-------
48
•MM
Mechanics of Using Hydrogeologic Setting Classification Methods. The first step is to
determine the availability of data for factors used hi the method. Based on this availability, the
user may choose to collect additional data, to modify the method, or to select another method.
The next step is to select the number of aquifer sensitivity classes. The selection is based
on both the number of plausible management options and the variety of hydrogeologic settings.
Fewer classes may be needed if only a limited number of management options exist. Likewise, if
the range or magnitude of a factor is small (such as depth to ground water less than 10 feet), the
number of sensitivity classes will likely be fewer than for areas where the range is great (e.g., from
5 to 300 feet).
Next, boundary values are defined to separate sensitivity classes for each hydrogeologic
factor. For example, hi a two-class system the depth-to-water boundary value distinguishing high
from low sensitivity might be 10 feet. Areas where the water table is shallower than 10 feet
would be highly sensitive for the factor. The selection of meaningful factors and boundary values
is best done by experts in hydrogeology and soil science.
Finally, a decision maker and/or the method must specify rules that establish how to assign
an overall sensitivity rating to an area with mixed ratings for key factors (e.g., recharge is rated
high, soil texture is medium, and depth-to-ground water is low). It is a judgment call whether all,
some, or only one of the classification criteria must be met. For example, if one factor is rated
high when all other factor ratings are low, conservative decision rules might call for this area to be
placed hi the highest sensitivity class, while less conservative rules might place the area in the
lower sensitivity class that fits most of the factors. Rule makers tend to select decision rules that
are most consistent with their protection needs, implementation constraints, and understanding of
the hydrogeologic setting.
Output. These methods produce classification designations — e.g., high sensitivity,
moderate sensitivity, or low sensitivity - for different subareas within a management area.
Classification output can be in the form of a map, list, or matrix. Table 3-5 presents a
classification method in which sensitivity classes are designated by number, based on soil
permeability. Soils classed as Number 4 are least sensitive and least permeable.
-------
49
Table 3-5. Leachability Classes of Kansas Soils (Kissel et al., 1982).
Class
1
2
3
4
Soil Texture
Sand, fine sand, coarse sand, & loamy coarse sand
Loamy fine sand, coarse sandy loam, & sandy loam
Loam, very fine sandy loam, silt loam & fine sandy loam
Clay loam, silty clay loam, silty clay, sandy clay, & clay
Soil
Permeability
6 to 20 in/hr
2 to 6 in/hr
0.6 to 2 in/hr
< 0.6 in/hr
Limitations. Relationships between factors may not be directly reflected in the
assessment method. In some cases factors may overlap, causing some parameters to be
represented twice. For example, including aquifer thickness and permeability in addition to
aquifer transmissivity is redundant. Because of the empirical nature of these methods, the
assessor's level of expertise and bias may significantly affect the results. These methods therefore
require users to have a high level of hydrogeological expertise. Decision rules used to classify
areas with characteristics of more than one class are developed by users, and therefore represent
another source of error or bias. Several different sets of decision rules can be found in the
literature, and there does not appear to be a technical basis for declaring one set of rules more
valid than another.
3.2.2 Scoring Methods
Definition and Uses. Aquifer sensitivity scoring methods are an extension of
hydrogeologic setting classification methods. In the literature, they may also be referred to as
ranking systems or numerical rating systems. Like classification methods, they tend to be based
on the evaluation of hydrogeologic factors.
Results of scoring methods can be presented in the form of maps or lists. The methods
also can be used at the field level if the data are sufficiently detailed and accurate. Resource
managers tend to use scoring methods to screen larger areas. As with classification methods,
these methods allow for the division of assessment areas into subareas with different levels of
sensitivity.
-------
50
Scoring methods reduce technical data to a form understandable by nontechnical people.
The score generated for a particular hydrogeologic setting or area can be compared easily to a
score for another setting or area. In addition, the use of factor weights provides a means to
account for the relative importance of each hydrogeologic factor. Because these methods
recognize a continuum of aquifer sensitivity, they facilitate differential protection of subareas
based on their sensitivity.
This type of method has a "track record" that can be reviewed. At least one scoring
method, Agricultural DRASTIC (AgDRASTIC) (Aller et al, 1985), or modifications of it, have
been employed in several states (including Idaho, Colorado, Wyoming, and North Carolina).
AgDRASTIC is described further in Chapter 5.
Mechanics of Using Scoring Methods. The first step is to determine the availability of
data for factors used in the method. Based on this data availability, the user may choose to
collect additional data, to modify the method, or to select another method.
For these methods, the range of each factor is subdivided into increments. For example,
depth to the water table could be subdivided into three levels: 0 to 10 feet, 10 to 20 feet, and 20
to 50 feet. Sensitivity values could then be assigned on a scale from 1 to 10, with 10 being the
most sensitive. The increments of depth to the water table would be placed on the sensitivity
scale as 10, 5, and 1, respectively.
After each factor is scored, the scores are combined to produce an overall sensitivity
score, index, or rating for the setting. Scoring methods may use either additive or multiplicative
formulas to combine factor scores into a final score for a setting. The multiplicative approach is
used to weight the relative importance of each factor. This results in larger scores and a wider
spectrum of scores unless values below 1.0 are used.
Output. Output for a sensitivity scoring method is typically a numerical score. Figure 3-1
is an example of a map showing DRASTIC scores for an assessment area. Scores also can be
listed for geopolitical units.
Sensitivity values do not have to be displayed numerically. For example, a user may
prepare a transparent map overlay for each factor, with sensitivity values reflected in gray shading
where the darkest shade corresponds to the greatest sensitivity value. Superimposing the overlays
-------
69W45'
69 W15'
- 41 N 30'
41N15
Legend
Hydrogeologic Setting
7Ba -Outwash
71 -Swamp/Marsh
7C -Moraine
7D-Buried Valley
7Aa - Glacial Till Over Bedded
Sedimentary Rocks
Pesticide
DRASTIC Score
206
149
134
126
108
Figure 3-1 DRASTIC Map of Bureau County, Illinois. Reproduced with Permission of
Monsanto Company. (From Sollar, 1992).
-------
for all factors produces a cumulative sensitivity map depicting relative hydrogeologic sensitivity in
shades of gray.
Limitations. Scoring methods may not include the particular factors that determine a
specific aquifer's sensitivity. Also, the weights assigned to various factors have been a subject of
controversy for users of this type of method. The objective of the weighting is to acknowledge
the theoretical contribution of each factor to overall sensitivity to contamination. However, the
exact relationship among factors is seldom understood. Statistical tools described in Section 3.5
may provide a basis for adapting a scoring method to a specific area.
Perhaps the most difficult aspect of using scoring methods is establishing a score that
triggers a decision to exercise protective management. The process of choosing the scores that
separate classes may be the same as that outlined for hydrogeologic setting classification methods
(Section 3.2.1). In other cases, the range of sensitivity scores may be naturally clustered or
segmented so that each cluster can be considered a separate sensitivity class.
33 GROUND WATER VULNERABILITY METHODS
Definition and Uses. Vulnerability methods go beyond considering only physical
hydrogeologic characteristics to include pesticide characteristics and agronomic practices. A
ground water vulnerability method provides a means to account for the interrelated processes that
govern the fate and transport of pesticides in the subsurface environment These methods are
used to determine areas with a high potential risk of ground water contamination, and thus are
helpful in selecting management practices.
Table 3-6 provides a summary of ground water vulnerability methods. With the exception
of the pesticide leaching subcategory discussed in Section 3.4, vulnerability methods are grouped
into two subsets, pesticide loading methods and simulation models. Two pesticide loading methods
are described in Table 3-6. These methods evaluate vulnerability by coupling an aquifer sensitivity
assessment with the quantity of pesticide applied to a subject area.
-------
*
1
s
&
S
.0
2
35
£
g
>
3
1
-a
1
1
H
1
Q
•B
1
1
®
CS
ff<
1 1
00 ^
CA
¥
|
<
H
•jj
If
i
ra ^" o
H .S -a j>" §
o -a j> JD e
*£€ 61
O oi bo C j_>
J2 « a> ~ :« -r, S -*
'5 S & H •«
3 S aa «j e
-S Q,S s w
g a 2 e b-
2 -5 s f?«
tplf
co *3 -5 h >
.a ^ ^ S a
^Qrt 1 §
U*
O tSi S
4~> T3 (O • •*
IJffll
t| |M|>
"o ^3 2
|sll"ls
3 *j c Q< *^ TV D-
q* so § g u « o
< » 8 >S Z, S H
g
3 0
0 C
5 "
cJ i-i C'
u
s
*o *ts
c o
03 J3 o w
S 5 -2 S
*O *y ca hi
1 S « « §
PL, ^ u 2 g
•g | - § J
3 ^ W 3 g
?^i$ .I
•c § S ^ <•?
EQ c c3 o ju
s2|ri
« ± a -a «
§ s*-° •«•§
sliSli'^r
s*l|ais
***** S "^ *
!|fifll»S*ti
s| si * iff
!f|lll!l«
l|!!i^f~
_. § W)Oji«S"g-S10
*4> PQ _c5 T3 •*-• O *s W S C
•3oou^)0S'-.2§
l^ilt^lll"
l*?-5*IS£|Sl.B
•S 'S a A^ c? *? 75 r! fi '•J -!P S
I 1
-------
•
f
« a
£•8
s
fa
I
I
1
o
£
fcf
B 1
I
I
VD
CO
D >-
•§ ^
.§
V)
-------
•B
1
3
es
o>
J
1
es
£
"§
S
g
0
o
1"
S
t»
VO
t
*«
SB
H
Method Description
IN
o
*^
0
^jg^
2 S
M
ta N
^^
f
1
1
If
"8 "§
a, 09
i
C8
^
• 1
^ S
§2
Si
C5 2
RJ ^»
1-^
W jj
3 g<
•0 S
11
11
73
JS
0)
•o
4o
•3
S3
CL,
J
^
c
i
o sr
Q CS
£.
1
D
S3
1
J^
•o
3 O3
0&
>H
s
-------
Method Descripti
of
1
4joS3;gHpfiBaT3s
-p>
'
s .«3-S~2«3Q9<'ScoT3OI&it>;ScL,8'?!<^
iliilllllliiiiiiisil
1
S.
•sir
SI,
I
t
Ol l
£
.3
c
1
3
«S
'I
•g
e
P
0 9
-------
57
*
i
t
0>
a
1
I
TS
c
3
2
0 '
«w
0
>»
a
£
so
«
o>
I
5
1
•c
u
1
i
s
tr
Js
£•
11
03 N
<£
i
•o
l_
^ tfl
1i fa
Is!
to o -S O^
Q °s
H'fcd
Ui
a)
— n
3 .9
PH :=
elpT
2 § >
S 3 ^
^ >^ ^w'
0
to
&°
•3
ra
S
i
-------
I
CO
1
-3
•g
0 3
.1
w
g
** ca .sF.2
**-< o ir PI
• o co e o
tfg?
D,
I
c
co
5
0
«si
"g
ro
ro
g
CO
-------
59
II
..« a-g
•-i .= *.n OT pn
! 73 u
_ 4> g
S~ 3 8 § '5 g' 2 E
08 - ° « "-1 CL o
so^ "S § "i w g cf'o1
tl III fl'IS.
*" 8 *? |*11 I 8 1 s
|5|l|l|"
HI,
8 <3 a> o
,1 S
.
q ej N fe S
U OS w H Q
2^
S g
Cfl
I
I II
-------
60
1
fe
Sen
a>
m c
I
<
o a
a, co
a
.o~
£1
¥8
'•3 .2
,-•3
.1.1
Il
*
§
I
.
I**
u &
-: o
H
I
™
1
I
"a
I
&
!
i
-------
61
e -a § s -
«12lgo
a) e c « S CL
QJ IS _._. S Q^
OH * 2 O 5, S
a > g" « a a
•a u 1 « ^ S
l§
!
I
K
pi
.ra.S § g §
?^ %£ <*
$ o S S t
rfc?
Q"'
«,
1
*
ST 1
13 ^
!S "ra
g
S
-------
^•c
I
I
I
.s
I
I 1
e §
-2 |
to S
o
CO
oJ
I
I
•a
I
e
I
-------
63
Q
1
?!
II
£8
§
l-s-tg
„. jr §> ! . -R so«
2.£ IS! sJSJL «
^•a 1 t: -s .g -sa
-ii^&ya
g-T-d'a.8
p n •—• X C
ca
SP —
21
8
.s
€
1
I
QC-
•§•
1
I
If
OH
CO
1
I
tg
CO
•<
a
CO
•g
e
s
1
u
s
-------
64
i
I
O _
O
I
II
03 N
e
O
vo
Jl
•§
1
I
_eo
"3
-------
65
Most reported vulnerability methods are simulation models, which are prepared as computer
codes (i.e., computer models). These models are theory-based, mathematical expressions of one or
more processes or phenomena related to the transport and fate of pesticides in soil/aquifer systems.
Some of these models simulate pesticide movement only through the soil zone; others simulate
movement through several zones. LEACHM, for example, simulates movement through the soil and
saturated zone. Several of the models simulate pesticide transport within different layers of soil,
unsaturated, or saturated zones. Some focus more on pesticide characteristics than on hydrogeological
processes, and vice versa.
Simulation models are related to pesticide leaching methods (see Section 3.4), which use
simplified fate and transport methods to establish the potential for a pesticide to leach. Unlike
simulation models, leaching methods do not quantitatively predict pesticide concentrations, travel times
or mass loading.
3.3.1 Pesticide Loading Methods
Definition and Uses. These methods combine pesticide use (loading) data with an aquifer
sensitivity assessment, and can be used for screening large areas. Pesticide sales data, commonly
available for counties in agricultural areas, can be used as proxies for pesticide use data. Results from
aquifer sensitivity analyses used in conjunction with pesticide use data are examined for the same unit
areas for which pesticide use data are reported.
An example of a pesticide loading method is the one used in North Carolina (Moreau and
Danielson, 1990). This method uses DRASTIC sensitivity scores in combination with loading rates for
pesticides with high reported leaching potentials. A map of county-wide DRASTIC scores is
superimposed on a map of county-wide pesticide use rates. Areas with high DRASTIC scores (see
Chapter 5) and high loading rates are predicted to be most vulnerable to pesticide contamination.
In this method, areas lacking significant pesticide application are considered to be least
vulnerable - regardless of their sensitivity. Also, only pesticides that are known to leach are considered.
Figures 3-2a - 3-2c illustrate the pesticide loading method used in North Carolina. Figure 3-2a shows
the rate of use of the teachable insecticide Ethoprop by county, and Figure 3-2b presents the
DRASTIC scores. Figure 3-2c shows the relative vulnerability of ground water to Ethoprop in North
Carolina, produced by overlaying the pesticide use map on the DRASTIC score map.
-------
5 TO 9.9
10 TO 14.9
15 TO 19.9
20 TO 25
Figure 3-2a Ethoprop Use by County (Moreau and Danielson, 1990)
DRASTIC INDEX
100 TO 125
125 TO 150
150 TO 175
175 TO 200
200 TO 225
Figure 3-2b Relative Sensitivity of Aquifers to Contamination (Moreau and Danielson, 1990)
VULNERABILITY
LEAST
NEXT LEAST
NEXT MOST
MOST
Figure 3-2c Vulnerability of Ground water to Contamination by Ethoprop (Moreau and
Danielson, 1990)
Figure 3-2 Example Output from the North Carolina Pesticide Loading Method.
-------
Expertise Needed. Pesticide loading methods should be used by individuals with strong
backgrounds in hydrogeology. Expertise requirements are similar to those for aquifer sensitivity
methods (see Table 4-1).
Data Requirements. Data requirements are similar to those for aquifer sensitivity
methods, except that pesticide loading data also are needed. Pesticide use data are grossly
translated as indications of mass loading over that portion of the land that receives the pesticide
application. However, pesticide use data may not be a satisfactory proxy for actual loading rates.
Crop acreage may also be used as a proxy for loading with the assumption that within a given
area, specific crops will require a specific pesticide.
Scale Considerations. These methods are subject to the same scale considerations that
apply to aquifer sensitivity methods. The assessment scale is generally dictated by the nature of
the aggregation of pesticide use data.
Mechanics of Using Pesticide Loading Methods. The steps in applying these methods are
identical to those for aquifer sensitivity methods, with the additional step of determining pesticide
loading. Pesticide manufacturers and retailers maintain sales data that may be compiled by state
agencies for all pesticides or for specific groups of pesticides. The sales data can be superimposed
onto a map of sensitivity assessment results, and may be used as a multiplier for the sensitivity
score to provide a range of ground water vulnerability scores. The North Carolina example
(Figure 3-2c) uses a classification system for designating areas from highest to lowest vulnerability.
Output. Results of pesticide loading methods can be depicted on maps or presented in
lists. The output typically uses vulnerability classes or scores similar to an aquifer sensitivity
method.
Limitations. Pesticide loading methods are prone to the same limitations as aquifer
sensitivity methods. In addition, they are limited by the difficulty in obtaining accurate loading
data. Pesticide sales data, aggregated for areas such as counties or subcounties, can serve as a
proxy for loading data, but they may not accurately represent actual use at any level, since they do
not include information about actual rates and locations of application. Other information, such
as permits for pesticide use at the farm or field level, would provide a more accurate
representation of mass loading. Crop acreage may also be used as a proxy for loading with the
assumption that within a given area, specific crops will require a specific pesticide. Because
-------
pesticide use for a county is often calculated by multiplying the number of acres planted in a
given crop by the manufacturer's recommended application rate (whether or not that pesticide
was used), the results of this method could deviate significantly from reality.
33.2 Simulation Models
A number of simulation models are available for evaluating pesticide fate and transport in
the soil, unsaturated and saturated zones (see Table 3-6). The methods vary primarily in the
number of processes incorporated into the computer codes and the number, and kinds of input
required.
Many simulation models are relatively simple models that incorporate only a few key fate
and transport processes and can be used on personal computers without extensive technical
support databases. For example, Ehteshami et al. (1991) used a one-dimensional model, CMLS,
to determine vulnerability for 16 potentially vulnerable sites identified from a previous sensitivity
assessment using DRASTIC. Other simulation models might best be referred to as research tools
because they require powerful computers or extensive amounts of storage, as well as substantial
technical support and several databases to operate.
Field-scale simulation models can be used to predict pesticide concentrations, loading, and
travel tunes at points of interest The accuracy of the prediction can be verified using a field
monitoring system. In addition, once a model has been validated and calibrated for a particular
hydrogeologic setting, it may be possible to test a multitude of management scenarios at rather
low cost. Simulation models can be used to estimate rates of movement and long-term impact of
pesticides in a wide range of hydrogeologic settings. A helpful overview of modeling approaches
to the environmental fate and transport of pesticides can be found in Jury and Ghodrati (1989).
Expertise. The level of technical expertise needed to operate a simulation model is
generally high, particularly in the areas of computer science and hydrogeology. The application of
comprehensive simulation models such as GLEAMS may involve multi-disciplinary teams of highly
skilled specialists. A good team must have expertise not only in hydrogeology, solute transport,
and numerical methods, but also in computer hardware and software. While the availability of
such a team is desirable, most users rely on support materials and "help tables" included hi the
software.
-------
; 69
Mechanics of Applying Simulation Models. Simulation models allow a wide range of
variation in their input parameters. Some, like GLEAMS and LEACHM, allow the user to vary
pesticide management practices to determine the effect of different strategies on mass loading or
concentration of pesticides exiting the unsaturated zone. These methods may be useful for areas
that have high aquifer sensitivity and are intensively fanned.
Simulation models can be run on both mainframes and microcomputers. In recent years,
however, the development of fast, relatively inexpensive microcomputers with large amounts of
random access memory made mainframes less necessary. The use of microcomputers for
hydrologic simulations requires sufficient memory for execution of the code as well as a math co-
processor for faster computations, an advanced processor, and a high resolution monitor and
graphics card. A printer and plotter also are needed to illustrate different simulations on paper.
A graphics package is recommended for post-processing and for analysis of results.
Scale Considerations. Simulation models discussed in this report are used for areas the
size of plots, fields, or other units considerably smaller than geopolitical units.
Output. The same simulation model can have many different kinds of output. .For testing
potential management practices, the output could be a number expressing the proportion of
applied pesticide that escapes the soil zone. Other outputs could be the total mass of a pesticide
expected to reach ground water in a field or the mean concentration of pesticide hi recharging
waters. For experimental purposes or model calibration, the output could be a graph of pesticide
concentration in soil versus soil depth at a selected point hi time after application. Another
output could illustrate pesticide leaching depth versus tune.
The more comprehensive simulation models typically will provide a complete listing of
inputs along with the desired output in their users guide. Readers should consult with modeling
experts when comparing their desired output with the types of output generated by different
models. A comparison of PRZM and GrLEAMS simulation results is shown in Figure 3-3 (Smith
et al., 1991).
Data Requirements. While there are differences in the number and kind of input
requirements among various simulation models, most require site- and pesticide-specific parameter
values related to the hydrogeologic system and to pesticide fate and transport processes. Data
may be needed for many of the factors listed hi Tables 3-3a to 3-3d. For the more complex
-------
70
HALF-LIFE (days)
O.
<
60-
50-
40-
30-
20-
«. GLEAMS-30
* GLEAMS-60
A GLEAMS-90
G--0--O PRZM-30
Q--H---E1 PRZM-60
A--A--A PRZM-90
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
60 110 160 210 260
Partition Coefficient (Koc), cm3/g
S
-------
71
mi
models, such data inputs are unlikely to be available for areas other than research areas. In some
cases, the number of inputs can be large.
