You are on page 1of 10

(This is a sample cover image for this issue. The actual cover is not yet available at this time.

This article appeared in a journal published by Elsevier. The attached


copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Environmental Pollution 177 (2013) 13e21

Contents lists available at SciVerse ScienceDirect

Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol

A three-dimensional water quality modeling approach for exploring the


eutrophication responses to load reduction scenarios in Lake Yilong (China)
Lei Zhao a, d, Yuzhao Li b, Rui Zou c, *, Bin He d, Xiang Zhu d, Yong Liu b, *, Junsong Wang d, Yongguan Zhu a
a
Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China
b
College of Environmental Science and Engineering, Key Laboratory of Water and Sediment Sciences (MOE), Peking University, Beijing 100871, China
c
Tetra Tech, Inc., 10306 Eaton Place, Ste 340, Fairfax, VA 22030, USA
d
Yunnan Key Laboratory of Pollution Process and Management of Plateau Lake-Watershed, Kunming 650034, China

a r t i c l e i n f o a b s t r a c t

Article history: Lake Yilong in Southwestern China has been under serious eutrophication threat during the past de-
Received 22 May 2012 cades; however, the lake water remained clear until sudden sharp increase in Chlorophyll a (Chl a) and
Received in revised form turbidity in 2009 without apparent change in external loading levels. To investigate the causes as well as
16 January 2013
examining the underlying mechanism, a three-dimensional hydrodynamic and water quality model was
Accepted 22 January 2013
developed, simulating the flow circulation, pollutant fate and transport, and the interactions between
nutrients, phytoplankton and macrophytes. The calibrated and validated model was used to conduct
Keywords:
three sets of scenarios for understanding the water quality responses to various load reduction in-
Hydrodynamic and water quality model
Lake Yilong
tensities and ecological restoration measures. The results showed that (a) even if the nutrient loads is
Water quality reduced by as much as 77%, the Chl a concentration decreased only by 50%; and (b) aquatic vegetation
Scenario analysis has strong interaction with phytoplankton, therefore requiring combined watershed and in-lake man-
Eutrophication agement for lake restoration.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction only reflects whether water quality could reach the targeted state
under certain assumptions, but also represents several ‘futures’ or
Eutrophication has been the primary problem facing most sur- different points of view simultaneously (Höjer et al., 2008).
face waters worldwide (Smith, 1983, 2009; Martins et al., 2008; Wu Employing scenario analysis could also evaluate all aspects of the
and Xu, 2011). Although the intricate mechanism for eutrophica- local decision making processes and it was thus widely used as an
tion has not been determined because of the nonlinear response of effective decision making support tool on water quality restoration
water quality to nutrient loading, it is certain that nutrient loads (Arnold et al., 1994; Miller et al., 2002). The integration of scenario
reduction is essential for water quality improvement and ecological analysis with the WQM has been used in a variety of studies
restoration (Rosenberga and Lars-Ove, 1988; Vieira and Lijklema, covering different areas of focus. Some scenarios were analyzed to
1989; Charpa, 1997; Liu et al., 2008b). Mechanistic models can investigate whether future agricultural land use would cause
reflect quantitative response relationships between load reduction deterioration of water quality (Rounsevell et al., 2005; Alcamo,
and water quality as well as be able to conduct scenario analyses for 2001; Ewert et al., 2005). For example, Kepner et al. (2004)
decision making. Water quality modeling (WQM) has therefore defined future land-use scenarios to demonstrate how modeling
been well developed in the past decades and successfully served as tools could be used to evaluate the spatial impacts of urban growth
scientific tools supporting decision makings for controlling the patterns on surface water hydrology. Similarly, scenario analysis
exogenous or ingenuous pollutions (Ahmad et al., 2001; Zou et al., was also employed to investigate other factors impacting water
2006, 2009; Liu et al., 2008a; Purandara et al., 2012). The WQM quality. In the water quality management for Lake Plastira, a
with scenario analysis is considered to effectively support strategic mathematical eutrophication-dissolved oxygen model was used to
decision making (Schoemaker, 1982; Wack, 1985). Compared with examine three water management scenarios based on alternative
other assessment frameworks, WQM-based scenario analysis not minimal operating water levels (Andreadakis et al., 2003; Kepner,
2004). The WQM-based scenario analysis was also used for
exploring the effects of climate change. A simple linear regression
* Corresponding authors. between air and water temperature is used to generate the sce-
E-mail addresses: rz5q@yahoo.com (R. Zou), yongliu@pku.edu.cn (Y. Liu). narios for river water temperature, and the results suggest that all

0269-7491/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.envpol.2013.01.047
Author's personal copy

