Академический Документы
Профессиональный Документы
Культура Документы
mODELLInG
ELSEVIER Ecological Modelling 97 (1997) 99-110
Abstract
The reservoir of the Keban Dam in eastern Anatolia is a very deep and large water body subject to significant seasonal
water level fluctuations within a given hydrological year. At the present time, different sections of the reservoir have
different trophic status and the upper portion of the reservoir, in particular, is eutrophic. In this regard, there is a need for the
assessment of alternative strategies to be implemented for phosphorus and thus eutrophication control in this segment of the
reservoir. In this study, mathematical modeling techniques are used to compare various alternatives in terms of their relative
effectiveness for phosphorus control. With this intention, a two dimensional depth-averaged hydrodynamic model and a
three-dimensional compartmental water quality model are coupled to represent the complex water body dynamics prevailing
in the reservoir. After calibration and verification, the coupled model was used to simulate dissolved oxygen and euphotic
chlorophyll-a concentrations under different scenario conditions corresponding to various strategies proposed for phosphorus
control. The modelling effort has yielded valuable information that can be used by decision makers for the evaluation of
different management alternatives including reduction of the phosphorus load coming from point as well as diffuse sources
within the catchment area, use of baffles and construction of pre-dam structures. © 1997 Elsevier Science B.V.
Keywords: Deep and large reservoirs; Hydrodynamic modelling; Water quality modelling; Eutrophication; Phosphorus control
In recent years, governmental agencies including started in order to develop a practical tool that can be
the Turkish Scientific and Technical Research Coun- used for water quality management purposes includ-
cil (TOBiTAK) and the General Directorate of State ing eutrophication control (Mukhallalati, 1994;
Hydraulic Works (DSi) initiated long-term joint re- Soyupak et al., 1994; Yemi§en and Soyupak, 1994).
search programmes to investigate water pollution As a result of these studies, it is concluded that
and eutrophication problems in the KDR. In this different segments of the KDR exhibit different
regard, an extensive field monitoring programme has trophic levels. In particular, the upper segment of the
been carried out for nearly four years. At the same KDR between the towns of Palu and Elaz:~ is classi-
time, mathematical modelling studies have also been fied as eutrophic. The other two main segments
• ,J \ Mnditerrnn.n.."~~'I ,
1
o, ~, ,O,~m " l
Outflow
Uluova~,
Pumping
vals. Temperature, pH, electrical conductivity, light eutrophication indicators (e.g., dissolved oxygen and
intensity and dissolved oxygen (DO) concentrations chlorophyll-a) in time and space by using such sim-
are measured in situ at many points along the depth plistic approaches.
of the water column by using automated field instru- Hence, the modelling approach adopted for the
ments. For the laboratory determination of other KDR is based on coupling a three-dimensional water
quality parameters, water samples are collected at quality model with a two-dimensional hydrodynamic
three different depths (0.2 m from the surface, at the model yielding water levels, flow rates and velocities
level of thermocline and at 1 m above the bottom) 4 required for the mass balance equations of the water
For each sample collected at a specific depth, quality model. For this purpose, a linkage program is
water quality parameters determined in the labora- developed to transfer the discretized outputs of the
tory include acidity, alkalinity and concentrations of hydrodynamic model into the water quality model.
nitrogen (nitrite, nitrate and ammonia), phosphorus
(total and orthophosphate), solids (suspended and
volatile), total iron, magnesium, calcium, potassium, 3.1. Hydrodynamic modelling
sodium, sulphate and chloride. Further, total Kjel-
dahl nitrogen as well as biological (BOD 5) and The depth-averaged hydrodynamic model used for
chemical (COD) oxygen demands are analyzed for the KDR is based on the following continuity equa-
depth-averaged composite samples. For each station, tion:
determination of chlorophyll-a concentrations and
identification of plankton species are carried out on Oqx Oqy
two different composite samples separately collected + +--=S-e (1)
Ox ay at
from the euphotic zone and the sublayer extending
from the bottom to the euphotic zone. In Eq. (1), qx and qe are water discharge compo-
All analyses are being conducted at the DSi labo- nents in longitudinal (x) and lateral (y) directions
ratories in Elazl~ according to the standard methods and t is time. The water surface elevation above still
(Clesceri et al., 1989). Details and results of t h e water is denoted by "0. The source term S represents
water quality surveys are given elsewhere (Yemi~en the rate of water gain (e.g., precipitation) to the
and Soyupak, 1994). water body; whereas, the sink term e represents the
rate of water loss (e.g., evaporation) from the water
body.