Limitations. There are significant limitations in using simulation models. These include:
- the large number of inputs required for more sophisticated models, which could
necessitate a data collection program;
- uncertainty when estimating input values where data are not available, which
introduces potential error and bias; and
- the need for a separate simulation model for each hydrogeologic setting in a study
area, which presents a problem in areas with many different settings.
Simulation models are based on a set of assumptions, and if those assumptions are not met the
results of the model will be invalid and misleading. In addition, the outputs of these models may not
always conform to reality. For example, reported pesticide concentrations in a field with a relatively
homogeneous soil at a specific depth and time can vary by at least two orders of magnitude between
the values predicted by the model and those measured in the field (Jones and Hanks, 1990). It is
therefore helpful to managers if model limitations and assumptions are presented with model results.
3.4 PESTICIDE LEACHING METHODS (Subcategory of Vulnerability Methods)
Definition and Uses. Pesticide leaching methods are a narrowly defined subcategory of
vulnerability methods that incorporate both physical and pesticide factors, but not agronomic factors.
These methods require both pesticide-specific and soil-specific information, and can be used to
determine the potential of a pesticide to leach through a particular soil for a given field condition.
Leaching methods are similar to sensitivity scoring methods in that they produce a relative index, score,
or classification. However, they also incorporate pesticide characteristics such as half-life, and so
appear to blend elements of both sensitivity and vulnerability methods.
The pesticide leaching methods reported in this document (see Table 3-6) were developed as
simplified substitutes for more complicated simulation models such as PRZM (Carsel et at.
-------
72_
1985). Leaching methods have the benefit of a theoretical base similar to those of simulation models.
Although they require fewer parameters than do simulation models, they may require significant
judgement when relating the parameters to physical settings.
Leaching methods may be used as screening tools for any size area for which input data are
available. If these data are overlaid on a pesticide loading map, the results could be used as a
pesticide-specific vulnerability map.
Leaching methods also can be used with a geographic information system (GIS) (Khan and
Liang, 1989). An attenuation factor index ~ an index for pesticide mass emission from the vadose zone
(after Rao et al., 1985) ~ was calculated with a GIS for 8-hectare parcels over a 1500 km2 area on the
island of Oahu in Hawaii. The resulting map identified the pesticide-specific attenuation potential of
soils and was used in designing monitoring and management programs. Figure 3-4 is the graphic
output of Khan and Liang's method.
Leaching methods can only be used under circumstances where the simplifying assumptions
are valid. If information on key input parameters, such as pesticide half-life or partitioning coefficients,
is incomplete or inappropriate for the soils of interest, the results will not be valid.
Leaching methods have gained acceptance by regulators as a tool in the pesticide registration
process. The Florida Department of Agriculture, for example, has used the leaching model developed
by Rao et al. (1985) in the evaluation of new active ingredients (NAIs). NAIs with a high potential to
leach based on standard soils data may be required to undergo more detailed evaluation using a
vulnerability method and/or field and laboratory studies to further refine their leaching characteristics.
Expertise. For the most accurate results, these methods require expertise in soil science and
pesticide chemistry. The use of methods that require standard input such as USDA soils data (Table
3-6) may significantly reduce the level of technical expertise needed.
Data Requirements. Generally, leaching methods do not require the extensive data inputs of
simulation models, and can be run using a hand calculator. Information on pesticide and soil
characteristics may be available in the literature or in publicly accessible databases. Many input
parameters can be estimated using sound professional judgement However, others may require site-
specific measurement
-------
73
21°35'
21°25'
Department erf Agricultural Engineering
University of Hawaii
EDB
Ground Water Contamination
Likelihood
very unlikely
moderately likely
likely
very likely
158° 15'
158° 05'
157°55'
157°45'
Figure 3-4
Ethyl Dibromide (EDB) Contamination Likelihood Map Based on EDB
Attenuation Potential for Soils on the Island of Oahu, Hawaii.
(Redrawn from Khan and Liang, 1989).
Scale. Leaching methods are one-dimensional, requiring input parameters for soils and
pesticide properties, etc., for a point (location) in space. These methods are applicable to any size area
but are commonly used at the field scale. The use of GIS and statistics to display and/or analyze
inputs, or to make them into a composite, facilitates the use of leaching methods at the scale of
counties and watersheds.
Outputs. The output of a leaching method is a numerical value that indicates either the
proportion of the pesticide that may be transmitted through the soil zone over time, or the likelihood
that a pesticide will migrate through the soil zone at detectable levels. Output can either be listed for
each soil-pesticide combination or mapped for a single soil-pesticide combination.
-------
74
MM
Limitations. As with any simplified approach to representing a complex phenomena such
as pesticide fate and transport, these methods are subject to distinct limitations for practical use.
Leaching methods do not predict pesticide travel time or concentration in soil or water as do
comprehensive simulation models.
3.5 STATISTICAL TOOLS
The literature includes several examples of the use of statistical tools in assessing the
potential for ground water contamination by pesticides. These tools include multiple regression
models and multivariate analysis. Some statistical tools are discussed in greater detail in
Chapter 5.
Steichen et al. (1988) used a multiple regression model to relate pesticide concentrations
in ground water to the age of the well, land use around the well, and the distance to the closest
possible source of pesticide contamination. Chen and Druliner (1987) and Druliner (1989)
employed multiple regression and non-parametric analyses in a Nebraska study. Teso et al.
(1988) used a multivariate statistical approach called Fisher's Linear Discriminate Analysis with
soil classification units to determine the probability of having a contaminated or a non-
contaminated site (binary approach) in one area as large as a county.
Definition and Uses. Statistical tools allow the user to analyze the relationship of one or
more hydrogeologic factors to known occurrences of pesticides in ground water. Statistical tools
involve the development of an equation, such as a regression equation, that explains observed
concentrations of a pesticide in soil and/or ground water. The equation is not necessarily related
to theoretical fate and transport relationships between hydrogeologic factors and observed
pesticide concentrations.
Statistical tools involve selecting appropriate independent variables (primarily
hydrogeologic factors) and defining an appropriate dependent variable (usually pesticide
concentration). Methods of statistical inference such as regression analysis or discriminant
analysis are employed to determine the relationship between the dependent variable and the
hydrogeologic variables.
-------
75
Statistical tools can identify which of the selected hydrogeologic factors associated with
ground water contamination are most important for a given location. Most of these tools have
been used to describe discrete components of the contamination process (i.e., movement of
pesticides through the root zone). Statistical tools may be useful in refining and aiding the
development of local sensitivity assessment methods or in validating those methods' prediction of
pesticide occurrence.
Expertise. Use of statistical tools generally requires a high level of expertise in
hydrogeology, soil science, pesticide fate and transport, and statistical techniques, because no off-
the-shelf statistical models are ready for use in all geographical areas. Users must determine
which parameters to evaluate, and must collect and quantitatively analyze regional and chemical-
specific data.
Mechanics of Using Statistical Tools. Statistical researchers use a process of empirical
study, which involves the following steps:
1. State the problem and list the hydrogeologic or other factors suspected to be
related to contamination of ground water by a given pesticide (e.g., depth to water,
soil type, recharge rates, and aquifer characteristics).
2. Sample wells in the study area (or validate an existing data set).
3. Relate the factors recognized in step 1 in a quantitative statistical analysis model
(such as a linear regression, correlation analysis, or factor analysis).
Hie product of these steps is typically an equation that explains the variability found in
the dependent variable.
Output Output from a statistical analysis depends on the particular technique and
computer software package used. For example, the result of a regression analysis typically is an
equation relating the dependent variable to the independent variable. Each independent variable
has a coefficient (i.e., coefficient of determination) that indicates the proportion of variation in
the dependent variable that is explained by the independent variable.
-------
76
MM
An example of a statistical tool developed by Chen and Druliner (1987) is presented in
Chapters.
Data Requirements. Statistical tools require pesticide monitoring data from sites (e.g.,
wells and springs) within the study area. To achieve the most representative results, a large
number of sampling sites is required. Where field data must be collected, quality
control/assurance programs will be needed to ensure the accuracy amd precision of the data used
in the models (this also is true for field data collected for assessment methods). The EPA Office
of Pesticide Program's Pesticide State Management Plan Guidance, Appendix B (EPA, 1993
Draft), provides guidance in this area. Monitoring data typically consist of information that
identifies the specific pesticides present and their concentrations, along with a measure of the
independent variables that have been defined.
The dependent variable, pesticide concentration, is assumed to be normally distributed for
normative statistics such as multiple regression analysis. For some data sets this may not be true,
and special treatment of the data (e.g., transformation to log values, nonparametric analysis, etc.)
will be required.
The same concerns about spatial distribution of the data exist for statistical models as for
aquifer sensitivity methods.
Limitations. Statistical tools require that pesticide contamination be present. They may
also require field sampling. In addition, the results of a statistical stiidy may not be transferable to
other settings; transferability depends on similarity of pesticide practices used in those areas and
on the hydrogeology of the setting. The statistical studies hi the literature reviewed for this
are snecific to soils and aauifers in distinct localities.
-------
Because statistical models rely on the proper specification of variables, model preparation
requires a considerable amount of time and technical expertise. Studies performed to develop a
statistical model may require years to instrument the study site, collect data, and prepare an
analysis.
Finally, it may be necessary to prepare several statistical studies to cover an entire
assessment area. This may be due to a lack of data for a consistent set of factors over the entire
area or to significant differences in hydrogeology or pesticide use. The results of each study
might be unique to the study subarea.
-------
3.6 SELECTED BIBLIOGRAPHY
Ahuja, L.R., D.G. DeCoursey, J.D. Hanson, M.J. Shaffer, R. Nash, K Rojas, and C.S. Hebson.
1992. Root Zone Water Quality Model. Version 1.0. U.S. Department of Agriculture,
Agricultural Research Service, Ft. Collins, CO. GPSR Technical Reports 2 and 3, 236 pp.
Aller, KT., J.H. Lehr, and RJ. Petty. 1985. DRASTIC: A Standardized System for
Evaluating Ground Water Pollution Potential Using Hydrogeologic Settings.
U.S. Environmental Protection Agency, 600/2-85/018, Ada, OK, 163 pp.
Berg, R.C., J.P. Kempton, and K. Cartwright. 1984. Potential for Contamination of Shallow
Aquifers in Illinois. Illinois Department of Energy and Natural Resources, State
Geological Survey Circular No. 532, pp. 30.
Bonazountas, M. and J. Wagner. 1984. SESOIL: A Seasonal Soil Compartment Model.
Consultant's Report to EPA, Office of Toxic Substances - EPA Contract No 68-01-6271.
Carsel, R.F., L.A. Mulkey, M.N. Lorber, and L.B. Baskin. 1985. The Pesticide Root Zone Model
(PRZM): A procedure for evaluating pesticide leaching threats to ground water.
Ecological Modeling 30: 49-69.
Gates, K.J. and F.W. Madison. 1990. Soil-Attenuation - Potential Map of Pepin County,
Wisconsin. University of Wisconsin - Extension. Madison, Wisconsin. Soil Map 10.
Chen, H.H. and A.D. Druliner. 1987. Nonpoint Source Agricultural Chemicals in Ground Water
in Nebraska - Preliminary Results for Six Areas of the High Plains Aquifer. USGS
Water-Resources Investigations Report 86 - 4338.
DeLuca, T. and P. Johnson. 1990. Rave: Relative Aquifer Vulnerability Evaluation. An
On-Farm Scoring System to Evaluate Aquifer Vulnerability to Pesticide Contamination.
Montana Department of Agriculture - Environmental Management Division. Helena,
MT. MDA Technical Bulletin 90-01.
-------
Druliner, A.D. 1989. Overview of the relations of nonpoint source agricultural chemical
contamination to local hydrogeologic, soil, land-use, and hydrochemical characteristics of
the high plains aquifer of Nebraska. U.S. Geological Survey Toxic Substance Hydrology
Program-Proceedings of 1988 Technical Meeting, U.S. Geological Survey Open File
Report 88-4220.
Ehteshami, M., R.C. Pealta, H. Eisele, H. Deer, and T. Tindall. 1991. Assessing pesticide
contamination to ground water: a rapid approach. Ground Water, Vol. 29, No. 6.
Enfield, CO., R.F. Carsel, S.Z. Cohen, T. Phan, and D.M. Walters. 1982. Approximating
pollutant transport to ground water. Ground Water 20 (6): 711-722.
Ferreira, V.A. and R.E. Smith. 1990. OPUS: An advanced simulation model for non-point
source pollution transport at the field scale. An overview. Proceedings of the
International Symposium on Water Quality Modeling of Agricultural Non-Point Sources,
Part 2. United States Department of Agriculture, Agricultural Extension Service. Logan,
UT, pp. 823-834.
Geologic Sensitivity Workgroup. 1991. Criteria and Guidelines for Assessing Geologic Sensitivity
of Ground Water Resources in Minnesota: Minnesota Department of Natural Resources,
Division of Waters, 122 pp.
Gibb, J.P., MJ. Barcelona, S.C Schock, and M.W. Hampton. 1983. Hazardous Waste in Olgle
and Winnebago Counties: Potential Risk Via Ground Water Due to Past and Present
Activities. Illinois Department of Energy and Natural Resources Document No. 83/26,
State of Illinois, Champaign, IL, 66 pp.
Goss, D.W. 1991. Screening procedure for soils and pesticides relative to potential water quality
impacts. Using Computer Simulation Models in Pesticide Registration Decision Making.
Symposium/Workshop, Weed Science Society of America, Louisville, KY.
Green, R.E., C.C.K. Liu, and N. Tamnakar. 1986. Modeling pesticides movement in the
unsaturated zone of Hawaiian soils under agricultural use. Evaluation of Pesticides in
Ground Water. American Chemical Society, Washington, D.C., pp. 366-383.
-------
80
mm
Grenney, W.J., C.L. Caupp, R.C. Sims, and T. Short. 1987. A mathematical model for the fate
of hazardous substances in soil: model description and experimental results. Hazardous
Waste and Hazardous Materials 4: 223-239.
Gustafcon, D.I. 1989. Ground Water Ubiquity Score: A sample method for assessing pesticide
leachability. Environmental Toxicology and Chemistry, 8: 339-357.
Harrigan, P. and A. Nold. 1989. Training Materials for GEMS and PCGEMS: Estimating
Chemical Concentrations in Unsaturated Soil and Groundwater. Unpublished U.S.
Environmental Protection Agency document, 54 pp.
Hearne, G.A., M. Wireman, A. Campbell, S. Turner, and G.P. IngersaU. 1992. Vulnerability of
the Uppermost Ground Water to Contamination in the Greater Denver Area, Colorado.
U.S. Geological Survey Water Resources Investigations 92-4143. 241 pp. 1 plate.
Hebson, C.S. and D.G. DeCoursey. 1987. A model for assessing agricultural management impact
on ground water quality. Agrichemicals Abstracts No. 23. 193rd ACS National Meeting.
Denver, CO, April 5-10.
Hutson, J.L. and R J. Wagenet. 1992. LEACHM - Leaching Estimation and Chemistry Model
(Version 3). Research Series No.. 92-3, Department of Soil, Crop, and Atmospheric
Sciences, Cornell University, Ithaca, New York.
Huyakorn, P.S., H.O. White, J.E. Buckley, and T.D. Wadsworth. 1987. Finite Element Code for
Simulating One-Dimensional Flow and Solute Transport in the Vadose Zone. Technical
report prepared by GeoTrans, Inc., for the U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens, GA.
Imhoff, John C, Paul R. Hummel, and John L. Kittle, Jr. 1992. Development of an automated
quantitative methodology for evaluating the leaching potential of pesticides-Final Report,
Year 1. Consultants report from AquaTerra Consultants to EPA, Office of Research and
Development. 96 pages.
-------
81
m
Jones, R.L. and RJ. Hanks. 1990. Review of unsaturated zone leaching models from a users
perspective. Proceedings of the International Symposium on Water Quality Modeling of
Agricultural Non Point Sources. Logan Utah, June 19-23.
Jury, W.A. and M. Ghodrati. 1989. Overview of Organic Chemical Environmental Fate and
Transport Modeling Approaches in Reactions and Movement of Organic Chemicals in
Soils. Soil Sci. Soc. Am. Special Publication No. 22.
Jury, W.A., W.F. Spencer, and WJ. Farmer. 1983. Behavior assessment model for trace organics
in soil: I. Model description. J. Environ. Qual. 12:558-564.
Jury, W.A., WJ. Farmer, and W.F. Spencer. 1984a. Behavior assessment model for trace
organics in soil: II. Chemical classification and parameter sensitivity. J. Environ. Qual.
13:567-572.
Jury, W.A, W.F. Spencer, and WJ. Farmer. 1984b. Behavior assessment model for trace
organics in soil: III. Application of screening model. J. Environ. Qual. 13:573-579.
Jury, W.A., WJ. Spencer, and W.F. Farmer. 1984c. Behavior assessment model for trace
organics in soil: -IV. Review of experimental evidence. J. Environ. Qual. 13:580-586.
Khan, M.A. and T. Liang. 1989. Mapping pesticide contamination potential. Environmental
Management 13: pp. 233-242.
Kissel, D.E., O.W. Bidwell, and J.F. Kientz. 1982. Leaching Classes of Kansas Soils. Kansas
State University, Kansas Agricultural Experiment Station, Manhattan, Kansas, Bulletin
641, 10 pp.
Knisel, W.G. ed. 1980. CREAMS: A Field-Scale Model for Chemical, Runoff, and Erosion
from Agricultural Management Systems. U.S. Deptartment of Agriculture, Science
Education Administration, Conservation Report No. 26.
Laskowski, D.A., C.A.I. Goring, PJ. McCall, and R.L. Swann. 1982. Terrestrial environment.
Environmental Risk Analysis for Chemicals. R.A. Conway ed. Van Nostrand Reinhold
Co., NY. pp. 198-240.
-------
LeGrand, H.E. 1983. A Standardized System for Evaluating Waste Disposal Sites. National
Water Well Association, Worthington, OH, 49 pp.
Lemme, G., C.G. Carlson, R. Dean, and B. Khakural. 1990. Contamination vulnerability indexes:
a water quality planning tool. J. Soil and Water Conservation. 2:349-351.
Lemme, G., C.G. Carlson, B.R. Khakural, L. Knutson, and L. Zavesky. 1989. Aquifer
Contamination Vulnerability Maps - A Water Resource Protection Planning Tool. Lake
Poinsett Pilot Project. Plant Science Department, South Dakota State University and
USDA - Soil Conservation Service. Plant Science Department Pamphlet #18.
Leonard, R.A, W.G. Knisel, and D.A Still. 1087. GLEAMS! Groundwater Loading Effects of
Agricultural Management Systems. Transaction of the ASAE, 30:1403-1418.
Liddle, S.K. and M.G. Ganley. 1988. The use of the DRASTIC classification system in surveys
of agricultural pesticides in drinking water wells. Proceedings of the Agricultural Impacts
on Ground Water Conference, Des Moines, Iowa.
Liddle, S.K., R.W. Pratt, and CM. Breidenbach. 1989. Ground Water Vulnerability to
Agricultural Chemicals Geographic Information System (GIS) Pilot Project. Research
Triangle Institute, Research Triangle Park, N.C. RTI/7848/002-02F. 18 pp.
McLean, J.E., R.C. Sims, WJ. Doucette, C.R. Caupp, and WJ. Grenney. 1988. Evaluation of
mobility of pesticides to soil using USEPA methodology. J. Environ. Engineering 114:
689-703.
Moore, J.S. 1989. SEEPAGE: A System for Early Evaluation of the Pollution Potential of
Agricultural Ground Water Environments. U.S. Department of Agriculture, Soil
Conservation Service. Geology Technical Note 5. (Revision 1). 23 pp.
Moore, J.S. 1988. SEEPAGE: A System for Early Evaluation of the Pollution Potential of
Agricultural Ground Water Environments. U.S. Department of Agriculture, Soil
Conservation Service, Chester, PA, 1 pp.
-------
Moreau, D.H. and L.R Danielson. 1990. Agricultural Pesticides and Ground Water in North
Carolina: Identification of the Most Vulnerable Areas. Water Resources Research
Institute of the University of North Carolina. North Carolina State University. Report
No. 252. 31 pp.
Mullins, J.A, R.F. Carsel, J.E. Scarbrough, and AM. Ivery. 1993. PRZM-2, A Model for
Predicting Pesticide Fate in the Crop Root and Unsaturated Soil Zones: User's Manual
for Release 2.0. U.S. Environmental Protection Agency, Office of Research and
Development, Environmental Research Laboratory - Athens, GA, EPA/600/R-93/046. 388
pp.
Nofziger, D.L. and AG. Hornsby. 1985. Chemical Movement in Soil. User's Guide. University
of Florida, Gainesville.
Nofziger, D.L., P.S.Q Rao and AG. Hornsby. 1988. CHEMRANK: User's Manual. Interactive
Software for Ranking the Potential of Organic Chemicals to Contaminate Groundwater.
Computer Series. Software in Soil Science. Florida Cooperative Extension Service, Univ.
of Florida, 58 pp.
Pacenka, S. 1984. Diagnosing the Causes of Ground Water Contamination Using a
Mathematical Model: Report on Technology Transfer Workshop. Report Sponsored by
U.S. Geological Survey, Report No. USGS/G-859(23), Water Resources Division, Reston,
VA, 42pp.