14 L. Zhao et al. / Environmental Pollution 177 (2013) 13e21

the hypothetical climate change scenarios would cause water EFDC was customized to represent the main water quality and
quality impairment. It has shown that there is a significant decrease ecological processes in Lake Yilong, enabling the predictive analysis
in DO levels due to the impact climate change has on temperature of multiple scenarios so as to provide possible explanations for the
and flow rates (Rehana and Mujumdar, 2009; Park and Lee, 2002). water deterioration as well as to establish a quantitative linkage
Lake Yilong is one of the nine largest plateau lakes in Yunnan between external nutrient loading and in-lake water quality.
Province, Southwestern China. The lake water quality deteriorated To produce more scientific support for water quality improve-
sharply in 2009 and a regime shift occurred from the macrophyte- ment management, three EFDC-based scenarios were designed and
dominated state to the phytoplankton-dominated state (Zhang analyzed in this study, including (a) response of algal blooms to
et al., 2010). Algal blooms occurred frequently since 2009 and aquatic vegetation existence in the lake, (b) the variation of load
caused great economic losses. To better understand the cause of the reduction requirement corresponding to different water quality
shift as well as facilitating effective decision making on eutrophi- standards, and (c) the eutrophication status of the lake corre-
cation control and ecological restoration in Lake Yilong, it is sponding to different water quality standard compliances.
necessary to establish a quantitative cause-and-effect relationship
between anthropogenic interference and lake response through 2. Modeling framework
mechanistic mathematical modeling. Considering the necessity of
resolving spatially variable hydrodynamics and complex water 2.1. Study area
quality and phytoplankton/macrophyte interactions, the Environ-
mental Fluids Dynamics Code (EFDC) was selected as the compu- Lake Yilong is one of the largest plateau lakes in Yunnan Province
tational platform for developing the model. EFDC is a widely (Fig. 1), with an average water surface elevation of 1414 m, an
applied 3-dimensional (3D) hydrodynamic and water quality average lake depth of 3.9 m, and a maximum depth of 5.7 m. The lake
simulation framework capable of simulating water circulation, area is 28.4 km2 and the volume can reach 114.9 million m3. There
temperature dynamics, and advanced eutrophication processes are seven main tributaries flowing into the lake. The watershed
involving nutrients, phytoplankton, macrophyte, and predation/ areas of the seven tributaries account for more than 70% of the entire
grazing processes. It has been applied for simulation and decision basin of Lake Yilong. Historical water quality data from 1998 to 2009
support analysis of surface water such as lakes, reservoirs, bays, and previous studies (i.e. the Lake Yilong Total Maximum Daily Load,
wetlands and estuaries (Li et al., 2011; Peng et al., 2011; Seo et al., YLTMDL) showed that the aquatic system in Lake Yilong experi-
2010; Shi et al., 2011; Wu and Xu, 2011). In the present study, the enced significant inter-annual fluctuations, while a clear signal of

Fig. 1. Lake Yilong watershed.


Author's personal copy

L. Zhao et al. / Environmental Pollution 177 (2013) 13e21 15

regime shift is identifiable for 2009. In 2009, the previously flour- that could reflect complex interactions between nutrients and
ishing macrophyte communities in the lake had nearly been elimi- phytoplankton, aquatic plants, DO and sediment, etc. It thereby
nated and a sharp increase in nutrient and phytoplankton requires a construct with maximum complexity in EFDC. Specif-
concentration occurred. According to the newly issued 12th Five- ically, the EFDC model for Lake Yilong should include all relevant
Year Plan for Water Pollution Control of Lake Yilong Watershed, over water quality drivers, such as carbon, nitrogen, phosphorus, algae
1 billion Chinese Yuan (approximately US$160 millions) are ex- and DO to fully characterize the process of eutrophication.
pected to be invested in the control of lake eutrophication from 2011 Furthermore, the interactions between the vascular aquatic plants
to 2015. To provide scientific basis for the decision making on and phytoplankton nutrients were also analyzed.
implementing the lake eutrophication control, it is necessary to The sediment diagenesis model can be used to predict changes
acquire a quantitative understanding between the watershed nu- in sediment nutrient flux historically or under certain scenarios,
trients load reduction and the lake water quality goals. Thus an such as watershed management and restoration. It can also char-
EFDC-based eutrophication modeling system was developed in this acterize the response of sediment oxygen (SOD) to watershed
study for Lake Yilong. The Shiping County Environmental Moni- nutrient load changes. A model with the ability to predict
toring Center is responsible for monthly water quality monitoring. sedimentewater interactions can overcome the key limitations of
Data were collected from three observation stations. Water samples non-predictive water quality models in analyzing the long term
were collected twice per month (Supporting Materials). response in water quality to watershed and in-lake management,
hence supporting more reliable watershed management decision-
2.2. EFDC model makings.
The following text describes the processes of developing the
The EFDC is a widely recognized simulation platform and a Lake Yilong model, including grid generation, configuration of
multi-task, highly integrated modular computational environ- initial and boundary conditions, model calibration and scenario
mental fluid dynamics package. The water quality model has 21 analysis.
state variables, and simulates the spatial and temporal distribution
of water quality parameters including dissolved oxygen (DO), (1) Grid generation
phytoplankton, various components of carbon, nitrogen, phos-
phorus and silica cycles, and fecal coliform bacteria (Hamrick and A curvilinear grid was developed to discretize Lake Yilong. The
Wu, 1997). EFDC also has a sediment diagenesis module which lake was represented by using a total of 241 horizontal grids where
simulates the diagenesis processes of organic matters in the bed the smallest grid was approximately 0.05 km2 and the largest was
and the resulting fluxes of inorganic substances and sediment ox- approximately 0.18 km2. The average depth at the shallowest grid
ygen demand back to the water column (Park et al., 2005). The was about 1.5 m while the average depth of the deepest grid was
hydrodynamic module of the model solves three-dimensional, about 5.3 m at the water surface elevation of 1414 m above sea
vertically hydrostatic, free surface and turbulent averaged equa- level. Although Lake Yilong is shallow, with no significant thermal
tions of motion for a variable-density fluid. Dynamically coupled stratification, phytoplankton and aquatic vegetation are still influ-
transport equations for turbulent kinetic energy, turbulent length enced by vertical distribution in light; therefore, it is desirable to
scale, salinity, and temperature are also solved (Hamrick and Wu, resolve variability in vertical light intensity and nutrient using a
1997). The general governing equations for EFDC are (Park et al., three-dimensional spatial resolution. In this model, the grid is
2005): vertically discretized into 5 layers, and a total of 1205 computa-
tional cells were generated from top to bottom to represent Lake

