Ecological Modelling 107 (1998) 171 188

A model for the effects of water hyacinths on water quality in

an experiment of physico-biological engineering in Lake Taihu,
China
Weiping Hu a,*, Jrgen Salomonsen b, Fu-Liu Xu b, Peiming Pu a
b

a
Nanjing Institute of Geography & Limnology, Chinese Academy of Sciences, 73 East Beijing Road, Nanjing 210008, China
Institute A, En6ironmental Chemistry, Royal Danish School of Pharmacy, Uni6ersity Park 2, DK-2100 Copenhagen , Denmark

Received 2 April 1997; accepted 1 December 1997

Abstract
A model for an experiment of physico-biological engineering purifying lake water in Lake Taihu by using water
hyacinths (Eichhornia crassipes (Mart.) Solms) was constructed. The model included 14 state variables. They are
NH4+ -N, NO3 -N, PO34 -P, nitrogen and phosphorus in detritus, phosphorus in the pore water, exchangeable
phosphorus and nitrogen in the sediment layer, density, nitrogen and phosphorus in phytoplankton, and nitrogen and
phosphorus in water hyacinths. The external forcing functions were solar radiation, water temperature, concentrations
of nutrients, phytoplankton and detritus in inflow water and the retention time of the water in physico-biological
engineering. The results of the model simulating the growth of phytoplankton and water hyacinths, and the cycling
of nitrogen and phosphorus inside physico-biological engineering were coincident with the results observed. In order
to decide which process would affect the water quality in the experiment, to which parameter the water quality
indexes such as phytoplankton, NH4+ -N, NO3 -N PO34 -P are sensitive has been analyzed by the use of the model.
It has been discussed how to culture and harvest the water hyacinths. The filtering effect has also been estimated. The
model could be used as a tool to guide physico-biological engineering design and its management. 1998 Elsevier
Science B.V. All rights reserved.
Keywords: Model; Physico-biological engineering; Water quality; Water hyacinths; Lake Taihu

* Corresponding author. Tel.: +86 25 3603715; fax: +86 25 7714759; e-mail: pupm@public1.ptt.js.cn
0304-3800/98/$19.00 1998 Elsevier Science B.V. All rights reserved.
PII S0304-3800(97)00219-6

172

W. Hu et al. / Ecological Modelling 107 (1998) 171188

1. Introduction
Drinking water has to be provided from surface water in the south eastern part of China due to the
high population density and overuse of ground water. This water is often highly eutrophicated and there
is therefore a need for cheap and ecologically neutral methods for purifying the surface water. In this
paper a model of the use of water hyacinths in an experimental ecosystem for improving the water quality
of drinking water was presented in the small Wulihu Bay in Lake Taihu.
Lake Taihu is located in the east part of China among 305% 328%N and 1198% 12155%E. It is one of
the five largest freshwater lakes in China, with a catchment area of 36500 km2, a surface area of 2338.1
km2, and an average water depth of 1.9 m. The annual average air temperature is 14.916.2C. The
climate is an SENW monsoon climate. The mean annual precipitation is 10001400 mm, the mean
annual evaporation 941 mm, the mean annual runoff into the lake 4100 million m3 (Sun et al., 1993).
There are six large cities and about 34.182 million people around the lake. The economic development of
this part of China has been fast in recent years, and the population increases quickly. Therefore, more and
more waste water was discharged into the lake which has become highly eutrophicated. In the summer of
1990, there were big problems with phytoplankton bloom taking place in the lake. About 119 factories
had to cease production and the people living in Wuxi City could not have sufficient drinkable water
during the period. This caused a loss of 112 million Chinese Yuan, i.e. approx. US$ 13330000 in Wuxi.
A new method for purifying the lake water was urgently needed at that time. So a series of experiments
on how to improve the water quality have been carried out in Lake Taihu (Pu et al., 1993; Dou et al.,
1995; Hu and Pu, 1995, Pu et al., 1995; Pu and Hu, 1996). Some experiments focused on the possibility
of using aquatic plants to remove the nutrients. Eichhornia crassipes (Mart.) Solms has been selected as
one important kind of water plant cultured in the experimental area for improving water quality, since it
can grow quickly and take up a lot of nutrients from water, such as phosphorus, nitrogen (Yan, 1986; Wu
et al., 1987a,b; Zhang, 1989; Dou et al., 1995). It is a kind of free-floating water plant whose leaves are
above the water surface. It reproduces rapidly by budding. Water hyacinths in the water surface can
minimize the light penetration to the water column below and therefore phytoplankton is outcompeted.
This is a favorable situation from the point of ecological management as the nutrients are concentrated
in water hyacinths instead of in phytoplankton, because water hyacinths can be more easily harvested
with nets or forks and used as feed for pigs and fish. Besides taking up nutrients, water hyacinths have
also been reported to take up heavy metal ions (Dai et al., 1991) from the water and to metabolize organic
substances such as phenol (Wu et al., 1987a,b). Culturing of water hyacinths can improve water
transparency. At the beginning of our experiment the water transparency was 40 cm, after 4 days
culturing it changed to 2.0 m. If this water flows to another place without too many water hyacinths
covering the water surface, submerged plants can increase growth. Because the water hyacinths can shade
the light on the water surface, a lower water temperature below water hyacinths can be observed. In
summer, with some water hyacinths covering the water surface (not too many), some submerged plants
can survive.

