Simulations with SESOIL are very poor when
ture profile resulting from a preceding modeling
run with laboratory-generated input data. For in-
period. Furthermore, the effects of snow and ice
stance, with laboratory input data, the model
melt on hydrological and sediment washload cy-
simulated that 65.5% of the chemical released mi-
cles under Alaskan conditions need to studied.
Validity of the sediment cycle has not been as
thoroughly tested as other cycles have. This cycle
showed only 4 to 8% of the chemical migrating to
ignores the chemical transport while simulating
sediment transport. The major limitation of the lay-
Since these studies were conducted, the model
ered approach is that when a chemical enters a soil
has been improved and tested in laboratory col-
layer, SESOIL considers it to be uniformly distribut-
umns with six organic compounds, and results
were compared with three field studies. The
ed throughout the layer. For instance, the model
modified model delivered better simulation re-
simulates low chemical concentration for large soil
sults for some compounds, but there was no im-
layers and high concentration for small soil layers.
provement for compounds with the lowest and
Hetrick et al. (1989) reported that modified
highest adsorption coefficients.
SESOIL predictions are in good agreement with ob-
The current version of the SESOIL model can
served laboratory and field data. However, the
simulate vadose zone contaminant concentration
model underestimates the concentration near the
in four soil columns, with each column divided
soil surface and they speculated that this may be a
into ten sub-layers. SESOIL's hydrological cycle
result of SESOIL ignoring the upward movement of
considers the soil column as one homogeneous
the chemical with the upward movement of water
compartment. Therefore, only one set of soil-
water content, porosity and core-disconnected-
nores the diffusive mobility of chemicals, which
ness parameters is used to describe the entire
may be important, depending on the Henry's Law
vadose zone. The model will not work if the soil
constant of the compound, and does not consider
has low permeability. SESOIL does not address
the volatilization enhancement when water evapo-
the soil-water spatial variability and water flow
rates. SESOIL should not be applied to a specific
in each compartment. Therefore, chemical trans-
port and distribution in the soil column could be
1989).
affected because the retardation coefficients and
volatilization fluxes of certain chemicals are
SENSITIVITY ANALYSIS
water-sensitive.
The model considers the internal soil moisture
The preliminary sensitivity analysis of SESOIL
at the beginning of each storm and in the inter-
for benzene was done by varying the few soil and
storm periods to be uniform, at its long-term
chemical input parameter values (Table 1) to below
spacetime average. This assumption may be a
or above the original values used by Drewett et al.
considerable departure from reality, as the soil
(1993). The input data file (SSOUT034.OUT) devel-
moisture profile in a later modeling period (i.e.,
oped by Drewett et al. (1993) in their SESOIL tests
1 month) cannot be influenced by the soil mois-
for Fort Greely, Alaska, was used in the sensitivity
Table 1. Summary of sensitivity analysis in the form of time to
predict peak and maximum leachate concentration.
Change in peak
Test*
Changed
Conc.
Time
(g/mL) (years)
Parameter
value
value
Soil density (g/cm3)
1.92
1.62
+0.12
5
1107
1104
2)
Intrinsic permeability (cm
+1.84
14
Disconnectedness index (unitless)
10
6.3
0.01
+4
Porosity (fraction)
0.25
0.35
0.04
+12
Organic carbon content (%)
0.13
0.09
+0.056
10
Adsorption coefficient on
83
69
0.013
5
organic carbon (koc)
5.48 104
2.74 104
Biodegradation rate in solid
+1.30
+2
phase (mg/kg per day)
5.48 104
2.74 104
Biodegradation rate in liquid
+0.28
+1
phase (mg/kg per day)
* Drewett et al. (1993)
5