3. Mathematical modelling studies In addition to Eq. (1), pertinent momentum rela-
tionships are used in the present study to account for
The KDR is subjected to a wide variation over the effects of numerous other factors including atmo-
both time and space. For instance, the flow rate of spheric pressure gradient, Coriolis' force, tidal force,
Murat fluctuates between 38 and 2000 m3/s, and the wind stress as well as bottom and lateral friction
water level of the reservoir exhibits changes up to 20 (Platzman, 1963; Reid and Bodine, 1968; Heaps,
m in time spans less than four months. Conse- 1969; Cheng and Tung, 1970; Simons, 1971; Abbott
quently, the dynamics of the reservoir is such that et al., 1973; Schuepfer et al., 1988; Jerome, 1990).
major water quality parameters show wide variations In order to account for vertical variation, the
longitudinally, laterally and vertically. Under such water column was divided into eight equal-depth
conditions, simple empirical models relating trophic layers. The vertical water velocity is kept constant
indicators and loading are not applicable. Further, it for each layer; but, the layers respond uniformly (by
is not possible to trace the concentrations of major changing equally) to changes in overall water depth
(increase or decrease).
The hydrodynamic model can predict water veloc-
ities and discharges at different time steps. In the
4 F o r the m o n t h s during w h i c h thermocline w a s not observed, present case, the model was applied to the segment
samples w e r e collected at the transect midpoint. of interest for different time periods during which
S. Soyupak et al. / Ecological Modelling 97 (1997) 99-110 103
water surface elevations and rate of inflows were In Eq. (2), C denotes the concentration of the se-
recorded by the DSi. The details pertinent to the lected constituent, while U, V and W are the water
spacial discretization and numerical solution of the velocities in x, y and z directions. And, ~x, 8y and
hydrodynamic model are given elsewhere (Mukhal- Ez are the turbulent diffusion (or dispersion) coeffi-
lalati, 1994). It should, however, be noted that the cients in x, y and z directions. In terms of the
agreement between the water levels measured by the source-sink terms, Sz is the total rate of kinetic
DSI and those predicted by the model is satisfactory transformations, SL is the point and diffuse source
(Mukhallalati, 1994). As a typical example, the pre- loading rate and SB stands for the boundary loading
dicted daily-averaged velocity field for a specific day rate given in units of (mass/volume-time).
within the simulation period is presented in Fig. 2. Numerical algorithms are available for solving
multi-dimensional transport expressions such as the
3.2• Water quality modelling one given in Eq. (2) above (Shanahan and Harleman,
1984). Unfortunately, such algorithms mostly as-
In terms of water quality modelling, the following sume control elements with constant (time invariant)
general advective-dispersive mass balance equation volumes. In a reservoir like the KDR, however,
is employed for the computation of the transient water levels (and thus control volumes) exhibit sig-
constituent concentrations: nificant temporal changes. In order to take this fact
0C O 0 0 into consideration, the solution technique utilized in
at - ax ( u c ) - T;y ( v c ) - Tz (we) this study is based on a computational grid system
0
+ Ox
(oc) o(O y) (
E~--~x +
~y
ey +
OZ 8 - -
z Oz
having control elements with variable depth and
volume. In this grid system, each control element is
treated as a continuously stirred tank reactor for
computation of constituent concentrations. Transport
+ SK + SL + SB (2) rates of a given water quality constituent, on the
~ : .- - - - : ." , . . ~
3.3. Kineticformulations Where C, C~, CBoo, CNm and Cphyt represent the
respective concentrations of DO, saturated DO, BOD,
ammonia and phytoplankton carbon in mg/l. On the
In water bodies, a myriad of transformation pro-
other hand k 2, k d, k n and k r are the reaeration,
cesses lead to changes in the aqueous concentrations
deoxygenation, nitrification and phytoplankton respi-
of water quality parameters of interest. Based on the
ration rates (1/day) 6. KBoo and Knit are the half-
information given in the pertinent literature, the fol-
saturation constants for oxygen limitation of deoxy-
lowing major processes are included in the water
quality model:
• Respiration, growth, death and settling of phyto-
plankton.
