl}\text{-}\mathrm{a}\times \left| {\%\mathrm{DO}} \right|\times \mathrm{DIN}\times P+k} \right)} \right]}}{m} $$
(1)

where P is the mineral or inorganic phosphorus (P–PO4 μmol ll−1), DIN is the mineral nitrogen—dissolved inorganic nitrogen (μmol l−1), chl-a is the chlorophyll-a concentration (μg l−1), and %DO is the absolute value of the oxygen saturation deviation from the oxygen calculated as |100 − %DO|. Parameters k = 1.5 and m = 1.2 are scale coefficients which were included to fix the lower limit value of the index.

Field data

Data were obtained from four sampling points located in HMLA. Sampling points were visited during the period from January 1 to December 31, 2010. Water samples were collected by using polyethylene bottles from depths of 1 and 1.5 m depending on the depth of the points, and samples were collected as close to the discharge points as possible.

Measured parameters in the sample points were physical parameters and biochemistry parameters. The physical parameters included boundary and initial flow velocity and flow flux. The biochemistry parameters included chemical oxygen demands (COD), total phosphorus (TP), total nitrogen (TN), and ammonia nitrogen (NH4–N). Samples were stored in ice boxes and shipped to the laboratory within 2–3 h. Water fluxes of each source were also measured during sampling. The flow rates in sample points were calculated from their cross-sectional areas and velocity of the outflows. Water flux data and concentrations were used to compute pollutant loadings into the lakes.

A thermometer was used to measure the temperature. The pH test paper was used to measure pH value. Dissolved oxygen (DO) was measured by iodometric method (Greenberg et al. 1985). Transparency was measured by the secchi disk. Samples for nutrient analysis (nitrate + nitrite, ammonia) were deep-frozen at −20°C and stored for no longer than 2 days. The COD was determined by dichromate titration method. Total nitrogen was obtained by UV spectrophotometric method–alkaline potassium persulfate digestion method. Ammonia was measured by Nessler's reagent colorimetric method. Total phosphorus was determined by ammonium molybdate spectrophotometric method. Chlorophyll-a concentrations were determined spectrophotometrically after extraction of the filtered samples by acetone (Greenberg et al. 1985).

Weekly mean actual measurement data of the first week in January 2010 of NH4–N, TN, TP, CODCr, SS, flux, water temperature, wind velocity, and pH are adopted as the initial data and background concentration. Weekly mean actual measurement data of the first week in April 2010 of NH4–N, TN, TP, CODCr, and SS are adopted as the inflow loading concentration. The data of contamination boundary conditions and background concentration are shown in Table 2.

Table 2 Boundary values and initial conditions (mg/L)
Full size table

Calibration and assessment

Modeling results of Fluent

Simulation results were first reviewed to understand the general flow behavior indicated by the model. Because the average depth of HMLA is only 1.2 m, it is a shallow lake and mixed quickly and evenly. According to this, a two-dimensional flow field model was established. The first step is constructing the mesh by program Gambit 2.0. Then, in the light of the design data of HMLA, water inlet and outlet were set in the model, and the flow velocity of the inlet was set in the boundary conditions. Lastly, a computation was done by the unstructured grid model and a flow velocity distribution figure was gained. In the figures, different colors correspond to different velocities (v, m/s), so the flow velocities in each part can be shown by the color distribution.

The result shown in Fig. 2 suggested that once these swirling regions become established, they remain in place, absorbing the energy of smaller eddies traveling downstream from the inlets. From the figure, the whole lake is almost motionless and the velocity of each region of the lake is lower than 0.01 m/s, excluding the inlet. There are several reasons for the low velocity. Firstly, some swirling regions are formed because there are several small islands in the front of the inlet and the velocity of the inlet is correspondingly faster. Secondly, some regions are far away from the inlet and outlet and hydraulic disturbance influences little on them. The artificial island of fast flow velocity is near to the inlet, and the nutrients are not easily accumulated. Except for one, the flow velocity is slow around the other two islands, and swirling regions are formed between the islands and the boundaries, so eutrophication easily takes place in those regions. Some effective measures should be adopted on these regions to improve the fluidity and avoid the backwater regions from forming because long-time interruptive hydraulic flow could lead to aggradation of contamination. These regions should be the key regions of water monitoring.

Fig. 2
figure 2

Flow flied of HMLA

Full size image

Modeling results of AQUATOX

Model performance was evaluated by comparing simulation results to HMLA's monitoring data from 2010, both visually and statistically. The visual comparison of observed against simulated NH4–N, TN, TP, and DO was generally good. Predicted chlorophyll-a concentrations were higher than the observed values, and the predicted phytoplankton and zooplankton dynamics were a little higher than the observed type. The chlorophyll-a, diatoms, and greens concentrations from 2010 to 2012 are shown in Fig. 3a, b.

