You are on page 1of 14

Ecological Modelling 144 (2001) 231– 244

www.elsevier.com/locate/ecolmodel

A GIS-based method of lake eutrophication assessment


Fu-Liu Xu *, Shu Tao, R.W. Dawson, Beng-Gang Li
Department of Urban and En6ironmental Sciences, MOE Laboratory for Earth Surface Process, Peking Uni6ersity,
Beijing 100871, People’s Republic of China

Received 18 September 2000; received in revised form 30 March 2001; accepted 7 May 2001

Abstract

A geographical information system (GIS)-based method of lake eutrophication assessment was undertaken to study
the spatial distribution of eutrophication conditions in lake environments. A trophic state index (TSI) consisting of
six physical, chemical and biological indicators including total phosphorus (TP), total nitrogen (TN), chemical oxygen
demand (COD), Secchi disk depth (SD), chlorophyll-a concentration (Chl-a) and phytoplankton biomass (CA) was
constructed to describe the eutrophication state of the lake environment. A 0 – 100 eutrophication scale was also
developed to indicate seven different trophic levels within the lake environment: oligotrophic, lower-mesotrophic,
mesotrophic, upper-mesotrophic, eutrophic, hypereutrophic and extremely hypereutrophic. A representation of the
spatial distribution of TSITP, TSITN, TSICOD, TSISD, TSIChl-a and TSICA was developed using the inverse distance
weighted (IDW) interpolation method. By categorizing the interpolated values, a clear illustration of the different
trophic levels was developed on six thematic maps. A GIS overlay technique was applied to synthesize the
information from the six thematic maps into a final map illustrating the spatial distribution of eutrophication
conditions within the study area. The different periods or levels associated with lake eutrophication assessment using
GIS were then discussed. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Eutrophication; Spatial distribution; Trophic state index; Thematic map; Overlay; Geographical information system
(GIS); Lake Chao

1. Introduction vided into abiotic and biotic aspects. Among the


abiotic parameters, plant nutrients (phosphate,
Since the 1960 – 70s, a number of attempts nitrate), oxygen demanded (BOD, COD) and
have been made to quantitatively evaluate the transparency were usually used to assess lake
trophic levels (e.g. Vollenweider, 1968, 1976;
trophic state of lakes using single-variable
Dobson et al., 1974; Schindler, 1977; Chapra,
trophic indices or multi-parameter approaches.
1980; Ritter, 1981; Gregor and Rast, 1982;
The single-variable trophic indices can be di-
Persson and Jansson, 1988; Thornton and Rast,
1988; Alvarez Cobelas et al., 1992; Boers et al.,
1993). Also, biotic parameters were often em-
* Corresponding author. Tel./fax: + 86-10-6275-1938. ployed to assess lake trophic conditions, given
E-mail address: xufl@urban.pku.edu.cn (F.-L. Xu). the sensitivity of aquatic organisms, especially

0304-3800/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 3 8 0 0 ( 0 1 ) 0 0 3 7 4 - X
232 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

algae and macro-invertebrates, to eutrophication for quantitative studies of the mechanisms be-
processes (e.g. Rosenberg and Resh, 1993; Whit- hind eutrophication. This effectively eliminates
ton and Kelly, 1995; Brodersen et al., 1998; the subjective labeling associated with the use of
Guntzel and Rocha, 1998; Canosa and Pinilla, oligotrophic, mesotrophic and eutrophic states
1999; Danilov and Ekelund, 2000). Phytoplank- as indicators. The trophic state index (TSI)
ton, both in running waters and lakes, turned based on several biological, chemical and physi-
out to be a reliable environmental tool when cal indicators, especially the Carlson-type TSI
estimating different levels of eutrophication (e.g. Carlson, 1977; Walker, 1979; Porcella et
(Tate, 1990; Paul et al., 1991; Proulx et al., al., 1980; Aizaki et al., 1981; Jin et al., 1990;
1996; Rolland et al., 1997; Watson et al., 1997; Swanson, 1998) offers the most suitable and ac-
Smoot et al., 1998). With the exception of ceptable method for evaluating lake
chlorophyll-a concentrations, phytoplankton cell eutrophication.
number, species number and biomass, some Mathematical methods play a very important
form of index, e.g. Hurlbert’s, Margalef’s, Men- role in lake eutrophication assessment in terms
hinick’s, Shannon’s, Simpson’s, McNaughton’s, of parameters chosen, weighting factor calcula-
has always been used in assessing eutrophication tion, and sample classification. Exploratory
conditions in aquatic environments (e.g. Wash- statistical regression analysis has been used to
ington 1984; Ludwing and Reynolds, 1988; investigate relationships between the related
Whitton and Kelly, 1995; Danilov and Ekelund, parameters and eutrophication levels (e.g. Vol-
1999). However, these relatively simple single- lenweider, 1968, 1976; Shannon, 1970; Shannon
variable trophic state criteria represent subjective and Brezonik, 1972a; Dillon and Riegler, 1974;
judgements, and may be limited spatially (e.g. Nicholls and Dillon, 1978; Rast and Lee, 1978;
Therriault and Platt, 1978; Reckhow and Chapra and Reckhow, 1979, 1983; Ciecka et al.,
Chapra, 1983; Powell et al., 1989; Boyle et al., 1980; Baker et al., 1981; OECD, 1982; Ahlgren
1990; Whitton and Kelly, 1995; Karydis and et al., 1988). Further, cluster analysis (e.g. Shan-
Tsirtsis, 1996; Danilov and Ekelund, 1999). Fur- non, 1970; Shannon and Brezonik, 1972a,b; Sch-
ther, the use of descriptive classifications for ernewski and Schulz, 1999; Danilov and
lake trophic states such as oligotrophic, Ekelund, 2000), fuzzy analysis (e.g. Cai, 1988;
mesotrophic, eutrophic and so on, could create Cao, 1991; Li et al., 1991), principal component
difficulties when attempting to describe continu- analysis (e.g. Li et al., 1990; Zitko, 1994), and
ous changes in a lake’s trophic state or in study- artificial neural networks (Cai et al., 1995; Li,
ing quantitatively the eutrophication mechanism 1995; Lu and Zhu, 1998), have proven to be
(Shannon and Brezonik, 1972b; Carlson, 1977; powerful tools in lake eutrophication assess-
Yoshimi, 1987). ment. Another important attempt at a multi-
The multidimensional nature of the eutrophi- parametric classification of trophic conditions
cation phenomenon means that no single vari- was undertaken by Zurlini (1996) by combining
able is representative of the eutrophication the exact probabilities from OECD (OECD,
status of a given water body (Shannon and Bre- 1982) of the log normal frequency distributions
zonik, 1972b; Carlson, 1977; Cruzado, 1987). of chlorophyll, nitrogen and phosphorus concen-
More robust trophic state criteria or indices us- trations as well as Secchi depths. Additionally,
ing multivariate approaches have been proposed several other researchers applied remote sensing
by a number of investigators (e.g., Shannon, technologies to the lake trophic classification
1970; Shannon and Brezonik, 1972a,b; Carlson, and assessment process (e.g. Borton et al., 1975;
1977; Walker, 1979; Porcella et al., 1980; Ritter, Novo and Godoy, 1989; Baban, 1996).
1981; Swanson, 1998). The contributions of The primary disadvantage of such methodolo-
Carlson, Walker and Porcella et al. offer a 0– gies, however, has to do with the spatial discon-
100 scale providing continuous numerical classes tinuity of their data, since their results were
of lake trophic states and a rigorous foundation based on fragmentary data collected during sam-
F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244 233