• Changes in phosphorus concentration due to 5 In terms of modelling, however, the present approach is quite
growth, death and settling of phytoplankton as different than classical models such as the WASP since the
well as mineralization of organic phosphorus. hydrodynamic model of WASP and its linkage to the water
• Changes in nitrogen concentration due to growth, quality model are one-dimensional.
6 The reaeration rate (k 2) is taken as the larger of the two
death and settling of phytoplankton as well as
values computed either by the flow-induced calculation methodol-
mineralization of organic nitrogen, nitrification ogy developed by Covar (1976) or the well-known wind-induced
and denitrification. calculation method developed by O'Connor (1983).
S. Soyupak et al. / Ecological Modelling 97 (1997) 99-110 105
DO Concentration ( m g / I ) DO Concsntrotion(mg/I )
6 B I0 12 14 2 4 t~ 8 I0 12 14
0! '1 | II I'111'l' I i , 2 i~ i I I I I I I ~
I0 IO
30 Q 3O
Station No: 6
40
o,.,, B, ,o Run date: 3 0 , 0 6 . 1 9 9 2
. . . . . . 1 . . . . . . . a_o.~_- Bottom
~0 50
DO Concsntratlontmg/I ) DO Concentration ( m g / I )
4 6 8 I0 12 i4 0 2 4 6 8 I0 12
, i , i ~ J l t i~ ~ , , i , , , | im, t ~ l , , ,, I 0 '''1 .... I ~~ ~ ' I J '.~L~' ' ' I' ' ' ' I
I0 ~ Jr'°""
JO
20
I
40 Station No: 6 40 1 Slotion No:6
Run dote : 20.07.1992 Run date ; 19.10.1992
4 Bottom
50 .... Bottom
genation and nitrification 7. The growth rate of phy- 3.4. Calibration and verification
toplankton is Gp w h i l e PNrt3 is the ammonia prefer-
ence factor. The temperature correction coefficients The adopted mathematical modelling approach
of deoxygenation, nitrification, sediment oxygen de- was subjected to a thorough calibration and verifica-
mand and respiration rates are denoted by 0d, 0nit, 0s tion procedure prior to its use for management pur-
and 0r with T being water temperature in degrees poses. With this intention, the results obtained from
Celsius. a series of calibration and verification runs were
compared with prescribed sets of field measure-
ments. The purpose of executing calibration trials for
7 Deoxygenation and nitrification rates are subject to saturation specified time spans was to find a suitable combina-
kinetics and hence decrease as the DO concentration increases (Di tion of numerical values for the model parameters
Taro and Connolly, 1980; Di Taro and Matystik, 1980; O'Connor, and hence to obtain an acceptable match between
1981; Thomarm and Fitzpatrick, 1982). Further, the model ac-
counts for oxygen production due to reduction of nitrate to
model predictions and field data. The results of
ammonia by phytoplankton when ammonia is not readily available calibration were checked using statistical tools such
for their growth. as constrained regression analysis and root mean
106 S. Soyupak et al. / Ecological Modelling 97 (1997) 99-110
25 25
®
2o z0
o J
o 15
o 1o 1o
=
2
Z5 25
m 20 2o
~ ~--®
o
L. 15
o
!
