RIVER WATER QUALITY MODEL NO.

1 (RWQM1):
CASE STUDY II. OXYGEN AND NITROGEN
CONVERSION PROCESSES IN THE RIVER GLATT

Peter Reichert

EAWAG, CH-8600 Dbendorf, Switzerland

ABSTRACT
Various simplifications of the river water quality model no. 1 are applied to data sets from the river Glatt in Swit-
zerland. In a first application, the biomass responsible for nitrogen and oxygen conversion processes is quantified
based on known reaeration rates, measured concentrations of ammonia, nitrite and oxygen and assumed growth pa-
rameters of algae and bacteria. In a second application, the model is extended to calculate chemical equilibria of
inorganic carbon compounds dissolved in the water and daily variations in pH. The influence of partially unknown
inflow concentrations and of calcite precipitation on fluctuations in electrical conductivity and pH are discussed. In
the last model, the processes of growth of sessile algae and bacteria, detachment of algae, and grazing by benthic
organisms are introduced. Due to lack of data for quantifying these processes, this last model application is specu-
lative. Nevertheless, it is interesting because it shows a direction to which river water quality modelling would have
to proceed in order to increase its predictive capabilities.

KEYWORDS
Ammonia, dissolved oxygen, nitrite, nitrification, pH, calcite precipitation, primary production, respiration,
rivers, water quality models.

INTRODUCTION
In a series of three papers, the river water quality model no. 1 was presented as a scientific and technical
base for formulating river water quality models (Shanahan et al., 2000; Reichert et al., 2000, Vanrolleghem
et al., 2000). It is the goal of the present case study to demonstrate the application of various simplified ver-
sions of this model to the river Glatt in Switzerland. In contrast to previous modelling studies for this river,
which were primarily devoted to identification of processes (Reichert et al. 2001; Uehlinger et al. 2000),
hypothetic applications of advanced model features are discussed in the present paper. Such applications can
show the direction to which river water quality modelling would have to proceed in order to improve its
predictive capabilities. In addition such applications can give impetus for more experimental work to be
done for testing emerging new hypotheses.

STUDY SITE
The river Glatt is a small river of about 35 km length flowing from lake Greifensee to the Rhine river. Its
characteristics change from an unpolluted lake outlet to a slowly flowing, heavily polluted river with a small
slope north of Zrich, and finally, to a relatively steep, highly aerated river with small drops every 50 m and
some cascades. This study, as well as the previous studies mentioned above, focus on the last part of the
river of about 10 km length. This river reach can best be used for the investigation of conversion processes
of substances transported with the river water, because there are no significant dry weather tributaries. For
this river reach, upstream and downstream on-line measurements of temperature, pH, ammonia, nitrite and
oxygen concentrations are available from six measurement campaigns with duration between a few days and
a few weeks (Berg, 1991, Kndler, unpublished). In addition, on-line measurements of temperature, pH and
oxygen concentration, and cumulative samples of many other substances are available at the downstream
site from the Swiss National River Survey Program (NADUF).
 REVIEW OF PREVIOUS MODELLING STUDIES
Because it has been shown experimentally that conversion processes in samples from the water column are
much slower than those observed in the river, all modelling approaches concentrate on conversion of sub-
stances dissolved in the water column caused by sessile bacteria and algae attached to the river bed.
The goals of the first study (Reichert et al. 2001) were to determine nitrification, production and respiration
rates in the river and to find correlation between environmental factors and oxygen and nitrogen conversion
rates. Stoichiometric coefficients of the simple model applied in this study were determined with the aid of a
mass composition for organic substances that considered only the elements C, H, O and N. For the dissolved
substances ammonia (SNH4), nitrite (SNO2), and oxygen (SO2), conversion rates were formulated with a lin-
ear dependence of algae growth on light intensity, a constant community respiration rate, and with Monod-
st nd
type limitation factors for oxygen and ammonia or nitrite for 1 and 2 stage nitrification, respectively. The
parameters of these transformation rates were estimated using upstream and downstream on-line measure-
ments for ammonia, nitrite and oxygen of all six measurement campaigns mentioned above (Berg, 1991,
Kndler, unpublished; the gas exchange coefficient for oxygen was determined independently with SF6
volatilisation experiments). The most important results of the study were the following:
For each of the six measurement campaigns a good fit was possible after adjusting the six model pa-
rameters. However, the values of the parameters differed from one measurement campaign to the other.
No significant correlation between environmental parameters and conversion rate parameters was found.
This led to the following conclusions:
The model is of an adequate complexity for the description of short term dynamics of ammonia, nitrite
and oxygen concentrations (daily variations within periods of a few days).
Over the time scale of a few days to a few weeks, nitrogen and oxygen conversion rate parameters do not
change significantly. Since these rates are caused by sessile algae and bacteria, this implies the presence
of a limiting factor for the accumulation of active algae and bacteria such as detachment, grazing, or a
limiting effect of active biomass in biofilms due to diffusion or shading.
Due to the absence of a mechanistic description of the limiting factor for the accumulation of active
biomass, the model cannot be used for long-term prediction or for the prediction of oxygen and nitrogen
dynamics under changed external driving conditions (e.g. improved wastewater treatment).
The most important problem of the investigation described above was the inability to find a significant cor-
relation between conversion rate parameters and environmental factors. One reason for this result was the
small number of measurement campaigns that could not be increased significantly.
In order to obtain a better statistical foundation at least for production and respiration rate parameters, the
model was simplified to a model only for dissolved oxygen. This model could be calibrated with the down-
stream on-line measurements of oxygen alone (Uehlinger et al. 2000). These are available over many years
from the Swiss National River Survey Program (NADUF). With this model, 123 dry weather periods of 3
days length within a period of 6 years could be evaluated. One respiration parameter (total community respi-
ration including nitrification) and two production parameters were fitted for all 123 investigation periods
and a correlation analysis with discharge, time, temperature, seasonal temperature gradient, time since last
flood, and global radiation was performed. This evaluation led to the following results:
The environmental parameters with most significant influence were temperature, seasonal temperature
gradient, time, and global radiation (used for production parameters only).
There was no significant effect of discharge or time since last flood.
The conclusions were, that in contrast to gravel-bed rivers (Uehlinger et al., 1996), floods have no signifi-
cant effect on production and respiration in the river Glatt with its more stable river bed and that production
and respiration activity is dominated by the seasonal effect.