pling surveys. They do not provide an explicit resulting spatial distributions of lake eutrophica-
view of the trophic status of lake eutrophication tion conditions.
nor any clear-cut information related to water
quality for lake researchers and managers. Lake
eutrophication assessment requires not only a 2. Materials and methods
large number of variables, but also a spatial dis-
tribution of eutrophication levels based on each 2.1. Study area and data collection
of the variables. The spatial assessment, how-
ever, of eutrophication levels may become quite Lake Chao is located in southeastern China
complicated, since the function and dynamics of (Fig. 1A). It has a mean surface area of 760
each parameter may lead to different eutrophi- km2, a mean depth of 3.06 m, and a mean re-
cation trends (e.g. Therriault and Platt, 1978; tention time of 136 days. Lake Chao is one of
Powell et al., 1989; Kitsiou and Karydis, 1998). the five largest fresh-water lakes in China, and
There seems to exist, then, a need for appropri- the most eutrophic. Prior to the 1950s, the lake
ate methodologies and tools capable of synthe- was well known for its scenic beauty and rich-
sizing spatially the eutrophication trends ness of aquatic life. Since that time, however,
presented by various parameters. By synthesizing the lake has suffered from serious eutrophica-
such trends, a final thematic map illustrating the tion. Increasing pressures from population
spatial distribution of eutrophication conditions growth and economic development in the
can be created. Fortunately, this can easily be drainage area are primarily responsible for the
performed through use of a geographic informa- lake’s current eutrophic state. The present con-
tion system (GIS). ditions have had negative ecological, health, so-
GIS can be used to perform a number of fun- cial, and economic effects on the lake and its
damental spatial analysis operations. Its major utilization. Lake Chao was selected some time
advantage is that it allows the user to identify ago for inclusion in a nation-wide study of lake
the spatial relationships between various map eutrophication conditions and processes. Ac-
features. More precisely, overlay techniques al- cordingly, comprehensive research on Lake
lows the synthesis of different map layers, based Chao has been carried out for more than 20
on a database where the information is stored years (see Tu et al., 1990; Wang et al., 1995;
as a whole. Comparisons, as well as further Xu, 1994, 1996, 1997; Xu et al., 1999a,b,c).
analyses, among and between both variables and Measurement and sampling of the lake’s wa-
layers can then be easily performed (GIS, 1994). ter for analytical purposes were performed
There is still much to do regarding the integra- monthly from April 1987 to March 1988. The
tion of environmental assessment and modeling water samples were collected using a van Dorhn
with GIS (Fedra and Goodchild, 1993). How- sampler from 34 stations (Fig. 1B) at a depth of
ever, there are some successful examples of ap- 0.5-m. The parameters chosen for measurement
plications of GIS to environmental modeling included both physic-chemical parameters (SD,
and assessment, e.g. modeling the effects of cli- pH, TN, TP, Si, COD, BOD, DO, etc.) and
mate change on water resources, surface water biological parameters (chlorophyll-a concentra-
flow analysis, groundwater flow analysis, urban tion (Chl-a), biomass concentration and the dry
air pollution regimens, regional fish species rich- weight of both phytoplankton and zooplankton,
ness modeling, community modeling, population and the number of phytoplankton cells). Ther-
and landscape ecology, and natural resource mal profiles were taken every 3 h at various
management (see Fedra and Goodchild, 1993 points in the lake by means of a thermograph.
for details). A weather station located on the site provided
The purpose of the present work is to inte- 3-h measurements of atmospheric pressure, wind
grate GIS techniques into the lake eutrophica- speed and direction, air temperature, humidity
tion assessment process and then to study the and solar radiation.
234 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