Passero, R.N., F.J. Cohen, S.J. Dulaney, P.M. Haff and H. Moaddel. 1988. Aquifer Vulnerability
Mapping in Southwest Michigan.
Rao, P.S.C., AG. Hornsby, and R.E. Jessup. 1985. Indices for ranking the potential for pesticide
contamination of groundwater. Symposium Soil and Crop Science of Florida, Proceedings
44:1-8.
Sacha, L., D. Fleming, and H, Wysocki. 1987. Survey of Pesticides Used in Selected Areas
Having Vulnerable Ground Waters in Washington State. U.S. Environmental Protection
Agency, EPA/910/9-87/169, Seattle, WA 324 pp.
-------
Seller, L.E. and L.W. Canter. 1980. Summary of Selected Ground Water Quality Impact
Assessment Methods. NCGWR Report No 80-3, National Center for Ground Water
Research, Norman, OK, 142 pp.
Shaffer, R.D. and D. Penner. 1990. Evaluation of leaching prediction models for herbicide
movement in the soil vadose zone. Proceedings of the Third National Research
Conference on Pesticides "Pesticides in the Next Decade: The Challenges Ahead."
Virginia Water Resources Research Center, Richmond, VA, November 8-9. 1990.
Smith, M.C., A.B. Bottcher, K.L. Campbell, and D.L. Thomas. 1991. Field testing and
comparison of the PRZM and GLEAMS models. Transactions of the American Society
of Agricultural Engineers, 34: 838-847.'
Steenhuif, T., S. Pacenka, and M. Van der Marl. 1984. Pragmatic model for diagnosing and
forecasting ground water contamination. Proceedings of Practical Application of Ground
Water Models. Natl. Water Well Assn., Worthington, OH. pp. 514-544.
Steichen, J., J. KoeUiker, D. Grosh, A. Heiman, R. Yearout, and V. Robbms. 1988.
Contamination of farmstead wells by pesticides, volatile organics, and inorganic chemicals
in Kansas. Ground Water Monitoring Review 8:153-160.
Teso, R.R., T. Younglove, M.R. Peterson, D.L. Sheeks, and R.E. Gallavan. 1988. Soil taxonomy
and surveys: classification of areal sensitivity to pesticide contamination of ground water.
J. Soil and Water Conservation 43:348-352.
Thornthwaite, S.W. and J.R. Mather. 1957. Instructions and Tables for Computing Potential
Evapotranspiration and the Water Balance. Drexel Institute of Technology, Laboratory of
Climatology, Publications in Climatology, Centerton, New Jersey, Vol. 10, No. 3, 311 pp.
U.S. Department of Agriculture, Soil Conservation Service. 1972. National Engineering
Handbook: Section 4, Hydrology. Washington, D.C., 548 pp.
US. Environmental Protection Agency. 1988. Comments on Preliminary Determination to
Cancel Registrations of Aldicarb Products. OPP - 30000/39A; FRL - 3407 - 2. Office of
Pesticide Programs. Washington, D.C.
-------
U.S. Environmental Protection Agency. 1986. Permit Guidance Manual on Hazardous Waste
Land Treatment Demonstrations. EPA-530/SW86-032. Office of Solid Waste,
Washington, D.C.
U.S. Environmental Protection Agency. 1983. Surface Impoundment Assessment National
Report. Report No. 570/9-84/002.
U.S. Environmental Protection Agency. 1993 Draft. "Pesticides State Management Plan
Guidance for Ground Water Protection" including Appendices A and B, "Review,
Approval, and Evaluation of State Management Plans," and "Assessment, Prevention,
Monitoring, and Response Components of Pesticide State Management Plans." Office of
Pesticide Programs. Washington, D.C.
Villeneuve, J.P., O. Banton and P. Lafrance. 1990. A probabilistic approach for the groundwater
vulnerability to contamination by pesticides: the VULPEST model. Ecol. Modelling,
51:47-58.
Villeneuve, J.P., O. Banton and P. Lafrance. 1987. Summary - Development of a Simple and
Practical Tool for the Evaluation of Groundwater Vulnerability to Contamination by
Pesticides. Scientific Report INRS No. 2336, University du Quebec, INRS, Canada,
54pp.
Wagenet, RJ. and J.L. Hutson. 1986. Predicting the fate of nonvolatile pesticides in the
unsaturated zone. J. Environ. Qual. 15:315-322.
Wisconsin Department of Natural Resources and Wisconsin Geological and Natural History
Survey. 1987. Ground Water Contamination Susceptibility in Wisconsin. Wisconsin
Department of Natural Resources and the Wisconsin Geological and Natural History
Survey.
Williams, J.R. and H.D. Berndt. 1977. Sediment Yield Prediction Based on Watershed
Hydrology. Transactions of ASAE 20: 1100-1104.
-------
-------
87
4.0 METHOD SELECTION, VALIDATION AND CALIBRATION
This chapter presents a basic framework for selecting and validating an aquifer sensitivity
or ground water vulnerability assessment method. Section 4.1 discusses administrative and
technical considerations that managers and staff members may want to consider in choosing a
method. Section 4.2 discusses how the methods can be evaluated as well as the uncertainty
involved in their outputs.
4.1 METHOD SELECTION
States should select an assessment method(s) based on the State's own management
objectives and on the requirements of State Management Plans (SMPs) for pesticides (See
Chapter 1). Several of the 12 program components of SMPs (EPA, 1993 Draft) will affect the
choice and implementation of methods. For example, Component 5, "Basis for Assessment and
Planning," specifically directs States to assess the character of hydrogeologic settings in areas
where pesticides are used and the potential for pesticide leaching exists on a subcounty level.
Such an assessment should consider aquifer sensitivity, pesticide usage data, and agronomic
methods. A State might therefore choose to select an aquifer sensitivity method(s) combined with
a vulnerability method(s); an aquifer sensitivity method combined with pesticide, agronomic and/or
ground water monitoring data; or a vulnerability method that includes pesticide usage and
agronomic factors.
Both a generic and a pesticide-specific SMP must:
• Discuss the State's efforts to characterize hydrogeology in areas where
pesticides are used, and describe how these efforts fit into the State's
comprehensive ground water protection effort. EPA encourages States to
integrate their ground water protection efforts under SMPs rather than
develop separate and possibly duplicative programs.
-------
• Discuss the State's 1) methods for assessing the potential for pesticide
leaching on a sub-county level in specific geographic areas where the State
intends to allow pesticide use, and 2) methods for identifying current or
reasonably expected sources of drinking water, ground water that supports
surface water ecosystems, or other highly valued ground water. The use of
monitoring (see EPA, 1992), modeling, other geographic planning tools
such as GIS, or work developed by other government programs in
evaluating ground water sensitivity and pesticide usage data should be
described.
• Discuss how the State's assessment of 1) current and reasonably expected
uses of ground water, 2) the value of ground water, and 3) the potential
for pesticide leaching into ground water will be used to set, determine, and
evaluate ground water protection priorities and management measures.
For example, the SMP may discuss how a combination of modeling and
monitoring will identify valuable ground water and determine its relative
vulnerability.1
Assessment results must be linked with other SMP components. For example, monitoring
and protection programs may be prioritized based on assessments that identify sensitive and/or
vulnerable aquifers. Similarly, public awareness and participation efforts may use assessment
results to help educate and inform the public.
States may wish to establish an advisory committee to help review the management,
technical, and policy considerations involved in choosing assessment methods. Because these
assessments are multi-disciplinary exercises, personnel from different agencies and institutions
could participate on the committee. Technical experts in geology, hydrology, soil science,
agronomy, water resources management, toxicology, pesticides, and other related disciplines
generally can be found in State and Federal agencies and in educational institutions. Officials
from resource management agencies and the agency implementing the assessment program also
could provide expertise.
1 For further discussion of monitoring and setting ground water protection priorities, see U.S.
Environmental Protection Agency. 1993 Draft.
-------
89
4.1.1 Administrative Considerations
Administrative considerations in choosing an assessment method include:
• Staffing requirements. Availability of experienced personnel is a significant
factor in selecting an assessment method. Table 4-1 identifies the type and
level of expertise needed for each method category. Experts may be
available in-house; in other State agencies and institutions, including
universities; or in Federal agencies such as the USDA Soil Conservation
Service or U.S. Geological Survey.
• Financial resources. The cost of performing assessments varies widely,
depending on such factors as staffing, data collection, and computer
support needs. Data collection, if necessary, can be a significant financial
consideration.
• Availability of support services. Some methods may require access to
computer hardware, software, graphic support, and database management
services. States need to consider both the availability of such support and
its cost.
4.1.2 Technical Considerations
Before choosing an assessment method, it is necessary to determine how useful it will be
with respect to particular hydrogeologic settings and pesticide fate and transport processes.
Technical considerations for evaluating methods include:
• Ease of application. States can determine the ease of applying a method by
consulting past users (authors are listed in Tables 3-4, 3-5 and 3-6) and by
reviewing the method's documentation. The contacts provided in Appendix A
should be able to supply this information for most of the methods listed in this
document. States also should consider a method's hardware or software
requirements, the need to modify the method based on their own circumstances
(including the availability of data) and the suitability of the output.
-------
90
•
1
i
§
1
s
3
o
^•n
1
•*•!
1
K
t/J
"g
I
o>
I
1*
§
to
1
5
*"* 0>
•8 w
f-3
oa ^
i
.j M
e '£3
_-. .J
S2 s
a 5o
a 1
t» t^
SB
o
s
V
r
•o
METHODS
J J s
J 3J ^
nJ J W
J J J
K g , g
K ffi ffi
^
n
•O -2
§ OT
WIT!
« *« ^-s
•i , i ii
11 1 *2 11
11 1 ! II
^ « 'c jg S
S^OT OT > -^
ffi
H-l
^
a
J
1 Pesticide Leaching
Methods Subcategory
••••••••nBaa^BBHBBBBBa^B^n
8
c
•8
>"? *J
2 c u M
'^ (U c« aj
«-
-------
91
Jurisdiction size. A screening-level sensitivity assessment may be best for
township or county-sized areas. Smaller areas requiring more accurate and
verifiable data may be assessed with a vulnerability method or a tiered
approach using multiple methods (see below). Field-level assessment may also
be appropriate where aquifer sensitivity varies greatly across a large area, where
data are sparse, or where there are dissimilar areas within a geological unit
Many pesticide leaching methods and simulation models are designed for areas
the size of a single field or small watershed. USDA Soil Survey Maps and
USGS topographic maps may be an appropriate base for these field-level
assessments.
Tiering or phasing. Sensitivity and vulnerability assessments can be conducted
in tiers, where different methods are used at different scales. For example, the
first tier might be an aquifer sensitivity assessment conducted at a small scale
such as 1:250,000 for screening purposes. The second tier might involve field-
level assessments for selected areas and pesticides. A different method could
be used in each tier, or the same method could be applied in phases as more
data become available. When using a tiered approach, States should consider
the compatibility of the different methods (e.g., scale, expertise needed, etc.).
Suitability to local conditions. Methods should be examined for their
appropriateness to a particular hydrogeologic setting. This consideration is
important because hydrogeology and pesticide management practices can vary
considerably across a State. State officials can consult local experts for a
description of local hydrogeology and key factors controlling aquifer sensitivity
and ground water vulnerability.
Suitability of supporting databases. The availability and format of data for key
method parameters should be a prime consideration in choosing a method.
Spatially configured databases such as a geographic information system (GIS)
will be very helpful for mapping data or method results. States may need to
compile available data as a first step toward understanding the spatial
distribution of those data. Some methods may also require that additional data
be collected, which could demand even more technical and financial resources.
If a method does require additional data, States should carefully weigh the
benefits of using that method against the added costs.
-------
92
Ease of quantifications. Some of the methods discussed in this document
will result in a quantitative measurement of aquifer sensitivity or ground
water vulnerability; others provide a qualitative measurement. The
mechanics of arriving at those measurements, scores, ratings, or indices vary
among the different categories of methods. Aquifer sensitivity methods, for
example, typically require a significant level of professional judgement.
Vulnerability methods, on the other hand, generally require more
computational skills and perhaps the intensive collection of data.
Extent of use. Information on the number of past applications of a method
and the hydrogeologic settings that have been assessed can be helpful in
evaluating whether that method is appropriate. States may also consider
the extent to which other studies have validated the results of assessment
methods. The contacts provided in Appendix A should be able to supply
this information for most of the methods listed in this document.
Need for monitoring. A State may choose to monitor ground water when
selecting or evaluating an assessment method. In addition to collecting its
own new data for this purpose, a State can obtain monitoring data required
by EPA or the State for pesticide registration as well as data from other
regulatory programs and private efforts. EPA discusses monitoring related
to SMPs in detail in its 1993 draft documents, "Pesticides State
Management Plan Guidance for Ground Water Protection" and Appendix
B, "Assessment, Prevention, Monitoring, and Response Components of
Pesticide State Management Plans." Vulnerability assessments also can be
used themselves to select areas for high-priority monitoring.
4.13 Making a Selection
The following box describes a process that may assist managers and technical experts in
choosing a suitable assessment method. This process does not imply the endorsement or
requirement of any particular method. States may choose other methods not presented here as
appropriate to the needs and resource base of the area to be assessed.
-------
HOW TO USE THIS DOCUMENT TO SELECT AN
ASSESSMENT METHOD
Review this Document: Read Chapters 2, 3, and 5 and Table 3-1 (Comparison of
Assessment Methods Categories) to become familiar with assessment methods and
concepts. Chapter 5 provides detailed descriptions of some methods.
Begin with an Aquifer Sensitivity Assessment Method for Screening Purposes: If
the State plans to use a screening method first, follow Steps 3 through 6 below. If
the State wishes to select a vulnerability assessment method, follow steps 7 through
10 below. &
Selection of a Sensitivity (Screening) Method
3.
4.
Table 3-2: Select methods from Table 3-2 (Summary of Methods According to
Hydrogeologic Zone) that reflect the hydrogeologic zone of interest or the zone for
which the most data are available.
Table 3-4: Refine selection based on more detailed information provided in Table 3-
4 (Summary of Documented Aquifer Sensitivity Assessment Methods).
5. Table 4-1: Refer to Table 4-1 (Level of Expertise Required to Use Various
Methods) to further reduce the list of possible assessment methods.
6.
Appendix A: For assistance in selecting a final method or methods, refer to
Appendix A, which contains method contacts.
Selection of a Vulnerability Assessment Method
7.
8.
9.
Table 3-2: Select a vulnerability assessment method(s) that reflects the hydrogeologic
zone of interest or the zone for which the most data are available by referring to
Table 3-2 (Summary of Methods According to Hydrogeologic Zone).
Table 3-6: Refine selection based on more detailed information provided in Table 3-
6 (Summary of Documented Ground Water Vulnerability Assessment Methods).
Table 4-1: Refer to Table 4-1 (Level of Expertise Required to Use Various
Methods) to further reduce the list of possible assessment methods.
10. Appendix A: For assistance in selecting a final method or methods, refer to
Appendix A which contains method contacts.
-------
4.2 METHOD EVALUATION AND UNCERTAINTY
Choosing an assessment method requires knowledge of how the output will be used and
what is the acceptable level of uncertainty in that output. For some vulnerability methods -
notably simulation models - there are generally accepted processes for evaluating uncertainty
through verification, validation, and calibration. There are, however, no generally accepted
processes for dealing with uncertainties arising from the use of aquifer sensitivity, pesticide
leaching, and pesticide loading methods.
According to the National Research Council (National Academy of Sciences, 1990),
vulnerability method prediction errors can be traced to three basic sources: (1) natural
heterogeneity that cannot be completely described with a limited number of field samples,
(2) measurement errors, and (3) structural differences between the real-world system and the
model used to represent it. The processes of verification, validation, and calibration are designed
to estimate these prediction errors. Much of the following text is extracted from the work of the
National Research Council.
4.2.1 Evaluatic
if Aquifer Sensitivity. Pesticide Leaching, and Pesticide Loading Methods
In evaluating aquifer sensitivity, pesticide leaching, or pesticide loading methods, it may be
helpful to compare outputs with field measurements or with outputs! from other models. There
are examples of each in the scientific literature. For example, Benton and Villenueve (1989)
compare results from DRASTIC and PRZM, while Ehteshami et al. (1991) supplemented
DRASTIC with CMLS. In a U.S. Geological Survey (USGS) report by Seller (1992),
AgDRASTIC results are depicted on maps and compared with results from other methods
developed by the Wisconsin Department of Natural Resources, Illinois State Geologic Survey
(ISGS), and ISGS-USGS (Soller, 1992).
EPA Region m currently is conducting two studies comparing DRASTIC results with
field data from parts of West Virginia, Virginia, Delaware, and Maryland (Koterba et al., 1993 (in
press), and EPA, 1993 (in press)). Each of these comparisons must be evaluated individually to
determine whether they have merit and whether they minimize uncertainty enough to meet user
requirements.
-------
4.2.2 Evaluation of Simulation Models
Uncertainty in simulation model output can be minimized by: (1) comparing the model's
output with that of another model or with a simplified solution (verification); (2) comparing the
model output with field or laboratory measurements (validation); and/or (3) adjusting appropriate
model input parameters until the outputs are acceptably similar to field measurements
(calibration). The appropriate use of simulation models is discussed further in a recent EPA
publication titled "Ground Water Modeling Compendium." (EPA, 1992)
There are no consensus descriptions of verification, validation, and calibration. The
American Society for Testing Materials (1984) defines these processes as follows:
Verification - Examination of the numerical technique in the computer code to ascertain
that it truly represents the conceptual model and that there are no inherent numerical problems
in obtaining a solution.
The objectives of verification are to: (1) demonstrate that the computer code can
accurately solve the governing equation, and (2) evaluate the sensitivity of the model to factors
contained in the analytical mathematical solutions. Comparing model output with simpler
solutions assumes that because the model is accurate for simple problems, it will be accurate for
complex simulations. To mitigate this assumption, the verification process also includes testing
the model against other complex models (benchmarking).
Validation - Comparison of model results with numerical data independently derived from
experiments or observations of the environment.
Field validation for any model is never possible for the model's full range of conditions.
Validation does, however, increase confidence that model results are reliable. Any field validation
should include the following: (1) data quality assurance, (2) model performance/acceptance
criteria, (3) sensitivity analysis, (4) comparison of output with acceptance criteria and (5) detailed
documentation of the validation exercise. Monte Carlo analysis is commonly used in the
validation process to investigate model prediction accuracy, A Monte Carlo analysis consists of
repeated simulations using predetermined variations in one or more input parameters. Prediction
confidence intervals and other statistical measures of uncertainty can be readily computed from
the complete suite of model analyses.
-------
mmmmmmmmmmmmmmmimmmmm^^^miiim^iim^m
Calibration - A test of a model using known input and output information, which is used
to adjust or estimate factors for which data are not available.
Kriging is commonly used in the calibration process to estimate the regional distribution of
a model input value based on scattered measurements. Kriging is a least-squares statistical
procedure that measures the accuracy of the regional estimate.
-------
43 SELECTED BIBLIOGRAPHY
American Society for Testing and Materials (ASTM). 1984. Standard Practices for Evaluating
Environmental Fate Models of Chemicals. Annual Book of ASTM Standards, E 978-84.
American Society for Testing and Materials, Philadelphia, PA
Benton, O., and J.P. Villenueve. 1989. Evaluation of ground water vulnerability to
pesticides: a comparison between the pesticide DRASTIC index and the PRZM
leaching quantities. Journal of Contaminant Hydrology. Vol. 4, pp. 285-296.
Ehteshami, M., R.C. Pealta, H. Eisele, H. Deer, and T. Tindall. 1991. Assessing pesticide
contaminants to ground water: a rapid approach. Ground Water. Vol. 29, No.6
Koterba, M.T., W.S.L. Banks, and R.J. Shedlock. 1993. Pesticides in Shallow Ground
Water in the DELMARVA Peninsula. J. Environ. Qual. 22:500-518.
National Research Council, Water Science and Technology Board, Committee on Ground
Water Modeling Assessment, Commission on Physical Sciences, Mathematics and
Resources. 1990. Ground Water Models, Scientific and Regulatory Applications.
National Academy Press. Washington, D.C.
Seller, David R. 1992. Applying the DRASTIC model - a review of county-scale maps.
U.S. Geological Survey Open-file Report 92-297. 36 pp.
Tsang, Chin-Fu. 1991. The Modeling Process and Model Validation. Ground Water,
Journal of the Association of Ground Water Scientists and Engineers. Vol. 29,
No.6. pp. 825-831.
U.S. Environmental Protection Agency. 1991a. Pesticides and Ground Water Strategy.
EPA Document 21T-1022. Office of Pesticides and Toxic Substances.
Washington, D.C.
U.S. Environmental Protection Agency. 1991b. Protecting the Nation's Ground Water:
EPA's Strategy for the 1990's. EPA Document 21Z-1020. Office of the
Administrator. Washington, D.C.
-------
98_
•••
U.S. Environmental Protection Agency. 1992. Ground Water Modeling Compendium. EPA
Document EPA-500-B-92-006. Office of Solid Waste and Emergency Response.
Washington, D.C.
U.S. Environmental Protection Agency. 1993 (in press). Ground Water Vulnerability to
Contamination by Agricultural Chemicals, Jefferson County, West Virginia. EPA
Region HI, Office of Ground Water, Philadelphia, PA.