I0 . 10
o 5
square error (RMSE) measuring goodness of fit verification runs as well as model predictions can be
(Mukhallalati, 1994). The calibrated values of major found elsewhere (Mukhallalati, 1994; Yemi~en and
model parameters are listed in Table 2. Even though Soyupak, 1994).
it was not possible to find kinetic information and
data for a reservoir of similar nature (in terms of
climatological and operational conditions), all pa- 4. Results and discussion
rameter values calibrated for the KDR are within the
general ranges reported in pertinent literature (Bowie It is generally accepted that, in cases where phos-
et al., 1985). phorus is the limiting nutrient, control of phosphorus
Following calibration trials, the response of the at source is the only long-term solution to the prob-
calibrated model to different initial conditions and lem of eutrophication. Other caretaking approaches
data sets acquired from the measurement and sam- which endorse tackling symptoms rather than the
pling programme was tested via a series of verifica- causes can only be acceptable when the feasibility of
tion runs. As exemplified in Fig. 3 and Fig. 4, the applying phosphorus control at source is question-
predictive capability and accuracy of the modelling able. In this regard, proposing viable strategies for
approach are satisfactory for the evaluation of future eutrophication control in a huge reservoir like the
management strategies. The details of calibration and KDR is not an easy task.
S. Soyupak et al. / Ecological Modelling 97 (1997) 99-110 107
Specifically, both the enormity and the socio-eco- were executed with the calibrated parameter values
nomic structure of the catchment area propagate and started with the same initial conditions. The
severe complexities. Moreover, sequential drying and subsequent effectiveness evaluations are based on
wetting occur due to seasonal water level fluctua- potential improvements in DO and euphotic chloro-
tions imposed not only by the operation of the phyll-a concentrations which are selected as the
hydro-power plant, but also by the highly variable primary water quality variables of choice. In other
meteoro-logical characteristics of the area which in- words, temporal and spatial concentration variations
duce significant temporal variations in the incoming expected from the implementation of the strategy of
hydraulic loads. Under these circumstances, four dis- interest are compared with the predictions obtained
tinct alternatives for phosphorus control at source are under the currently existing conditions (see Fig. 5
evaluated in terms of overall effectiveness in mitiga- and Fig. 6). The DO comparisons are made for the
tion of eutrophication: month of October when oxygen content of the reser-
1. Phosphorus control within the catchment area of voir is at the lowest level.
the Murat River.
2. Tertiary treatment of Elazl~ wastewaters. 4.1. Phosphorus control within the catchment of
3. Construction of baffles in the eutrophic segment Murat River
of the reservoir.
4. Construction of pre-dam structures on the Murat Water quality surveys revealed that the majority
River. of the nutrients reaching the KDR is being carried by
Each mitigation strategy is assessed separately in the Murat River (Yemi§en and Soyupak, 1994). In
the following paragraphs. For this purpose, the this context, this strategy is the only one among all
adopted modelling approach was used to represent the proposed alternatives which truly pertains to
the behaviour of the eutrophic reservoir segment phosphorus control at source. However, regional so-
under different scenario conditions. All simulations cio-economic characteristics and the size of the
Dissolved oxygen ( m g / i )
0 2 4 6 8 I0 12
0 i , i , | I i i , I ! I , ' I i , ! , I I~: 11
It ~ W l ; l i l I
IO
: q-
•
:;t I
.
Station N o : 8
Run dote s 1 9 . i 0 . 1 9 9 2
30
• ; I I
I! i .;oi:
,.,,...~-[.,
i I :
40
~i : I' BottQm
SO
F i g . 5. E f f e c t o f a l t e r n a t i v e s t r a t e g i e s o n D O profiles.
108 S. Soyupak et al. / Ecological Modelling 97 (1997) 99-110
25 | |
agricultural management techniques within the
catchment area since application of a treatment pro-
~ 2O
cess (physico-chemical or biological) to the waters
of Murat is not deemed feasible due to the magni-
tude of the flow ranging in between 38 and 2000
m3/s.
As illustrated in Fig. 5 and Fig. 6, reduction of the
|
B j" \. phosphorus load transported by the Murat River will
lead to significant improvements in water quality.
For instance, based on a comparison of the predicted
o chlorophyll-a concentrations with the limits estab-
g s
lished by the OECD (1982), a 25% reduction of the
Stotion ~ o : S
phosphorus load will upgrade the existing eutrophic
I J conditions to eutrophic-mezotrophic levels. Based on
Hoy 1 9 9 2 :uncl992 ,July 1992 the same criteria, the conditions will be further up-
graded to mezotrophic levels as a result of a 50%
reduction of the incoming phosphorus load.