The flow chart of the GIS-based method for the total phosphorus (TP, in mg/l of P), total nitrogen
lake eutrophication assessment is shown in Fig. 2. (TN, in mg/l of N), chemical oxygen demand
(COD, in mg/l), Secchi disk depth (SD, in m),
2.2. Assessment indicators and their chlorophyll-a concentration (Chl-a, in mg/m3), and
eutrophication scales phytoplankton biomass (CA, as C in mg/m3) (see
Table 1). The 0–100 scale was divided into ranges,
Consulting Carlson (1977) and Porcella et al. each representing a particular trophic state: 0–30
(1980), a trophic state index (TSI) on a scale from representing oligotrophic, 30–40 lower-mesotro-
0 to 100 was constructed. The TSIs was based on phic, 40–50 mesotrophic, 50–60 upper-mesotro-

Fig. 1. Geographical location (A) and distribution of sampling points (B) of Lake Chao.
F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244 235

Fig. 2. The flow chart for the GIS-based method for lake eutrophication assessment.

Table 1
The scale of the trophic state index (TSI) and the evaluation standards for Lake Chaoa

TSI Eutrophication level TP (mg/l) TN (mg/l) COD (mg/l) SD (m) Chl-a (mg/m3) CA (mg/m3)

0 Oligotrophic 0.0004 0.010 0.06 48 0 /


10 0.0009 0.070 0.12 27 0.10 B50
20 0.0020 0.150 0.24 15 0.26 50
30 0.0046 0.300 0.48 8.0 0.66 100
40 Lower-mesotrophic 0.0100 0.600 0.96 4.4 1.60 150
50 Mesotrophic 0.0230 1.000 1.80 2.4 4.10 200
60 Upper-mesotrophic 0.0500 1.500 3.60 1.3 10.0 250
70 Eutrophic 0.1100 2.000 7.10 0.73 20.0 300
80 Hypereutrophic 0.2500 3.000 14.0 0.40 40.0 500
90 Extremely 0.5500 4.600 27.0 0.22 100 800
hypereutrophic
100 1.2000 10.00 54.0 0.12 200 \800

a
Consulting the trophic state index and evaluation standards for lake eutrophication of OECD (1982) and Japan National
Environmental Institute (Aizaki et al., 1981), and for Lake Tai eutrophication in China (Jin et al., 1990). TP, total phosphorus; TN,
total nitrogen; COD, chemical oxygen demand; SD, Secchi disk depth; Chl-a, chlorophyll-a concentration; CA, phytoplankton
biomass.

phic, 60– 70 eutrophic, 70– 80 hypereutrophic, the OECD (1982) and the Japanese National
and 80 –100 the extremely hypereutrophic. The Environmental Institute (Aizaki et al., 1981),
assessment standards for each indicator were and those used in assessing the eutrophication
based on those constructed for the Evaluation of Lake Tai in China (Jin et al., 1990) (see
Standards for Lake Eutrophication designed for Table 1).
236 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

2.3. Calculation of eutrophication le6els and tion map. The following steps describe the synthe-
generation of thematic maps sizing procedure used to analyze the six thematic
maps:
The following expression was used to calculate 1. Development of an trophic state index (TSI)
the lake eutrophication levels for each of the scale from 0 to 100 to label the different
indicators: trophic levels (see Table 1).
2. Application of this ordinal scale to all the
TSIi = (TSIk − 1 +((Ci −Si,k − 1)/(Si,k −Si,k − 1)) pixels/cells on each thematic map; each pixel
*(TSIk − TSIk − 1)) (1) was assigned a value from 0 to 100 based on a
comparison between its initial value and the
where Ci is the measured concentration of the i-th eutrophication scales in Table 1.
indicator (i=TP, TN, COD, SD, Chl-a and CA), 3. Analysis of the six thematic maps on a cell-by-
TSIk and TSIk − 1 are the k-th and (k −1)-th scales cell basis. As a result, a final map illustrating
of the i-th indicator, Si,k and Si,k − 1 are the evalu- the spatial distribution of eutrophication levels
ation standards of k-th and (k − 1)-th scales of the was produced. The following expression was
i-th indicator (see Table 1). used in the overlay operation to produce the
The inverse distance weighted (IDW) interpola- TSI values:
tion method (Lam, 1983) with a spatial resolution
of 500×500 m and ArcView Version 3.1 (ArcView * WTP + TSITN
TSI = (TSITP * WTN + TSICOD
* WCOD
3.1, ESRI, Inc.) were used to generate the six
+ TSISD
* WSD + TSIChl-a
* WChl-a
thematic maps indicating the spatial distribution of
eutrophication levels based on each indicator. The + TSI*
CAWCA) (3)
IDW interpolation method is based on the principle
of assigning higher weights to data points closest where TSITP, TSITN, TSICOD, TSISD, TSIChl-a and
to an unvisited point relative to those which are TSICA are the eutrophication levels for TP, TN,
further away (Weber and Englund, 1992, 1994). In COD, SD, Chl-a and CA on the six thematic
other words, the assigned weight is a function of layers; WTP, WTN, WCOD, WSD, WChl-a and WCA
inverse distance as represented in the following are the weighting factors for each indicator (as-
formula (Lam, 1983): sumed as 1/6 for each indicator in the operation).