U.S. Environmental Protection Agency. 1993 Draft. "Pesticides State Management Plan
Guidance for Ground Water Protection" including Appendices A and B, "Review,
Approval, and Evaluation of State Management Plans," and "Assessment, Prevention,
Monitoring, and Response Components of Pesticide State Management Plans." Office of
Pesticide Programs. Washington, D.C.
Van der Heijde, P.K.M., and R.A. Park. 1986. Report of Findings and Discussion of
Selected Ground Water Modeling Issues, U.S. EPA Ground Water Modeling
Policy Study Group. International Ground Water Modeling Center, Holcomb
Research Institute, Butler University, Indianapolis, IN.
-------
5.0 SELECTED METHOD DESCRIPTIONS
The assessment methods described in this chapter represent each of the two main
categories of methods: sensitivity and vulnerability: Descriptions of statistical tools also are
included at the end of the chapter. These methods were chosen because they are well
documented and have been applied in the context of pesticide management. EPA does not
endorse or recommend the following methods over other methods listed in Chapter 3 or more recent
methods not presented in this document.
5.1 DESCRIPTIONS OF SELECTED SENSITIVITY METHODS
5'1-1 Hvdrogeologic Setting Classification Methods
5.1.1.1 Leachability Classes of Kansas Soils
According to Kissel et al. (1982), many soil materials that overlie aquifers-in Kansas offer
protection from contaminants that might be transported by infiltrating waters. In some areas of
the state, soil materials allow contaminants to reach ground water more easily than in other areas,
and the authors devised a classification system to account for the spatial variation in the
attenuation potential of these overlying soils.
In general, the greater the percentage of sand in soils (coarse-textured soils), the more
susceptible they were judged to be to pesticide leaching. Accordingly, this method grouped
Kansas soils into four classes of leaching susceptibility (see Table 3-5) based on soil texture and
soil permeability (expressed as the water infiltration rate).
The classification decision rule focused on the limiting soil horizon in the soil profile. For
example, a soil with a top layer (horizon) permeability of 10 inches per hour and a subsoil horizon
permeability of 4 inches per hour will be placed in leaching susceptibility Class 2, where
permeabilities range from 2 to 6 inches per hour.
-------
The final class determination is based on the factor most limiting to pesticide leaching
losses - either texture or permeability. For example, a loam soil with a texture listed in Class 3
but a permeability listed in Class 2 would be classified as Class 3, since Class 3 is less susceptible
to leaching.
A leaching susceptibility map of Kansas soils was prepared based on the general soils map
available from the National Cooperative Soil Survey at a scale of 1 cm:40 km. An initial attempt
was made to classify only those soils with more than approximately 50,000 mapped acres published
in Soil Survey reports through May 1981. However, some soils with less acreage were included if
they were known to be highly cultivated, particularly if they were Class 1 or 2 soils.
5.1.1.2 Wisconsin Soil Attenuation Potential
Researchers at the University of Wisconsin-Extension (Gates and Madison, 1990) devised
a classification system to characterize the attenuation potential of Wisconsin soils to surface-
applied pesticides. They attribute the capacity of soils to attenuate the migration of pollutants
into ground water to seven physical and chemical properties: texture of surface (A) horizon;
texture of subsoil (B) horizon; organic matter content; pH of surface (A) horizon; depth of soil
solum (the upper and most weathered part of the soil profile; the A & B horizons); permeability
of subsoil (B) horizon; and soil drainage class.
The soil data used in this method were obtained from published Wisconsin county soil
surveys and assigned weighted values. Soil drainage was divided info the following four classes
and their respective weighted values: well-drained (10), well-drained to moderately well-drained
(7), moderately well-drained (4) and somewhat poorly, poorly, very poorly and excessively poorly-
drained (1). These values, ranging from 1 (poorest attenuation) to 10 (best attenuation), were
determined subjectively based on experience.
Scores generated from the summation of the weighted values were grouped into the
following four attenuation potential classes:
1 (least), with weighted value totals ranging from 0 to 30;
2 (marginal), with totals ranging from 31 to 40;
3 (good), with totak ranging from 41 to 50; and
4 (best), with totals greater than 51.
-------
To date, 14 out of the 72 counties in Wisconsin have been mapped at a 1:100,000 scale.
The county maps represent the four attenuation potential classes and indicate areas where the
bedrock is within 10 feet of the surface. The maps also show the distribution of soils based on
their potential for contaminant attenuation. The soils are grouped into soil series based on
similar chemical and physical characteristics, type of parent material, and the horizon
arrangement.
This method has been expanded recently to assess contamination risk as part of the
Wisconsin Farmstead Assessment System, which is based on information provided in county soil
survey reports. The expanded method enables determination of the subsurface attenuation
potential of farmsteads by assigning class values to geological material/depth to ground water
combinations.1 The values range from 1 to 4.
The method also determines the ground water contamination potential on farmsteads by
combining soil ranks and subsurface ranks. The combined ranking is then assigned one of the
following risk levels: 1 (high-risk), 2 (high-moderate risk), 3 (low-moderate risk), or 4 (low-risk).
This new expanded method is presently being revised and has not yet been published. It
will be refined to accommodate organic matter soil test results, as opposed to using less reliable
and less accurate data recorded for soil groups in soil surveys.
5.1.2 Scoring Methods
5.1.2.1 Agricultural DRASTIC
Agricultural DRASTIC (Aller et al., 1985 and 1987), or AgDRASTIC, is a standardized
method for evaluating aquifer sensitivity to pesticide pollution by using a parameter
weighting/scoring system. It is important to note that DRASTIC/AgDrastic was designed to
predict contamination potential in ground water rather than in wells. Many investigators have
applied the AgDRASTIC method as originally written or have modified it to suit their own
hydrogeologic settings and needs. AgDRASTIC is intended for use in areas 100 acres or larger in
size.
Geological material is defined as material below soil overburden.
-------
The hydrogeologic parameters used in this system, which make up the acronym
"DRASTIC," are:. Depth to water table, net water Recharge, Aquifer medium, Soil medium, land
surface Topography, Impact of the unsaturated zone, and water hydraulic Conductivity of the
aquifer. In the AgDRASTIC method, a prescribed weight is given to each parameter, and a
numeric value for the parameter is assigned to every point in the study area. For each point, the
value of each of the seven parameters is multiplied by the appropriate weighting factor; then the
seven products are summed to produce the AgDRASTIC Index (i,e., the score). High index
scores represent areas most sensitive to ground water contamination by pesticides.
DRASTIC was developed as a general model to predict potential ground water
contamination by any chemical. Agricultural DRASTIC has a slight modification: higher weights
are assigned to soil and topography factors because they are considered to be more important
than pther factors in determining the leaching of pesticides to ground water. The soil factor is
particularly important because all processes such as pesticide filtration, biodegradation, sorption,
and volatilization occur in the top soil media. Topography is a measure of land slope that helps
determine whether a pesticide would be part of runoff water or sediment or would remain on the
surface long enough to penetrate the soil profile and possibly pollute ground water.
The DRASTIC method indicates only relative sensitivity among different areas, not
absolute values for potential contamination levels. The method does not define "sensitive" or
"non-sensitive" areas. The user must select index values that are appropriate to define highly
sensitive local areas. The advantages of DRASTIC are that it is simple for people with a
moderate level of expertise to use; it makes effective use of available information; and it can be
used as a planning device prior to detailed site-specific investigations.
An early effort of the EPA National Pesticide Survey (U.S. EPA, 1992) was to broadly
assess ground water sensitivity in each of the 3,137 counties in the United States using
AgDRASTIC (Alexander and Liddle, 1986). This initial, small-scale application of AgDRASTIC
over a large area allowed EPA to select a subset of counties, which had been classified as
sensitive, for more focused study in later stages of the NFS.
The AgDRASTIC method, according to Alexander and Liddle, can be used to map a
county with an area of about 1,000 square miles, with an expenditure of between 200 and 300
person-hours. This estimate was based on the National Water Well Association's estimates of
how long it took to assess the first 10 demonstration counties (Liddle, 1992). The cost, including
-------
———————————————— 103
the amount of time and effort required, varies depending on the complexity of regional geology,
availability of data, and county size.
Alexander and Liddle estimated that between 300 and 450 person-years would be required
to evaluate all counties in the United States using AgDRASHC. Another estimate, based on the
1988 National Pesticide Survey Evaluation Technical Report (1988), says that a county assessment
takes two to five weeks. This estimate translates to 126 to 514 person-years required to assess all
3,144 U. S. counties (Liddle, 1992).
AgDRASTIC,.a true aquifer sensitivity method, does not account for loadings of pesticides
or for the specific chemical properties of potential contaminants. The method is based only on
the likelihood of a non-sorbed, non-degrading solute leaching into ground water. The Water
Resources Research Institute of the University of North Carolina and North Carolina State
University (Moreau and Danielson, 1990) consequently recommended that DRASTIC be applied
in combination with pesticide use patterns to assess the relative vulnerability of ground water to
pesticides. Moreau and Danielson proposed a four-group classification system: (1) most
vulnerable, (2) next most vulnerable, (3) next least vulnerable, and (4) least vulnerable.
Applications of AgDRASTIC
Several applications of the AgDRASTIC method have been reported in the literature
(Alexander and Liddle, 1986; Liddle and Gaiiley, 1988). The National Pesticide Survey (NFS)
classified 3,137 counties of the United States into three categories of ground water sensitivity
(low, moderate, and high). A second application was conducted for the NFS pilot survey that
consisted of determining intracounty AgDRASTIC scores for seven individual counties.
In another application conducted for the National Alachlor Well Water Survey (NAWWS)
(Holden, et al., 1992), AgDRASTIC scores were calculated for 90 counties in 26 states. The
NAWWS scores were similar to those recorded for the NPS.
Banton and Villeneuve (1989) found no significant correlation between DRASTIC index
values and leached pesticide quantities calculated using the PRZM simulation model (see Section
5.2.2.2) for non-sorbed and sorbed degradable solutes. The authors reported, however, that the
only factors common to both methods - depth to ground water and vadose zone media type -
influenced the correlation. They concluded that the chemical characteristics of contaminants are
-------
104
•MM
essential factors in assessing contamination potential. The authors noted that chemical properties
are not considered in DRASTIC.
David Soller (1992) evaluated several reports published by the Monsanto Company that
used AgDRASTIC to assess sensitivity in several counties located in the agricultural midcontinent
of the United Sates. He found that "DRASTIC cannot, by itself, adequately predict
contamination" (page 2), and concluded that "It is likely that future studies will more clearly
define the predictive abilities of DRASTIC in different hydrogeologic settings and at different
map scales" (page 33).
The USGS and EPA (Hearne and Wireman et al., 1992) also used an ARC-INFO GIS to
store spatial and attribute data to produce a shallow ground water vulnerability map for the
greater Denver area. This system accounted for all seven DRASTIC parameters.
DRASTIC remains a widely used method. A few attempts are now being made to
correlate ground water contamination information with DRASTIC scores. The Research Triangle
Institute (Liddle et al., 1989) recently proposed procedures to test, calibrate, and revise the
DRASTIC model where necessary. Their investigation was based on an established, large-scale,
spatial GIS database (pc ARC/INFO) that included DRASTIC mapping data, agricultural land use
data, digital base map data, and ground water monitoring data.
AgDRASTIC and the National Pesticides Survey
In 1990, EPA completed its five-year National Survey of Pesticides in Drinking Water
Wells (NFS), a statistically-based nationwide survey to assess the presence of pesticides, pesticide
degradates, and nitrates in public and rural domestic drinking water wells in the United States.
The Phase I report provides estimates of the occurrence of nitrates and pesticides in those wells.
As part of Phase n of the NFS, EPA assessed the relationships between pesticide
detections in wells and AgDRASTIC scores and subscores,2 as well as between nitrate detections
and AgDRASTIC scores and subscores. Prior to NPS sampling, overall scores and subscores for
each of the AgDRASTIC parameters were calculated for each county in the United States.
2 A subscore is the value a user assigns to an individual AgDRASTIC parameter. The
subscore is multiplied by the weight assigned to that parameter by the AgDRASTIC model.
-------
105
The Phase II Report concluded that, "At the county level...the overall DRASTIC score
either was not associated with detections in individual wells, or, in the case of nitrate detections
for Community Water Supply (CWS) wells, was associated in a manner contrary to the design of
DRASTIC..."
"Five of the seven DRASTIC subscores for county-level scoring (depth to ground water,
net recharge, aquifer media, topography, and hydraulic conductivity) were found to be associated
with detections, but primarily in the wrong direction. These results do not indicate that the
subscores are good indicators of contamination. Relatively few associations were identified,
considering the number of tests that were performed, and the associations that were identified
were not consistent over both CWS wells and rural domestic wells for both pesticides and nitrates.
"For four of the seven statisticaEy significant results...the possible effect was contrary to
that expected from the underlying hypothesis of the DRASTIC system." (EPA, 1992, pp. 108-9).
For the 90 counties in which rural domestic wells were sampled, scores and subscores were
developed on a sub-county scale for areas that showed similar sensitivity on a county scale. Tests
of association were made between sub-county-level scores and detections in rural domestic wells
only. According to the report, "No overall sub-county area DRASTIC score for underlying
aquifers was found to be associated with detections in individual wells, and only three sub-county
subscores — aquifer media, impact of vadose zone, and hydraulic conductivity — were related to
detections. Of these, scores for aquifer media and hydraulic conductivity were related to nitrate
detections in the opposite manner to that expected....
"....higher DRASTIC subscores for impact of the vadose zone were found to reflect a
greater likelihood of pesticide detections, in rural domestic wells" (EPA, 1992, page 110).
It is important to note that DRASTTC/AgDRASTIC was designed to predict the potential
for contamination in ground water rather than in wells. Because the NFS sampled wells in either
confined aquifers or water-table aquifers, and at different depths below the land surface and water
table, it is impossible to use the results of the Phase II report alone to validate or invalidate
DRASTIC as an indicator of potential ground water contamination. Lack of detection in deeper
wells indicates only that contamination may not have migrated into the deeper part of the aquifer,
not that it failed to enter the aquifer.
-------
106
AgDRASTIC and Results of the National Alachlor Well Water Survey
From mid-1987 through late 1989, the Monsanto company (with participation from EPA)
conducted the National Alachlor Well Water Survey (NAWWS), which included a statistically-
based sampling survey of alachlor occurrence targeted to private, rural domestic wells in counties
where alachlor is used (Holden et ah, 1992). Results of the NAWWS were presented to EPA in
February, 1990. The survey used county-level DRASTIC scores developed for the National
Pesticides Survey as a measure of relative aquifer sensitivity.
Wells located in sample strata (groups) corresponding to higher alachlor use and higher
sensitivity (based on the county DRASTIC score) had a higher probability of being selected for
sampling. The three-stage, stratified survey resulted in a final sample of 1,604 wells in 89
counties. Water samples were collected from 1,430 wells. Well-specific DRASTIC scores were
developed for each of the hydrogeologic settings in which the 1,430 sample wells were located.
The survey found that nitrate was common in rural wells, and that about 5% of the wells
in counties where alachlor was used contained nitrate concentrations greater than or equal to 10
mg/L. A significant association between high nitrate concentrations (levels of 10 mg/1 or greater)
and herbicide detections was found for the herbicides alachlor, atrazine, and simazine. Atrazine
was the most frequently detected herbicide in well water.
The relationship between aquifer sensitivity (well-specific DRASTIC scores) and chemical
occurrence was not clear. "The NAWWS found little, if any, association between DRASTIC
scores and herbicide occurrence....Simpler classification systems, focusing on only a few major
vulnerability factors, look to be more useful than DRASTIC in the short term" (Holden, 1992, p.
19).
5.1.2.2 Aquifer Vulnerability Index
South Dakota State University and the USDA Soil Conservation Service have initiated a
pilot project in Lake Poinsett in eastern South Dakota to develop sensitivity indexes for assessing
potential surface water and aquifer contamination (Lemme et al., 1989 and 1990). The method
accounts for soil, topographic, and geologic data. It uses a revised version of the Universal Soil
-------
Loss Equation3 to determine sensitivity to surface runoff and to establish likely locations for
buffer zones for sensitive areas in the vicinity of all bodies of water.
Four sensitivity classes representing the thickness of overlying material were selected,
ranging from very high sensitivity (0 to 5 feet) to low sensitivity (greater than 60 feet). The
depths to the gravelly sand aquifer material are represented on a map. Data for the project were
derived from a total of 159 borehole logs.
The index method used in South Dakota is a combination classification/scoring method.
The equation that calculates the index incorporates the sorptive capacity of the surface layer with
the effectiveness of the overlying material in protecting the aquifer from contamination. Soil
organic matter content and permeability also are considered. Information from the state soil
survey database was used to calculate the index. The resulting equation integrates soil
information with the thickness of the overlying material to a depth of 60 feet.
Vi = 2.0 - [ (0.2666) (SOM) (St) ] + £^(0.0067 x Lpd)
Where:
Vi = Sensitivity index
SOM = Surface organic matter (%) [(0 to 5)]
St = Surface soil profile thickness (<; 1.5 ft)
Lpd = Layer permeability at depth d (0 to 20 inph/h)
d = Depth (0 to 60 ft)
The resulting sensitivity index values range from 0 to 10. A sensitivity class map was
generated using geologic borehole data based on the following four sensitivity classes: very low (0
to 2), 2 to 6 (low), 6 to 8 (medium), and greater than 8 (high). The authors report that this
method identifies areas requiring intensive management planning. They also suggest that the
chemical characteristics of a pesticide should be considered. In addition, they report that the
indexes or class threshold values developed in this pilot project could be modified for use in
different settings.
o
The equation takes these factors into account: rainfall, soil credibility, slope length, and
slope gradient.
-------
108
mmam
5,2 DESCRIPTIONS OF SELECTED VULNERABILITY METHODS
This section includes one pesticide loading method (RAVE), four simulation models
(GLEAMS, PRZM/PRZM-2, LEACHM, and CMLS), and three pesticide leaching methods
(Ground Water Contamination Likelihood, Jury's Benchmark Approach, and SPISP).
5.2.1 Pesticide Loading Method
5.2.1.1 RAVE
RAVE (Relative Aquifer Vulnerability Evaluation) was developed by the Environmental
Division of the Montana Department of Agriculture (DeLuca and Johnson, 1990). This on-farm
scoring system encompasses the following nine factors: cropping practice; depth to ground water
(in feet); distance to surface water (in feet); percentage of organic matter; pesticide application
frequency; pesticide application method; pesticide leachability; topographic position; and soil
texture. Each factor is assigned a numerical value. Depth to ground water, for example, is
assigned one of four values: 0 (greater than 50 feet); 5 (25-50 feet); 12 (10-15 feet); or 20 (2-10
feet). The method warns against pesticide applications in areas where the water table is within 2
feet of the land surface.
The values assigned to the factors vary. Depth to ground water, soil texture, topographic
position, and pesticide teachability are considered more important factors, and higher numerical
values are assigned to them. The summation of the nine factor values produces an aquifer
vulnerability score ranging from 30 to 100 on the RAVE score card. Scores of 65 and above
indicate a high potential for pesticide contamination. Scores of 80 and above constitute a warning
against the use of a particular pesticide. Scores under 45, and from 45 to 65, indicate a low or
moderate potential for contamination, respectively.
5.2.2 Simulation Models
5.2.2.1 GLEAMS
GLEAMS (Groundwater Loading Effects of Agricultural Management Systems) was
developed by the U.S. Department of Agriculture's Agricultural Research Service (USDA-ARS)
for application to field-sized areas. The method was designed to study the effects of agricultural
-------
management systems on the movement of agricultural chemicals within and through the root zone
(Leonard et al., 1987).
This comprehensive model incorporates rainfall, infiltration, and runoff processes. Runoff
is estimated from daily rainfall using the curve number technique (USDA-SCS, 1972). The
erosion components are relatively sophisticated; they include overland flow, channel flow and
impoundments, as well as the effects of particle-size distribution and the specific gravity of each
particle-size class (Foster et al., 1985). Solute transport is modeled using a storage routing
procedure (i.e., simplified water balance). Seepage from the root zone occurs only when
saturation exceeds field capacity.
GLEAMS divides the root zone into a minimum of three and a maximum of 12
computational layers. Pesticide mass transport is based on a complete mix model coupled with
the soil/water submodel. Other pesticide processes handled by GLEAMS are: degradation,
volatilization, plant uptake, extraction into runoff, metabolites, and transport with sediments. Soil
properties and degradation rates can be varied with depth.
The GLEAMS model accounts for the difference between pesticide application on plant
foliage and application on the soil surface. Different degradation rates also can be specified for
chemicals on foliage and within the soil. GLEAMS will simulate up to 10 chemicals
simultaneously. This feature provides information on the changes in chemical properties. The
model simulates the formation, fate, and transport of pesticide degradation products.
GLEAMS was developed as an offshoot of an earlier USDA model, CREAMS, which
modeled chemicals, runoff, and erosion from agricultural management systems (Knisel, 1980).
CREAMS is a non-point source model for predicting sediment, nutrient, and pesticide losses with
surface runoff from agricultural management systems. GLEAMS builds on CREAMS by adding
components to simulate the movement of water and chemicals within the crop root zone. The
major difference between GLEAMS and CREAMS is the treatment of soil particle-size classes in
the pesticide submodel. The principal focus of the GLEAMS/CREAMS models is to allow for
evaluation of alternative management practices, such as two different tillage methods.