v mx my HC v  v v  Yilong in its entirety (Fig. 2).
þ my HuC þ ðmx HvCÞ þ mx my wC
vt vx vy vz
      (2) Initial conditions
v my HAx vC v mx HAy vC v Az vC (1)
¼ þ þ mx my
vx mx vx vy my vy vz H vz
Initial conditions represent the starting point for the model
þ mx my HSc simulation. In this study, the period simulated was from the sum-
mer of 2008 to the summer of 2009. The initial temperature was set
where C is the concentration of a water quality state variable; u, v, w as 20.0  C based on a value observed in early July. The three velocity
are velocity components in the curvilinear, sigma x-, y-, and z-di- vectors were initialized at 0.0 m/s following standard convention in
rections, respectively; Ax, Ay, Az are the turbulent diffusivities in the hydrodynamic modeling. The hydrodynamic model was then run
x-, y-, and z-directions, respectively; Sc is the internal and external
sources and sinks per unit volume; H is the water column depth;
mx, my are the horizontal curvilinear coordinate scale factors. Water
temperatures are needed for computation of the water quality state
variables, and they are provided by the internally coupled hydro-
dynamic model (Park et al., 2005).

2.3. Configuration for Lake Yilong EFDC model

The first necessary step in establishing long-term water quality


restoration efforts in Lake Yilong would be to develop a quantita-
tive assessment via a sophisticated water quality model to aid in
determining the cause of the sharp increase in algal bloom intensity
in 2009, as well as to determine the effective management levels
that would result in compliance of water quality standards. To
simulate the dynamic factors causing the drastic change in the
aquatic ecosystems of Lake Yilong, it is essential to adopt a model Fig. 2. Computational grids of hydrodynamic water quality simulation in Lake Yilong.
Author's personal copy

16 L. Zhao et al. / Environmental Pollution 177 (2013) 13e21

with a spinning-up period of 1 month to represent the initial


condition for the coupled hydrodynamics-water quality model.
Initial water quality conditions were relatively complex
compared to hydrodynamic model because the former requires a
relatively long period of time to eliminate the impact of initial
conditions during the run process of model. To get a more accurate
water quality model initialization, the observational data from
August 22, 2008 was spatially interpolated to each grid cell to form
a best-available estimate of the initial condition (Supporting
Materials).

(3) Boundary conditions Fig. 3. Comparison of simulated and observed elevations of Lake Yilong.