Fig. 3
figure 3

Chlorophyll-a (a) and algal concentration (b) of HMLA from 2010 to 2012

Full size image

The modeling result shows that the algal biomass distinctly changed with different seasons. In the wet season, the chlorophyll-a concentration and the algae concentrations gradually increase, and the chlorophyll-a concentration reaches a peak in August of 2010, 2011, and 2012. For different algae categories, their highest concentration in a year happens in different periods. The diatom grows quickly in the wet season, while the greens grow quickly from July to December. The growing time of Peri High-Diatom is earlier than last year's.

The water discharges caused by anthropogenic activities could produce the rise of algae and lead to eutrophication. In HMLA, there are NH4–N, nitrate (NO3–N), and TN, and their existent forms are nitride in suspended detritus (N Mass Susp. Detritus), nitride in animals (N Mass Animals), and nitride in plants (N Mass Plants). In the shallow landscape lake, the soluble nitride is the main form of TN, and it can support the nutrient requirement for the pelagic algae. The changes of concentration and distribution in the ecosystem are shown in Fig. 4a. The concentration and distribution figure of TP is shown in Fig. 4b. There are four existent forms of phosphide: soluble phosphide, phosphide in suspended detritus (P Mass Susp. Detritus), phosphide in animals (P Mass Animals), and phosphide in plants (P Mass Plants). The absorption of TP by aquatic animals is very little. In considering the modeling results, eutrophication is serious with algal blooms. The high concentrations of nutrients cannot be adequately absorbed by the lake water (Le et al. 2010).

Fig. 4
figure 4

Distribution of nitrides (a) and phosphides (b) in HMLA

Full size image

Eutrophication assessment result

The length of the related trophic scale of TRIX is from two to eight, of which the meaning of the values are shown in Table 3. Results of the TRIX index application of TRIX values calculated from January 2010 to December 2012 of different seasons are shown in Table 4, and the TRIX values of mean value of each period were classified according to the proposed scale (Table 5). Only in the period of the first dry season (Jan–Mar 2010) has the value been characterized as oligotrophic. In the period Dec 2010–Mar 2011, the value has been characterized as mesotrophic. However, the values of other periods have been characterized as eutrophic, except in the period Apr–Aug in 2010 and 2012 when the values have been characterized as hypereutrophic. The assessment result suggested that: (1) in the dry season, the eutrophication state is more acceptable than in wet seasons and (2) the eutrophication state tends to be serious from 2010 to 2012.

Table 3 General ranking for TRIX (Penna et al., 2004)
Full size table
Table 4 Statistical description of parameters and indices
Full size table
Table 5 Classification of eutrophication state symbol in different seasons in HMLA
Full size table

Management scenarios results

An improvement of changing water circulation and three scenarios of changing eutrophication state are developed to reduce the eutrophication risk of HMLA.

Flow field optimization of HMLA

As regard the small landscape lake in the city, improving the state of hydrodynamic flow is a rapid and highly efficient method. On the basis of the modeling results and analysis, the velocity of 0.25 m/s was chosen to optimize the flow field (Imteaz et al. 2003). Two points, A and B, were chosen to supply fresh water into the lake. The flow field figures of supplying for 3 h are shown in Fig. 5, and the modeling results show that the fluidity of the lake with water supplied by point A is better than point B, which can entirely improve the hydrodynamic flow of HMLA. So, supplying water by point A is effective and made to work on dilution and degradation of the contamination in HMLA.

Fig. 5
figure 5

Flow flied of HMLA by a water compensation from A and b water compensation from B

Full size image

Although the velocity could be raised by spring water at the proper point, the water quality of HMLA is not clear. There were still some regions with bad fluidness. There is a serious need for appropriate water quality monitoring for future planning and management of clean water resources (Ghumman 2011). Therefore, the regions with bad fluidness need attention by monitoring and water quality management.

Scenarios

Scenario 1—implementation of ecoremediation

Aquatic vegetation can improve the removal efficiency of nitrogen from water. In the former instance, vegetation lowers nitrogenous loads by direct uptake and incorporation into plant biomass (Hoagland et al. 2001; Silvan et al. 2004). Fish can dramatically change their behavior within short periods of time from foraging to fighting to mating and even to periods of inactivity (Tsanis et al. 1998). From the condition and modeling results of HMLA, a Fontinalis plant of dry weight 1 g/m2 has been put into HMLA since the first dry season and the chub of dry weight 5 g/m2 were used to cut down the concentration of the deleterious algae. The results are shown in Figs. 6 and 7.