-------
110
5.2.2.2 PRZM/PRZM-2
EPA developed the Pesticide Root Zone Model or PRZM (Carsel et al., 1985) to
evaluate potential pesticide leaching under field conditions. It is often used to assess generic
hydrogeologic settings rather than specific fields. The model describes the complete rainfall-
infiltration-nmoff process as it relates to the fate of pesticides in the environment. Surface runoff
is estimated from daily rainfall using an adaptation of the curve number technique (USDA-SCS,
1972). Erosion and sediment yield are modeled with a modified version of the Universal Soil
Loss Equation (USLE) developed by Williams and Berndt (1977). Solute transport is solved
using a water balance approximation. The model defines percolation or recharge as occurring
only when soil water is computed to be in excess of field capacity. Water is then routed vertically,
one layer at a time, based on the assumption that excess moisture drains from the entire soil
profile within one day. Transport is modeled using a vertical discretization of 5 cm and a time
step of one day.
PRZM allows soil degradation rates to vary with depth. It also includes estimation
routines for some input parameters. Estimates for many other input parameters can be derived
from other databases such as the Natural Resources Inventory Database (NRI) (U.S. Department
of Agriculture, 1987). Daily weather data needed for model runs are available for many localities
across the United States.
PRZM has been used commonly in EPA regulatory assessments and in other research
applications. The users' manual provides good documentation and explains how input parameters
can be obtained or estimated.
Operationally, this flow and transport scheme is sufficient for considering "worst case"
potential leaching of pesticides. The advantages of PRZM are that it is readily available; it is
relatively inexpensive to run; its mathematics are appropriate for the field-size scale; and it has
been applied in many field verification studies. Most of these studies involved aldicarb and
compared model predictions to aldicarb residue concentrations in soil. Field verification tests
have been conducted recently for atrazine, but the model has yet to be tested for a diversity of
pesticides on a site-specific basis.
PRZM has been successfully adapted to employ Monte Carlo techniques for assigning
probabilities of leaching. Dean et al. (1987) compared the results of Monte Carlo stochastic and
-------
deterministic PRZM simulations of aldicarb leaching under citrus plantations in Florida. In a
, Monte Carlo simulation, input values for the parameters are randomly selected from known
distributions of values. The authors concluded that the stochastic approach can improve PRZM
results.
More recently, Carsel et al. (1988 a, b) applied PRZM using a Monte Carlo stochastic
method to make regional assessments of pesticide leaching. The predicted frequencies of
pesticide concentrations compared favorably to concentrations obtained from more than 100 wells
in study areas under peanut fields in North Carolina.
The newest pesticide fate model from EPA is PRZM-2 (1993). The new model is a union
of PRZM and VADOFT - a one-dimensional, finite-element, flow and transport model that
simulates the soil profile from the bottom of the root zone to the top of the water table
(Huyakorn et al., 1987). PRZM describes flow, transport, runoff, and erosion in the root ?one
and at the soil surface. The version of PRZM incorparted into PRZM-2 has a modified
hydrology component more representative of flow within soils than the original version's.
PRZM-2 also includes volatilization of pesticides from soil and plant surfaces. Field verification
Of the linked system (PRZM and VADOFT) has yet to be published.
5.2.23 LEACHM
Leaching Estimation and Chemistry Model (LEACHM) (Wagenet and Hutson, 1986) is a
finite-difference model for simulating the. fate of nonvolatile pesticides in the unsaturated zone.
It is one of the few models that calculates water balance using Richards' equation:
h
Where:
K(0) = Hydraulic Conductivity (CM SEC'1)
0 = Soil Water (CM3 CM'3) Content
Z = Direction, Positive Downward
-------
112
Because it is not in the public domain, users are not able to access the computer model
code for LEACHM. The model is research-oriented and was not intended to consider the effects
of management practices. The LEACHM model also does not consider surface hydrology or
erosion processes. It is flexible regarding boundary and initial conditions and can model the
effects of layered soils, multiple rainfall or evapotranspiration cycles, application of chemicals in
wet or dry form, and plant growth. It also can perform simulations over multiple years. The
model, which has been verified against field data, can simulate multiple metabolites as well as
parent pesticides.
5.2.2.4 CMLS
Chemical Movement in Layered Soils (CMLS) is a one-dimensional simulation model that
locates the leading edge of non-polar chemicals in the unsaturated. zone (Nofziger and Hornsby,
1985). The distribution of chemicals remaining in the soil is portrayed using a simplified water
balance.
The model uses known infiltration and evaporation inputs. Linear adsorption/desorption
isotherms are used to describe chemical affinity for the soil matrix. Dispersion is not considered.
Degradation and retardation are considered, however, and can be varied with depth.
The model is intended as an instructional and screening tool, with simulations easily
repeated for numerous soil types. Infiltration can be varied to assess potential leaching impacts
on ground water quality and runoff. Erosion losses are not calculated. The model also lacks the
ability to examine interrelationships between management practices. For areas of Florida for
which CMLS was developed, or other areas with extremely porous soils and limited topographic
relief, the assumption of no runoff or associated erosion is probably valid.
5.23 Pesticide Leaching Methods
5.23.1 Ground Water Contamination Likelihood
In this method, described in Rao et al. (1985), ground water vulnerability to pesticide
contamination is considered to be dependent on the surface soil horizon and on other parameters
combining characteristics of a given pesticide and soil type. The authors considered in their
evaluation the following factors: (1) travel time and persistence in the crop root zone and in the
-------
intermediate vadose zone; (2) pesticide mass emissions from the vadose zone; (3) pesticide
concentration in the ground water; and (4) pesticide persistence and dilution in the saturated
zone. The authors' objective was to evaluate pesticide behavior in soils using pesticide properties
and site-specific information. They proposed a simple method to rank the relative potentials of
different pesticides to intrude into ground water, and presented a series of equations to define the
totaj amount of pesticide leaching past the crop root zone and vadose zone, based on a
calculation of attenuation and retardation factors.
The equation to calculate the attenuation factor (AF) is:
= Exp
x RF x
The equation to calculate the retardation factor (RF) is:
RF = 1 + BD x PC x Koc + ACxKh
(FC) (FC)
Where:
Where:
AC = P-FC
P = i -BD
PD
BD = bulk density
OC = organic carbon
KQC = sorption coefficient
FC = field capacity fraction
AC = air-filled porosity fraction
PD = particle density
KJJ = Henry's Law Constant
ll/2 = pesticide half-life
^gw = distance from surface to ground water
q = ground water recharge
P = porosity
-------
Values in this index range from 0 to 1. The value indicates the fraction of applied
pesticide that reaches the aquifer after passing through soil surface horizons. A value of 0 means
that none of the applied pesticide has the potential to contaminate ground water, whereas a value
of 1 indicates that all the applied pesticide potentially contaminates the aquifer. The index is
divided into classes indicating the relative likelihood of pollution. Pesticides with values between
0 and 0.0001 are least likely to contaminate an aquifer, while those with values of 0.25 and above
are most likely to cause contamination.
The attenuation factor (AF) and retardation factor (RF) can be used as simple indices for
ranking a number of pesticides. Ranking pesticides based on RF values is recommended in the
absence of information on pesticide half-lives, either in the root zone or in the intermediate
vadose zone.
Rao et al. (1985) calculated ground water contamination potential indices for a total of 41
pesticides commonly used in the United States, then ranked them. The top 10 pesticides using
the AF index are water soluble with half-lives exceeding 40 days. The authors found little
difference between pesticide rankings using either AF or the LEACH index (Laskowski et al.,
1982; see Table 3-6).
A comparison of the AF and RF pesticide ranking permits the evaluation of pesticide
losses by degradation. Volatile pesticides (Kh>10'4) were found to reach the water table faster
than anticipated based on the time for vapor-phase diffusion (tD) and AF ranking schemes.
Diffusion ranking alone overestimates the pesticide's potential to contaminate ground water. It
does not account for pesticide volatile losses at the soil surface or losses that can be attributed to
degradation in the soil. In the AF scheme the soil-water flux in the crop root zone is assumed to
be constant with depth and time.
The AF index is designed to rank pesticides on the basis of their relative potential to
intrude into ground water. Attenuation index ranges can be displayed visually on maps. The
following four assumptions are made to facilitate and simplify use of the index:
(1) The vadose zone properties are independent of depth;
(2) The average ground water recharge rate can be calculated using local
rainfall, irrigation, and evapotranspiration data;
-------
(4)
The KOC value for each pesticide can be estimated assuming that the
hydrophobic interaction is dominant; and
The average t1/2 value for each pesticide can be estimated.
The authors recommend that users of this method exercise caution when testing the
validity of these assumptions. The authors also have proposed an equation to estimate the
average pesticide concentration in ground water, which depends on a determination of mixing
depth. The equation does not account for degradation processes.
5.2.3.2 Juiy's Benchmark Approach
In a series of studies Jury et al. (1983; 1984 a, b, and c) presented procedures to rank
pesticides using a "benchmark" approach. They indicated that the following chemical and physical
pesticide characteristics were critical in assessing pesticide loss through volatilization, leaching,
and/or degradation in soil: phase partitioning coefficients, degradation coefficients, diffusion
coefficients, and the influence of hydrodynamic dispersion. The authors also considered the
pesticide organic carbon partition coefficient, saturated vapor density, solubility, and degradation
half-life. They showed that a chemical with a linear equilibrium partitioning between its vapor,
liquid, and sorbed phases moves with a convective velocity (VE).
Where:
VE = JW/RL = Jw/ (BD Kd + O + aKn)
RL = Ratio of the total concentration to the liquid concentration.
JW = Water flow
Kd = Sorbed-dissolved distribution coefficient
BD = Soil bulk density
KH = Henry's constant
a = Volumetric air content
O = Volumetric water content
This model may be used to define the convection time (to) needed to move a distance Q.
in the presence of a water flux (Jw).
tc =
= (BD Kd + O + aKH) (/Jw
-------
When sorption is relatively high, tc becomes proportional to Kd and BD. In the presence
of non-ionic pesticides, Kd = f^ K^ where f^ = organic carbon fraction and K^. = the organic
carbon partition coefficient.
The authors presented the index, 10 = 60^1^ H/Jw, to describe the relative mobility
of pesticides. Rao et al. (1985) reviewed the work of Jury et al. (1983; 1984 a, b, c) and indicated
that ranking pesticides by their travel time is equivalent to ranking these compounds by their
retardation factor (RF) (see Section 5.2.3.1).
In situations where mass flow by convection is small or negligible, chemicals move through
the soil primarily by liquid or vapor diffusion. The author presented an equation to describe the
diffusion tune (tD) to move a distance fc tD = F/DE
where the soil diffusion coefficient
where:
DE
KD
Do3*1"
Dlwater
a
0
PbKD + etaKD
Henry's constant
gaseous diffusion coefficient in air
liquid diffusion coefficient in water
air content
porosity
Chemicals that move primarily in the vapor phase are characterized by a small tD
tr> = (hfoc Koc + 6 taKtf)
DGair
Pesticides with high diffusion coefficients or Henry's constants are characterized by shorter
diffusion tunes. This index is strongly dependent on the water content. As the air-filled porosity
decreases, pesticide movement will occur predominantly in the aqueous phase.
-------
5.2.3.3 Soil Pesticide Interaction Screening Procedure (SPISP)
SPISP (Goss, 1991) evaluates the potential for a specific pesticide to be lost to surface
runoff and leaching. It is used for screening, and is intended to be the first tier in a tiered
assessment.
In the SPISP method, soil and pesticide leaching factors are ranked based on algorithms
developed during several thousand runs of the GLEAMS model. Two types of leaching potentials
are determined: 1) soil leaching potentials, which take into account only soil factors; and 2)
pesticide leaching potentials, which are based on pesticide properties. Factors used to determine
soil leaching potential are: hydrologic soil group (based on the USDA's National Soils
Handbook); fraction of organic matter content in soil; and soil K factor, which describes soil
erodeability. Pesticide leaching potential is based on the pesticide's organic carbon partition
coefficient (K^), solubility, and half-life.
The overall leaching loss potential is obtained by constructing a matrix showing soil
leaching potential on one axis and pesticide leaching potential on the other axis (Table 5-1). The
method also can be used to determine the potential for an applied pesticide to be lost to surface
runoff either in solution (dissolved phase) or by adsorption to particles (due to erosion). The
leaching loss potential is a relative score ranging from 1 to 4. Scores of 1 or 2 indicate a higher
potential for a pesticide to be lost to leaching. In this case, a second-tier evaluation of the
pesticide using a more comprehensive method would be suggested. Overall leaching potential
scores of 3 or 4 indicate that the pesticide under evaluation would not leach. The method's
boundary of consideration for loss to leaching is the bottom of the root zone.
Several considerations not directly incorporated into SPISP must be taken into account
when interpreting the results:
The pesticide application method. Foliar applications will result in a smaller
amount of pesticide available for leaching. Pesticides incorporated into the soil or
applied in a band beneath the soil surface are more available for leaching.
Rainfall following application. The method assumes that considerable rainfall will
occur immediately after application. If the post-application period is dry, the
method may overestimate leaching potential.
-------
Hydrogeologic features. Hydrogeologic features, such as perched and/or shallow
ground water, fractured bedrock, macroporosity, karst topography, aquifer recharge
areas, and lithic soils, must be accounted for.
Table 5-1. Pesticide and Soil Leaching Potentials (Goss, 1991).
Soil Leaching
Potentials
High
Intermediate
Low
Very Low
Large
1
1
2
3
Pesticide Leaching
Medium
1
2
3 '
3
Potentiate
Small
2
3
3
4
Extra Small
3
4
4
4
The higher the number (e.g., 4), the lower the overall pesticide leaching potential.
53 DESCRIPTIONS OF SELECTED STATISTICAL TOOLS
This section presents two examples of the use of statistical methods in assessing the
potential for ground water contamination by pesticides. Chen and Druliner (1987) presented a
multiple regression model to relate pesticide concentrations in ground water in Nebraska to
several factors. Teso et al. (1988) used a multivariate statistical approach called Fisher's Linear
Discriminant Analysis with soil classification units to determine the probability of having a
contaminated or noncontaminated site (binary approach) in one area as large as a county.
5.3.1 Multiple Regression Analysis
Chen and Druliner (1987) presented a multiple regression method and described results
after a one-year study of existing ground water contamination within six study areas in Nebraska.
The report constitutes the "reconnaissance phase" of a national USGS effort to develop methods
useful for evaluating ground water contamination and for determining the effects of human
activity and local hydrology on ground water quality.
-------
The researchers focused on nine independent variables suspected of affecting triazine-
herbicide concentrations, including: hydraulic gradient; hydraulic conductivity; specific discharge;
depth to water; well depth; annual precipitation; soil permeability; irrigation-well density; and
nitrate concentration. The researchers applied nonparametric statistics and stepwise multiple
regression analyses to data for each of these variables as a way to identify and explain the relative
significance of factors affecting triazine concentrations in ground water. The regression equation
applied to the data was:
Where:
Y
a =
brb9 =
X, =
X3 =
X4 =
X =
X
7 -
X$> =
Y = a + bjXi + b2X2 + . . . +
predicted value of dependent variable (triazine-herbicide concentration)
regression constant
regression coefficients
hydraulic gradient
hydraulic conductivity
specific discharge
depth to water
well depth
annual precipitation
soil permeability
irrigation-well density
nitrate concentrations
Statistical analyses suggested that two variables - specific discharge and well depth - explained
a significant amount of the variation in triazine-herbicide concentrations. The presence of nitrate
concentration as an independent variable also appeared to be important The best regression equation
giving the highest multiple correlation coefficients (R2 = 0.84) used two variables, nitrate
concentrations (Xp) and specific discharge (X3): Y = -0.11 + 0.027(X9) + 7.693(X3).
Druliner (1989) presented a final report on the Nebraska project (based on three years of
data) following the preliminary paper by Chen and Druliner (1987). The complete database
included 24 independent variables. The multiple regression analysis with the larger database
considered factors for: average hydraulic conductivity of the unsaturated zone; specific
conductance; irrigation-well density; pesticide application date; average screened well depth; and
-------
depth to water. Taken together, these factors accounted for the largest variations in atrazine
concentration in ground water. The results of the three-year data model are shown in Table 5-2.
Another project used a random survey to determine the degree and extent of drinking
water contamination in farmstead wells by pesticides, volatile organics, heavy metals, and other
inorganic compounds (Steichen et al., 1988). this study used a multiple regression equation to
predict nitrate concentrations and suggested another equation relating pesticide concentrations in
ground water to the age of a well, land use around the well, and distance to the closest possible
source of pesticide contamination.
Multiple regression analyses require a high level of expertise and knowledge in
hydrogeology to identify appropriate independent variables. They also require expertise in the
design, application, and interpretation of statistical analyses. They may also demand field data
collection (e.g., water sampling and laboratory analyses) if such data are not available.
Table 5-2.
Multiple Regression
Model (Druliner, 1989).
Dependent variable: Triazine concentration
Independent
Variable8
Hydraulic
conductivity
Specific
conductance
Irrigation well
density
Pesticide
application
date
Well depth
Depth to water
Multiple
R-Squared
0.431
0.671
0.752
0.781
0.805
0.838
Change in
R-Squared T-Ratio
0,431 6.45
0.240 3.44
0.981 3.64
0.029 -2.04
0.024 3.09
0.033 -2.40
Regression
Coefficient
0.00569
0.0005
0.1021
-0.0054
0.0023
-0.0041
a This list includes only those independent variables that explained significant amounts of
variation in the dependent variable (T-ratio greater than 2.00).
-------
121
••••••••••••••••••
5.3.2 Discriminant Statistical Analvsis/Soil Taxonomy and Surveys
Teso et al. (1988) presented a method that predicts ground water contamination by
pesticides using the Bureau of Land Management's U.S. Rectangular Coordinate Survey System,
the Soil Conservation Service's Cooperative Soil Survey,4 and a regional well inventory. ITie
method was verified with data from another location and was found to be an excellent predictive
tool. It can predict sensitivity to contamination in any location that has soil mapping units and an
inventory of well water analyses.
The method uses Fisher's Linear Discriminant Functions, a multivariate statistical tool
whereby soil classification names and their areal extent in field maps are used to discriminate
between sites of existing and potential contamination. Teso et al. used soil survey reports to
evaluate contamination problems in an area as large as a county and as small as a 0.03 square-
mile section of a township-range survey system (Cazier, 1976).
5.4 METHOD CONTACTS
Readers interested in obtaining more information about the methods and models
described in this document can refer to Appendix A, which includes a list of authors and contact
addresses.
Soil survey reports are a valuable source of information about land and soil characteristics.
Modern U.S. surveys, most of which are published at a scale of 1:15,840 by the USDA Soil
Conservation Service, are geared toward providing information at the farm level.
-------
122
5.5 SELECTED BIBLIOGRAPHY
Alexander, WJ. and S.K. Liddle. 1986. Ground water vulnerability assessment in support of the
first stage of the National Pesticide Survey. Proceedings of the Agricultural Impacts on
Ground Water, Aug. 11-13, 1986, Omaha, NE; National Water Well Association, Dublin,
OH, pp. 77-87.
Aller, L.T., J.H. Lehr, and R.J. Petty. 1985. DRASTIC: A Standardized System for
Evaluating Ground Water Pollution Potential Using Hydrogeologic Settings.
U.S. Environmental Protection Agency, 600/2-85/018, Ada, OK, 163 pp.
Aller, L.T., T. Bennett, J.H. Lehr, RJ. Petty, and G Hackett. 1987. DRASTIC: A
Standardized System for Evaluating Ground Water Pollution Potential Using
Hydrogeologic Settings. U.S. Environmental Protection Agency, 600/2-87/035,
Ada, OK, 622 pp.
Bates, R.L., and J.A. Jackson. 1980. Glossary of Geology. American Geological Institute, Falls
Church, VA.
Becker, R.S., D. Herzfeld, K.R. Ostlie and E.J. Stamm-Katovich. 1989. Pesticides: Surface
Runoff, Leaching and Exposure Concerns: Minnesota Extension Service Publication
AG-BU-3911, University of Minnesota.
Brady, N.C. 1974. The Nature and Properties of Soils, 8th Edition. Macmillan Publishing Co.,
Inc.
Carsel, R.F., R.S. Parrish, R.L. Jones, J.L. Hansen, and R.L. Lamb. 1988a. Characterizing the
uncertainty of pesticide leaching in agricultural soils. Journal of Contaminant Hydrology
2:111-124.
Carsel, R.F., R.L. Jones, J.L. Hansen, R.L. Lamb, and M.P. Anderson. 1988b. A simulation
procedure for groundwater quality assessments of pesticides. Journal of Contaminant
Hydrology 2(2):125-138.
-------
Carsel, R.F., LA Mulkey, M.N. Lorber, and L.B. Baskin. 1985. The Pesticide Root Zone Model
(PRZM): A Procedure for Evaluating Pesticide Leaching Threats to Ground Water.
Ecological Modeling 30:49-69.
Gates, K.J., and F.W. Madison. 1990. Soil-Attenuation-Potential Map of Pepin County,
Wisconsin. University of Wisconsin - Extension. Madison, Wisconsin. Soil Map 10,
Cazier, L. 1976. Surveys and Surveyors of the Public Domain: 1785-1975. Bureau of Land
Management, U.S. Department of the Interior, Washington, D.C. 227 pp.
Chen and Druliner. 1987. Nonpoint Agricultural Chemicals in Ground Water in Nebraska.
U.S.Geological Survey Water Resources Investigation Report 86-4338.