The horizontal and surface boundary conditions in the model


represent external driving forces to the lake dynamics. The lateral level, indicating an overall balance in water mass related processes
boundary conditions consisted of tributary inflow rates and asso- such as inflows, outflows, direct precipitation, and evaporations.
ciated temperature and water quality constituent concentrations. After the flow balance was verified, the hydrodynamic model
The surface boundary conditions were described by temporally was further evaluated using observed water temperature data.
variable meteorological conditions. In the Lake Yilong model, Model simulated temperature was output for every 6 h and
tributary boundary conditions were configured based on the compared against observed data at three stations across the lake
monitoring data collected during the 2008e2009 study period. For (Fig. 4). The comparison shows that the model accurately repro-
those tributaries with available inflow rate and water quality con- duced the observed spatial and temporal variability in water tem-
centrations, boundary conditions were set directly based on perature in the lake, suggesting that the model has reached
observed data, while for those ungaged inflows, the flow rates were reasonable representation of the hydrodynamic and thermal dy-
derived based on a flow balance analysis using the hydrodynamic namics processes in the lake, hence forming a foundation for
model. In addition to tributaries, atmospheric deposition was further calibrating the water quality processes in the lake.
another major source of nutrient loading. The concentrations of
total phosphorous (TP) and total nitrogen (TN) in the study area 3.2. Water quality simulation and calibration
from dry deposition were 0.00019 g/m2/day and 0.0069 g/m2/day,
respectively, whereas concentrations of TP and TN from wet The water quality model calibration was conducted through
deposition were 0.039 mg/L and 0.95 mg/L, respectively. These fine-tuning the model parameters to reproduce observed water
loadings were configured in the Lake Yilong model using the cor- quality at the three monitoring stations. The constituents used for
responding input data slots in EFDC model. calibration include Chl a, DO, NHþ 4 , TN and TP (Fig. 5). The water
Atmospheric boundary conditions include hourly data of at- quality model calibration involved an iterative process where key
mospheric pressure, air temperature, relative humidity, precipita- parameter values were adjusted and simulated water quality con-
tion, evaporation, solar radiation, cloud cover, wind speed and wind centrations evaluated against observed data. This process was
direction. In the modeling process, hourly data were obtained from repeated many times until the simulated values could reproduce
the Shiping County Weather Station (SPWS) and were converted the observed temporal and spatial patterns. In the eutrophication
into the compatible format of EFDC to represent the atmospheric model, the key parameters related to phytoplankton, nitrogen,
boundary conditions. phosphorus, and carbon would need to be calibrated. In addition to
the dynamic process included in the eutrophication model
(4) Simulation and calibration described in modeling framework, a key process in the Lake Yilong
model was to simulate the impact of fish predation on aquatic
Model calibration is an essential process in water quality model vegetation. In the early spring of 2009, local authorities had been
development (Lung, 2001). Calibration of hydrodynamic and water stocking a large number of herbivorous fish (grass carp). Limited by
quality model in Lake Yilong was implemented in a phased manner: the observation data and the complexity of the process, it was not
the hydrodynamic model was developed and calibrated first before feasible to simulate grass carp stocking, growth, predation and
running the water quality model for fast computing. After the hy- excretion kinetics unless the process was simplified in the EFDC
drodynamic model was calibrated, the water column nutrient- model; therefore, an additional predation parameter was intro-
phytoplankton-macrophyte simulation modules as well as the duced into the EFDC mathematical equation to parameterize the
sediment diagenesis module were activated to begin the water impact of grass carp. The predation pressure on the macrophyte
quality calibration process. from the grass carp was parameterized as a predation coefficient
linearly increasing from the beginning to the end of the stocking
period, which then maintains a constant value after the stocking
3. Results and discussions period is completed.
The model performance was evaluated based on data from all
3.1. Hydrodynamic simulation and calibration the monitoring stations, while only the results for the West Lake
Station are plotted in Fig. 5. As shown, the model reproduced both
The model simulation was implemented with a time step of the observed trends and general magnitude of water quality in the
270 s, which was determined to be small enough to guarantee both lake. For example, according to the model’s predictions, the Chl a
stability and accuracy through multiple sensitivity analysis. The concentration in the lake would be much higher in 2009 than that
hydrodynamic model for Lake Yilong was calibrated through in 2008, which matches the observed data very well. In the
comparing the simulated water surface elevation and water tem- meantime, the model also predicted a significant increase in TN and
perature against observed data (Fig. 3). As shown, the simulation TP concentrations from 2008 to 2009, which is also consistent
results were in good agreement with the observed daily water with observational data. In addition to the trends in nutrient
Author's personal copy

L. Zhao et al. / Environmental Pollution 177 (2013) 13e21 17

30.0
R2 =0.7308

COD1(mg/L)
20.0
Simulated value
10.0 Observed Value

0.0
08/8/22 08/11/30 09/3/10 09/6/18 09/9/26
Date
300
250 R2 =0.6547

Chl a (ug/L)
200
150 Observed value
100
Simulated value
50
0
08/8/22 08/11/30 09/3/10 09/6/18 09/9/26
Date
20.0
15.0 R2 =0.1313

DO(mg/L)
10.0
5.0
0.0
08/8/22 08/11/30 09/3/10 09/6/18 09/9/26
Date
1.0
0.8
NH4(mg/L)

0.6
0.4
0.2
0.0
08/8/22 08/11/30 09/3/10 09/6/18 09/9/26
Date
8.0

6.0 R2 =0.7955
TN(mg/L)

4.0

2.0

0.0
08/8/22 08/11/30 09/3/10 09/6/18 09/9/26
Date
Fig. 4. Comparison of simulated and observed temperatures.
0.3

concentrations and algal blooms, the simulated DO values also 0.2 R2 =0.0619
TP(mg/L)

reproduced the observed data for seasonal and short-term changes


very well. 0.1

0.0
4. Discussion 08/8/22 08/11/30 09/3/10 09/6/18 09/9/26
Date
The observed data showed that the water quality in 2009 was
significantly worse than that in the past few years. The EFDC model Fig. 5. Comparison of water quality variables at West Lake Station.