25
Station 5 and thus creating a uniform DO distribu- tertiary treatment of Elazl~ wastewaters, provision of
tion with depth. The results of model runs for this boundary flow mixing with baffles and construction
scenario demonstrated that conception of such a of pre-dam structures) will only result in marginal
boundary of uniform bio-chemical composition will improvements.
considerably improve the bottom DO concentrations The modelling approach utilized for the analysis
downstream of Station 8 (see Fig. 5). of the eutrophication problem in the KDR has yielded
sufficiently accurate information on the overall effec-
4.4. Effect of pre-dam structures tiveness of alternative management strategies to be
implemented in the future. Such modelling ap-
Rather than being an original strategy, this alter- proaches are especially useful to decision makers
native merely refers to future DSi plans regarding when other simplistic methods are not applicable due
dams on the Murat River and in particular to the to the complex properties of the system under con-
proposed Palu Dam situated immediately upstream sideration.
of the KDR. It is clear that, as a pre-dam structure,
the reservoir of the Palu Dam will eliminate a por-
tion of the incoming pollutants including phospho- Acknowledgements
rus. Based on the principles of methods developed
by Benndorf and P~tz (1987) for the estimation of This study has b~en supported by the Turkish
phosphorus removal in pre-dam structures, potential Scientific and Technical Research Council
(TOBiTAK) through DEBAG contracts 1/G and
phosphorus reductions in the reservoir of the planned
Palu Dam will range from 0% during times of high 83/6. The authors thank the related personnel of the
State Hydraulic Works (DSi) and Environmental En-
flow to 27% in a relatively short period during fall.
gineering Department of Middle East Technical Uni-
The mean annual phosphorus reduction provided by
the Palu Reservoir is expected to be about 7.5% versity.
(Karul, 1995). Thus, it is obvious that, without im-
plementing proper control measures in the upper References
catchment of Murat, only marginal water quality
improvements can be anticipated from the construc- Abbott, M.B., Damsgaard, A. and Rodenhuis, G.S., 1973. A
tion of the proposed Palu Dam. design system for two-dimensional nearly horizontal flows. J.
Hydraul. Res., 11: 1-28.
Ambrose, R.B., Wool, T.A. and Martin, J.L., 1988. WASP4 A
hydrodynamic and water quality model: Model theory, user's
5. Concluding remarks manual, and programmer's guide. EPA/600/3-87-039,
USEPA, Athens, GA.
A mathematical modelling approach, based on Benndorf, J. and Fritz, K., 1987. Control of eutrophication of lakes
and reservoirs by means of pre-dams, Part I: Mode of opera-
coupling a compartmental water quality model with
tion and calculation of the nutrient elimination capacity. Water
a hydrodynamic model has been applied to the eu- Res., 21: 829-838.
trophic segment of the KDR in order to determine Bowie, G.L., Mills, W.B., Porcella, D.B., Campbell, C.L., Pa-
the effectiveness of various eutrophication control genkopf, J.R., Rupp, G.L., Johnson, K.M., Chan, P.W.H.,
alternatives. For this purpose, the dynamic behaviour Gherini, S.A. and Chamberlain, C.E., 1985. Rates, constants,
and kinetics formulations in surface water quality modeling,
of the reservoir segment of interest has been exam-
2rid ed. EPA/600/3-85/040, USEPA, Athens, GA.
ined under different scenarios related to phosphorus Chang, R.T. and Tung, C., 1970. Wind-driven lake circulation by
control. the finite-element method. Proc. 13th Conf. on Great Lakes
According to simulation results obtained by using Research, Int. Assoc. of Great Lakes Research, Ann Arbor,
a set of calibrated model parameters, significant MI, pp. 891-903.