 N
n 
f(x, y)= % w(di )zi } % w(di )
N
n (2) 3. Results
i=1 i=1

where f(x, y) is the interpolated value at point (x, y); The thematic maps of TSITP, TSITN, TSICOD,
w(di ) is the weighting function; zi is the data value TSISD, TSIChl-a and TSICA developed using the IDW
at point i; and di is the distance from point (x, y). interpolation method and the eutrophication scales,
The interpolated values of any point within the are illustrated in Fig. 3a–f, respectively. The final
dataset are bounded by min(zi ) B f(x, y) B max(zi ), TSI map developed as a result of the overlay
as long as w(di )\ 0 (Lam, 1983). The IDW inter- technique is given in Fig. 4.
polation method has been widely used on many Fig. 3a shows the spatial distribution of TSITP.
types of data because of its simplicity in principle, From it, one can see that the western part of Lake
speed in calculation, easiness in programming, and Chao is characterized mainly as severely eutrophic
credibility in interpolating surfaces (Lam, 1983). (TSITP 70–80), while the eutrophic field (TSITP
60–70) is distributed in the eastern part of the lake.
2.4. The o6erlay of the thematic maps The eutrophication levels near the river mouths are
representative of eutrophic to extremely hyper-
The overlay technique, widely used in GIS appli- eutrophic (TSITP 60–90) conditions, while the up-
cations (GIS, 1994), was applied to synthesize the per-mesotrophic and eutrophic (TSITP 50–70)
six thematic maps and develop the final eutrophica- conditions can be observed in the open water areas.
F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244 237

Fig. 3b illustrates the spatial distribution of eastern parts, near the river mouths. Most of the
TSITN. The hypereutrophic field (TSITN 70 – 80) is study area can be characterized as eutrophic
mainly distributed in the northwestern and north- (TSITN 60–70) centered in the southwestern and

Fig. 3. Spatial distribution of the lake trophic state index based on each indicator: (a) TSITP; (b)TSITN; (c)TSICOD; (d)TSISD;
(e)TSIChl-a; (f) TSICA.
238 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

Fig. 3. (Continued)

southeastern parts and upper-mesotrophic (TSITN is extended from the central to the southeastern
50–60) in the central part. part of the lake, with most of the remainder of the
Fig. 3c shows the TSICOD spatial distribution. study area being characterized as eutrophic
Here, the upper-mesotrophic field (TSICOD 50 – 60) (TSICOD 60–70).
F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244 239

Fig. 3d reveals the spatial distribution of TSISD. areas in the northwestern part of the lake. The
The extremely hypereutrophic field (TSISD 80 – hypereutrophic field (TSICA 70–80) is only found
100) covers the entire lake. The eutrophication near the mouth of the Zhegaohe River in the
level near the river mouths in the western part of eastern part of the lake.
the lake is higher (TSISD 90 – 100) than that found Finally, the overall TSI spatial distribution is
elsewhere in the study area (TSISD 80 – 90). illustrated in Fig. 4. Eutrophic conditions (TSI
Fig. 3e demonstrates the spatial distribution of 60–70) cover most of the study area, while the
TSIChl-a. The central and southwestern parts of hypereutrophic conditions (TSI 70–80) are dis-
the lake are characterized by mesotrophic and tributed mainly near the river mouths in the
lower-mesotrophic fields (TSIChl-a 30 – 50), with a northwestern and northeastern parts of the lake.
strong mesotrophic field (TSIChl-a 40 – 50) domi- A very limited area in the central part of the study
nance. The northwestern and eastern parts of the area is characterized as upper-mesotrophic (TSI
study area are characterized as primarily upper- 50–60).
mesotrophic (TSIChl-a 50 – 60), with the eutrophic
field (TSIChl-a 60–70) extending mainly near the
river mouths in the same area. The severely eu- 4. Discussions
trophic (TSIChl-a 70– 80) and the extremely hyper-
eutrophic fields (TSIChl-a 80 – 90) are limited to The spatial distribution of Lake Chao’s eu-
small areas in the central and eastern parts of the trophication levels derived from this study is
lake. closely correlated with the actual conditions of the
Fig. 3f indicates the spatial distribution of lake. The northwestern and northeastern parts of
TSICA. The mesotrophic field (TSICA 40 – 50) ex- the lake, especially the river mouths near the
tends from the southwestern to central parts of Nanfeihe and Zhegaohe Rivers, receive much
the lake, while the lower-mesotrophic field (TSICA more wastewater because of their nearness to
30–40) is limited to small areas in the same Hefei City, the capital of Anhui Province, and
region. The upper-mesotrophic (TSICA 50 – 60) Chaohu City, the second largest city in the lake’s
covers most of the northwestern and the eastern watershed (see Fig. 1A). The nutrient contents
parts of the study area. The eutrophic range which are primarily responsible for eutrophication
(TSICA 60–70) is found principally in the eastern (Rast and Holland, 1988; Ryding and Rast, 1989;
part of the lake and in limited number of small Cooke et al., 1993) both in the lake’s water and