reproduced the trends accurately. Before 2008, the Chl a concen-


tration never exceeded 150 mg/L; however, in the summer of 2009,
its concentration reached a maximum of 300 mg/L. Meanwhile, the external sources; therefore, the sharp increase in water column
biomass of macrophyte in 2009 declined sharply and is much lower nutrient concentration cannot be attributed to external loading
than that in 2008. changes. It means that although the TN and Chl a data demonstrate a
The data showed that Chl a concentration sharply increase in statistically significant correlation between algal bloom outbreaks
2009 from the much lower level in 2008, accompanied by significant and increases in nitrogen concentrations in 2009, the statistical
increase in TP and TN concentration. Based on watershed survey in analysis could not explain the origin of the nitrogen that caused the
2009, there was no sudden increase in nutrient loading from the algal blooms. While our model results revealed that the sharp
Author's personal copy

18 L. Zhao et al. / Environmental Pollution 177 (2013) 13e21

increase in TN and TP concentration in the lake was not from Table 1


external sources, but from internal sources of nutrients which were Water quality concentration goals in Lake Yilong.

stored in the macrophyte biomass. The previous study (i.e. YLTMD) TN TP CODMn NH4eN
has also proved that (a) the macrophyte was consumed by herbiv- Class III (mg/L) 1.0 0.05 6 1.0
orous fishes and nutrients were then released to the water column Class IV (mg/L) 1.5 0.10 10 1.5
to form high concentrations; and (b) the diminish of macrophyte as Class V (mg/L) 2.0 0.20 15 2.0
Allowed exceedance (%) 50% 50% 50% 50%
well as the resultant high nutrient concentration form positive force
to stimulate blue-green algae blooms in the lake.
significantly lower when macrophyte vegetation is restored. In
5. Scenario analysis on eutrophication responses to load other words, the management methods for controlling algal bloom
reduction in Lake Yilong might include options of watershed management and
in-lake restoration. The results of this scenario indicate that the in-
After the model was developed and calibrated, it was used to lake ecological restoration alone can result in significant depression
analyze a series of water quality compliance scenarios to provide of algal bloom without any watershed management measures. The
quantitative information regarding the magnitude of watershed result also reveals the risk of un-informed decision making in a lake
load reduction. The above analysis showed that aquatic ecosystem watershed, and highlights the importance of scientifically sound
structure plays an important role in maintaining water quality. planning for any actions in lake watersheds.
Therefore, three basic scenarios were designed to quantify (a) the
importance of aquatic vegetation to algal bloom control; (b) the 5.2. Scenario II: load reduction requirement for water quality
load reduction requirement under different water quality targets; standards compliance
and (c) the relationship between nutrients load reduction and
eutrophication condition as represented by algal blooms. The results of Scenario I indicate that under existing watershed
loading condition, the water quality of Lake Yilong would not be
5.1. Scenario analysis I: impact of aquatic vegetation on algal bloom able to attain the targeted water quality standards even if the
control macrophyte vegetation were restored. Therefore, it is necessary to
determine the magnitude of watershed load reduction required to
To analyze the impact of aquatic vegetation on algal blooms, two achieve compliance of water quality standards for the lake. Table 1
scenario simulations were conducted. The first simulation is named lists three different levels of water quality targets for Lake Yilong.
the baseline loading without macrophyte (BLWM) scenario, which The three targets are based on the Class III, IV, and V standards of
represents the baseline condition after the 2009 regime shift where the China Surface Water Environmental Quality Standards (GB3838-
macrophyte vegetation was destroyed. To represent the long term 2002). In GB3838-2002, Class III is corresponding to beneficial use
impact, the model was driven with the same boundary conditions of drinking water sources, Class IV for industrial water supply and
as in the 2008e2009 simulation but continue for 20 years to reach Class V for agricultural irrigation purpose. The reason of setting
steady-state in water quality. The other simulation is named three levels of targets instead of one was to determine the ach-
baseline loading plus macrophyte (BLPM) scenario, which repre- ievability and feasibility of each targets based on their respective
sents the condition where macrophyte vegetation is fully recovered load reduction requirement.
to the condition before the regime shift, while all the dynamic To evaluate the compliances of water quality targets, two as-
processes, boundary conditions, and simulation period are the pects were considered: (a) the water quality standards should to be
same as in the BLWM simulation. With this design, any difference in achieved on an annual average basis; and (b) the instantaneous
algal bloom intensity between the two simulations would be concentration is not allowed to exceed the water quality target
attributed to existence of the macrophyte vegetation. concentrations for more than 50% of the time. Therefore, the final
Fig. 6 shows the simulated Chl a concentration at the West Lake compliance was established based on the more stringent of the two
Station for the 20th year of both the scenarios. The results clearly aspects.
show that even with the same watershed loadings and meteoro- The scenario analysis was conducted via an iterative process
logical conditions, the algal bloom intensity in Lake Yilong can be where the baseline model in Scenario I, i.e., BLWM, was used as the
basis, but the watershed loadings being scaled down to a pre-
400 specified ratio. The model was then run for 20 years and the re-
sults of the 20th year were used to evaluate the compliance from
the aforementioned two aspects. After multiple trials, load reduc-
Chl a concentration ( ug/l)

