7.1. The Classification ofGroundwater Quality Models
199
For practical problems, the artificial recharge rate R, solute concentration
C R contained in the recharging water, and pumping rate P are usually
known. The remaining three parameters in Eqs. (7.1.26) and (7.1.27) are the
natural infiltration rate N, the solute concentration of infiltration water CN
and the reaction time t R. These three parameters can be obtained by calibration of the model with historical observation data.
Gelhar and Wilson (1974) reported using the one element model to study
the impact of highway-frost-proof salt on groundwater quality. Mercado
(1976) introduced the application of the lumped parameter model in the
management of groundwater quality, which was quoted in a book by Bear
(1979).
7.1.5 Criteria of Model Selection
We have already introduced the lumped parameter model and the distributed parameter model of water quality. The laUer can be further divided
into the pure advection type and the advection-dispersion type. Which type
should we select in a practical application? We now present some general
principles on model selection.
To select a model, we must make sure what our purpose iso If we want to
study the pollutant concentration distribution in an aquifer and to simulate
the advance of the concentration front with the model, then the concentration variation with space is a factor which must be taken into consideration.
In this case, we have no alternative but to adopt the distributed parameter
model. If the problem requires knowing only the variation of the mean solute
concentration over time and not the solute distribution in an aquifer, then we
can choose the lumped parameter model. It should be noted that the distributed parameter model gives time-space changes of the solute concentration,
and therefore, depicts the real situation in a more authentie and more meticulous way. On the other hand, the lumped parameter model is relatively
rough. The concentration obtained by a lumped parameter model does not
represent the solute concentration at a certain part of an aquifer or in a
certain weIl.
The second consideration for model selection is the quantity and quality of
data available. To build up a distributed parameter model requires many
distributed parameters, for example, porosity, hydraulic conductivity, dispersion coefficients, and so forth. Moreover, we should also know the boundary
conditions of water flow and water quality, and the distribution of sourcej
sink terms. The acquisition of these data depends on field observation and
parameter identification (see the next paragraph for details). If there is only a
small amount of reliable information, it will not be worthwhile to establish a
complex model because we do not have sufficient and reliable data for the
model calibration. Without reliable calibration data, even a complex model
will not be accurate. In that case, it would be better to use a simple mass
balance model with only one or a few elements.
199
For practical problems, the artificial recharge rate R, solute concentration
C R contained in the recharging water, and pumping rate P are usually
known. The remaining three parameters in Eqs. (7.1.26) and (7.1.27) are the
natural infiltration rate N, the solute concentration of infiltration water CN
and the reaction time t R. These three parameters can be obtained by calibration of the model with historical observation data.
Gelhar and Wilson (1974) reported using the one element model to study
the impact of highway-frost-proof salt on groundwater quality. Mercado
(1976) introduced the application of the lumped parameter model in the
management of groundwater quality, which was quoted in a book by Bear
(1979).
7.1.5 Criteria of Model Selection
We have already introduced the lumped parameter model and the distributed parameter model of water quality. The laUer can be further divided
into the pure advection type and the advection-dispersion type. Which type
should we select in a practical application? We now present some general
principles on model selection.
To select a model, we must make sure what our purpose iso If we want to
study the pollutant concentration distribution in an aquifer and to simulate
the advance of the concentration front with the model, then the concentration variation with space is a factor which must be taken into consideration.
In this case, we have no alternative but to adopt the distributed parameter
model. If the problem requires knowing only the variation of the mean solute
concentration over time and not the solute distribution in an aquifer, then we
can choose the lumped parameter model. It should be noted that the distributed parameter model gives time-space changes of the solute concentration,
and therefore, depicts the real situation in a more authentie and more meticulous way. On the other hand, the lumped parameter model is relatively
rough. The concentration obtained by a lumped parameter model does not
represent the solute concentration at a certain part of an aquifer or in a
certain weIl.
The second consideration for model selection is the quantity and quality of
data available. To build up a distributed parameter model requires many
distributed parameters, for example, porosity, hydraulic conductivity, dispersion coefficients, and so forth. Moreover, we should also know the boundary
conditions of water flow and water quality, and the distribution of sourcej
sink terms. The acquisition of these data depends on field observation and
parameter identification (see the next paragraph for details). If there is only a
small amount of reliable information, it will not be worthwhile to establish a
complex model because we do not have sufficient and reliable data for the
model calibration. Without reliable calibration data, even a complex model
will not be accurate. In that case, it would be better to use a simple mass
balance model with only one or a few elements.