Fig. 4. Final map for the spatial distribution of the lake eutrophication based on overlay technique
240 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

sediments, were far higher in these two regions ing a complete picture of the trophic state of a
than anywhere else in the lake (see Fig. 3a,b and lake. A GIS-based method to spatially assess lake
Xu et al., 2000). The concentration of Chl-a and eutrophication states was proposed in this paper.
phytoplankton biomass, two of the more obvious GIS technology is mainly adequate for distributed
symptoms of eutrophication (Rast and Holland, data which are within the resolution of the GIS
1988; Ryding and Rast, 1989; Cooke et al., 1993), grid, e.g. land use, land cover (both can be de-
both followed similar trends in spatial distribution duced from sattelite inages), DEM, etc. The ap-
(see Fig. 3e,f). proach in this paper is not the same— the authors
Lake eutrophication, however, cannot be sim- used rather limited number of observation points
ply evaluated by a single physical, chemical, or and a special technique (IDW) to calculate the
biological parameter. It is a multidimensional fea- GIS mesh values of variables. The methodology
ture (Shannon and Brezonik, 1972b; Carlson, could have three different approaches depending
1977; Cruzado, 1987). These single indicators on integration style and date exchange methodol-
cover different aspects of the lake eutrophication ogy between the GIS approach and the TSI calcu-
phenomenon. It is necessary, therefore, to apply lations (see Fig. 5). The first is shown in Fig. 5a.
several indicators simultaneously (including physi- Here, both the GIS and the TSI calculations
cal, chemical, and biological) to derive a more belong to two separate systems, each having their
complete lake eutrophication assessment. Only own identifiable user interface. The GIS is used
through such a multidimensional approach can simply to display the results of the TSI calcula-
one capture all the features needed to yield a fully tions and their integration is limited to file ex-
informative assessment of the eutrophic condition change activity. The second approach is
of a lake. Unfortunately, the sensitivity of a sin- illustrated in Fig. 5b. Here, the GIS and the TSI
gle-parameter and/or its weighting factor may be calculations still belong to two separate systems.
different in different lakes depending on the However, they possess a common user interface
parameters used in the TSI (e.g. Therriault and that is used to activate the GIS and TSI calcula-
Platt, 1978; Reckhow and Chapra, 1983; Powell et tions as a single system, in addition to managing
al., 1989; Boyle et al., 1990; Whitton and Kelly, common data and file exchanges. This integrating
1995; Karydis and Tsirtsis, 1996; Danilov and style reduces errors associated with both data and
Ekelund, 1999). The choice of suitable indicators file exchanges between separate systems. The third
and their weights is critically important in the approach is presented in Fig. 5c. The GIS and the
eutrophication assessment of a specific lake. TSI calculations are now completely integrated as
Hooper (1969) identified four important criteria a Decision-Making Support System for Lake Eu-
useful in the development of eutrophication in- trophication Assessment (DMSSLEA), having a
dices: (1) it should discriminate between changes uniform user interface and data sharing. In this
associated with nutrient level and those associated approach, the TSI calculations are treated as an
with other categories of environmental change; (2) analysis function of the integrating system. The
it should have considerable sensitivity to levels of calculations are programmed using a specific GIS
enrichment; (3) it should have properties which programming language, e.g. Avenue in ArcView,
are widespread geographically and short-lived ge- SML in Arc/Info PC version, MapBasic in Map-
ologically; and (4) it should be suitable for long- Info, or Genius in GanaMap. Here the GIS man-
term surveillance and monitoring, i.e. indices ages the spatial and attribute data, in addition to
should both document past changes and serve as manipulating and displaying the results of TSI
a predictive function. calculations.
Most of the current research has focused on the The first approach was applied in the case study
identification and selection of suitable indicators presented in this paper. The TSI calculations were
relative to assessing the eutrophic condition of a developed using MS Excel 97, while the genera-
lake. The spatial distribution of eutrophication tion and synthesis of the six thematic maps was
levels, however, is equally important in develop- performed using the IDW interpolation method
F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244 241

Fig. 5. The different periods or levels of GIS-based method for lake eutrophication assessment: (a) the first periods or levels; (b) the
second periods or levels; (c) the third periods or levels.

and overlay technique within the framework of a framework was applied to analyze the informa-
Geographic Information System (ArcView 3.1, tion from the thematic maps in order to develop a
ESRI). The results of the TSI calculations were final map illustrating the spatial distribution of
stored in Dbase format. The map indicating the eutrophication conditions in the lake. Results
boundary and sampling points in the lake served from the study indicate that the boundaries asso-
as the basis for visualizing the resulting spatial ciated with different trophic levels (upper-
data (Fig. 1B). A DMSSLEA completely inte- mesotrophic, eutrophic and hypereutrophic) could
grated with a GIS approach to the TSI calcula- be clearly defined in a final eutrophication map.
tions is currently under development with the This result comprises the principal advantage of
financial support of the Natural Sciences Founda- the proposed methodology when compared with
tion of China (39970121and 40024101). other attempts based on a multi-parametric clas-
sification and assessment of lake trophic trends.
The latter approach does not lead to a clear
5. Conclusions definition of the boundaries associated with differ-
ing trophic levels. As such, the proposed method-
A GIS-based method of lake eutrophication ology could be of special interest to policy makers
assessment was proposed with the purpose of involved in lake management and planning, since
studying the spatial distribution of eutrophication policy makers need a more explicit view of trophic
conditions in the lake environment. The method status and clear information on water quality.
included the integration of GIS methodology into
lake eutrophication assessment using a trophic
state index. The IDW interpolation method was Acknowledgements
used for generating thematic maps indicating the
spatial distribution of each of the Trophic State The authors are greatly indebted to Professor
Indices. An overlay technique within the GIS S.E. Jgensen and Dr S.N. Nielson of the Royal
242 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

