8.2. Seawater Intrusion
271
time derivative to turn Eq. (8.2.33) into a system of algebraic equations:
1
[A]::'+1/2H::W + M[B]::'+l(H::':l- Hn) + F = 0,
(8.2.38)
where subscript n denotes the nth time step, superscript m denotes the mth
iteration, tlt the length of the time step, Hn+l the hydraulic heads to be solved
at time n + 1, and H n the known vector of hydraulic heads at time n. Several
iterations may be required for each time step. Ouring each iteration, the
values of the coefficients are replaced with new values obtained from the last
iteration until convergence is achieved.
Pinder and Page (1976) used this method to consider the problem of
seawater intrusion in an aquifer near Long Island, New York. The domain
considered was divided into 144 tri angular elements, with a total of 85 nodes.
The observation data from 1973 to 1975 in seven weHs were used to calibrate
the model. The inputs of the model include the known rainfaH infiltration and
artificial recharge. The flow boundary conditions along the co ast were determined during the model calibration. The calibrated model was then used to
predict the changes of the seawater-fresh water interface in the case of increased extraction.
Mercer et al. (1980) adopted the FOM to solve Eqs. (8.2.25) and (8.2.26).
They also tried various iteration methods for solving the resulting nonlinear
difference equations, and concluded that the method ofblocked linear successive over relax iteration was the most effective. In their paper, an example
is given which considers a problem of sewage irrigation in a coastal aquifer
in Hawaii. The se wage was injected into the seawater zone at the lower part
of the aquifer. The objective of the study was to estimate the seawater-fresh
water interface caused by the sewage. The simulated area covered 13 km 2 •
The parameters of the model were calibrated by the observed steady-state
heads.
8.2.4 Determination of the Transition Zones
When the transition zone is relatively wide, advection-dispersion models must
be used to describe the problem of seawater intrusion. In this case, the
unknown variable is the distribution of salt concentration and not the location of the interface. The salinity decreases graduaHy from the seawater
region to the fresh water region, and thus a natural transition zone is formed.
If the impact of salt concentration on groundwater flow is neglected, then
the problem of seawater intrusion will be the same as that of groundwater
contamination mentioned in Section 8.1. For instance, Gupta and Yapa
(1982) adopted the advection-dispersion model in their study of seawater
intrusion near Bangkok, Thailand, without considering the impact of concentration changes on flow velocity. Hydraulic conductivity, porosity and
dispersivity are determined through fitting the observed chloride ion concentration curves. Using this model, they found that a major cause of salination
271
time derivative to turn Eq. (8.2.33) into a system of algebraic equations:
1
[A]::'+1/2H::W + M[B]::'+l(H::':l- Hn) + F = 0,
(8.2.38)
where subscript n denotes the nth time step, superscript m denotes the mth
iteration, tlt the length of the time step, Hn+l the hydraulic heads to be solved
at time n + 1, and H n the known vector of hydraulic heads at time n. Several
iterations may be required for each time step. Ouring each iteration, the
values of the coefficients are replaced with new values obtained from the last
iteration until convergence is achieved.
Pinder and Page (1976) used this method to consider the problem of
seawater intrusion in an aquifer near Long Island, New York. The domain
considered was divided into 144 tri angular elements, with a total of 85 nodes.
The observation data from 1973 to 1975 in seven weHs were used to calibrate
the model. The inputs of the model include the known rainfaH infiltration and
artificial recharge. The flow boundary conditions along the co ast were determined during the model calibration. The calibrated model was then used to
predict the changes of the seawater-fresh water interface in the case of increased extraction.
Mercer et al. (1980) adopted the FOM to solve Eqs. (8.2.25) and (8.2.26).
They also tried various iteration methods for solving the resulting nonlinear
difference equations, and concluded that the method ofblocked linear successive over relax iteration was the most effective. In their paper, an example
is given which considers a problem of sewage irrigation in a coastal aquifer
in Hawaii. The se wage was injected into the seawater zone at the lower part
of the aquifer. The objective of the study was to estimate the seawater-fresh
water interface caused by the sewage. The simulated area covered 13 km 2 •
The parameters of the model were calibrated by the observed steady-state
heads.
8.2.4 Determination of the Transition Zones
When the transition zone is relatively wide, advection-dispersion models must
be used to describe the problem of seawater intrusion. In this case, the
unknown variable is the distribution of salt concentration and not the location of the interface. The salinity decreases graduaHy from the seawater
region to the fresh water region, and thus a natural transition zone is formed.
If the impact of salt concentration on groundwater flow is neglected, then
the problem of seawater intrusion will be the same as that of groundwater
contamination mentioned in Section 8.1. For instance, Gupta and Yapa
(1982) adopted the advection-dispersion model in their study of seawater
intrusion near Bangkok, Thailand, without considering the impact of concentration changes on flow velocity. Hydraulic conductivity, porosity and
dispersivity are determined through fitting the observed chloride ion concentration curves. Using this model, they found that a major cause of salination