GOALS OF THE PRESENT STUDY

The focus of the previous studies summarised above was to quantify nitrogen and oxygen conversion rates.
This was done with parsimonious models that use coefficients in empirically formulated conversion rates as
model parameters. These models do not describe changes in algal and bacterial densities mechanistically.
Therefore, they are not able to predict conversion rates under changed external driving conditions (e.g. im-
proved wastewater treatment). The goal of the present study is to perform speculative applications of ade-
quate submodels of the river water quality model no. 1 in order to demonstrate how an increase in the pre-
dictive capability of river water quality models could be achieved. This goal is approached by discussing
three submodels of increasing complexity. The three submodels of the river water quality model no. 1 ap-
plied in these three steps as well as the results of their application to the river Glatt are described in the fol-
lowing three sections.
All results shown in the following sections were gained by integrating the partial differential equations for
river hydraulics, substance transport and substance conversion with the simulation and parameter identifica-
tion program AQUASIM (Reichert 1994; 1995; http://www.aquasim.eawag.ch).

SUBMODEL FOR OXYGEN, NITROGEN, AND PHOSPHORUS CONVERSION BY

CONSTANT BENTHIC BIOMASS
The first submodel of the river water quality model no. 1 is described qualitatively in Table 1. In addition to
the processes shown in this matrix, gas exchange of dissolved oxygen at the cascades and along the river is
considered. Bacterial densities are assumed to be constant and are used as model parameters.
The submodel shown in Table 1 is similar to the model used in the first case study mentioned above (Rei-
chert et al. 2001). The major difference to this model is that here the values of growth parameters and yields
of algae and bacteria are assumed and that surface densities of sessile algae and bacteria are used as model
parameters instead of more direct conversion rate parameters for nitrification, production and respiration.
This leads to more speculative results but it is a good first step towards a model with better predictive capa-
bilities that tries to simulate the dynamics of algae and bacteria. A difference of minor importance is the
consideration of nitrate, SNO3, dissolved organic substrate, SS, and phosphate, SHPO4, as additional state
variables.