300 tion ratios required for Class III, IV, and V targets were determined
to be 77%, 55%, and 35%, respectively (Table 2 and Fig. 7). These

200
Table 2
Load reduction rates to meet certain water quality goals.

100 Monitoring sites TN (mg/L) TP (mg/L) COD (mg/L) Load reduction


ratio
Lake West 1.00 0.042 4.56 35%
0 1.50 0.067 7.08 55%
A S O N D J F M A M J J A 2.00 0.101 9.28 77%
Lake Mid 0.79 0.034 3.58 35%
Months 1.22 0.055 5.58 55%
1.62 0.082 7.49 77%
Fig. 6. Simulated Chl a concentration (mg/L) on scenario I at West Lake Station. The top
Lake East 0.73 0.034 3.23 35%
blue line represents the BLWM scenario and the bottom red one for BLPM scenario.
1.18 0.056 5.23 55%
(For interpretation of the references to colour in this figure legend, the reader is
1.63 0.084 7.21 77%
referred to the web version of this article.)
Author's personal copy

L. Zhao et al. / Environmental Pollution 177 (2013) 13e21 19

16
Ave TN Med TN TN Standard Ave TP Med TP
14
TP Standards Ave COD Med COD COD Standard
Concentration (µg/L) 12

10

0
Class III Scenario Class IV Scenario Class V Scenario
West Lake Station

(a) West lake station

16 Ave TN Med TN TN Standard

14 Ave TP Med TP TP Standard

Ave COD Med COD COD Standard


12
Concentration ( µg/L)

10

0
Class III Scenario Class IV Scenario Class V Scenario

(b) Middle lake station

16
Ave TN Med TN Standard N
14 Ave TP Med TP Standard P

12 Ave COD Med COD Standard COD


Concentration (µg/L)

10

0
Class III Scenario Class IV Scenario Class V Scenario

(c) East lake station


Fig. 7. Water quality conditions after load reduction at three monitoring stations.

results suggest that the current watershed loading far exceed the compliance of water quality targets? As the modeling results show,
allowable levels for attaining the pre-set targets, preventing Lake significant further load reduction in addition to what have previ-
Yilong from meeting the desired beneficial uses. This helps explain ously accomplished is required to allow the compliance, therefore,
a puzzling question the government often confronts: why the it is not reasonable to expect that the compliance of the water
previous watershed management efforts didn’t produce expected quality targets can be reached in near future. On the contrary, a
Author's personal copy

20 L. Zhao et al. / Environmental Pollution 177 (2013) 13e21

long-term watershed management as well as in-lake restoration 200

planning is needed to be developed based on the modeling analysis 180 Average Chla

for continuously improving the water quality and ecological con- 160 Median Chla

dition in the lake. 140 Maximum Chla

Concentration (ug/L)
120
5.3. Scenario III: eutrophication response 100

80
To further investigate the impact of watershed load reduction on 60
Chl a concentration in Lake Yilong, the simulated Chl a concen- 40
tration of the 20th year of the BLPM are compared against that of
20
the reduction scenario with 77% reduction of watershed loading on
0
the top of the in-lake restoration corresponding to the BLPM sce- BLWM 77R-ER 77R BLWM 77R-ER 77R BLWM 77R-ER 77R
nario. Fig. 8 plots the results for the West Lake Station, showing that West Lake Station MIddle Lake Station East Lake Station
with the 77% reduction in watershed loading, the phytoplankton in
the lake is significantly depressed due to lower nutrient concen- Fig. 9. Comparison of average, median and maximum Chl a concentrations (mg/L) in
Lake Yilong under BLWM, 77R and 77R-ER scenarios.
tration in the water column resulted from watershed management.
However, it is also apparent that even when the lake water attain
Class III water quality target along with a full restoration of the
macrophyte vegetation, the 77% reduction of watershed loads only 6. Conclusions
led to approximately 50% of Chl a concentration reduction. Since
the extreme reduction of 77% in watershed loading would still (1)This study developed a 3-D hydrodynamic and water quality
result in peak Chl a concentration of approximately 80 mg/L that is model for Lake Yilong, China. The model accurately reproduced
highly liable to algal bloom, it is apparent that the eutrophication the observed water surface elevation, water temperature, and
problem is not controlled even with the compliance of the most nutrient and algal conditions, indicating a reasonable numerical
stringent water quality target (Class III) for Lake Yilong. representation of the actual hydrodynamics and eutrophication
In scenario I, we presented the results representing the impact of dynamics in the lake.
ecological restoration under existing loading level. To further (2)The model results showed that the existence of aquatic
investigate the effects of ecological restoration under much reduced vegetation had significant impacts on algal blooms in Lake
loading condition, another analysis was conducted through Yilong. Even though the watershed load remained under the
comparing the Chl a concentration under 77% reduction and same conditions, algal bloom intensity in the lake can be
ecological restoration (77R-ER) and that of 77% reduction alone significantly depressed under the vegetation restoration con-
(77R). The averages, median and maximum value of Chl a concen- dition than under the condition where aquatic vegetation
tration at the three monitoring stations were compared in Fig. 9. It is diminished. Therefore, both watershed load reduction and in-
shown that the Chl a concentrations for 77R are much higher than lake restoration are necessary for effective eutrophication con-
those of 77R-ER, indicating that the benefit of in-lake ecological trol in Lake Yilong.
restoration is significant not only under current high loading con- (3)The study analyzed the watershed load reduction require-
dition, but also under the future condition where watershed load is ment for achieving compliance of various water quality targets.
highly reduced. Therefore, it is recommended that for Lake Yilong, The results indicated that to achieve the water quality targets
both watershed management and in-lake restoration need to be corresponding to Class III, IV, and V of the national water quality
implemented to control the eutrophication impairment. Given the standard, the watershed load reduction ratios would need to be
data and funding limitation, this study did not assess economic ef- 77%, 55% and 35%, respectively.
ficiency, social development and specific ecological restoration (4)The study also suggested that the adaptation of national
technologies. Future research is expected to be focus on these areas water quality standards for nutrients as lake water quality tar-
to provide more effective decision support to the eutrophication gets might not be consistent with the eutrophication control
control of Lake Yilong. targets. The model results indicate that even if the most strin-
gent nutrient target was met with 77% reduction of watershed
loading, the peak Chl a concentration in the lake can still reach
80 mg/L, a concentration that is highly liable to algal bloom.
200
77% reduction