2. Background of the experiment

2.1. Physical and chemical characteristics of the experimental site

The experimental area, 2000 m2 and a mean depth of 2.0 m in summer, is situated in the north-east part
of the small Wulihu bay in Lake Taihu (see Fig. 1a). It is in the intake area of Wuxi Zhongqiao Drinking
Water Plant near Wuxi city. It is divided into 10 channels, each 40 m long and 5 m wide, enclosed by
nontoxic chemical fiber cloth. Water from the outside can only enter the southern side of the channel (see

W. Hu et al. / Ecological Modelling 107 (1998) 171188

173

Fig. 1. (a) The topographical map of the Lake Taihu. (b) The schematic diagram of the engineering experiment for improving water
quality.

174

W. Hu et al. / Ecological Modelling 107 (1998) 171188

Fig. 1b). The water quality in the experimental area was worse than that in the other parts, with a high
average density of chlorophyll a and concentrations of nutrients, which are shown in Table 1. The water
quality parameter COD in this area was even worse in 1994 (see Table 1) (Li, 1996a). TP (in P2O5) and
TN concentrations in the sediments were high, up to 0.20 and 0.21% of the dry weight. Exchangeable
phosphorus in sediment is up to 2.64% of total phosphorus in the bottom sediment. Its concentration in
pore water was high, up to 52.7 mg/l (in P2O5). The pH of bottom water varied between 6.8 and 7.2. The
sediment surface was anaerobic during the sampling period with Eh values from 120 to 356 mV.

2.2. Purpose of the experiment

Water hyacinths were cultured in the experiment to minimize the density of phytoplankton, improve
water quality and increase the water transparency. The water quality parameters inside and outside the
channel were measured every day or every second day. The water in the channel was pumped out to
simulate the water pumping of Zhongqiao Water Plants from Wulihu Bay.

3. Ecological model construction of the experiment

3.1. Conceptual diagram of model

The main physical, chemical and biological processes are: (1) nutrients (phosphorus, ammonium
nitrogen, nitrate nitrogen uptake by phytoplankton and water hyacinths); (2) growth of phytoplankton
and water hyacinth; (3) mortality of phytoplankton and water hyacinths; (4) settling of detritus; (5)
mineralization of detritus; (6) nutrient release from sediment; (7) oxidation of ammonium nitrogen. The
external control functions are: water temperature, solar radiation, water flow into and out the experimental ecosystem, and nutrient concentrations of water flowing in and out of the experimental ecosystem.
Therefore, the ecological model should include: phytoplankton, water hyacinths, NH4+ -N, NO3 -N,
PO34 -P, nitrogen and phosphorus in sediment and detritus. Its conceptual diagram is shown in Fig. 2.

Fig. 2. Conceptual diagram of the ecological model.

1991
1992
1993
1994

34.5
30.1
23.3
38.2

Chl.a (mg/l)
14.9
18.2
9.2
57

COD (mg/l)
2.21

BOD (mg/l)
2.1
46.9
6.64
10.9

TN (mg/l)

Table 1
Annual mean water quality of Wulihu Bay from 1991 to 1994

0.37
0.46
0.53
0.56

0.03
0.19
0.13
0.25

0.53
2.75
1.99
4.12

74.1
397
186
150

+
3
NO
3 (mg/l) NO2 (mg/l) NH4 (mg/l) TP (mg/m )

13.9
31.8
16.7
91.7

PO3
(mg/m3)
4

9.1
5.1

DO (mg/l)

0.47
0.34
0.41
0.48

SD (m)

W. Hu et al. / Ecological Modelling 107 (1998) 171188

175

W. Hu et al. / Ecological Modelling 107 (1998) 171188

176

3.2. The mathematical equations in model

3.2.1. Equations for growth of phytoplankton and water hyacinths
The mass conservation law can be used to establish the equations of the model. For phytoplankton the
following equations are used.
dBph
= GABMA +INA OUTA
dt