Component i (1) (3+4) (5) (6) (7+8) (9)

j Process SS SNH4 SNO2 SNO3 SHPO4 SO2
(1a) Aerobic Growth of Heterotrophs - ? ? -
with NH4
(1b) Aerobic Growth of Heterotrophs - - ? -
with NO3
(2) Aerobic Respiration of + + -
Heterotrophs
(5) Growth of 1st-stage Nitrifiers - + - -
(6) Aerobic Respiration of 1st-stage + + -
Nitrifiers
(7) Growth of 2nd-stage Nitrifiers - + - -
(8) Aerobic Respiration of 2nd-stage + + -
Nitrifiers
(9a) Growth of Algae with NH4 - - +
(9b) Growth of Algae with NO3 - - +
(10) Aerobic Respiration of Algae + + -
(11) Death of Algae (+) (+) (+)
(15) Hydrolysis + (+) (+) (+)

Table 1: First submodel of the river water quality model no. 1 as it is used for the present investigation.
Anoxic processes, processes and state variables for dissolved inorganic carbon compounds and all
particulate state variables are omitted. Instead, algal and bacterial surface densities are used as
model parameters.
Table 2 summarises the input data used for the simulations with the model shown in Table 1. In addition to
these data, light intensity and atmospheric pressure were used from a nearby measurement site of the Swiss
Meteorological Institute and water temperature from Berg (1991).

Variable Input concentration Comment

July 90 Nov. 90 Unit
3
SS 7.7 6.6 gCOD/m based on cumulative sample of river survey program
3
SNH4 measured time series of NH3+NH4 gN/m see Fig. 1
3
SNO2 measured time series gN/m see Fig. 1
3
SNO3 4.5 4.5 gN/m based on cumulative sample of river survey program
3
SHPO4 0.28 0.21 gP/m based on cumulative sample of river survey program
3
SO2 measured time series. gO/m see Fig. 1
Table 2: Input concentrations used for the model described in Table 1.
All parameter values of the model were taken from the numerical example given in Reichert et al. (2000)
with the exception of KI, the value of which was increased to 2000 W/m2 in order to account for the use of
light intensity at the water surface instead of light intensity at the river bed (there was no significant light
limitation or inhibition observed). The bacterial densities that are model parameters and not state variables
in this simplified model were used to adapt the model results to the measurements. Table 3 gives the values
of the model parameters for two measurement campaigns in July and November 1990 and Fig. 1 shows
measured and calculated upstream and downstream concentrations of oxygen, ammonia and nitrate.

Parameter July 1990 Nov. 1990 Unit Parameter July 1990 Nov. 1990 Unit
XALGs 650 800 gCOD/m XN1s 20 50 gCOD/m
XHs 350 0 gCOD/m XN2s 3.5 15 gCOD/m
XSs 250 0 gCOD/m
Table 3: Parameter vales adjusted for adapting the simulated results to the downstream measurements. The
width of the river is about 17 m (biomass surface densities can be obtained by a division by 17 m).

As already stated in Reichert et al. (2001) the good agreement of calculated with measured concentrations
for this simple model indicates that there is no significant change of active biomass in the river during the
measurement periods.
Oxygen, July 1990 Oxygen, November 1990
16 12
14
12

gO/m3
gO/m3

10 11
8
6
4 10
7/17/90 7/18/90 7/19/90 7/20/90 11/21/90 11/26/90 12/1/90 12/6/90

Ammonia, July 1990 Ammonia, November 1990

0.8 1.5

0.6 1
gN/m3
gN/m3

0.4
0.5
0.2
0 0
7/17/90 7/18/90 7/19/90 7/20/90 11/21/90 11/26/90 12/1/90 12/6/90

Nitrite, July 1990 Nitrite, November 1990

0.8 0.25
0.6 0.2
gN/m3

gN/m3

0.4 0.15
0.2 0.1
0 0.05
7/17/90 7/18/90 7/19/90 7/20/90 11/21/90 11/26/90 12/1/90 12/6/90

Figure 1: Measured (markers) and calculated upstream (thin lines) and downstream (thick lines) concen-
trations of oxygen (top), ammonia (middle), and nitrite (bottom) for the measurement cam-
paigns of July 1990 (left) and November 1990 (right).