BLPM scenario Therefore, future studies are required to investigate more


reasonable water quality targets to support future eutrophica-
150
tion control efforts for Lake Yilong.
Chl a concentration (µg/L)

100
Acknowledgments

This work was supported by the National Natural Science


50
Foundation of China (Grant No. 41101180 and 41222002) and the
“Major Science and Technology Program for Water Pollution Con-
0
trol and Treatment” (2013ZX07102-006).
A/08 S/08 O/08 N/08 D/08 J/09 F/09 M/09 A/09 M/09 J/09 J/09 A/09
months
Appendix A. Supplementary data
Fig. 8. Simulated Chl a concentration (mg/L) on scenarios of baseline (BLPM) and ul-
timate (77%) reduction. The top blue line represents the BLPM scenario and the bottom
red one for 77% reduction scenario. (For interpretation of the references to colour in Supplementary data related to this article can be found at http://
this figure legend, the reader is referred to the web version of this article.) dx.doi.org/10.1016/j.envpol.2013.01.047.
Author's personal copy

L. Zhao et al. / Environmental Pollution 177 (2013) 13e21 21

References geospatial watershed assessment tool. In: Proceedings of the 2nd Federal
Interagency Hydrologic Modeling Conference, Las Vegas, NV, USA, p. 12.
Park, S.S., Lee, Y.S., 2002. A water quality modeling study of the Nakdong River,
Ahmad, S., Khan, I.H., Parida, B.P., 2001. Performance of stochastic approaches for
Korea. Ecol. Model. 152 (1), 65e75.
forecasting river water quality. Water Res. 35 (18), 4261e4266.
Park, K., Jung, H.S., et al., 2005. Three-dimensional hydrodynamic-eutrophication
Alcamo, J., 2001. Scenarios as Tools for International Environmental Assessments.
model (HEM-3D): application to Kwang-Yang Bay, Korea. Mar. Environ. Res.
Environmental Issue Report No. 24. European Environmental Agency,
60 (2), 171e193.
Copenhagen.
Peng, S., Fu, G.Y.Z., Zhao, X.H., Moore, B.C., 2011. Integration of Environmental fluid
Andreadakis, A., Noutsopoulos, C., Gavalaki, E., 2003. Assessment of the water
dynamics Code (EFDC) model with geographical information system (GIS)
quality of lake plastira through mathematical modelling for alternative man-
platform and its applications. J. Environ. Inform. 17, 75e82.
agement scenarios. In: Proceedings of the 8th International Conference on
Purandara, B.K., Varadarajan, N., Venkatesh, B., Choubey, V.K., 2012. Surface water
Environ Sci and Technol. Oral Presentations, vol A, pp. 17e24.
quality evaluation and modeling of Ghataprabha River, Karnataka, India. Envi-
Arnold, J.G., Williams, J.R., Srinivasan, R., King, K.W., Griggs, R.H., 1994. SWAT: Soil
ron. Monit. Assess. 184 (3), 1371e1378.
Water Assessment Tool. U. S. Department of Agriculture, Agricultural Research
Rehana, S., Mujumdar, P., 2009. An imprecise fuzzy risk approach for water quality
Service, Grassland, Soil and Water Research Laboratory, Temple, TX, USA.
management of a river system. J. Environ. Manage. 90 (11), 3653e3664.
Charpa, S.C., 1997. Surface Water-quality Modeling. McGraw-Hill, New York.
Rosenberga, Rutger, Lars-Ove, Looa, 1988. Marine eutrophication induced oxygen
Ewert, F., Rounsevell, M.D.A., Reginster, I.R., Metzger, M.J., Leemans, R., 2005. Future
deficiency: effects on soft bottom Fauna, Western Sweden. Ophelia 29, 213e
scenarios of European agricultural land use. I. Estimating changes in crop
225.
productivity. Agric. Ecosyst. Environ. 107, 101e116.
Rounsevell, M., Ewert, F., Reginster, I., Leemans, R., Carter, T., 2005. Future scenarios
Hamrick, J., Wu, T., 1997. Computational Design and Optimization of the EFDC/
