Environmental Engineering Reference
In-Depth Information
system. It explicitly accounts for topography
and soil and vegetation characteristics and is
driven by conventional meteorological data
(Dawes
et al
.,
1997
). Bare soil evaporation and
plant transpiration are calculated separately
by using Penman-Monteith type equations.
Runoff is generated when precipitation exceeds
the infiltration capacity of the surface soils.
Water movement through soils is assumed to
be vertical and is simulated by using the one-
dimensional Richards equation. Drainage from
the base of the simulated soil zone was assumed
to be recharge.
The field study was conducted from May
1992 through December 1994. A meteorologi-
cal station was established at the study site to
measure air temperature, relative humidity,
wind speed, radiation, and rainfall. Leaf-area
index was measured on three 1-m
2
plots. Soil-
water content was measured biweekly at four
depths with time-domain reflectometry probes
at nine sites. Water levels in three wells were
measured at the same frequency. Hydraulic
properties of the sediments were determined
from analysis of soil cores. Streamflow from the
watershed was measured with a V-notch weir
with an automatic stage recorder.
Model input data consisted of daily values of
precipitation, air temperature, vapor pressure
deficit, and solar radiation. Land-surface con-
tours and catchment boundaries were derived
with a digital elevation model. The watershed
was divided into 1977 computational elem-
ents with an average area of 835 m
2
. Drainage
was calculated for each individual element
and ranged from 10 to 200 mm for the period
of simulation. Sensitivity analysis indicated
that soil type was the most important param-
eter influencing recharge because vegetation
was fairly uniform. Average drainage for the
watershed was 98 mm, about 5% of total rain-
fall. Drainage typically occurred in late winter
and spring (June through November); within
that period, large rainfall events triggered brief
sharp increases in drainage rates (
Figure 3.9
).
Little recharge occurred in late 1994 because
of a severe drought. Simulated soil-water con-
tents compared favorably with measured
data (
Figure 3.10
); simulated streamflow and
evapotranspiration were also similar to meas-
ured values. Zhang
et al
. (
1999
) used the model
to predict how planting trees on all or parts of
the watershed would affect recharge rates.
3.5 Groundwater-flow models
Groundwater-flow models are used to predict
aquifer response, in terms of head (groundwater
level) and fluxes into and out of an aquifer, to
natural and human-induced stresses; they are
important tools for resource and environmen-
tal management, and they provide groundwater
velocities needed for simulation of subsurface
contaminant transport. A common form of the
groundwater-flow equation (Harbaugh,
2005
)
can be written as:
∂ ∂∂∂+∂ ∂∂∂
+∂∂∂∂++=∂∂
(
K Hx x K Hy y
KHz zQ RSHt
/
)/
(
/
)/
xx
yy
(3.8)
(
/
) /
'
/
zz
s
where
K
xx
,
K
yy
, and
K
zz
are values of saturated
hydraulic conductivity along the
x
,
y
, and
z
coordinate axes (and it is assumed that the
major axes of the hydraulic conductivity ten-
sor are aligned with the coordinate axes),
H
is
total head,
Q'
represents all sources and sinks
except recharge, including water withdrawn
from wells and groundwater evapotranspira-
tion, and
S
s
is specific storage. Equation (
3.8
)
is derived from the continuity equation and
the Darcy equation and is, therefore, a water-
budget equation.
Groundwater-flow modelers may be reading
this topic for insight on independent estimates
of recharge for use as input for a groundwater-
flow model, but estimates of recharge can be
obtained indirectly from a groundwater-flow
model through the calibration process if actual
measurements of water levels and groundwater
discharges are available (
Section 3.2
). Values
of recharge used in the model are adjusted to
bring simulated heads and fluxes (such as base
flow) into agreement with measured values.
The best estimates of recharge rates are those
that produce the best model results, in terms
of minimized objective function. This approach