(1)

dNph
= NHuph + NOuph +Nphi Npho Nphm
dt

(2)

dPph
= Puph +Pphi Ppho Pphm
dt

(3)

where Bph, Nph, Pph, GAB, MA, INA, OUTA are phytoplankton density in mg/l, nitrogen and phosphorus
in phytoplankton in mg/l, phytoplankton growth, phytoplankton mortality, input of phytoplankton due
to water inflow, loss of phytoplankton due to water outflow respectively. Npho and Ppho are the loss of
nitrogen and phosphorus in phytoplankton due to the outflow of water; Nphi and Pphi the input of
nitrogen and phosphorus in phytoplankton due to the water inflow; Nphm and Pphm the loss of nitrogen
and phosphorus in phytoplankton due to mortality of phytoplankton. Mortality is a source of nitrogen
and phosphorus in detritus which will be described later; NHuph, NOuph and Puph are the uptakes of
phytoplankton for NH4+ -N, NO3 -N and PO34 -P, respectively.
GAB can be written in following form:
GAB= grph Bph
where grph is the growth rate of phytoplankton.
The limitation of the growth rate of phytoplankton is described in the classical two steps. The first step
is uptake of nutrients in accordance with the MichaelisMenten kinetics, and the second step is
determined by the internal nutrient concentration. The uptakes of nitrate nitrogen, ammonium nitrogen
and phosphorus are (Bendoricchio et al., 1993, 1994; Cai, 1995):
NHuph = VmNHph NHpre Npro ph NH Bph/(NH + Kmnh ph)
NOuph = VmNOph (1 NHpre) Npro ph NO Bph/(NO + Kmno ph)
Puph = VmPph Ppro ph P Bph/(P + Kmp ph)
where VmNHph, VmNOph and VmPph are the nutrient maximum uptake rates of phytoplankton for
ammonium nitrogen, nitrate nitrogen and phosphorus, respectively; Npro-ph and Ppro-ph the normalized
saturation values for the intracellular nutrient; Kmnh-ph, Kmno-ph and Kmp-ph, the half-saturation constants
for each of these three nutrients; NH, NO, P concentrations of NH4+ -N, NO3 -N and PO34 -P in water.
NHpre ratio of NH4+ -N of the soluble nitrogen in the water, it is:
NHpre =NH/(NH + NO)
The normalized saturation constants for nutrients are expressed as:
Npro ph =(Nmax ph Nph/Bph)/(Nmax ph Nmin ph)
Ppro ph =(Pmax ph Pph/Bph)/(Pmax ph Pmin ph)
where Nmax ph and Pmax ph are the maximum quotas of phytoplankton for nitrogen and phosphorus,

respectively, Nmin ph and Pmin ph the minimum quotas of phytoplankton for nitrogen and phosphorus,
respectively.

W. Hu et al. / Ecological Modelling 107 (1998) 171188

177

The growth rate of phytoplankton can now be written as follows:

grph =Gmax ph F(Nph,Pph) F(T) F(I),
Gmax ph is the intrinsic value for growth rate of phytoplankton; F(Nph,Pph), F(T), F(I) are the functions

of Nph and Pph, water temperature (T) and light (I) affecting the phytoplankton growth, which are
formulated as:
F(Nph,Pph)=min(1 Nmin
F(T)= e
F(I)=

ph

Bph/Nph,1 Pmin

ph

Bph/Pph),

0.007T 25

I0 {I/1 exp[ 2(1.7/TSP + 0.014Bph)]} max(1 Bwh/320.0,0.0)

,
2(1.7/TSP +0.014Bph) IK

where T is water temperature (C); TSP is Secchi Disc depth (m); I0 is daily average of incident light
strength (uE/m2/s), Bwh is the biomass density of water hyacinths, IK is the saturation constant of
phytoplankton for light. The term max(1Bwh/320.0,0.0) reflects the influence of water hyacinths on light
in water.
The mortality of phytoplankton is composed of two parts. One is the general mortality of phytoplankton without the influence of water hyacinths, while the other part is influenced by water hyacinths because
of the shading effect. From this the following equation can be obtained:
MA= Rm ph Bph +(1 Rm ph) Bph (1 F(I))
where Rm-ph is the normal death rate of phytoplankton. The losses of nitrogen and phosphorus in
phytoplankton due to its death can be written as:
Nphm = MA Nph/Bph
Pphm =MA Pph/Bph
For water hyacinths analogous equations are formulated:
dBwh
=GWH MWH OUTWH BHAR
dt

(4)

GWH=grwh Bwh
dNwh
= NHuwh +NOuwh Nwhm Nwhh
dt

(5)

dPwh
=Puwh Pwhh Pwhm
dt

(6)

NHuwh = VmNHwh NHpre Npro wh NH Bwh/(NH + Kmnh wh)

NOuwh = VmNOwh (1 NHpre) Npro wh NO Bwh/(NO + Kmno wh)
Puwh =VmPwh Ppro wh P Bwh/(P + Kmp wh)
Npro wh = (Nmax
Ppro wh = (Pmax
grwh =Gmax

wh

wh

wh

Nwh/Bwh)/(Nmax wh Nmin wh)