EXTENSION OF THE SUBMODEL TO THE CALCULATION OF DISSOLVED

CARBONATE EQUILIBRIA, CALCITE PRECIPITATION, AND pH
Although the first submodel shown in Table 1 could describe the behaviour of the most important dissolved
compounds in the river, it is interesting to extend the dissolved variables in the model to the calculation of
the pH value. This would make it possible to calculate the concentration of NH3 (toxic for fish) and to use
pH dependent rate expressions (of special importance for nitrification, see Quinlan 1984 and Boon and
Laudelout 1962). The second submodel of the river water quality model no. 1 shown in Table 4 extends the
submodel shown in Table 1 by the state variables and processes required for this task. Gas exchange of car-
bon dioxide is taken into account in addition to gas exchange of oxygen.
 Component i (1) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)
j Process SS SNH4 SNH3 SNO2 SNO3 SHPO4 SH2PO4 SO2 SCO2 SHCO3 SCO3 SH SOH SCa
(1a) Aerobic Growth of Heterotrophs - ? ? - + ?
with NH4
(1b) Aerobic Growth of Heterotrophs - - ? - + ?
with NO3
(2) Aerobic Respiration of + + - + -
Heterotrophs
(5) Growth of 1st-stage Nitrifiers - + - - - +
(6) Aerobic Respiration of 1st-stage + + - + -
Nitrifiers
(7) Growth of 2nd-stage Nitrifiers - + - - - -
(8) Aerobic Respiration of 2nd-stage + + - + -
Nitrifiers
(9a) Growth of Algae with NH4 - - + - -
(9b) Growth of Algae with NO3 - - + - -
(10) Aerobic Respiration of Algae + + - + -
(11) Death of Algae (+) (+) (+) ? ?
(15) Hydrolysis + (+) (+) (+) ? ?
(16) Equilibrium CO2 <-> HCO3 -1 1 +
(17) Equilibrium HCO3 <-> CO3 -1 1 +
(18) Equilibrium H2O <-> H + OH 1 1
(19) Equilibrium NH4 <-> NH3 -1 1 +
(20) Equilibrium H2PO4 <-> HPO4 1 -1 +
(21) Equilibrium Ca <-> CO3 + 1

Table 4: Second submodel of the river water quality model no. 1 as it is used for the present investiga-
tion. The model is extended by the inorganic carbon compounds and the calcium ion used for pH
calculation and by the corresponding chemical equilibrium processes.
In addition, constant values, SK, SMg, SNa, SCl and SSO4, for the concentration of potassium, magnesium, sodium,
chlorine and sulfate were assumed in order to make a calculation of electrical conductivity possible. Note
that only the most essential variables for the calculation of pH were introduced. More complete pH models
can be found in Chapman (1982), Yeh and Tripathi (1991) and Runkel et al. (1996a and b).

Variable Input concentration Comment

July 90 Nov. 90 Unit
3
SS 7.7 6.6 gCOD/m based on cumulative sample of river survey program
3
SNH4 SNH / ( 1 + Keq,N / SH ) gN/m SNH is the measured time series of NH4+NH3 (Fig. 1)
3
SNH3 SNH / ( 1 + SH / Keq,N ) gN/m SNH is the measured time series of NH4+NH3 (Fig. 1)
3
SNO2 measured time series gN/m see Fig. 1
3
SNO3 4.5 4.5 gN/m based on cumulative sample of river survey program
3
SHPO4 SPO/(1+ SH /Keq,P) gN/m SPO is the measured cumulative sample for phosphate (Tab. 4)
3
SH2PO4 SPO/(1+Keq,P/SH) gN/m SPO is the measured cumulative sample for phosphate (Tab. 4)
3
SO2 measured time series gO/m see Fig. 1
3
SCO2 SH SHCO3 / Keq,1 gC/m
3
SHCO3 see equation (1) below gC/m calculated from a charge balance
3
SCO3 Keq,2 SHCO3 / SH gC/m
-pH 3
SH 1000 10 gH/m based on measured time series of pH
3
SOH Keq,w / SH gH/m
3
SCa 65 73 gCa/m Based on cumulative sample of river survey program
Table 5: Input concentrations used for the model described in Table 5. In addition SK (4.2, 3.9 gK/m3), SMg
(16, 16 gMg/m3), SNa (17, 16 gNa/m3), SCl (23, 24 gCl/m3), and SSO4 (22, 23 gSO4/m3), are intro-
duced as constant substance concentrations in order to consider the most important ions for elec-
trical conductivity (values are given for the two simulation runs for July and November 1990).
Because pH dependence of process rates is neglected in the model, this second submodel leads to the same
results for nutrients and oxygen as the first submodel (see Fig. 1). However, dissolved inorganic carbon
compounds, protons (and pH), hydroxyl ions and calcium ions can be calculated additionally. This leads
also to an increased number of inflow concentrations to be provided at the upstream end of the river section.
Table 5 gives an overview on the origin of the data used for the input. Because there were no quasi-
continuous measurements available for some of the modelled substances, the concentrations of some sub-
stances were assumed to be constant at a value measured in cumulative samples of the river survey program.
The inflow concentration of bicarbonate ions was then calculated from a charge balance according to
12 S S S S 2S 2S 2S 2S S S S (1)
S HCO 3 = S H S OH + NH 4 NO 2 NO 3 H 2 PO 4 HPO 4 SO 4 + Ca + Mg + K + Na Cl
1 + 2 K eq , 2 / S H 14 14 14 31 31 96 40 24.3 39 23 35.5
 pH, July 1990 Conductivity, July 1990
9.5 600