of European agricultural land use: II. Projecting changes in cropland and
HEM3D Surface Water Hydrodynamic and Eutrophication Models. Society of
grassland. Agric. Ecosystems Environ. 107 (2), 117e135.
Industrial and Applied Mathematics, Philadelphia, pp. 143e161.
Schoemaker, P.J.H., 1982. The expected utility model: its variants, purposes, evi-
Höjer, M., Ahlroth, S., Dreborg, K.H., Ekvall, T., Finnveden, G., Hjelm, O.,
dence and limitations. J. Econ. Lit., 529e563.
Hochschorner, E., Nilsson, M., Palm, V., 2008. Scenarios in selected tools for
Seo, D., Sigdel, R., Kwon, K.H., Lee, Y.S., 2010. 3-D hydrodynamic modeling of
environmental systems analysis. J. Clean. Prod. 16 (18), 1958e1970.
Yongdam Lake, Korea using EFDC. Desalin. Water Treat. 19, 42e48.
Kepner, J., 2004. High performance computing productivity model synthesis. Int. J.
Shi, J.H., Li, G.X., Wang, P., 2011. Anthropogenic influences on the tidal prism and
High Perform. Comput. Appl. 18 (4), 505e516.
water exchanges in Jiaozhou bay, Qingdao, China. J. Coast. Res. 27, 57e72.
Kepner, William G., Semmens, Darius J., Bassett, Scott D., Mouat, David A.,
Smith, V.H., 1983. Low nitrogen to phosphorus ratios favor dominance by blue-
Goodrich, David C., 2004. Scenario analysis for the san pedro river, analyzing
green algae in lake phytoplankton. Science 221 (4611), 669e671.
hydrological consequences of a future environment. Environ. Monit. Assess. 94,
Smith, V.H., Schindler, D.W., 2009. Eutrophication science: where do we go from
115e127.
here? Trends Ecol. Evol. 24 (4), 201e207.
Li, Y.P., Acharya, K., Yu, Z.B., 2011. Modeling impacts of Yangtze River water transfer
Vieira, J.M.P., Lijklema, L., 1989. Development and application of a model for
on water ages in Lake Taihu, China. Ecol. Eng. 37, 325e334.
regional water quality management. Water Res. 23 (6), 767e777.
Liu, Y., Yang, P.J., Cheng, H., Guo, H.C., 2008a. Water quality modeling for load
Wack, P., Nov./Dec. 1985. Scenarios: shooting the rapids. Harv. Bus. Rev. 6, 139e150.
reduction under uncertainty: a Bayesian approach. Water Res. 42, 3305e3314.
Wu, G.Z., Xu, Z.X., 2011. Prediction of algal blooming using EFDC model: case study
Liu, Y., Guo, H.C., Yu, Y.J., Dai, Y.L., Zhou, F., 2008b. Ecologicaleeconomic modeling as
in the Daoxiang Lake. Ecol. Model. 222, 1245e1252.
a tool for lakeewatershed management: a case study of Lake Qionghai
Zhang, J.L., Zhao, L., Nie, J.F., 2010. Research on total amounts of major pollutants of
Watershed, China. Limnologica, 89e104.
wastewater from the fish ponds near the Lakeshore of Lake Yilong. Environ. Sci.
Lung, W.S., 2001. Water Quality Modeling for Wasteload Allocations and TMDLs.
Surv. 29 (2), 36e38 (in Chinese).
John Wiley & Sons, Inc., 333 pp.
Zou, R.., Carter, S., Shoemaker, L., Parker, A., Henry, T., 2006. An Integrated hydro-
Martins, G., Ribeiro, D.C., Pacheco, D., Cruz, J.V., Cunha, R., Goncalves, V.,
dynamic and water quality modeling system to support nutrient TMDL devel-
Nogueira, R., Brito, A.G., 2008. Prospective scenarios for water quality and
opment for Wissahickon Creek. J. Environ. Eng. ASCE 132 (4), 555e566.
ecological status in Lake Sete Cidades (Portugal): the integration of mathe-
Zou, R., Lung, W.S., Wu, J., 2009. Multiple-pattern parameter identification and
matical modelling in decision processes. Appl. Geochem. 23 (8), 2171e2181.
uncertainty analysis approach for water quality modeling. Ecol. Model. 220,
Miller, S.N., Semmens, D.J., Miller, R.C., Hernandez, M., Goodrich, D.C., Miller, W.P.,
621e629.
Kepner, W.G., Ebert, D., 2002. GIS-based hydrologic modeling: the automated

You might also like