Danish School of Pharmacy for their valuable concept to the Great Lakes. In: Loehr, C., Martin, C.S.,
suggestions. The project (39970121 and 40024101) Rast, W. (Eds.), Phosphorus Management Strategies for
Lakes. Ann Arbor Science Publishers, Ann Arbor, MI, pp.
was supported by National Natural Science Foun-
135 – 152.
dation of China, and the State Education Min- Chapra, S.C.&, Reckhow, K., 1979. Expressing the phospho-
istry of China. rus loading concept in probabilistic terms. J. Fish. Res. Bd.
Can. 36, 225 – 229.
Chapra, S.C., Reckhow, K., 1983. Phosphorus Management
Strategies for Lakes, Mechanistic Modelling. Butterworth,
References Woburn, MA.
Ciecka, J., Fabian, R., Merilatt, D., 1980. Eutrophication
Ahlgren, I., Frisk, T., Kamp-Nielsen, L., 1988. Empirical and Measures for Small Lake Water Quality Management.
theoretical models of phosphorus loading, retention and Water Resources Bull. 16 (4), 681 – 689.
concentration vs. lake trophic state. Hydrobiologia 170, Cooke, D.G, Welch, E.B., Peterson, S.A., Newroth, P.R.,
285 – 303. 1993. Restoration and Management of Lakes and Reser-
Aizaki, M., Iwakuma, T., Takamura, N., 1981. Application of voirs, Second Edition. Lewis Publishers, p. 548.
modified Carlson’s trophic state index to Japanese lakes Cruzado, A., 1987. Eutrophication in the pelagic environment
and its relationship to other parameters related to trophic and its assessment. In eutrophication in the Mediterranean
state. Res. Rep. Natl. Inst. Environ. Stud. 23, 13 – 31. Sea: Receiving capacity and monitoring of long term ef-
Alvarez Cobelas, M., Munoz Ruiz, P., Rubio Olmo, A., Prat fects. UNESCO Reports in Marine Science 49, 57 – 66.
i Fornells, N., 1992. Current state of eutrophication in Danilov, R., Ekelund, N.G.A., 1999. The efficiency of seven
spanish inland waters. Eur. Water Pollut. Cont. 2 (5),
diversity and one similarity indices based on phytoplank-
27– 32.
ton data for assessing the level of eutrophication in lakes in
Baban, S.M.J., 1996. Trophic classification and ecosystem
central Sweden. Sci. Total Environ. 234 (1 – 3), 15 – 23.
checking of lakes using remotely sensed information. J. Sci.
Danilov, R., Ekelund, N.G.A., 2000. The use of epiphyton
Hydrol. 41 (6), 939 –958.
Baker, L.A., Brezonik, P.L., Kratzer, C.R., 1981. Nutrient and epilithon data as a base for calculating ecological
Loading-Trophic State Relationships in Florida Lakes. indices in monitoring of eutrophication in lakes in central
Springfield VA 22161. Publication No 56, p. 126. Sweden. Sci. Total Environ. 248, 63 – 70.
Boers, P.C.M., Cappenberg, T.E., Van Raaphorst, W., 1993. Dillon, P.J., Riegler, F.H., 1974. A test of a simple nutrient
Proceeding of the Third International Workshop on Phos- budget model predicting the phosphorus concentration in
phorus in Sediments. Hydrobiologia 253, 17 –376. lake water. J. Fish. Res. Board Can. 31, 1771 – 1778.
Borton, T., Wezernak, C.T., Raney, R.K., 1975. Inland Lakes Dobson, H.F.H., Gilbertson, M., Sly, P.G., 1974. A Summary
Water Quality and Watershed Planning: Remote Sensing and Comparison of Nutrients and Related Water Quality
Technology Applications. The USA National Science in Lake Erie, Ontario, Huron and Superior. J. Fish. Res.
Foundation, Report NSF-RA-E-75-036, p. 188. Bd. Can. 31, 731 – 738.
Boyle, T.P., Smillie, G.M., Anderson, J.D., 1990. Beeson Fedra, K., Goodchild, M.F., 1993. Environmental Modeling
DR.A sensitivity analysis of nine diversity and seven simi- with GIS. Oxford University Press, Oxford, p. 358.
larity indices. Res. J. WPCF 62, 749 –762. GIS by ESRI, 1994. Cell-based modeling with grid. Environ-
Brodersen, K.P., Dall, P.C., Lindegaard, C., 1998. The fauna mental Systems Research Institute Inc., USA.
in the upper stony littoral of Danish lakes: Macroinverte- Gregor, D.J., Rast, W., 1982. Simple Trophic State Classifica-
brates as trophic indicators. Freshwater-Biol. 39 (3), 577 – tion of the Canadian Nearshore Waters of the Great
592. Lakes. Water Res. Bull. 18 (4), 565 – 573.
Cai, Q., 1988. Fuzzy cluster analysis of pollution in Lake
Guntzel, A., Rocha, O., 1998. Relationships between the
Donghu ecosystem. Acta Hydrobiol. Sinica 12 (3), 193 –
zooplankton community and the trophic state of Caconde
198.
Lake, Osorio, RS, Brazil. Iheringia Serie Zool. 84, 65 – 71.
Cai, Y.D., Wang, L., Yao, L.S., 1995. Artificial neural net-
Hooper, F.F., 1969. Eutrophication indices and their relation
work model for lake eutrophication assessment. China
Environ. Sci. 15 (2), 123 –127. to other indices of ecosystem change. In: Eutrophication:
Canosa, A., Pinilla, G., 1999. Bacteriological eutrophication Causes, Consequences, Correctives. USA National
indicators in four Colombian water bodies (South Amer- Academy of Sciences, Washington, D.C, pp. 225 – 235.
ica). Lakes and Reservoirs: Res. Manage. 1 – 2, 23–27. Jin, X., Liu, H., Tu, Q., Zhang, Z., Zhu, X., 1990. Eutrophica-
Cao, P., 1991. Fuzzy analysis for lake eutrophication assess- tion of Lakes in China. China Environmental Sciences
ment. Environ. Sci. 12 (5), 88 –91. Press, Beijing, p. 614.
Carlson, R.E., 1977. A trophic state index for lakes. Limnol. Karydis, M., Tsirtsis, G., 1996. Ecological indices: a biometric
Oceanog. 22 (2), 361 –369. approach for assessing eutrophication levels in the marine
Chapra, S.C., 1980. Application of the phosphorus loading environment. Sci. Total Environ. 186, 209 – 219.
F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244 243