Pwh/Bwh)/(Pmax wh Pmin wh)

F(Nwh,Pwh) Fwh(T) Fwh(I0)

F(Nwh,Pwh)= min(1 Nmin

wh

Bwh/Nwh,1 Pmin wh Bwh/Pwh)

W. Hu et al. / Ecological Modelling 107 (1998) 171188

178

Fwh(T)=1.05 exp( 0.024T 25.8)

Fwh(I0)=I0/1300 p
MWH = Rm wh Bwh
Nwhm = MWH Nwh/Bwh
Pwhm =MWH Pwh/Bwh
The symbols for water hyacinths above have analogous meanings, as in the equations for phytoplankton.
The formulations of dynamics of water hyacinths differ from the phytoplankton dynamics at one point.
There is neither input nor output of water hyacinths with water flow. However, there is loss of biomass,
nitrogen and phosphorus of water hyacinths due to the harvesting of water hyacinths. The terms related
to this are expressed as:
BHAR=max(Bwh Kbwh,0.0)/DT
Nwhh =BHAR Nwh/Bwh
Pwhh =BHAR Pwh/Bwh
where Kbwh, Nwhh, Pwhh are the density of water hyacinths for beginning harvest, the outputs of nitrogen
and phosphorus due to the harvesting; DT is the time interval for calculation.

3.2.2. Equations for phosphorus cycling

The nutrient cycle includes chemical, physical and biological processes which are shown in Fig. 2.
Mathematical equations for phosphorus and nitrogen can be achieved by using the mass conservation
law. The equations for phosphorus cycling are:
dP
=Pi +Pd m +Ps r Po Puwh Puph

dt

(7)

dPd
= Pdi Pd m +Pwhm Pdo Pds +Pphm
dt

(8)

dPpw
=Pber Ps r
dt

(9)

dPbe
= Pds Pber
dt

(10)

where Pd is the phosphorus concentration in water in the form of detritus; Ppw the phosphorus
concentration in pore water in the upper layer of bottom that is 20 cm thick. It is converted to the amount
in the whole water column for calculation convenience. Pbe is the exchangeable phosphorus in the upper
layer of bottom divided by water volume; Pi and Po are the inorganic phosphorus input and output due
to the water inflow and output, respectively; Pd m is the phosphorus release from detritus mineralization;
Ps r is the phosphorus exchange between the inorganic phosphorus in the water column and in the pore
water; Pds is the loss of phosphorus in the form of detritus due to the sink of detritus; Pber is the release
of the exchangeable phosphorus of the upper layer of the bottom sediment to the pore water; Pdi, Pdo are
the phosphorus input and output in form of detritus, respectively.
Pd m can be written as (Jrgensen, 1988):
20
Pd m =Pd Vmd K Tmdt

According to Nielsens mud water exchange experiment (1974), Ps r can be written in the following form:

W. Hu et al. / Ecological Modelling 107 (1998) 171188

! 

T +273
Ph7

V mpp (10.0Ppw P)
1.0
280 0.7
Phmax Phmin
Ps r =

0.0

"

179

when DO= 0.0

when DO\ 0.0

where Ph is the value of the pore water and Phmax, Phmin are its maximum and minimum values. Because
the sediment of the experimental site is anaerobic, Ps r can be simplified as:

Ps r = Vmpp (10Ppw P) (T +273)/280/0.7

Pber is expressed by using a first-order kinetic equation and can be written as:
Pber = Vrex K Text 20 Pbe

3.2.3. Equations for nitrogen cycling

Three forms of nitrogen compounds are very important water quality parameters for drinking water, as
well as for general water quality. Because the conversion from NH4+ -N, to NO2 -N and then to NO3 -N
is the rate-limiting process in nitrification, and NO2 -N is unstable, this model does not include NO2 -N.
It is assumed that ammonium nitrogen is oxidized to nitrate nitrogen in one step. By using this
simplification, the first kinetic equation for ammonium nitrogen and nitrate nitrogen can be derived:
dNO
=NOi + NOd m +NHO NOo NOuwh NOuph
dt
dNH
= NHi + NHd m NHO NHo NHuwh NHuph + Ns r

dt

(11)
(12)

where NOd m, NHd m are the releases of nitrate and ammonium nitrogen, respectively, due to the
mineralization of detritus, which are formulated as:
NOd

20
= 0.45Vdm K Tmdt
Nd

20
NHd m = 0.55Vdm K Tmdt
Nd

where Nd is the nitrogen concentration in detritus and Ns r is the release of ammonium nitrogen from the
sediments. It is related to pH, in this paper it is given the value of e0.151(T 20) (0.00004Ns + 0.00008). The
oxidization of ammonium nitrogen is
NHO = Vnho NH
The equations for nitrogen in detritus and in the upper layer of bottom are:
dNd
= Ndi +Nwhm +Nphm Nds NHd m NOd m Ndo
dt
dNs
=Nds Ns r Nsl
dt

(13)
(14)

where Nsl is the denitrification in bottom sediments. The other terms not described in Eq. (11), Eq. (12),
Eq. (13) and Eq. (14) have analogous meaning as in equation Eq. (7), Eq. (8), Eq. (9) and Eq. (10).