Dean, J.D, P.P. Jowise, AS. Donigian Jr., R.F. Carsel, and LA Mulkey. 1984. Rapid assessment
methodology for leaching of agricultural chemicals. Innovative means of dealing with
potential sources of ground water contamination. Proceedings of the Seventh National
Ground Water Quality Symposium, September 26-28, 1984, Las Vegas Nevada. National
Water Well Association, 500 W. Wilson Bridge Road, Worthington, OH pp. 246-272.
Dean, J.D., E.W. Stecker, AM. Salhotra, and L.A Mulkey. 1987. Exposure Assessment for the
Pesticide Aldicarb in Florida. Environmental Protection Agency, EPA Report No.
EPA/600/D-87/190, June 1987, 14 pp.
DeLuca, T., and P. Johnson. 1990. Rave: Relative Aquifer Vulnerability Evaluation. An
On-Farm Scoring System to Evaluate Aquifer Vulnerability to Pesticide Contamination.
Montana Department of Agriculture - Environmental Management Division. Helena,
MT. MDA Technical Bulletin 90-01.
Driscoll, F.G. 1986. Groundwater and Wells, Second Edition. Johnson Division, St. Paul, MN.
Druliner, AD. 1989. Overview of the relations of nonpoint source agricultural chemical
contamination to local hydrogeologic, soil, land-use, and hydrochemical characteristics of
the high plains aquifer of Nebraska. U.S. Geological Survey Toxic Substance Hydrology
Program-Proceedings of 1988 Technical Meeting, U.S. Geological Survey Open File
Report 88-4220.
-------
124
Foster, G.R., R.A. Young, and W.H. Neibling. 1985. Sediment composition for nonpoint source
pollution analyses. Transactions of the ASAE 28(1):133-139, 146.
Friesel, P., M. Kiper, G. Milde, D. Muelhausen, V. Neumayr, J. Pribyl, and V. Schinz. 1983.
Detection and assessment of groundwater contaminations by organic chemicals. Ground
water in water resources planning; proceedings of an international symposium convened
by UNESCO in cooperation with the National Committee of the Federal Republic of
Germany for the International Hydrologic Programme, Aug. 28-Sept. 3, 1983;
OHPUNESCO, Vol. 2, pp. 1043-1053.
Goss, D.W. 1991. Screening procedure for soils and pesticides relative to potential water quality
impacts. Using computer simulation models in pesticide registration decision making.
Symposium/Workshop, Weed Science Society of America, Louisville, KY.
Hearne, G., M. Wireman, A. Campbell, S. Turner, G.P. Ingersall. 1992. Vulnerability of Ground
Water in the Greater Denver Area. U.S. Geological Survey Water Resources
Investigation Report 92-4143. 241 pp. 1 plate.
Holden, L. R., J. A. Graham, R. W. Whitmore, W. Joseph Alexander, R. W. Pratt, S. K Liddle,
and L. L. Piper. 1992. Results of the National Alachlor Well Water Survey.
Environmental Science and Technology. Vol. 26 No. 5, pp. 935-943.
Huyakorn, P.S., H.O. White, J.E. Buckley, and T.D. Wadsworth. 1987. Finite Element Code for
Simulating One-Dimensional Flow and Solute Transport in the Vadose Zone. Technical
report prepared by GeoTrans, Inc., for the U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens, GA.
Jury, W.A., W.F. Spencer, and WJ. Farmer. 1983. Behavior assessment model for trace organics
in soil: L Description of model. J. Environ. QuaL 12:558-564.
Jury, W.A., WJ. Farmer, and W.F. Spencer. 1984a. Behavior assessment model for trace
organics in soil: n. Chemical classification and parameter sensitivity. J. Environ. Qual.
13:567-572.
-------
Jury, W.A., W.J. Spencer, and W.F. Farmer. 1984b. Behavior assessment model for trace
organics in soil: HI. Application of screening model. J. Environ. QuaL 13:573-579.
Jury, W.A., WJ. Spencer, and W.F. Farmer. 1984c. Behavior assessment model for trace
organics in soil: IV. Review of experimental evidence. J. Environ. Qual.. 13:580-586.
Kissel, D.E., O.W. Bidwell, and J.F. Kientz. 1982. Leaching Classes of Kansas Soils. Kansas
State University, Kansas Agricultural Experiment Station, Manhattan, Kansas, Bulletin
641,10 pp.
Knisel, W.G., ed. 1980. CREAMS: A Field-Scale Model for Chemical, Runoff, and Erosion
from Agricultural Management Systems. U.S. Department of Agriculture, Science
Education Administration, Conservation Report No. 26.
Laskowski, D.A., C.A.I. Goring, PJ. McCall, and R.L. Swann. 1982. Terrestrial environment.
Environmental Risk Analysis for Chemicals. R.A. Conway ed. Van Nostrand Reinhold
Co., NY. pp. 198-240.
Lemme, G., C.G. Carlson, R. Dean, and B. Khakural. 1990. Contamination vulnerability indices:
A water quality planning tool. J. Soil and Water Conservation. 2:349-351.
Lemme, G., C.G. Carlson, B.R. Khakural, L. Knutson, and L. Zavesky. 1989. Aquifer
Contamination Vulnerability Maps - A Water Resource Protection Planning Tool. Lake
Poinsett Pilot Project. Plant Science Department, South Dakota State University and
USDA - Soil Conservation Service. Plant Science Department Pamphlet #18.
Leonard, R.A., W.G. Knisel, and D.A. Still. 1987. GLEAMS: Groundwater Loading Effects of
Agricultural Management Systems. American Society of Agricultural Engineers,
Transactions TAAEAJ, Vol. 30, No. 5, pp. 1403-1418, September-October 1987.
Liddle, S.K. 1992. Personal communication to U.S. EPA, Office of Pesticide Programs, July 7,
1992.
-------
Liddle, S.K., and M.G. Ganley. 1988. The use of the DRASTIC Classification System in surveys
of agricultural pesticides in drinking water wells. Proceedings of the Agricultural Impacts
on Ground Water Conference, Des Moines, Iowa.
Liddle, S.K, and R.W. Pratt, and CM. Breidenbach. 1989. Ground Water Vulnerability to
Agricultural Chemicals Geographic Information System (GIS) Pilot Project. Research
Triangle Institute, Research Triangle Park, N.C. RTI/7848/002-02F. 18 pp.
Lohman, S.W. 1972. Ground Water Hydraulics. U.S. Geological Survey Professional Paper 708,
70 pp.
Mason, R.E., L.L. Piper, WJ. Alexander, R.Wi Pratt, S.K. Liddle, J.T. Lessler, M.C. Ganley,
D J. Munch and G. Langner. 1988. National Pesticide Survey Pilot Evaluation Technical
Report No. RTJ/7801/06-02F, 299 pp.
Montgomery, J.H., and L.M. Welkom. 1990. Groundwater Chemicals Desk Reference. Lewis
Publishers, Chelsea, ML
Moreau, D.H., and L.E. Daniekon. 1990. Agricultural Pesticides and Ground Water in North
Carolina: Identification of the Most Vulnerable Areas. Water Resources Research
Institute of the University of North Carolina. North Carolina State University. Report
No. 252. 31 pp.
Nofziger, D.L., and A.G. Hornsby. 1985. Chemical Movement in Soil. User's Guide. University
of Florida, Gainesville.
Rao, P.S.C., A.G. Hornsby, and R.E. Jessup. 1985. Indices for ranking the potential for pesticide
contamination of groundwater. Symposium Soil and Crop Science of Florida, Proceedings.
44:8.
Soller, David R. 1992. Applying the DRASTIC Model -- a Review of County-Scale Maps. U.S.
Geological Survey Open-file Report 92-297. 36 pp.
-------
Steichen, J., J. Koelliker, D. Grosh, A. Heiman, and R. Yearout. 1988. Contamination of
farmstead wells by pesticides, volatile organics, and inorganic chemicals in Kansas.
Ground Water Monitoring Review GWMRDU, Vol. 8, No. 3, pp. 153-160, Summer 1988.
Teso, R.R., T. Younglove, M.R. Peterson, D.L. Sheeks HI and R.E. Galavan. 1988. Soil
taxonomy and surveys: Classification of areal sensitivity to pesticide contamination of
groundwater. Journal of Soil Water Conservation, Vol. 43, No. 4, pp. 348-352.
U.S. Department of Agriculture, Soil Conservation Service (USDA-SCS), 1972. National
Engineering Handbook: Section 4, Hydrology. Washington, D.C. 548 pp.
U.S. Department of Agriculture. 1987. "Basic Statistics - 1982 National Resources Inventory."
Statistical Bulletin 756. Soil Conservation Service and Statistical Lab. Iowa State
University, Ames IA.
US. Environmental Protection Agency. 1985. DRASTIC: A Standardized System for Evaluating
Ground Water Pollution Potential Using Hydrogeologic Settings. Robert S. Kerr
Environmental Research Laboratory. EPA 600/2-85/018, 163 pp.
U.S. Environmental Protection Agency. 1987. Guidelines for Delineation of Wellhead Protection
Areas. Office of Ground Water Protection, Washington, D.C. EPA 440/6-87-010.
U.S. Environmental Protection Agency. 1989. Transport and Fate of Contaminants in the
Subsurface. Seminar Publication, Center for Environmental Research Institute,
Cincinnati, OH.
U.S. Environmental Protection Agency. 1990. National Survey of Pesticides in Drinking Water
Wells, Phase I Report. Office of Water and Office of Pesticides and Toxic Substances.
EPA 570/9-90-015. 530pp.
U.S. Environmental Protection Agency. 1992. National Survey of Pesticides in Drinking Water
Wells, Phase H Report. Office of Water and Office of Pesticides and Toxic Substances.
EPA 579/09-91-020. 176pp.
-------
U.S. Geological Survey, Ground Water Subcommittee of the Federal Interagency Advisory
Committee on Water Data. 1989. Federal Glossary of Selected Terms, Subsurface Water
Flow and Transport, 38 pp.
Verschueren, J. 1983. Handbook of Environmental Data on Organic Chemicals, Second Edition.
Van Nostrand Reinhold Co., New York,'NY.
Villeneuve, J.P., P. Lafrance, O. Banton, P. Frechette and C. Robert. 1988. A sensitivity analysis
of adsorption and degradation parameters in the modeling of pesticide transport in soils.
Journal of Contaminant Hydrology, Vol. 3, No. 1, pp. 77-96.
Wagenet, R J., and J.L. Hutson. 1986. Predicting the fate of nonvolatile pesticides in the
unsaturated zone. J. Environ. Qual. 15:315-322.
Williams, J.R., and H.D. Berndt. 1977. Sediment yield prediction based on watershed hydrology.
Transactions of ASAE 20:1100-1104.
-------
6.0 TWO CASE STUDIES
This chapter presents examples of how two States, Idaho and Minnesota, selected and
implemented aquifer sensitivity/ground water vulnerability assessment methods. The States
followed many of the steps described in this document for choosing a method. The reader is
cautioned, however, that assessment programs in both States are continuing to evolve and may be
different from the descriptions provided here.
6.1 SNAKE RIVER PLAIN, IDAHO
6.1.1 Introduction
The Idaho Department of Health and Welfare's (IDHW) Division of Environmental
Quality initiated a ground water vulnerability project (the Idaho Project) in 1988. * Although
IDHW defines vulnerability as a combination of aquifer sensitivity and contaminant loading, the
mapping methods developed by IDHW at the time of this writing assessed aquifer sensitivity only
and did not take contaminant loading into account. Future phases of the project will include
contaminant loading potential information as well as other criteria to improve the reliability of the
assessment methods.
The general goal of the Idaho project is to focus the State's efforts related to prevention
of ground water contamination. IDHW anticipates that vulnerability maps developed in this
project may be used in a broad range of programs, including those focused on pesticides and
other agricultural chemicals. The objectives of the Idaho Project are to:
1 Assisting IDHW are the Idaho Department of Water Resources (IDWR), the U.S. Geologic
Survey (USGS) and the U.S. Department of Agriculture (USDA) Soil Conservation Service
(SCS). .
-------
130 _
mm
1) Assign priorities for ground water management and monitoring programs;
2) Build public awareness of the vulnerability of ground water to contamination;
3) Assist in the development of regulatory programs; and
4) Provide access to technical data through the use of a computerized geographic
information system (GIS).
The project will include sensitivity/vulnerability maps at a 1:100,000 scale in an area of the
Snake River Plain and surrounding tributary valleys covering 33,980 square miles. The area
requires 20 different USGS maps (1:100,000 scale).
6,1.2 Selection of an Assessment Method
IDHW chose to develop a sensitivity assessment method specific to the Snake River Plain
that used DRASTIC as the initial method. The department evaluated other available methods
but regarded them as inappropriate for the conditions and data available in the Snake River Plain.
The resulting assessment method is different from DRASTIC in several ways: it used different
sources of information; the mapping scale was finer; a scoring system more appropriate to local
conditions was devised; and a GIS2 was used (DRASTIC was developed using standard
cartographic methods).
6.13 Selection of Key Factors
IDHW selected three of the seven DRASTIC factors to calculate hydrogeologic
sensitivity: depth-to-water, recharge, and soil characteristics. Each factor is represented by a
separate GIS layer. The depth-to-water factor incorporates water-level information from more
than 1,200 wells. The soil characteristic factor contains detailed and soil-specific data obtained
from two Soil Conservation Service (SCS) databases. Due to the low precipitation on the Snake
River Plain, the recharge factor was determined by adding irrigation, the largest contributor to
ground water recharge, to precipitation.
2 The project used ARC/INFO GIS software.
-------
Other DRASTIC factors were not considered in the project because the data were not
available and/or the cost to gather data and develop additional GIS layers was high. For example,
topography was not considered because it was judged to be the least important factor for future
improvement of the initial assessment. However, another DRASTIC factor that was not
considered, impact of the vadose zone layer, was judged to be critical in determining sensitivity in
Idaho, so IDHW plans to compile the necessary data to add a GIS layer for that factor when
additional funds become available. Data availability and funding are the driving forces behind the
IDHW's phased approach.
In cooperation with the Idaho Water Resource Research Institute, the State plans to
assess pesticide vulnerability as a combination of previously mapped sensitivity factors plus
pesticide loading. Further refinement of the soil characteristics layer - such as adding organic
matter content as a factor to reflect pesticide sorption potential - also is under consideration.
Depending on the results of this future study, the factor rating schemes and the criteria used to
distinguish sensitivity may be adjusted. The long-term objective of the Idaho method is to
continually refine maps as more information becomes available.
6-1.4 Idaho's Modified Scoring System
The Idaho Project uses a modified DRASTIC scoring system. Individual-factor ratings
were revised to reflect conditions in Idaho, and one factor, soil characteristics, was divided into
four subfactors. The initial ratings, which are summarized in this section, represent best scientific
judgement and may be adjusted in the future based on comparison of the sensitivity maps with
monitoring data from wells.
The depth-to-water GIS layer, prepared by the USGS, used water-level data from a
network of more than 1,200 wells. USGS used a geostatistical technique called Kriging (see
Section 4.2.2) to generate a water-level surface. Depth-to-water information was obtained by
subtracting the water-level surface from land-surface elevations using a FORTRAN program.
Depth-to-water ratings were based on the ranges listed below:
-------
FACTOR
Pepth-to-Water Factor
1 to 25 feet
26 to 50 feet
51 to 100 feet
101 to 250 feet
> 250 feet
Rating (Points')
50
35
20
10
1
The recharge GIS layer prepared by the Idaho Department of Water Resources combined
subfactors for irrigated areas (dry vs. irrigated), irrigation type (sprinklers vs. gravity-fed), and
land cover types (range lands, agricultural lands, forest, lava flows, and riparian areas). Different
conditions were assigned the recharge ratings listed below:
Recharge Class
Gravity-fed irrigated land
Riparian areas
Sprinkler-fed irrigated land
Forests
Dry land agriculture
Range land
Bare rock (lava flows)
Urban areas
Surface water
Rating (Points')
50
50
40
30
20
20
10
No rating
No rating
The soils GIS layer incorporated data from the State Soil Geographic Database
(STATSGO) and the USDA SCS SOILS-5 database. Four soil-land subfactors chosen from the
information in the SOILS-5 database were rated for each soil serksj: permeability of the most
restrictive layer, depth-to-bedrock (from land surface), depth-to-water-table (from land surface),
and flooding frequency. The permeability of the most restrictive layer depends on the thickness
of the layer (e.g., a thin restrictive layer underlain by gravels would not be considered restrictive).
The ratings for each soil subfactor based on professional judgement are listed below.
-------
SOIL CHARACTERISTICS
Permeability (in/hr)
No soil
Very Rapid (>20.0)
Rapid (6.0 - 20.0)
Moderately rapid (2.0 - 6.0)
Moderate (0.6 - 2.0)
Moderately slow (0.2 - 0.6)
Slow (0.06 - 0.2)
Very slow (<0.06)
Limiting layers (e.g., duripan)*
Depth to-Water Table
Within 60 inches
Greater than 60 inches
Rating (points)
20
20
16
12
8
6
4
2
2
Rating (points)
8
0
Depth-to Bedrock finches)
Absent (no soil)
Very shallow (0 - 10)
Shallow (10 - 20)
Moderately deep (20 - 40)
Deep (40-60)
Very deep (>60)
Rating (points)
10
9
8
5
2
1
Flooding Frequency
Frequent
Occasional
Rare
None
5
4
2
0
Duripan is a mineral soil horizon that is cemented by silica, usually in
its opal or microcrystalline forms, to the point that air-dry fragments
will not slake in water or hydrochloric acid. It may also have accessory
cement such as iron oxide or calcium carbonate.
-------
134
Ratings for each subfactor were summed and multiplied by three to get the overall soil
factor score. For mapping units known to contain more than one dominant soil series, the rating
for each series was weighted according to the amount of area it represented in the mapping unit.
The final hydrogeologic sensitivity score is the sum of the three factor ratings for depth-to
water, recharge, and soil. A hydrogeologic sensitivity map was generated by merging these three
layers. The map was then subdivided into four sensitivity classes: low, moderate, high, and very
high, which accounted for 30, 30, 30, and 10 percent, respectively, of the area mapped. A pilot
mapping program was carried out in the Lake Walcott l:100,000-scale quadrangle (Figure 6-1)
and mapping of the entire State is planned (excluding wilderness areas). Additional layers to
represent the remaining four DRASTIC factors could be developed as resources become
available.
Sensitivity maps for the Snake River Plain were published by IDHW with the following
caveats:
1) Because of the small scale of mapping, maps should be used only for regional
planning purposes and not for site-specific decisions (e.g., the application of a
restricted-use pesticide on a farm field). An area may have an inclusion with
dissimilar ratings that is too small to depict at the scale of the map.
2) The maps illustrate the relative sensitivity of areas and should not be interpreted
as showing areas that will be, cannot be, or already have been contaminated. The
maps do not take into account factors such as potential contaminant sources or the
individual characteristics of the contaminants.
3) If an area has low ratings, it does not assure that unrestricted land-use activities in
that area are acceptable. Because virtually any ground water resource can become
contaminated when subjected to improper land-use practices, restricted-use
pesticides should not be applied in low-rated areas unless an effective ground
water monitoring program is in place.
-------
H
I
I
1
I
-------
In the second phase of the Idaho Project, results of the assessment method are being
selectively compared to monitoring data. This validation study compares ground water sensitivity
maps for the Snake River Plain with monitoring data from the USGS water quality database,
WATSTORE. Data representing about 2,300 sample analyses for wells located in the Snake
River Plain were collected.
Data selected from the database included latitude and longitude of each well location, the
sample date, the depth from a measuring point near the land surface to the water table, the depth
of the well, dissolved nitrate (filtered), and total nitrate (unfiltered). Very few wells had data for
total nitrate because USGS requires samples for nitrate analysis to be filtered. Furthermore, few
wells had data for depth-to-water-table or total well depth. Nitrate was selected for the validation
study "because it more closely resembles the movement of water, which is what the vulnerability
model is designed to predict."3
A statistical comparison of the monitoring data with the sensitivity maps is in progress.
Preliminary conclusions include the following:
• Based on analysis of 2,307 samples in the database, only 1.6% of the sampled wells
in areas assigned low and medium ratings had nitrate concentrations above 10
milligrams per liter (mg/L);
• There were only two locations :on the Snake River Plan where sensitivity maps did
not closely correlate the sensitivity of the area to nitrate contamination. At both
locations it is likely that previously identified poor waste management activities
preclude the use of this assessment method; and
• Based on this validation study, it is evident that the Idaho sensitivity maps are
useful for prioritizing ground water management activities. Figure 6-2 shows the
comparison of nitrate concentrations to sensitivity ratings for the Snake River
Plain.
3 Written communication from IDHW to EPA Region 10, February 26, 1992.
-------
137
il
f
I
ssl
-o
I
.S 5
Bfi
-------
138
6.2 MINNESOTA AQUIFER SENSITIVITY AND VULNERABILITY ASSESSMENT
PROGRAMS
6.2.1 Introduction
The Minnesota Ground Water Protection Act of 1989 requires the Minnesota
Department of Natural Resources (MDNR) to develop sensitivity criteria, to map sensitive
areas,4 and to indicate the "type of risk of ground water degradation that may occur from
activities at or near the surface."
In order for Minnesota to implement the law, a set of criteria for identifying and
delineating sensitive areas has been developed.