Kitsiou, D., Karydis, M., 1998. Development of categorical Ritter, W.F., 1981. Survey and Classification of Delaware’s
mapping for quantitative assessment of eutrophication. J. Public Lakes. Report prepared for U.S. Environmental
Coast Conserv. 4, 35 –44. Protection Agency, Philadelphia, PA, March, p. 459.
Lam, N.S.N., 1983. Spatial interpolation methods: a review. Rolland, T., Fayolle, S., Cazaubon, A., Pagnetti, S., 1997.
Am Car. Tographer 10 (2), 129 –149. Methodical approach to distribution of epilithic and drift-
Li, Z., 1995. Eutrophication assessment based on B-P artificial ing algae communities in a French subalpine river: infer-
neural network and its effectivity. Acta Sci. Circumstant. ences on water quality assessment. Aquat. Sci. 59, 57 – 73.
15 (2), 186 – 191. Rosenberg, D.M., Resh, V.H., 1993. Freshwater Biomonitor-
Li, Z., Deng, X., Hong, J., 1990. Application of principal ing and Benthic Macroinvertebrates. Chapman & Hall,
component analysis on comparison of lake eutrophication London, p. 325.
assessment. Acta Sci. Circumstant. 10 (3), 311 – 317. Ryding, S.O., Rast, W., 1989. The Control of Eutrophication
of Lakes and Reservoirs. Man and the Biosphere Series,
Li, Z., Deng, X., Zhang, H., Hong, J., Jin, X., 1991. Applica-
vol. 1. UNESCO, Paris, p. 256.
tion of fuzzy degree concept to eutrophication assessment
Schernewski, G., Schulz, U., 1999. Changes of water quality in
of lake. Res. Environ. Sci. 4 (2), 32 –36.
north German lakes between 1983 and 1993: an applica-
Lu, W.X., Zhu, T.C., 1998. Artificial neural network evalua-
tion of cluster analysis. Limnologica 29 (2), 146 – 159.
tion of lake eutrophication. Chinese J. Appl. Ecol. 9 (6),
Schindler, D.W., 1977. Evolution of phosphorus limitation in
645– 650. lakes. Science 195, 260 – 262.
Ludwing, L.A., Reynolds, J.F., 1988. Statistical Ecology. A Shannon, E.E., 1970. Eutrophication-trophic state relation-
Primer on Methods and Computing. Wiley, New York, p. ships in north and central florida lakes. Doctoral disserta-
337. tion for Ph.D., University of Florida, p. 256.
Nicholls, K.H., Dillon, P.J., 1978. An evaluation of phospho- Shannon, E.E., Brezonik, P.L., 1972a. Eutrophication:
rus– chlorophyll – phytoplankton relationships for lakes. Cause—Effect Relationships. Dept. of Environmental En-
Int. Revue Ges. Hydrobiol. 63, 141 – 154. gineering, Florida University, Gainesville, p. 59.
Novo, E.M.L.M., Godoy, M. Jr, 1989. Eutrophication Assess- Shannon, E.E., Brezonik, P.L., 1972b. Eutrophication analy-
ment Through Remote Sensing Techniques, vol. 5. IEEE, sis: a multivariate approach. J. San. Eng. Div. ASCE 98
Piscataway, NJ, USA, pp. 2825 – 2828. (1), 37 – 57.
OECD, 1982. Eutrophication of Waters. Monitoring, Assess- Smoot, J.C., Langworthy, D.E., Levy, M., 1998. Periphyton
ment and Control. OECD, Paris. growth on submerged artificial substrate as a predictor of
Paul, B.J., Duthie, H.C., Taylor, W.D., 1991. Nutrient cycling phytoplankton response to nutrient enrichment. J. Micro-
by biofilms in running waters of different nutrient status. J. biol. Methods 32, 11 – 19.
North Am. Benthol. Soc. 10, 31 –41. Swanson, E.R., 1998. Trophic state index revisited. LakeLine
Persson, G., Jansson, M., 1988. Phosphorus in freshwater 18 (4), 18 – 20.
ecosystems. Hydrobiologia 170, 1 –340. Tate, C.M., 1990. Patterns and controls of nitrogen in tallgrass
Porcella, D.B., Peterson, S.A., Larsen, D.P., 1980. Index to prairie streams. Ecology 71, 2007 – 2018.
evaluate lake restoration. J. Environ. Eng. Div. ASCE 106 Therriault, J.C., Platt, T., 1978. Spatial heterogeneity of phy-
(EE6), 1151 – 1169. toplankton biomass and related factors in the near-surface
Powell, T.M., Cloern, J.E., Huzzey, L.M., 1989. Spatial and waters of an exposed coastal environment. Limnol.
temporal variability in south San Francisco Bay USA.I. Oceanogr. 23, 888 – 899.
Horizontal distributions of salinity suspended sediments Thornton, J.A., Rast, W., 1988. Preliminary observations on
comparative limnology of north and south semi-arid man-
and phytoplankton biomass and productivity. Estuar.
made lakes and possible implications for eutrophication
Coastal Shelf Sci. 28, 583 –597.
assessment. 8 Annual International Symposium on Lake
Proulx, M., Pick, F., Mazumder, A., Hamilton, P., Lean, D.,
and Watershed Management. p. 33.
1996. Experimental evidence for interactive impacts of
Tu, Q.Y., Gu, D.X., Yi, C.Q., Xu, Z.R., Han, G.Z., 1990. The
human activities on lake algal species richness. Oikos 76,
Researches on the Lake Chao Eutrophication. The pub-
191– 195. lisher of University of Science and Technology of China.
Rast, W., Lee, G.F., 1978. Summary of analysis of the North Hefei, pp. 225 (in Chinese).
American (US portion) OECD Eutrophication Project: Vollenweider, R.A., 1968. The scientific basis of lake eutrophi-
Nutrient loading-lake response relationships and trophic cation, with particular reference to phosphorus and nitro-
state indices. Environmental Research Laboratory, Corval- gen as eutrophication factors. Tech. Rep. DAS/DSI/68.27,
lis, Oregon, USA, p. 454. OECD, Paris.
Rast, W., Holland, M., 1988. Eutrophication of lakes and Vollenweider, R.A., 1976. Advances in defining critical loading
reservoirs: a framework for making management decisions. levels for phosphorus in lake eutrophication. Mem. Ist.
Ambio 17, 2 – 12. Ital. Idrobiol. 33, 53 – 83.
Reckhow, K.H., Chapra, S.C., 1983. Engineering Approaches Walker, W.W., 1979. Use of hypolimnetic oxygen depletion
for Lake Management: Dara Analysis and Empirical Mod- rate as a trophic state index for lakes. Water Res. 15 (6),
eling, vol. 1. Butterworth Publishers, Boston, p. 340. 1463 – 1470.
244 F.-L. Xu et al. / Ecological Modelling 144 (2001) 231–244