3.3. Values of the parameters in the model

There are about 30 parameters in this model. Some of them are achieved by measurement, some of
them are found from literature studies, and the rest are achieved by calibration which will be described
in the next section; they are listed in Table 2.

W. Hu et al. / Ecological Modelling 107 (1998) 171188

180
Table 2
Model parameters
Symbol

Description

Unit

Value used

Source

1/d

0.384

1/d

0.24

1/d

0.01

mg/l

0.176

Jrgensen et al.,
1991
Jrgensen et al.,
1991
Jrgensen et al.,
1991
Calibration

mg/l
mg/l

0.164
0.015

Maximum growth rate of phytoplankton

1/d

0.11

Maximum uptake velocity of water hyacinths for ammonium

nitrogen
Maximum uptake velocity of water hyacinths for nitrate nitrogen

1/d

0.002

Calibration
Jrgensen et al.,
1991
Jrgensen et al.,
1991
Calibration

1/d

0.0078

Calibration

VmNHph Maximum uptake velocity of phytoplankton for ammonium

nitrogen
VmNOph Maximum uptake velocity of phytoplankton for nitrate nitrogen
VmPph

Maximum uptake velocity of phytoplankton for phosphorus

Kmnh-ph Half saturation constant of ammonium nitrogen for

phytoplankton
Kmno-ph Half saturation constant of nitrate nitrogen for phytoplankton
Kmp-ph Half saturation constant of phosphorus for phytoplankton
Gmax

ph

VmNHwh
VmNOwh
VmPwh
Kmnh-wh

1/d
mg/l

0.0025
0.3

Calibration
Jin, 1994

Kmno-wh
Kmp-wh
Gmax wh

Nmin ph

Pmax ph

Pmin ph

Nmax ph

Nmin wh

Pmax wh

Pmin wh

Maximum uptake velocity of water hyacinths for phosphorus

Half saturation constant of ammonium nitrogen for water
hyacinths
Half saturation constant of nitrate nitrogen for water hyacinths
Half saturation constant of phosphorus for water hyacinths
Maximum growth rate of water hyacinths
Minimum quota of phytoplankton for nitrogen
Maximum quota of phytoplankton for phosphorus
Minimum quota of phytoplankton for phosphorus
Maximum quota of phytoplankton for nitrogen
Minimum quota of water hyacinths for nitrogen
Maximum quota of water hyacinths for phosphorus
Minimum quota of water hyacinths for phosphorus

mg/l
mg/l
1/d
mg N /mg d.w.
mg P/mg d.w.
mg P/mg d.w.
mg N/mg d.w.
mg N/mg d.w.
mg P/mg d.w.
mg P/mg d.w.

0.15
0.01
0.1181 5
0.007
0.015
0.001
0.085
0.0158
0.016
0.001

Nmax

Maximum quota of water hyacinths for nitrogen

mg N/mg d.w.

0.096

Velocity of detritus mineralization at temperature 20C

A constant for the influence of temperature
Maximum diffusion coefficient of phosphorus from pore water
Release velocity of phosphorus from exchangeable phosphorus to
pore water
Constant for the influence of temperature
Oxidization velocity of ammonium nitrogen

1/d
1/d
1/d
1/d

0.022
1.15
0.005
0.013

Calibration
Calibration
Measured
Calibration
Cai, 1995
Cai, 1995
Calibration
Dou et al., 1995
Dou et al., 1995
Jrgensen et al.,
1991
Jrgensen et al.,
1991
Cai, 1995
Cai, 1995
Calibration
Cai, 1995

1/d
1/d

1.13
0.1

Cai, 1995
Calibration

Vmd
Kmdt
Vmpp
Vrex
Kext
Vnho

wh

3.4. Calibration of the model

The result of an experiment of the culturing of water hyacinths in a small square with an area of 25 m2
and a depth of 2.1 m, carried out in the summer of 1996, is used to calibrate this model. The experimental
time was 7 days. At the beginning of the experiment, the water quality parameters (chl-a, DO, NH4+ ,
NO3 , PO34 , TP, TN, BOD5, detritus, primary productivity, water transparency, pH), water temperature
inside and outside the square, and solar radiation were measured. After that, these parameters were
measured every second day. The model and the parameters were checked by comparing the calculated