log10(mol/l)
9

uS/cm
8.5 500
8
7.5 400
7/17/90 7/18/90 7/19/90 7/20/90 7/17/90 7/18/90 7/19/90 7/20/90

Alkalinity, July 1990 Calcium, July 1990

4.2 80
4
60

gCa/m3
mol/m3

3.8
3.6 40
3.4
20
3.2
3 0
7/17/90 7/18/90 7/19/90 7/20/90 7/17/90 7/18/90 7/19/90 7/20/90

HCO3, July 1990 CO2, July 1990

50 1.5
45
gC/m3

gC/m3
40
0.5
35
30 0
7/17/90 7/18/90 7/19/90 7/20/90 7/17/90 7/18/90 7/19/90 7/20/90

Figure 2: Measured (markers) and calculated upstream (thin line) and downstream (thick line) pH value,
electrical conductivity, alkalinity and the concentration of calcium, bicarbonate and CO 2.

Figure 2 shows the results for some of the additional state variables and derived variables for the second
submodel. For these simulations, the gas exchange efficiency for CO2 at the cascades and the kinetic coeffi-
cient for calcium dissolution/precipitation had been adjusted. Calculated values of pH (Fig. 2, top left) are
qualitatively correct, however, there is a time shift in comparison to the measurements. Electrical conduc-
tivity is in the correct order of magnitude (Fig.2, top right). The differences in the daily variations is proba-
bly due to (unknown) variations in some ion concentrations at the upstream end of the river reach (for the
calculation, due to lack of data, most upstream ion concentrations were set to constant values, see Table 5).
The plots for calcium and CO2 in Fig. 2 show that the river is significantly supersaturated with both compo-
nents. The dashed lines show the equilibrium levels for calcium based on the current CO3 concentration and
for CO2 the equilibrium with the atmosphere. Assumption of equilibration of calcium with calcite through
precipitation (by increasing the value of keq,s0) would significantly diminish the amplitude of pH variations.