Fig. 6
figure 6

Chlorophyll-a comparison (a) and algal concentration (b) of HMLA from 2010 to 2012 by implementing ecoremediation

Full size image
Fig. 7
figure 7

Distribution of nitrides (a) and phosphides (b) by implementing ecoremediation; the nutrient comparison between implementing ecoremediation and control state (c) from 2010 to 2012 in HMLA

Full size image

From Fig. 6, the concentration of chlorophyll-a declined from 2011; in 2012 the concentration is obviously lower than without ecoremediation, and particularly in the wet season of 2012, the concentration of the chlorophyll-a nearly will not lead to eutrophication. For the nutrient, from comparing Fig. 7a, b with Fig. 4, the distribution suggested that the plants and fishes absorbed more nutrients in 2011 and 2012. Figure 7c shows the nutrient concentration of scenario 1 and control state. TN, NO3–N, and TP decreased from Nov 2011, and in the wet season of 2012, they declined obviously. However, NH4–N rose, with the others decreasing. Overall, implementing ecoremediation to improve the eutrophication state is an efficient method; especially in 2012, the effect is more obvious than in 2011, which means that it should be implemented for a long time.

Scenario 2—implementation of BMPs

Best management practices (BMPs) encompass a wide variety of appropriate technologies intended to minimize the effect of runoff. Examples of BMPs that achieve storm flow peak reduction and rainwater storage include detention, retention, and bioretention basins. HMLA, the landscape lake, could be treated as a detention-based storm water basin on the basis of its function, but the initial runoff of rainwater directly entering into the lake would have serious influence to the water quality. However, infiltration approaches reduce runoff volumes and the concentration of the nutrient of the initial runoff; the two methods could be used effectively in combination with each other (Perez-Pedini et al. 2005). As for that, this scenario represents a combination of the two methods by building grassland bank protection around HMLA. According to the study results of some researches and investigation in Tianjin, the removal rate of the nutrient is defined as 35 % (Ackerman and Stein 2008).

The change of concentration of chlorophyll-a is little (as shown in Fig. 8). On one side, due to the discontinuous, small number of infiltration runoff and the lower concentrations of the nutrients and, on the other side, due to the long-term influence of infiltration runoff on the ecosystem, the change of chlorophyll-a is not clear in 2011. However, in the wet season of 2012, the peak concentration of chlorophyll-a declines more distinctly. The difference in nutrient concentration is not obvious too (shown in Fig. 9). However, building grassland bank protection to decrease the runoff's nutrient could play an effect on decreasing the concentration of nutrients. According to the comparison, TP decreases by about 6.5 %, TN decreases about by 2.5 %, NO3–N decreases by about 2.8 %, and NH4–N decreases by about 2.3 % in peak concentrations. Therefore, the effect of using grassland bank protection can improve the water quality of HMLA in most seasons, but there is almost no difference of chlorophyll-a values in 2010 and 2011.

Fig. 8
figure 8

Chlorophyll-a concentration of HMLA from 2010 to 2012 by implementing BMPs

Full size image
Fig. 9
figure 9

Nutrient concentration of HMLA from 2010 to 2012 by implementing BMPs

Full size image

Scenario 3—implementation of ecoremediation and BMPs

This scenario tests the effect of combining the implementations of scenario 1 and scenario 2. From the above researches, the implementation of scenario 1 could decrease the concentration of chlorophyll-a in 2011 and 2012. Scenario 2 has not changed the state of the HLMA directly and obviously, but it declined the risk of ecosystem from the non-point pollution and restores water resource. Hence, this scenario will take advantage of scenarios 1 and 2.

The results show that there was obvious deviation between the control and scenarios for the concentration of chlorophyll-a from the wet season of 2011. According to a comparison of chlorophyll-a between scenarios 1 and 3, its decrease is obviously in 2012 (as shown in Fig. 10). It suggested that the influence of the implementations would emerge after a long time. Otherwise, the deviation of nutrients (shown in Fig. 11) is not as obvious as of chlorophyll-a, but the concentration of TN decreases particularly in the wet seasons of 2011 and 2012. For comparison purposes, Table 6 presents the TRIX results for scenario 3, and it shows the eutrophication state of HMLA in the combined implementations scenario. The calculated TRIX index of scenario 3 suggested that in 2010 the eutrophication state is serious, particularly in the wet season, the state of which is almost the same as the control state. In 2011, the value of the TRIX index decreased slightly. From the dry season of Dec 2011–Mar 2012, the eutrophication states have dropped to oligotrophic. For the wet season of 2012, the control state was hypereutrophic, and it changes to oligotrophic by the implementation of scenario 3. From the whole modeling period, it is suggested that the implementations of ecoremediation and BMPs cannot improve the water quality immediately; nevertheless, they will finally and effectively work in a long time (compare Table 5).