W. Hu et al. / Ecological Modelling 107 (1998) 171188

181

Table 3
The calculated and observated results of the experiment for calibration
Day

1
2
3
4
5
6
7

NH+
4 N (mg/l)

NO
3 N (mg/l)

Cal

Obs

Cal

0.32
0.29
0.28
0.31
0.30
0.31
0.39

0.32

0.27

0.44

0.23
0.25
0.25
0.26
0.27
0.30
0.40

3
PO3
4 P (mg/m )

Phytoplankton (mg/l)

Obs

Cal

Obs

Cal

Obs

0.23

0.24

0.65

3.50
3.25
3.40
3.81
4.19
4.32
4.50

3.50

3.80

3.00

2.73
2.14
2.00
1.98
1.84
1.74
1.50

2.73
2.16

2.20

, not observated; obs, observated value; cal, calculated value.

results with the experimental values. The parameters achieved by calibration are shown in Table 2. The
calculated and the observed results are shown in Table 3. It is shown that the result of the model is
consistent with experiment for calibration.

4. Results

4.1. Initial 6alues of the state 6ariables and external function

In the dynamic experiment, the water was pumped out at the north end of the channel and the outside
water entered at the southern end to simulate the water pumping of the water plant. In the experiments,
the density of phytoplankton and water hyacinth, concentration of chlorophyll-a, NO2 -N, NO3 -N,
PO34 -P, DO, TP, TN, BOD5, water transparency inside and outside the experimental site, solar radiation
in air and in water, pH and temperature were measured. Values for Nph, Pph, Nwh, Pwh, Nd, Pd are
calculated from these measurements. The initial values for Pbe, Ppw, Ns are obtained from research reports
from Nanjing Institute of Geography and Limnology, Chinese Academy of Science (1990) and Li (1996b).
The initial values for state variables are shown in Table 4. The retention time (Ret) of water in the
experiment is 1.2 days.

Table 4
Initial value of state variables (mg/l)
Variable

Value

Variable

Value

Bph
Nph
Pph
Bwh
Nwh
Pwh
NH

1.1565
0.11565
0.011565
285.625
9.02575
1.473825
0.274

NO
Nd
Pbe
Ppw
Ns
Pd
P

0.163
0.71768
3.96
2.2704
159.75
0.115105
0.001

182

W. Hu et al. / Ecological Modelling 107 (1998) 171188

Fig. 3. Observed and calculated concentration of ammonium nitrogen.

4.2. Comparison between calculation and reality

Figs. 3 and 4 are the diagrams of water quality parameters versus time with the outside water entering
the channel without filtering. It is shown that the calculated results of water quality parameters NH4 -N
and NO3 -N are consistent with the observed experimental results. At the beginning of the pumping
experiment, the concentrations of NH4+ -N and NO3 -N in the channel began to increase quickly because
of the high concentrations of NH4+ -N and NO3 -N in the inflow, and as the concentrations of NH4+ -N
and NO3 -N in the inflow decreased the concentration of NH4+ -N and NO3 -N in the channel increased
slowly until maximum values were reached. They then began to decrease, but the calculated results
achieved a maximum value earlier than the observed results. The differences were induced by the average
of the whole channel. From Fig. 5, we can see that the calculated value of PO34 -P in the channel varied

Fig. 4. Observed and calculated concentration of nitrate nitrogen.

W. Hu et al. / Ecological Modelling 107 (1998) 171188

183

Fig. 5. Observed and calculated concentration of phosphorus.

slowly, but there were disturbances of the observed value of PO34 -P. The deviation was caused by the
difficulty of measuring very low concentrations of PO34 -P in the channel. However, the calculated value
is consistent with the observed average value PO34 -P of the experiment. In Fig. 6 we show curves of
phytoplankton concentration. Although the calculated values are not consistent with the observed values,
they show the same trend. Therefore, it can be said that the model is consistent with the reality.

4.3. Sensiti6ity analysis of the model

The results of the model sensitivity analysis are shown in Table 5. It is found that state variable P was
most sensitive to the retention time and then to parameter VmNHwh; state variable NO was most sensitive
to the retention time, then to VmNOwh, and then to VmNHwh; state variable P was sensitive most to Vmpp,

Fig. 6. Observed and calculated concentration of phytoplankton.