HYPOTHETICAL SIMULATION OF THE DYNAMICS OF BENTHIC BIOMASS

The third submodel of the river water quality model no. 1 used for a more speculative simulation is shown
in Table 5. In addition to the dissolved state variables of the first model shown in Table 1, it contains par-
ticulate state variables for sessile algae and bacteria (XH,s, XN1,s, XN2,s, XALG,s), for benthic consumers
(XCON,s), for sedimented organic material (XS,s, XI,s) and for suspended organic material (XS, XI).
Modelling of sessile or benthic organisms requires some additional considerations. In order to make it easy
to fulfil mass balances even in the case of fluctuating water levels, it is advantageous to introduce sessile or
benthic biomass per unit river length instead of per unit river bed surface. In Table 6 such masses per unit
length are indicated with an index s (e.g. XH,s denotes the mass of heterotrophic bacteria per unit river
length). The coefficients in the stoichiometric matrix that convert from substances dissolved or suspended in
the water column to substances attached to the river bed must then be multiplied with the wetted cross-
sectional area A of the river (this is the water volume per unit river length).
As an additional physical process, detachment of attached biomass to the water column and sedimentation of
suspended organic material must be taken into account. The process rates in Reichert et al. (2000) are for-
mulated with in-situ nutrient and oxygen concentrations and with in-situ light conditions. If the benthic bio-
film in the river is modelled on the basis of surface densities or masses per unit river length only, limiting
effects due to self-shading and diffusion of nutrients into the biofilm are not considered. An empirical con-
sideration of this effect is possible with a limiting factor for growth of algae and bacteria with increasing
algal or bacterial biomass. This can be implemented by a factor
1
(2)
k gro
1+ X
rmax
Component i (1) (3+4) (5) (6) (7+8) (9) (16s) (17s) (18s) (19s) (20s) (21s) (22s) (21) (22)
j Process SS SNH4 SNO2 SNO3 SHPO4 SO2 XH,s XN1,s XN2,s XALG,s XCON,s XS,s XI,s XS XI
(1a) Aerobic Growth of Heterotrophs - ? ? - A
with NH4
(1b) Aerobic Growth of Heterotrophs - - ? - A
with NO3
(2) Aerobic Respiration of + + - -A +
Heterotrophs
(5) Growth of 1st-stage Nitrifiers - + - - A
(6) Aerobic Respiration of 1st-stage + + - -A +
Nitrifiers
(7) Growth of 2nd-stage Nitrifiers - + - - A
(8) Aerobic Respiration of 2nd-stage + + - -A +
Nitrifiers
(9a) Growth of Algae with NH4 - - + A
(9b) Growth of Algae with NO3 - - + A
(10) Aerobic Respiration of Algae + + - -A +
(11) Death of Algae (+) (+) (+) -A + +
(12a) Growth of Consumers on XALG (+) (+) - - A +
(12b) Growth of Consumers on XS (+) (+) - A -
(12c) Growth of Consumers on XH (+) (+) - - A
(12d) Growth of Consumers on XN1 (+) (+) - - A
(12e) Growth of Consumers on XN2 (+) (+) - - A
(13) Aerobic Respiration of + + - -A +
Consumers
(14) Death of Consumers (+) (+) (+) -A + +
(15) Hydrolysis + (+) (+) (+) -A

Table 6: Third submodel of the river water quality model no. 1 as it is used for the present investigation.
In addition to the first submodel described in Table 1, this submodel contains algae, bacteria,
consumers and dead organic material as state variables. Algae, bacteria and consumers are as-
sumed to be only effective at the sediment surface (sessile algae, attached bacteria and benthic
consumers).
in the growth rate of algae and bacteria. A similar factor was also introduced for consumers. A final exten-
sion of the model concerns the assumption of a homogeneous mixture of the food of the consumers. This
may not be realistic for sessile algae and bacteria. Because in contrast to algae, bacteria can grow without
light, they can grow on places that may be difficult to access by consumers. This can be considered by mul-
tiplying all bacterial concentrations in the rate expression for consumer growth by an empirical factor BAC.
Biomass per unit river length 1990/91 Respiration Rates 1990/91
1500 XALG 40
XCON
30
gCOD/m

gO/m2/d

1000 XH
XS 20
500
10
0 0
1/1/90 1/1/91 1/1/92 1/1/90 1/1/91 1/1/92

Figure 3: Calculated biomass densities (biomass per river length) and respiration rates (oxygen consump-
tion per surface area) for the third submodel described in Table 6. The dots in the right hand plot
show the seasonal pattern of respiration calculated from oxygen data of the years 1992-1997 by
Uehlinger et al. (2000). Some of the scattor of this data is due to plotting all six years.
All these additional processes introduce a number of additional model parameters that cannot be identified.
Figure 3 shows a simulation of the benthic population based on average nutrient concentration in the water
and on actual light, discharge and temperature data in the river Glatt. This Figure demonstrates, that simula-
tions can lead to reasonable results (the typical seasonal change of the respiration rate shown in Uehlinger et
al. 2000 is roughly reproduced). However, much more data on the algal and bacterial population at the river
bed would be necessary in order to identify the parameters of such a river model.
 SUMMARY AND CONCLUSIONS
This case study demonstrated the applicability of the river water quality model no. 1 in three steps:
The application of the first (simplest) submodel demonstrated the capability of reproducing previous re-
sults obtained with even simpler models.
The second application showed the usefulness of the inorganic carbon submodel for modelling pH. With
the consideration of pH-dependent rates this could lead to an interesting model extension.
Finally, the third speculative application demonstrated that the model has the potential to be used for
predicting algal and bacterial densities and nutrient and oxygen conversion rates also for changed exter-
nal driving conditions. However, such a prediction needs much more experience with model parameters
and formulations that must be gained by data evaluations for a wide spectrum of river types.
As shown with this case study, prediction of nutrient and oxygen conversion rates in rivers is a difficult task
which requires a lot of experience with model formulations and model parameters for a given type of river.
We hope the river water quality model no. 1 can help facilitate the exchange of experience among river wa-
ter quality modellers and thereby can accelerate improvements in predictive power of river water quality
models.