6.2.2 Criteria and Approach for Minnesota Time of Travel Sensitivity Assessments
The Minnesota multi-agency workgroup that developed these criteria determined that they
should be based on properties of the geologic materials overlying the ground water. The
•workgroup based the sensitivity of the overlying materials on the known or estimated time it
would take for a generic water-borne contaminant to travel vertically from its source at or near
the land surface to the aquifer (Geologic Sensitivity Workgroup [GSW], 1991, page 1). The
workgroup assumes the travel tune for the generic contaminant to be equal to that for water.
Areas considered most sensitive to ground water contamination have the shortest travel times,
while the least sensitive areas have the longest travel times.
The workgroup has defined sensitivity criteria as "five overlapping ranges of travel times
that have been assigned relative sensitivity ratings from Very High to Very Low" (MDNR, 1991,
and GSW, 1991). Travel time was selected as the criterion because it can be measured and
because it clearly conveys the concept of ground water contamination to the public. Sensitivity
ratings and travel times are illustrated in Figure 6-3. The overlap of travel times illustrates the
imprecision inherent in measuring geologic factors that affect travel time.
4 The Act defines a sensitive area as "a geographic area defined by natural features where
there is a significant risk of ground water degradation from activities conducted at or near the
land surface."
-------
139
1 TIME RANGE- DYE TRACE |
VERY HIGH
1 TW
1
IE RANGE -Tl
HIGH
HTIUM
I
| MOD
|
=]
[LOW
(TIMERANGE-14C|
1
f VERY LOW P^»
Ground Water Travel Times
Log10 Hours
Figure 6-3 Geologic Sensitivity Ratings and Ground Water Travel Times for Minnesota. The
upper bars show time ranges for verification using dye tracing, tritium, and
carbon 14 studies (figure from MDNR, personal communication, March 20, 1992
and GSW, MDNR, 1991).
The Minnesota time-of-travel (TOT) sensitivity assessment method is intended as a starting
point for assessing aquifer sensitivity and ground water vulnerability for a specific aquifer. Sensitivity
assessment maps produced by the method are interpretations of geologic and hydrologic data and are
intended as advisories to local governments fnr ormmH waiw nmtft-tinn ^t^M** n* ,,,»n „„ „ 1
-------
140
6.23 Discussion of Assessment Methods
The Minnesota workgroup developed a three-level approach to applying these time-of-
travel criteria. Each of the levels may be considered a distinct assessment method, and a local
government may choose to apply one or more of the levels. Table 6-1 provides information about
each level, including the benefits, limitations, and suggested applications.
A Level 1 assessment is a preliminary evaluation of surficial geologic sensitivity using
available information. (MDNR, 1991, Summary Report) (Figure 6-4). In many cases, the only
available information is from geologic maps and soil surveys. A Level 1 assessment does not
require additional data gathering if a surficial geologic map and/or soil survey is available. If this
information is not available, a Level 1 assessment cannot be completed. A Level 1 assessment
estimates the travel time in the vadose zone based on the geologic material present at the ground
surface (GSW, 1991, Criteria and Guidelines, page 12). Minnesota suggests that a Level 1
assessment is the minimum level for: 1) planning, 2) regulation/management, 3) program
implementation, and 4) education, including local comprehensive plan development, public
facilities systems planning, county-wide water resources planning, and local water management
planning (e.g., support for zoning and subdivision regulations).
-------
o
i
B
«
8 -
II
'S
If
if
o co
'•& u
If
11
a o
-------
142
g
cq
vo
I
^
1
1^
I
§
i "s
in >
o--5 S^ —
1111111
o
•c
«2 J2g<2 "• .Q g o
5 -S Z? 8 >
IIII i I
r~f ™»N ^ -^ *«rf
1 1
jf
11
.a S
II
™ OJU
8 2.2
.£ -O S
i-^l
(O *_4 1_<
E SS
C^ O .B
•s i
-------
-------
144
A Level 2 assessment evaluates the geologic sensitivity of materials in the vadose zone
(Figure 6-4). The Level 2 assessment conservatively assumes that contaminants behave similarly
or identically to water, and that they are introduced at the land surface. It also estimates the
vertical travel time of contaminants which, for this assessment, depends only on vertical
permeability and the thickness of the vadose zone. In a Level 2 assessment, vertical permeability
in the vadose zone is considered to be influenced by the geology of the material (grain size
distribution, iithology) and the presence of fractures, joints, and solution features in bedrock.
Minnesota prefers Level 2 assessments for most of the State's regulation and management
activities.
A Level 3 assessment evaluates the sensitivity of aquifers below the water table aquifer by
determining the presence of low permeability (confining) layers that reduce aquifer recharge, and
by evaluating the cumulative thickness of confining layers (Figure 6-4). Sensitivity ratings for the
deeper aquifers reflect the thickness of these confining layers, which in Minnesota usually consist
of clay, clayey till, clay loam till, or shale. Deeper aquifers are considered confined if one or more
ten-foot-thick intervals of confining material are present below the water table aquifer.
Site-specific sensitivity assessments require rigorous evaluation of local variations in the
confining characteristics of geologic materials. In order to assess the sensitivity of a confined
aquifer, site-specific lithologic data must be gathered to define these characteristics, and geologic
mapping must be performed to identify the spatial distribution of confining layers. "If more than
one deeper aquifer occurs in an area,: each deeper aquifer is rated separately....The lower deeper
aquifer will be less sensitive because the cumulative thickness of overlying confining layers will be
greater for the lower deeper aquifer" (GSW, 1991, page 62).
A Level 3 assessment is recommended for relatively few types of activities in Minnesota;
however, it generally is recommended for activities that may have an impact on ground water
quality over a long period of time. Level 3.assessments also are recommended when identifying
areas for facilities that handle or produce toxic materials, such as a sanitary or hazardous waste
landfill, or for water resource protection programs, including wellhead protection programs.
-------
6.2.4 Limitations of Assessment Methods
The three assessment levels described above are intended to assist local governments in
applying the general time-of-travel criteria in relation to ground water sensitivity. Minnesota
cautions users about the limitations of these methods, however. They do not account for:
• Physical and chemical characteristics of contaminants (including mobility,
persistence, and/or transformation over time) and their interactions in earth
materials;
• Whether the contaminant source is at or near the land surface (e.g., pesticides) or
below the land surface (e.g., fuels from leaking underground storage tanks);
• Moisture content of the vadose zone; or
• Differences in behavior of contaminants in the saturated and unsaturated zones.
Additional limitations inherent in each level of assessment are provided in Table 6-1. The
MDNR report explains that the guidelines for completing sensitivity assessments "represent a
qualitative approach to the assessment of geologic sensitivity of ground water resources. They are
designed to use data that are already available, or can be obtained reasonably, in most parts of
the State" (GSW, 1991, page 16). The MDNR report warns that the data used for sensitivity
assessments are not rigorously quantitative; i.e., the criteria are based on time-of-travel estimates
that may be very broad and overlapping.
Furthermore, Minnesota's methods are not intended to take the place of site-specific
studies that may assess ground water sensitivity more accurately. "A study which uses exact and
detailed information, including field measurements, supersedes a study which uses less information
or only estimates of local conditions." (GSW, 1991, page 17).
6.2.5 Pesticide Assessments in Minnesota
The introduction of pesticides to the vadose and saturated zones is of special concern in
Minnesota. The State has developed an assessment method that estimates the potential for a
specific pesticide to leach through a specific soil. This method is discussed in the Minnesota
-------
146
mmm
report on geologic sensitivity, and may be combined with one of the three geologic sensitivity
assessment methods (i.e., Levels 1, 2, 3) listed above.
The Minnesota Cooperative Extension System (MES) worked with the USDA Soil
Conservation Service and Agricultural Extension Service (ARS) to develop a system for rating
soils according to the potential for pesticides to leach through them. Each soil mapping unit is
assigned a rating of "nominal," "intermediate," or "high." Similarly, each pesticide receives a rating
of "small," "medium," or "large" based on its solubility, persistence, and soil adsorption value.
(GSW, 1991, pages 104 and 111). Ratings for the soil mapping unit and for the pesticide are
combined in a matrix that provides a third rating for leaching potential through the soil. The
potentials range from Potential 1 (largest leaching potential) to Potential 3 (smallest).
This method is intended as a guide for managing individual fields in Minnesota. Most soil
and pesticide data are readily available in the State. This method and one of the sensitivity
assessment methods described above may be applied in the same area. For example, combining
assessment methods could result in a soil being rated nominal in leaching potential, but high in
sensitivity, based on a Level 1, 2, or 3 assessment.
In cases where such a discrepancy has been analyzed, according to the MDNR, "It is the
result of a relatively low permeability parent material, for example, loess over outwash, or over
carbonate bedrock. In such a case both systems are accurate in their application. A contaminant
such as a pesticide used in an area rated nominal is unlikely to leach through the root zone, but if
it does, it is likely to reach ground water if the area has high geologic sensitivity." (GSW, 1991, p.
Ill) In such a situation, which does not normally occur, decision criteria for assigning a final
combined rating must be carefully selected.
-------
6.3 SELECTED BIBLIOGRAPHY
Geologic Sensitivity Workgroup. 1991. Criteria and Guidelines for Assessing Geologic Sensitivity
of Ground Water Resources in Minnesota: Minnesota Department of Natural Resources,
Division of Waters, 122 pp.
Idaho Department of Health and Welfare, Division of Environmental Quality. 1991. Ground
Water Vulnerability Assessment, Snake River Plain, Southern Idaho: Idaho Department
of Health and Welfare, Idaho Department of Water Resources, U.S. Geological Survey
25pp.
Idaho Department of Health and Welfare. 1992. Written communication from Paul Jehn
(IDHW) to Bill Mullen (EPA Region 10), February 1992.
Idaho Department of Health and Welfare, Division of Environmental Quality. 1992. Written
communication from Gerry Winter (IDHW) to Jane Marshall (EPA Headquarters), April
1992.
Minnesota Department of Natural Resources. 1991. A Summary Report to the Legislative
Commission on Minnesota Resources, Water Resources Management - Ground Water
Sensitivity, Minnesota Department of Natural Resources - Division of Waters. 10 pp.
Preliminary Document Subject to Revision
-------
.•>•. ?'.".<
-------
APPENDIX A
METHOD CONTACTS
-------
-------
W3
I
-------
C/3
1
z
o
u
Q
O
(A
S
•n
Contact Adi
S
(M
1
6
^^*
^
o
i
•g
§
SO)
w
«i,
g
f
s e
w? C
as o
05 '5
IM 2
USDA-ARS, Southeast Wate
Laboratory
Coastal Plain Experimental S
P.O. Box 946
Tifton, GA 31793
a-
o
"2 _ ^
« "3 c
5 j >2 35 .2
m *? j §
S ^ . ^ .5
o c5 £ eS, &
t
.O
1
c
1
^^^
1
. 1
i I •
5 g.
s
1
2
5T.8
> s
United States Geological Sur
Water Resources Division, F
Denver, CO 80225
e s
8 1-2
S 5 g-
E ^ t*5
NM W
02^
S*
!§
•"S
i
1,
i
1 c
9 s
0 i
i &
P ts
§1
ll
o
United States Environmental
Agency, Region VIII, Office
Water, Denver, CO 80202
kd f/5
QJ M
g S>
a S3
r^
co O
^
•c
G3
1
1
Soil Science Department
University of Rorida
Gainesville, FL 32611
of*
IE ^ *—*
(J • . fi
(fl W W Q\
a; '
-------
A-5
(A
i
•e
e
a
I
OD CO
f
fe ^ *
o 53 «^
le-a
r/^ r» ^
ISO
I
S
I
B
2
#£«
»- BS
o y -t
« .£ -^
S3 C s^
1DZ
l^rf
s.e I
S6«
II
1
1
ark
ces
I
V
ng Indi
Leac
•o
I
ao
(Pesticide Leac
CHM
Simulation Mode
-------
(/)
M
M
D CONTA(
0
52
VJ
Contact Addre
/_s
CO
^
o
j?
"e
•g
i
*i
o> 5£,
1
1
e
t:
§.
Division of Water, Minnesota D
of Natural Resources
500 Lafayette Road
St. Paul, MN 55155-4032
o 2
£>* °3 *3
^* qj *•»
"5 c3 >ffi
*"* :§
*jj2 *5- *tt
c5 ^^
Ills
III!
O ^ Q 05
.|>
2
g
«
'SB
.S
8
CO.
fi I
S 5 .2?
If 1
il 1
g .1
Montana Department of Agricul
Environmental Management Dii
Helena, MT 59602-0205
=j
g S
£ 2
Ji 2
^7 C
Hi ^* O
cO ^ ^
ill
re
i
3
i o
Hi
11 S
§i i
s£ ^
1
CT<
Center for Environmental Resei
Cornell University
Ithaca, NY 14853
2
a
fli x— s.
o ^^
03 OO
OH C^
co d'
1
i
3
I
c
o
S «
CO *S
» §
0 |
•^
1
8
e
(for distribution and informatio
Manin Sailus
Northeast Regional Agricultura
Engineering Service
Riley Robb Hall
Cornell University
Ithaca, NY 14853
g
pj
1
ep
United States Department of A
Northern Plains Area
Ft. Collins, CO, 80522
at
t-4
iis-
o5 ^ ^
t
"2
3
}
o
co "3
O Q
-------
A-7
CO
o a
» O
1
l~l a
co .o
I
I
§
1
CO
-------
D CONTA
O
en
1 "
1
Contact
/w
o
A
•**
^
s
I«ri
^^
11
"3
i
•g
1^
•si
a ^
S
2:
a
o
&" "
o £3
||
•g 1
t-i g
8 S
11 1
S « <°
|i S
g CO W
llfff
is &|
w P < <
t^
a-^ »-5 ^
2 . » i^
• j !Z * ®0
?5»
3
.0
ca
1
p
1
S
t=
€O
~
g 1
£ '^
a
o
•<3
>. Q
u< a>
O «
11
31
imental Research
States Environme
, GA 30605-2720
o "S? fr %
% 1 § 1
g o SbS
[3D «
JS
, oo
P
tg t Q
sill'
•^ 3 3 £
^ OT M
<3gw^
»-, cs »-> <;
&
i
xi
2
M
1
>
s-*.
%
O
S
g
M 1
§ i
£ S-
Q % <* '"'
§ "If •§
Q & «j M
o^oil
O^ JCt
CO
t
I
I
I
1
CO
cc
-------
•
1
»D CONTA<
w
8
Contact Addrei
'£
„§
*
2
H
•§
A
i
1
i?
•s-g
di CO
v ^,
z
ADL. Inc.
Cambridge, MA 02140
S
§ i1—-
« || S>
£*£
*
^M
s
V
J3
>
j^
1
d I
I I
gg
'^ -
$£ o5
United States Environmental Pro
Agency, Office of Toxic Substanc
Washington, DC 20460
l-o
S
o
3 Q
Q
-------
A-10
8
£
B
Q
O
Contact Address
/w>
1
5
.
f*
•g
£
g
Name of Metbod
(Subset)
Universite' du Quebec Institute
National de la Recherche Scientifique
P.O. Box 7500
Sainte-Foy Quebec, Canada G1Z4C7
£ 1
Jail
3 S -; Q PH ^
J
JD
1
"3
}
kt O
o- "g
1 ^
Nutrient and Pest Management, Universit
of Wisconsin-Extension
Madison, WI 53706
e
8\
,a
0
en
Si
^ P-<
•g
2
g
CO
1
1 .s
1 1
CO 5 O
a ^5 '3
•S « i>3
a 53 o ic
1
1 1
8 «- S s-<
*•* €3 ON *•!£
1 SI b
& 3 n 2
« o •§ ^sa
& S S S ~
Q 05 O •- go
^ O ? e« CO
&
'?
;•§
u
CO
u '2
5 o
w O
C8 f/\
If I
*! 1
€§ s
l'| |
^ " ...^.._ ]T -- -
-------
APPENDIX B
GLOSSARY OF TERMS
-------
-------
APPENDIX B
GLOSSARY OF TERMS
Absorption - Physical or chemical attraction and attachment within the matrix of a material (solid
or liquid).
Adjuvants - Constituents of pesticide formulations which enhance pesticide activity (WSSA, 1982).
Examples: wetting agents, emulsifiers, and dispersing agents.
Adsorption - The attraction of ions or compounds to the surface of a solid (Brady, 1974).
Aeolian sand - Sand deposited by wind, e.g., the sand found in dunes (Brady, 1974).
Aerobic - Molecular oxygen is present; active or occurring in the presence of oxygen. For
example, aerobic biodegradation of some pesticides occurs in shallow soil, where sufficient oxygen
is likely to be available.
Agronomy - A specialization of agriculture concerned with the theory and practice of field-crop
production and soil management. The scientific management of land (Brady, 1974).
Alluvial sand - Detrital sand deposited during relatively recent geologic time by a stream or river
as sediment in the bed of the stream or on its floodplain or delta, or as a cone or fan at the base
of a mountain slope (Driscoll, 1986).
Anaerobic - Molecular oxygen is not present; active or occurring in the absence of free oxygen.
For example, anaerobic biodegradation of some pesticides occurs in deep subsoil or ground water
where oxygen is likely to be limited.
Aquifer - A formation, group of formations, or part of a formation that contains sufficient
saturated permeable material to yield sufficient, economical quantities of water to wells and
springs (EPA, 1987).
-------
B-4
Aquifer sensitivity - The relative ease with which a contaminant (for the purpose of this
document, a pesticide) applied on or near the land surface can migrate to the aquifer of interest.
Aquifer sensitivity is a function of the intrinsic characteristics of a hydrogeologic setting and is not
dependent on agronomic practices or pesticides characteristics.
Aquifer sensitivity methods - Methods of assessing aquifer sensitivity that consider only
hydrogeologic factors.
Attenuation - The process of diminishing constituent concentrations in ground water, due to
filtration, biodegradation, dilution, dispersion, sorption, volatilization, and other processes (EPA,
1987).
Benchmark soils - Classes of soils that are representative of significant areas of a state.
Benchmarking - A technique for model or method verification in which a new model is compared
to a similar model that is accepted to be correct. Close agreement of the results of both models
may be taken as evidence for verification of the new model.
Best management practices - Agricultural practices that have been identified as reducing the
potential for agricultural chemicals to contaminate ground water, including pesticide application,
soil tillage, and irrigation practices.
Biodegradation - Molecular degradation of compounds resulting from the complex action of living
organisms (Verschueren, 1983).
Bulk density - The mass of dry soil per unit bulk volume including the air space. The bulk
volume of the soil is determined before drying (Brady, 1974).
Cation - Positively charged ion. An example of a pesticide that forms cations in water is
Paraquat, a herbicide.
Confined aquifer - An aquifer bounded above and below by confining units of distinctly lower
permeability than the aquifer media. Confining units are typically high in clay content (EPA,
1987).
-------
——————MK^^^MMMMMMHMHMi
Conservation tillage - any tillage or planting system that maintains at least 30 percent of the soil
surface covered by residue after planting to reduce soil erosion by water: or where soil erosion by
wind is the primary concern, maintains at least 1000 pounds of flat, small grain residue equivalent
on the surface during the critical erosion period. Three major types of conservation tillage are:
no-till, ridge-till, and mulch-till or other tillage techniques that result in meeting the residue cover
requirements of conservation tillage (US Dept. of Ag, 1989).
DBCP - Dibromochloropropane; a neinaticide.
Degradation - The transformation of a compound into one or more simpler chemicals. Carbon
dioxide is the final aerobic degradational product of most pesticides. Degradation includes the
processes of photolysis, anaerobic and aerobic degradation, and field dissipation.
Discriminant analysis - Multivariate method of statistical inference in which samples may be
categorized into two or more distinct groups based on similar properties or characteristics
(Lachenbruch and Goldstein, 1979).
Duripan - A mineral soil horizon that is cemented by silica, usually opal or microcrystalline forms
of silica, to the point that air-dry fragments will not slake in water or HC1. May also have
accessory cement such as nonoxide or calcium carbonate.
Fractured bedrock - Rock, containing cracks, fissures, joints or faults, which may underlie soil or a
thin layer of unconsolidated deposits (EPA, 1987).
Geographic information system (GIS) - A computerized database/mapping system that may be
used to store, retrieve, and analyze information, such as soils and hydrogeologic data, based on
geographic location.
Geopolitical units - Land delineations based on politically defined geographic areas such as
counties, cities, and townships.
Glacial outwash plains - Deposits, usually in valleys and on plains, left by streams heavily laden
with glacial sediment (Brady, 1974).
-------
Ground truthing - Technique of verifying mapped data whereby a factor shown on the map is
field-checked.
Ground water discharge - Row of water released from the zone of saturation (EPA, 1987).
Ground water vulnerability - The relative ease with which a contaminant (for the purpose of this
document, a pesticide) applied on or near the land surface can migrate to the aquifer of interest
under a given set of agronomic management practices, pesticide characteristics and aquifer
sensitivity conditions.
Ground water vulnerability methods - Methods that integrate hydrologic characteristics, pesticide
characteristics, and agronomic practices.
Half-life - The time required for a compound and/or its transformation products to be reduced to
50 percent of its original amount.
Henry's law constant - Henry's law is associated with solutions of volatile solutes, usually gases in
liquids. It relates the mole fraction of the solute to its partial (vapor) pressure. For a given gas
dissolved in water, the relationship at a given temperature is a constant (Chang, 1981).