W. Hu et al. / Ecological Modelling 107 (1998) 171188

184

Table 5
Results (in %) of the sensitivity analysis of the model by 9 25 change
Parameter

NH

NO

Bph

Bwh

Nwh

Pwh

VmNHwh

B0.1

B0.1

B0.1

B0.1

B0.1

B0.1

B0.1

B0.1

VmNHph
VmNOph
VmPph

B0.1
B0.1
B0.1

B0.1
B0.1
B0.1

B0.1
B0.1
B0.1

B0.1
B0.1
B0.1

B0.1
B0.1
B0.1

+0.75
0.42
B0.1
B0.1
B0.1

Ret

11.12
+12.50
B0.1

7.08
10.5
B0.1

B0.1

B0.30

B0.1

Vmpp

9.00
10.17
B0.1

97.10
+51.16
B0.1
B0.1
0.64
+0.74
B0.1

1.88
+1.87
+2.10
2.45
B0.1

B0.1

VmPwh

1.62
+1.95
11.3
+9.30
B0.1

B0.1

VmNOwh

3.46
+3.58
B0.1

B0.1

B0.1

Vrex

B0.1

B0.1

B0.1

B0.1

B0.1

10.95
9.93
+0.93
0.60

+68.06
100.0
+6.70
6.60

B0.1

i.e. to the diffusion coefficient of phosphorus from pore water, then to VmPwh, then to Vrex, and then to
VmPph; phytoplankton density was only sensitive to retention time; water hyacinth biomass was not
sensitive to these parameters, but Pwh was sensitive to Vmpp and the increment would increase versus time.

5. Discussion
The results show that the model can simulate well the reality of the experiment, so the model can be
used as a guide for the design and management of physico-biological engineering for purifying lake water
using water hyacinths. For example, the results of the model sensitivity analysis show that retention time
has an important influence on the nutrient nitrogen concentration. So, increasing retention time by 25%
can improve the purifying capacity of engineering for nitrogen by 10% and can cut down phytoplankton
density by 10%. From Table 5, the concentration of phosphorus can be lowered by decrease of the release
from sediment. It is found that the nutrient concentrations are sensitive to the nutrient uptake velocities
of water hyacinths, but not to that of phytoplankton, which indicates that the growth of water hyacinths
determines the water quality. Thus, the big problems which should be solved are how to culture and
harvest water hyacinths and how many water hyacinths should be left on the surface of water.
From a long-term point of view, the purifying effect of the ecosystem should be assessed by the net
uptake of water hyacinths for nutrients. The net uptake is the nutrient uptaken by water hyacinths minus
the nutrient entering water due to mortality. Fig. 7 shows the net uptake curves of water hyacinths for
nutrient nitrogen and phosphorus versus the starting density of water hyacinths. The data which were
used to draw the figure are from the calculation of the model. It shows the net uptake increase, according
to the increase of the starting density, to its maximum and then beginning to decrease as the starting
density increases. The optimum starting density of fresh water hyacinths for nitrogen is 9.6 kg/m2, the
optimum starting density for phosphorus is 5.12 kg/m2. Because nitrogen pollution is more serious than
phosphorus pollution, the optimum beginning density of fresh water hyacinths is 9.6 kg/m2. Fig. 8 shows
the net nutrient uptake curves of water hyacinths versus the density at which they are harvested. From the
figure, the optimum fresh water hyacinth density for nitrogen is 12.24 kg/m2, and the optimum density for

W. Hu et al. / Ecological Modelling 107 (1998) 171188

185

Fig. 7. Influence of initial density of water hyacinths on net nutrient uptake.

phosphorus is 11.4 kg/m2. These two densities are close, so the optimum density for start of harvesting
can be considered as the average value of the two: 11.82 kg/m2.
Figs. 912 are the curves of phytoplankton, ammonium nitrogen, nitrate nitrogen, phosphorus versus
time, with the outside water being filtered by a filter cloth as it enters the channel. From these figures, it
is found that the filtering action improved the purifying effect greatly, except for phosphorus which is
controlled by release from the bottom; the phytoplankton, especially, is cut almost to zero.

Fig. 8. Influence of harvesting density of water hyacinths on net nutrient uptake.

186

W. Hu et al. / Ecological Modelling 107 (1998) 171188

Fig. 9. Comparison of concentration of ammonium nitrogen.

6. Conclusion
An ecological model for experiments proposed for lake water purification using water hyacinths is
presented. The calibration and validation show that the results of the model are consistent with the
reality. From the model, it is found that P is sensitive to Vmpp, Vrex; NH to Ret, VmNHwh; NO to
VmNHwh, VmNOwh and Ret; phytoplankton to Ret. The optimum initial fresh water hyacinth density is
9.6 kg/m2. The optimum fresh water hyacinth density for start of harvesting is 11.82 kg/m2. The filtering
measures can cut down phytoplankton density greatly and thus improve the water quality in the channel.

Fig. 10. Comparison of concentration of phytoplankton.

W. Hu et al. / Ecological Modelling 107 (1998) 171188

187

Fig. 11. Comparison of concentration of nitrate nitrogen.

Fig. 12. Comparison of concentration of phosphorus.