Wang, S.Y., Jin, C.S., Meng, R.X., Xu, F.L., 1995. Environ- ecosystem. Ecol. Model. 99, 41 – 49.
mental Research for Lake Chao in Anhui Province. In: Jin, Xu, F.-L., Jorgensen, S.E., Tao, S., 1999a. Ecological indica-
X.C. (Ed.), Lakes in China (1). China Ocean Press, China, tors for assessing freshwater ecosystem health. Ecol.
p. 580. Model. 116, 77 – 106.
Washington, H.G., 1984. Diversity, biotic and similarity in- Xu, F.-L., Jorgensen, S.E., Tao, S., Li, B.G., 1999b. Modeling
dices. Water Res. 18 (6), 653 –694. the effects of macrophyte restoration on water quality and
Watson, S., McCauley, E., Downing, J., 1997. Patterns in ecosystem of Lake Chao. Ecol. Model. 117, 239 – 260.
phytoplankton taxonomic composition across temperate Xu, F.-L., Tao, S., Xu, Z.R., 1999c. The restoration of wet-
lakes of different nutrient status. Limnol. Oceanogr. 42, lands and macrophytes in the Lake Chao: possibility and
487– 495. effects. Hydrobiologica 405, 169 – 178.
Weber, D., Englund, E., 1992. Evaluation and comparison of Xu, F.-L., Tao, S., Xu, Z.R., Dawson, R.W., 2000. TN, TP
spatial interpolators. Math. Geol. 24 (4), 381 – 391. and OM in the sediments of a shallow eutrophic Chinese
Weber, D., Englund, E., 1994. Evaluation and comparison of lake: distributions, limiting factors and effects. Biogeo-
spatial. Interpolators II. Math. Geol. 26 (5), 589 – 603. chemistry (accepted).
Whitton, B., Kelly, M., 1995. Use of algae and other plants Yoshimi, H., 1987. Simultaneous construction of single-
for monitoring rivers. Aust. J. Ecol. 20, 45 – 56. parameter and multi-parameter trophic state indices. Wat.
Xu, F.-L., 1994. Scientific decision-making system for environ- Res. 21 (12), 1505 – 1611.
mental management of the Lake Chao Watershed. Envi- Zitko, V., 1994. Principal component analysis in the evalua-
ron. Protect. 21 (5), 36 –39. tion of environmental data. Mar. Pollut. Bull. 28 (12),
Xu, F.-L., 1996. Ecosystem health assessment of Lake Chao, a 718 – 722.
shallow eutrophic Chinese lake. Lakes & Reservoirs: Res. Zurlini, G., 1996. Multi-parametric classification of trophic
Manage. 2, 101 – 109. conditions. The OECD methodology extended combined
Xu, F.-L., 1997. Exergy and structural exergy as ecological probabilities and uncertainties application to the North
indicators for the development state of the Lake Chao Adriatic Sea. Sci. Total Environ. 182, 169 – 185.

You might also like