HRS - Hazard Ranking System
Hydrogeologic factors - Factors that deal with geology-related aspects of ground water and surface
water, such as depth to ground water, net recharge, and aquifer permeability.
Hydrogeologic setting - Composite description of all the major hydrogeologic factors that affect
and control ground-water movement into, through and out of an area (Aller et aL, 1987).
Hydrolysis - The direct reaction of dissolved compounds with water molecules, resulting in
transformation products (EPA, 1989). Hydrolysis reactions are important degradation processes
for certain pesticides.
Hydrophobic - Tendency of non-polar compounds not to dissolve in water.
Inclusions - Small areas of dissimilar sensitivity located within a delineated area on a map.
-------
Karst - Areas composed of limestone and gypsum characterized by sinkholes, caves, and solution
channels, formed by dissolution by ground and surface water (EPA, 1987).
Kriging - A technique in which sampling data is used to estimate the values (such as water levels)
at unsampled locations. Kriging is frequently used to make contour maps.
Leaching - Removal of constituents in solution from rock, soil or waste (USGS, 1989); separation
or dissolution of soluble constituents from a porous medium by percolation of water (EPA, 1987).
Macroporosity - The presence within a geologic material of pores too large to hold water by
capillarity. Macroporosity may result in enhanced migration of pesticides into ground water.
Map delineations - The smallest units of resolution on an aquifer sensitivity or ground water
vulnerability map.
Mapping scale - See scale.
Mass loading - The amount of a pesticide applied per unit land area.
Mineralization - The conversion of an element from an organic form to an inorganic state as a
result of biodegradation. The mineralization, under aerobic conditions, of hydrocarbons
(molecules containing only carbon and hydrogen atoms) results in the formation of carbon dioxide
and water molecules.
Mobility - Ability of a pesticide or its transformation produces) to be transported by runoff,
volatilization, wind erosion, leaching and ground-water flow.
Monte Carlo procedure - Procedure that randomly samples a suite of input parameters in a
computer model simulation. The user selects the suite of input parameters (e.g., normal
distribution, binomial distribution, log-normal distribution).
Multiple regression - Method of statistical inference in which two or more independent variables
are used as predictors of the dependent variable.
-------
B-8
Non-parametric analyses - Statistical techniques that do not depend on data being drawn from a
particular distribution; distribution-free techniques.
Nomogram - A graph having three coplanar curves, each graduated for a different variable so that
a straight line cutting all three curves intersects the related values of each variable.
Normative statistics - Statistical methods which assume that data are normally distributed.
Octanol-water partition coefficient, K^ - A characteristic property of a chemical, expressed as the
ratio of the solute concentration in the water-saturated n-octanol phase to the solute
concentration in the n-octanol-saturated water phase (Montgomery and Welkom, 1990). KOW is
used in several of the simulation models as an indicator of a pesticide's tendency to avoid
dissolution in water, and an indicator of its sorption potential.
Organic carbon partition coefficient, K^ - A characteristic property of a pesticide, expressed as
the ratio of sorbed pesticide per unit weight of sediment organic carbon to the aqueous pesticide
concentration. K^. is used in hybrid and vulnerability assessment methods to quantify a pesticide's
potential to sorb to soil organic matter.
Oxidation - A chemical reaction in which a chemical loses an electron(s), resulting in an increase
in valence; the process of combining with oxygen.
Partition coefficient, K$ - The ratio of the concentration of a chemical sorbed to the solid phase
to its concentration in the aqueous phase (EPA, 1989). K
-------
B-9
Pesticide - A naturally occurring or synthetic chemical used to kill pests, including insects, weeds,
fungus, nematodes, rodents, and bacteria.
Pesticide leaching methods - A narrowly defined subcategory of vulnerability methods that
incorporate both hydrogeologic and pesticide factors.
Photolysis - The physical reaction(s) that may occur when a molecule absorbs light, causing it to
transform to a higher- or lower-molecular-weight chemical(s). Photolysis reactions are not
important degradation processes for pesticides occurring in soils below the surface or in ground
water.
Polar - Having an uneven distribution of charge, so that one end of a molecule is predominantly
positive and the other end is predominantly negative. Polar pesticide molecules tend to be water-
soluble.
Pore space - Voids within a rock or soil.
Pore water - Subsurface water in the voids of a rock or soil (Bates and Jackson, 1980).
Recharge - Water derived mainly from precipitation or irrigation that percolates from the ground
surface through the soil and vadose zones into an aquifer (EPA, 1987).
Reduction - Chemical reaction in which a chemical gains an electron(s), resulting in a decrease in
valence. An example of a reduction reaction includes the reductive dehalogenation of certain
chlorinated pesticides.
Riparian - Pertaining to or situated on a body of water, such as a river or stream.
Root zone - The soil layer six to sixty inches in depth from the soil surface capable of supporting
crop growth.
Saturated zone - Subsurface zone in which ground water fills all pore space (EPA, 1987).
Scale - A progressive classification of size. In mapping, scale is usually expressed as a proportion
relating the size of the map to the size of the area represented by the map.
-------
B-10
••MM
Scoring method - Method that characterizes sensitivity and vulnerability to pesticide pollution by
using a numerical weighting system (i.e., a score). The score reflects relative sensitivity and
vulnerability.
Simulation model - A conceptual description of a physical system (i.e., a simplified version of a
physical system, including soils, unsaturated and saturated geologic materials) and the associated
mathematical representation that is used to estimate responses to changes in the initial system as
a function of internal or external changes and as a function of time and space (modified from
U.S. Geological Survey, Ground Water Subcommittee of the Federal Interagency Advisory
Committee on Water Data, 1989, and Bear and Verruijt, 1987).
Solubility (of water) - The total amount of solute species that will be dissolved in water at a given
temperature and pressure.
Sorption - Processes that remove solutes from the fluid phase and concentrate them on the solid
phase of a medium (USGS, 1989); a term used to encompass absorption and adsorption (EPA,
1987). Sorption of pesticides by soils and aquifer materials reduces their mobility.
Surfactant - A chemical or mixture of chemicals that reduces the surface tension of liquids, or
reduces the interfacial tension between two liquids, or between a liquid and a solid. A surfactant
is an example of an adjuvant.
Transformation - The physical, chemical, and biological processes by which a pesticide molecule is
altered to form a higher- or lower-molecular-weight chemical(s).
Transmissivity - The rate at which water of the prevailing kinematic viscosity is transmitted
through a unit width of an aquifer under a unit hydraulic gradient (Lohman, 1972). Transmissivity
is a measure of the ability of an aquifer to transmit water.
Vadose zone - Unsaturated zone, the zone where pore spaces are not completely filled with
water.
Van der Waals' forces - Forces between the molecules of a non-polar compound -- forces include
the dipole-dipole, ion-induced dipole, dipole-induced dipole, and London interactions
(Chang, 1981).
-------
B-ll
Volatilization - The conversion of a chemical from the solid or liquid state into the gaseous state.
Water table - Level below which the soil or rock unit is saturated with water and at which the
hydraulic pressure is equal to atmospheric pressure (Brady, 1974).
-------
B-12
References Cited
Aller, L-, T. Bennett, J.H. Lehr, RJ. Petty, and G Hackett. 1987. DRASTIC: A
Standardized System for Evaluating Ground Water Pollution Potential Using
Hydrogeologic Settings. U.S. Environmental Protection Agency, 600/2-87/035,
Ada, OK, 622 pp.
Bates, RJL, and J.A. Jackson. 1980. Glossary of Geology. American Geological Institute, Falls
Church, VA.
Bear, J. and A. Verruijt. 1987. Modeling groundwater flow and pollution. Theory and
applications of transport in porous media. Ed. J. Bear. D. Reidel Publishing Co.,
Dordrecht, Netherlands.
Brady, N.C. 1974. The Nature and Properties of Soils. Macmillan Publishing Co., Inc, New
York, NY. 639pp.
Chang, R., 1981. Physical Chemistry with Applications to Biological Systems (Second Edition)
Macmillan Publishing Co, Inc. New York, NY. 659 pp.
Driscoll, F.G. 1986. Groundwater and Wells, Second Edition. Johnson Division, St. Paul, MN.
Lachenbruch and Goldstein. 1979. Discriminant Analysis. Biometrics. University of Iowa
Department of Preventative Medicine and Environmental Health, Iowa City, IA.
Lohman, S.W. 1972. Ground Water Hydraulics. U.S. Geological Survey Professional Paper 708,
70pp.
Montgomery, J.H., and L.M. Welkom. 1990. Groundwater Chemicals Desk Reference. Lewis
Publishers, Chelsea, MI.
Morrison, R.T. and R.N. Boyd. 1977. Organic Chemistry. Allyn and Bacon, Inc. Boston, MA.
-------
B-13
tiSBSSSSSBSSBBt
U.S. Department of Agriculture. 1989. Water Quality Survey, Interviewer's Manual.
Washington, DC.
U.S. Environmental Protection Agency. 1987. Guidelines for Delineation of Wellhead Protection
Areas (EPA 440/6-87-010). Office of Ground Water Protection, Washington, DC.
U.S. Environmental Protection Agency. 1989. Transport and Fate of Contaminants in the
Subsurface. Seminar Publication, Center for Environmental Research Information,
Cincinnati, OH.
U.S. Geological Survey, Ground Water Subcommittee of the Federal Interagency Advisory USGS,
Committee on Water Data, 1989
Verschueren, J. 1983. Handbook of Environmental Data on Organic Chemicals, Second Edition.
Van Nostrand Reinhold Co., New York, NY.
-------
-------
C-l
APPENDIX C
MAPPING CONSIDERATIONS
-------
-------
C-3
APPENDIX C
MAPPING CONSIDERATIONS
A variety of tools exist for representing data from sensitivity/vulnerability assessments,
including maps, graphs, tables, and geographic information systems (GIS). Each of these tools has
advantages and limitations.
Mapping Uncertainties
Mapping uncertainties arise from extrapolation between data points or from the inclusion
of small areas in a larger area of dissimilar character (e.g., a localized soil within a broad area of a
different soil type). These inclusions may occur because of scale constraints or because of a lack
of data to identify them. Smaller scale maps tend to have more inclusions than larger scale maps.
Additionally, maps have limitations that need to be included with assessment results, either
in the form of a margin of error or an explanation of the limitations. For example, even though a
drafted line represents an infinitely thin boundary, the line itself has some thickness, which causes
uncertainty in the mapped position of the boundary. The degree of this uncertainty is related to
the scale of the map.
Scale Considerations
Selection of map scale is an important decision. Pesticide management programs focused
on field-by-field decision-making will require a much larger scale map than programs that manage
pesticides on a county or subcounty level. Map scales in the range of 1:12,000 to 1:24,000 are
capable of delineating areas as small as 3 to 5 acres. These scales are consistent with those used
in USDA soil surveys and USGS topographic mapping. Smaller scale maps (1:100,000 to
1:250,000) capable of displaying areas no smaller than 100 to 600 acres in size have been used for
assessing county and subcounty areas.
-------
C-4_
MM
Small scale assessments are usually conducted by resource managers who:
• Need a screening assessment before proceeding to more detailed studies of
priority areas (aquifer sensitivity assessments are usually conducted on a
regional scale);
• Manage pesticides on a county-wide or region-wide basis;
• Expect only a few hydrogeologic settings to occur within the region;
• Expect only a limited range of agricultural practices or pesticide-use
practices within the region; or
• Do not have enough data to conduct more detailed assessments or lack the
resources to acquire the necessary data.
If the resource manager chooses to manage pesticides at a larger scale such as field scale
(e.g., 1:24,000) but the data in the regional assessment area are insufficient for this purpose, the
manager must:
1) Accept a higher level of uncertainty in the results of the assessment than is
preferable; or
2) Collect more accurate, site-specific information or additional data.
All maps should specify the scale at which they were developed. Maps developed at small
scales should not be enlarged to meet the needs of managers at the local level, because the data
required for larger-scale interpretation may not have been used in the development of the maps.
Use of Geographic Information Systems (GIS)
GIS is a tool that provides readily accessible, two-dimensional (x,y) spatial data that can be
accessed and overlaid in a variety of different formats (coverages). GIS is less practical where
depth (z direction) is a substantial part of the database.
-------
A distinction should be made between GIS and sensitivity/vulnerability assessments. GIS is
a valuable tool for integrating data and displaying results visually, but it is not itself an assessment
method. Because of the ease of map enlargement by GIS, users of GIS should be particularly
cautioned about changing scales. Mapping accuracy is significantly reduced when a map compiled
at a smaller scale is displayed at a larger scale because only the scale is increased, not the
resolution of the data.
Political Units and Boundaries
Aquifer sensitivity and ground water vulnerability boundaries seldom coincide with political
boundaries. To facilitate resource management, a map of political units can be superimposed
over a sensitivity or vulnerability map and the degree of sensitivity/vulnerability for each political
unit can be assigned based on some preselected criterion.
The sensitivity/vulnerability of political units also can be determined directly. For example,
a political unit could be divided into polygons. Evenly distributed data points could be located
across the political unit. The sensitivity/vulnerability determination at each data point could then
be averaged over each polygon to produce a sensitivity or vulnerability map.
-------
-------
D-l
APPENDIX D
OTHER EPA AND NATIONAL RESEARCH COUNCIL EFFORTS
RELATED TO AQUIFER SENSITIVITY AND
GROUND WATER VULNERABILITY METHODS
-------
-------
D-3
APPENDIX D
OTHER EPA AND NATIONAL RESEARCH COUNCIL EFFORTS
RELATED TO AQUIFER SENSITIVITY AND
GROUND WATER VULNERABILITY METHODS
Information Systems for State Use
EPA's Office of Research and Development (ORD), in cooperation with the agency's
Office of Pesticide Programs (OPP) and EPA Region HI, has completed a project titled
"Prevention of Ground Water Contamination From Pesticides: Information Systems for State
Use." The project consists of two efforts:
PATRIOT
A vulnerability assessment method called PATRIOT (Pesticide Assessment Tool for
Rating Investigations of Transport) has been developed as a simple procedure for State and local
agencies to use hi identifying areas where pesticides are likely to leach to ground water.1 The
method makes use of current ORD data sources and software capabilities developed for use on a
personal computer (PC).
PATRIOT is a dynamic modeling system that incorporates:
• A combined flow and transport model;
• National databases for rainfall, geographic distribution of soils, soil
properties, pesticide properties, and cropping practices;
PATRIOT uses another ORD product called DBAPE (Database and Parameter Estimator),
which is a support program for analyzing soils databases and providing parameter estimation for
RUSTIC and other models.
-------
D-4
• Database management;
• A soil water retention parameter estimator; and
• Ranking procedures for comparing leaching potentials for various
combinations of hydrologic conditions, soil, and pesticide.
PATRIOT analyzes the pesticide transport potential of geologic materials down to the
water table. The flow and transport model used - PRZM-2 - is based on Richards' equation for
modeling flow in unsaturated material. Chemical transport and transformation is modeled using
the convective/dispersive equation. The chemodynamic database contains data on solubility,
organic carbon distribution coefficient, and molecular weight. Soil water retention properties are
estimated based on empirical relationships with previously measured soils properties contained in
the soils properties database.
Ranking procedures in PATRIOT are based on unit soils analyses and on acreage-
weighted geographic information aggregated at various scales - including States, USDA's major
land resource areas, counties, and hydrologic cataloging units. The method is designed to be
applicable at county or subcounty levels if digitized surveys are available.
The system software allows full screen interaction among database, flow and transport
model, parameter estimation and ranking procedures. PATRIOT has been designed specifically
for minimal user input and rapid analysis in a PC environment
DELMARVA Study
EPA Region m, in coordination with the USGS NAWQA. (National Water Quality
Assessment) program, built upon USGS's ongoing hydrologic field work on the DELMARVA
(Delaware, Maryland, Virginia) peninsula. Key elements of the study included evaluation of the
DRASTIC sensitivity assessment method for pesticides, and development and verification of a
monitoring strategy.
-------
D-5
The specific objectives of the Region III study were:
• To relate landscape and shallow subsurface features and activities to "the
spatial occurrence and depth distribution of selected pesticides and nitrates
in shallow ground water;
• To determine if the DRASTIC model identifies and accounts for those
landscape and subsurface features that best correlate with the occurrence
of pesticides and elevated nitrates;
• To evaluate the effect of the scale at which landscape and near-subsurface
features are expressed in relation to the occurrence of pesticides and
nitrate in ground water. The results of this portion of the study will be
published in the "Journal of Environmental Hydrology" in 1993.
The DELMARVA study also has produced:
• Guidance on monitoring strategies based on temporal, spatial, and
environmental factors affecting pesticide occurrence and transport; and
• Additional data for OPP's Pesticide Information Network (PIN). The
results of this portion.of the study will be published in 1993.
GIS Projects
EPA Region Hi's Ground Water Protection Section currently is working in two pilot
counties (Jefferson County, WV and Lancaster County, PA) to demonstrate how States can use
GIS to develop ground water management plans.2 Interagency workgroups representing federal,
State, and local environmental, agricultural, and health agencies were established in both counties
to carry out the projects.
s*
U.S. EPA. 1993 (in review). Use of GIS to Determine Ground Water Vulnerability to
Contamination by Agricultural Chemicals: Jefferson County, West Virginia Pilot Study. Region
TTTT T^. • 1 • *t 9 . l^m m » - _ - _ JO
HI Drinking Water/Ground Water Protection Branch.
-------
D-6_
Detailed data on pesticide use and drinking water wells were obtained by the West
Virginia Region 9 Planning and Development Council in Jefferson County through interviews
with 118 farmers, and by the Lancaster County Conservation District through interviews with 250
fanners. Collected data include: pesticide application rates; crop type and acreage; pesticide
storage, disposal, mixing, loading, and equipment cleaning practices; and well location and
construction.
In Jefferson County the following data exist in GIS: digitized coverages of property
boundaries; SSURGO (county level) soil types (obtained from USDA); soil teachability potential
(using the Soil Pesticide Interaction Screening Procedure (SPISP) developed by USDA); sinkhole
locations; geology; fracture and fault locations; hydrology; water table contours (obtained from
USGS); a DRASTIC map showing relative ground water vulnerability (obtained from the W. Va.
Department of Natural Resources); and all the pesticide usage information obtained from on-farm
interviews.
In Lancaster County, the work is focusing on one watershed, the Pequea/Mill Creek basin.
Data for this area are being input by several project participants. The Temple University GIS lab,
under contract to EPA Region HI, is digitizing property boundaries; the Soil Conservation Service
is digitizing data on soils (SSURGO level with associated attributes and SPISP leachability
rankings); and USGS is entering data on geology, hydrologic units, springs and wells, aquifer
parameters, sinkholes, water table contours, and depth of the unsaturated zone into a GIS. A
digital DRASTIC map for the county was obtained from the Monsanto Chemical Company. In
addition, the pesticide usage survey data are being entered into Dbase files for automation to
GIS.
Using the coverage described above in GIS, EPA Region DI is conducting several
different types of analyses for Jefferson County. These analyses also are planned for Lancaster
County. The analyses identify locations where highly leaching pesticides are being applied in
areas with permeable soils and vulnerable ground waters. Drinking water wells that have a risk of
contamination also are being identified. In addition, the proximity of pesticide use and disposal
areas to potential ground water conduits such as sinkholes and abandoned wells is being analyzed.
Sampling occurred over the summer of 1992 in wells and springs identified as potentially
at risk to determine if contamination had occurred. The monitoring data will be used to help
validate the susceptibility criteria used for the analysis. In addition, pesticide use data collection
-------
D-7
methods --such as landowner interviews and County Agriculture Agent estimates - will be
compared using GIS.
The final products of this project include: a report, Ground Water Vulnerability to
Contamination by Agricultural Chemicals, Jefferson County, West Virginia; a documentation of
mettiods used to obtain, analyze, and manipulate data, and of methods used to convert software;
and final maps of individual coverages and of the results of analyses. Additional information is
available from U.S. EPA Region HI, Office of Ground Water, 841 Chestnut Street, Philadelphia,
Pennsylvania 19106.
National Research Council Study
Another study has been conducted by the Water Science and Technology Board (WSTB)
of the National Research Council. The Council appointed a Committee on Techniques for
Assessing Ground Water Vulnerability. This committee, which is supported by the EPA, USGS,
and USDA, was charged with identifying opportunities for improving regional scale ground water
vulnerability assessment techniques. The committee examined approaches to vulnerability
assessment from three perspectives: model capabilities, data availability, and management needs.
In addition, it examined the use of GIS as a tool to facilitate assessments. The committee's report
was released in late 1993.
In the area of modeling capabilities, the committee identified and critically evaluated the
scientific basis for the range of model types commonly used in regional assessments. Problems of
uncertainty and error propagation also were considered. In addition, problems specific to regional
scale assessments, such as spatial aggregation and validation, were addressed.
With regard to data availability, the committee identified and briefly characterized the
types of databases available at the national, State, and county levels that could be appropriate for
use in regional-scale vulnerability assessments. Database characteristics such as data sources,
quality, scales, and aggregation are discussed with respect to their impacts, on assessments. New
data collection needs were identified, and GIS was examined as a tool that can ease the
management of large amounts of information and simplify the presentation of results.
-------
The committee considered the use of vulnerability assessment techniques in the context of
management needs, and concluded that different needs may lead decision makers to select
different assessment techniques. The committee examined current and potential applications of
regional scale assessment techniques and discussed how management decisions may be affected.
This portion of the study addressed concerns such as scale and resolution, confidence and
uncertainty, and flexibility and robustness.
-------