Acknowledgements
This work is a part of China Denmark joint study on the biological engineering for removing
phytoplankton and improving water quality in Lake Taihu supported by EU. The authors are very
grateful to Professor S.E. Jrgensen and Dr. S. Nors-Nilsen of the Royal Danish School of Pharmacy for
their constant support and valuable suggestions and other help. The authors thank Wang Guoqiang, Hu
Chunhua and Li Wangchun of Nanjing Institute of Geography & Limnology for their help doing the
experiment.

188

W. Hu et al. / Ecological Modelling 107 (1998) 171188

References
Bendoricchio, G., et al., 1993. Modeling the photosynthetic efficiency for Ul6a rigida growth. Ecol. Model. 67, 221 232.
Bendoricchio, G., et al., 1994. A trophic model for Ul6a rigida in the Lagoon of Venice. Ecol. Model. 75/76, 485 496.
Cai, Q.-H., 1995. A Mathematical Model for Ecosystem Dynamics of the Donghu Lake, Studies on the ecosystem of Donghu Lake,
Science Press, China, 470483.
Dai, Q., Chen, Y., Pi, Yu., 1991. Study on the accumulation amount of silver in waster water and its utilization by water hyacinth.
Chin. J. Appl. Ecol. 2 (2), 159167.
Dou, H.S., Pu, P.M., Zh, Sh.Zh., Hu, W.P., Pang, Y., 1995. An experimental study on culture of Eichhornia crassipes (Mart.) Solms
on open area of Taihu Lake. J. Plant Resources Environ. 4 (1), 54 60.
Hu, W.-P., Pu, P.-M., 1995. Study on the ecological efficiency of soft enclosure in lake water. In: Proceedings of 6th International
Conference on the Conservation and Management of Lakes, Kasumigaura 95, vol. 3, pp. 1886 1889.
Jin, S.-D., et al., 1994. Uptake by potamogeton crispus of nitrogen and phosphorus from water and some affecting factors. Acta
Ecol. Sinica 14 (2), 168173.
Jrgensen, S.E., 1988. Fundamentals of Ecological Modeling. Elsevier, Amsterdam.
Jrgensen, S.E., Nielsen, S.N., Jrgensen, L.A., 1991. Handbook of Ecological Parameters and Ecotoxicology. Elsevier, Amsterdam.
Kamp-Nielsen, L., 1974. Mud-water exchange of phospate and other ions in undisturbed sediment cores and factors affecting the
exchange rates. Arch. Hydrobiol. 73 (2), 218237.
Li, W.-CH., 1996a. Biological and environmental succession in Wuli bay of Taihu Lake along with the eutrophication processes. J.
Lake Sci. 8 Suppl., 3745.
Li, W.-CH., 1996b. On the adaptiveness of aquatic macrophyte to the sediment in Wuli Lake. J. Lake Sci. 8 suppl., 30 36.
Nanjing Institute of Geography & Limnology, Chinese Academy of Science, 1990, Control of the total amounts of main kinds of
pollutants entering the Lake Taihu from the main rivers, Research report.
Pu, P.-M., Hu, W.-P., 1996. Key strategies for controlling eutrophication of large shallow lakes in China. In: 28th International
Geographical Congress, 1996, Utrecht.
Pu, P.-M., et al., 1993. Physico-ecological engineering for improving Taihu Lake water quality in intake area of Mashan Drinking
Water Plant. J. Lake Sci. 5 (2), 171180.
Pu, P.-M. et al., 1995. The high-efficient artificial ecosystem in local water bodies and biological enterprises for environment
conservation, their roles in eutrophication control for Taihu Lake, China. In: Proceedings of 6th International Conference on the
Conservation and Management of Lakes, Kasumigaura 95. Vol. 3, pp. 1882 1885.
Sun, Sh.-C., Huang, Y.-P. et al., 1993. Taihu Lake. Chinese Ocean Press, pp. 1 19.
Wu, Zh.-B., Qiu, Ch.-Q., Xia, Y.-Ch., Wang, D.-M., 1987a. A study on the purifying the Yanshan petroleum chemical industry
waster water by using Eichhornia crassipes (Mart.) Solms, I. Dynamic simulation experiment. Acta Hydrobiol. Sinica 11 (2),
139 150.
Wu, Zh.-B., Qiu, Ch.-Q., Xia, Y.-Ch., Wang, D.-M., 1987b. A study on the purifying the Yanshan petroleum chemical industry
waster water by using Eichhornia crassipes (Mart.) Solms, II. Stable purifying experiment. Acta Hydrobiol. Sinica 11 (4),
299 308.
Yan, J.S., 1986. The main principles and types of ecological engineering for the conversion of waste water into resource. Rural
Eco-Environ. 4, 19 23.
Zhang, Y.-Q., 1989. Effect of the carrying capacity of Eichhornia crassipes on its utilization and purification function. Rural
Eco-Environ. 1, 40 43.