Clesceri, L.S., Greenberg, A.E. and Trussell, R.R. (Editors), 1989.
improvements in water quality can only be achieved
Standard methods for the examination of water and wastewa-
by applying proper phosphorus reduction strategies ter, 17th ed. APHA, AWWA and WPCF, Washington, DC.
within the catchment of the Murat River. Sole appli- Connolly, J.P. and Winfield, R., 1984. A user's guide for
cations of other end-of-pipe control measures (i.e., WASTOX: A framework for modeling the fate of toxic chemi-
110 S. Soyupak et aL / Ecological Modelling 97 (1997) 99-110
cals in aquatic environments, Part 1: Exposure concentration. Organisation for Economic Co-operation and Development
EPA-600/3-84-077, USEPA, Gulf Breeze, FL. (OECD), 1982. Eutrophication of Waters: Monitoring, Assess-
Covar, A.P., 1976. Selecting the proper reaeration coefficient for ment and Control. OECD, Paris.
use in water quality models. USEPA Conf. on Environmental Platzman, G.W., 1963. Dynamic prediction of wind tides on Lake
Simulation and Modeling, Cincinnati, OH. Erie. Meteorol. Monogr., 4: 1-43.
Di Toro, D.M. and Cormolly, J.P., 1980. Mathematical models of Reid, R.O. and Bodine, B.R., 1968. Numerical model for storm
water quality in large lakes, Part 2: Lake Erie. Ecol. Res. Ser., surges in Galveston Bay. ASCE J. Waterways Harbour Div.,
EPA-600/3-80-065: 90-101. 94: 33-37.
Di Toro, D.M. and Matystik, W.F., 1980. Mathematical models of Schuepfer, F.E., Lennon, G.P., Weisman, R.N. and Gabriel, R.,
water quality in large lakes, Part 1: Lake Huron and Saginaw 1988. Hydrodynamic model of Great Sound, New Jersey. Mar.
Bay. Ecol. Res. Ser., EPA-600/3-80-065: 28-30. Geol., 82: 1-15.
Di Tort>, D.M., Fitzpatrick, J.J. and Thomann, R.V., 1983. Water Shanahan, P. and Harleman, D.R.F., 1984. Transport in lake water
quality analysis simulation program (WASP) and model verifi- quality modeling. ASCE J. Environ. Eng., 110: 42-57.
cation program (MVP) documentation. Hydroscience Inc., Simons, T.J., 1971. Development of numerical models of Lake
Westwood, NY (for USEPA, Duluth, MN. Contract No. 68- Ontario. Proc. 14th Conf. on Great Lakes Research, Int.
01-3872). Assoc. of Great Lakes Research, Ann Arbor, MI, pp. 654-669.
Heaps, N.S., 1969. A two-dimensional numerical sea model. Soyupak, S., G~kqay, C.F., Yerli, S., Erden, S., Akbay, N. and
Philos. Trans. R. Soc. A, 265: 93-137. (~ilesiz, A.F., 1994. Assessment of eutrophication for Keban
Jerome P., 1990. An efficient horizontal two-dimensional hydro- Dam Reservoir, Fresenius Environ. Bull., 3: 181-186.
dynamic model. Coast. Eng., 14: 1-18. Thomann, R.V. and Fitzpatrick, JJ., 1982. Calibration and verifi-
Karul, C., 1995. Assessment of management strategies for the cation of a mathematical model of the eutrophication of the
Keban Dam Reservoir. M.Sc. Thesis, Environmental Engineer- Potomac Estuary. D.C. Dep. of Environmental Services,
ing Department, Middle East Technical University, Ankara. Washington, DC.
Mukhallalati, L., 1994. A dissolved oxygen model for reservoirs Yemi~en, D. and Soyupak, S., 1994. The physical, chemical and
and lakes with highly varying inputs and outputs, Ph.D. biological properties of the Keban Dam Reservoir:. Eutrophica-
Thesis, Environmental Engineering Department, Middle East tion problem in the Keban Dam Reservoir, the results of
Technical University, Ankara. hydrodynamic and water quality modelling and proposed solu-
O'Connor, D.J., 1981. Evaluation of factors influencing the tem- tions. Final Report for the project on environmental problems
poral variation of DO in the New York Bight--Phase II. of the Keban Dam Reservoir and its catchment, DEBAG-
Manhattan College, Bronx, NY. 124/G, Elazl~ (in Turkish).
O'Connor, D.J., 1983. Wind effects on gas-liquid transfer coeffi-
cients. ASCE J. Environ. Eng., 109: 731-752.