ACKNOWLEDGMENTS
This study could not have been made without the measurements and data evaluations performed in previous
Glatt river investigations. I thank Michael Berg, Peter Holzer, Jan Kndler, Christoph Knig, Urs Uehlinger
and Oskar Wanner for data collection, data evaluation and stimulating discussion. The chemical part of the
study has profited from discussions with Maria Dittrich, Beat Mller and Bernhard Wehrli. In addition, data
has been provided by the Swiss National River Survey Program (NADUF) and the Swiss Meteorological
Institute (SMA). Finally, I thank the other members of the IWA task group on river water quality modelling,
Dietrich Borchardt, Mogens Henze, Wolfgang Rauch, Pete Shanahan, Lszl Somlydy and Peter Vanrolle-
ghem, for important discussions and Pete Shanahan for very carefully reviewing the manuscript.

REFERENCES
Berg, M. (1991), Stickstoffdynamik in der Glatt, Gas-Wasser-Abwasser 12/1991, 822-830.
Boon, B. and Laudelout, H. (1962), Kinetics of nitrite oxidation by Nitrobacter winogradskzi, Biochem. J. 85, 440-447.
Chapman, B.M. (1982), Numerical simulation of the transport and speciation of nonconservative chemical reactants in rivers,
Water Resourc. Res. 18(1), 155-167.
Quinlan, A.V. (1984), Prediction of the optimum pH for ammonia-N oxidation by Nitrosomonas europaea in well-aerated natural
and domestic-waste waters, Water Res. 18(5), 561-566.
Reichert, P. (1994), AQUASIM - A Tool for Simulation and Data Analysis of Aquatic Systems, Water Sci. Tech. 30(2), pp. 21-30.
Reichert, P. (1995), Design Techniques of a Computer Program for the Identification of Processes and the Simulation of Water
Quality in Aquatic Systems, Environmental Software 10(3), 199-210.
Reichert, P., Berg, M, Holzer, P., Kndler, J. and Wanner, O. (2001), Quantification of ammonia, nitrite and oxygen transforma-
tion rates in the river Glatt, in preparation.
Reichert, P., Borchardt, D., Henze, M., Rauch, W., Shanahan, P., Somlydy, L. and Vanrolleghem, P. (2000), River Water Quality
Model No. 1: II. Biochemical Process Equations, Wat. Sci. Tech., this volume.
Runkel, R.L., Bencala, K.E., Broshears, R.E. and Chapra, S.C. (1996a), Reactive solute transport in streams 1. Development of an
equilibrium-based model, Water Resourc. Res. 32(2), 409-418.
Runkel, R.L., McKnight, D.M., Bencala, K.E. and Chapra, S.C. (1996a), Reactive solute transport in streams 2. Simulation of a
pH modification experiment, Water Resourc. Res. 32(2), 409-418.
Shanahan, P., Borchardt, D., Henze, M., Rauch, W., Reichert, P., Somlydy, L. and Vanrolleghem, P. (2000), River Water Quality
Model No. 1: I. Modeling Approach, Wat. Sci. Tech., this volume.
Uehlinger, U., Bhrer, H. and Reichert, P. (1996), Periphyton dynamics in a floodprone prealpine river: evaluation of significant
processes by modelling, Freshwater Biology 36, 249-263.
Uehlinger, U., Knig, Ch. and Reichert, P. (2000), Variablility of gross primary production and ecosystem respiration in a small
river, Freshwater Biology, in print.
Vanrolleghem, P., Borchardt, D., Henze, M, Rauch, W., Reichert, P., Shanahan, P. and Somlydy, L. (2000), River Water Quality
Model No. 1: III. Submodel Selection Criteria, Wat. Sci. Tech., this volume.
Yeh, G.T. and Tripathi, V.S. (1991), A model for simulating transport of reactive multispecies components: Model development
and demonstration, Water Resourc. Res. 27(12), 3075-3094.