Fig. 10
figure 10

Chlorophyll-a concentration of HMLA from 2010 to 2012 by implementing ecoremediation and BMPs

Full size image
Fig. 11
figure 11

Nutrient concentration of HMLA from 2010 to 2012 by implementing ecoremediation and BMPs

Full size image
Table 6 Classification of eutrophication state symbol in different seasons in HMLA
Full size table

When comparing the TRIX index results to AQUATOX results, it is not surprising to see that the period of index values decreasing is almost the same for chlorophyll-a and the nutrients. Chlorophyll-a makes a significant influence on the TRIX index.

Discussion

According to the comparison and analysis of the three scenarios and the control state, the evaluation of the effects of different implementations has shown some water reuse and management methods on the landscape lake of multiple pollution sources. There are a number of issues and implications that model results demonstrate for controlling and managing the water quality of the landscape lake:

  1. 1.

    Control of the PS and NPS pollution: Firstly, the information collected on the pressure of nutrients and their impacts is used in deciding the environmental objectives for water bodies and in the development of water basin management plans (Santhi et al. 2006). The quantitative aspect helps in the evaluation of the precise needs for pollution control to make a water body meet its environmental objective (Gupta et al. 2012). The PS pollution is the most direct and serious impact on the water quality. As for that, before inflowing into the landscape lake, the water quality must meet its standard. Secondly, NPS pollution has become a serious environmental problem in the aquatic systems throughout the world. NPS pollution is much more complicated as it is driven by multiple factors and includes diffuse pollution (Novotny and Olem 1994). Above all, for directly improving the water quality and aquatic system, the sources should be emphasized, and some tools to analyze problems related to environmental monitoring and planning could help in decision making and support for better management of the water resources.

  2. 2.

    Improving nutrient retention with the implementation of ecoremediation methods: The establishment of environmental risk assessment systems provides sustainable environmental management solutions that contribute to the preservation of biodiversity and pollution reduction, increase the quality of water and soil, and can be applied in protected and sensitive areas (Bulc and Slak 2009). As our study emphasized that the ecoremediation methods can impact the water quality and the distribution of the nutrients, the eutrophication state can be changed to a better state.

  3. 3.

    Water resources reusing: Runoff water is a kind of water which deserves reuse but has other potential impacts of storm water runoff on surface water bodies. In the north of China, the water resources are seriously precious because of the weather; nevertheless, in recent years, in the wet season, the frequency of a rainstorm to happen increases. Thus, the runoff water should be collected well and reused by the water resource planners and managers (Jang et al., 2010). The case study of HMLA according to use of BMPs in scenario 3 shows both initial rainwater treatment and water reuse. Although it has not made obvious impact on the improvement of water quality, it is a good case of runoff reuse and could be extended to other water bodies. Otherwise, the diverse pollutants including heavy metals, hydrocarbons, and pesticides, which were not mentioned in this study, should be researched as other study objectives.

Conclusions

Methodologies are needed to jointly evaluate water quality, hydraulic flow, and water ecosystem to control the situation of the water comprehensively. The combined modeling for evaluating water quality, improving hydraulic flow, and restoring the water ecosystem has been described in this paper. Taking a landscape lake HMLA with multiple pollution sources as the study object, Fluent and AQUATOX as the modeling tools are used. The TRIX index is used in this paper as the evaluation tool of the eutrophication state. Based on the initial state and predicted state of HMLA as the control state, other three scenarios were built to improve the water quality and the eutrophication state, reuse the runoff water, and reduce the impact of runoff water to the water body.

Firstly, the entire procedure–process simulation and restoration shows that the modeling tools could not only ascertain the effects of improving landscape water's quality but also orientate the areas with low velocity. Secondly, the results present that implementation of ecoremediation and BMPs could decrease the eutrophication risk and pollutions in the water body. Lastly, the TRIX value results of control state and scenario 3 suggest that the combined implementation of ecoremediation and BMPs could bring significant eutrophication state improvements in a later period. This evaluation and management method is a strong argument for combining both eutrophication assessment and modeling techniques to a multi-stressor field site, and it could be used in other water regions like reservoir and pint-size lakes to prove and improve.