UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
WASHINGTON,
D.
C.
20460
OFFICE
OF
PREVENTION,
PESTICIDES
AND
TOXIC
SUBSTANCES
Date:
September
21,
2004
Subject:
Clarification
of
the
Trifluralin
Drinking
Water
Assessment
for
the
Health
Effects
Division
(
HED)
Tolerance
Reassessment
(
PC
Code:
036101)
and
characterization
on
relative
differences
of
USGS
NAWQA
ground
water
monitoring
data
and
its
comparison
to
SCI­
GROW
model
predictions
as
presented
in
the
NRDC
objection
(
Imidacloprid
FR)
and
the
trifluralin
TRED
(
D308490)

To:
Richard
Griffin,
Risk
Assessor
Reregistration
Branch
I
Health
Effects
Division
(
7509C)

John
Pates,
Chemical
Review
Manager
Laura
Parsons,
Chemical
Team
Leader
Susan
Lewis,
Branch
Chief
Reregistration
Branch
II
Special
Review
and
Reregistration
Division
(
7508C)

From:
Michael
Barrett,
Ph.
D.
Senior
Chemist
James
Hetrick,
Ph.
D.
Senior
Physical
Scientist
Environmental
Fate
and
Effects
Division
(
7507C)

Approved
By:
Sid
Abel,
Branch
Chief
Mah
Shamim,
Ph.
D.
Branch
Chief
Environmental
Fate
and
Effects
Division
(
7507C)

This
memorandum
provides
clarification
and
characterization
on
the
relative
difference
in
USGS
NAWQA
ground
water
monitoring
data
for
trifluralin
and
its
comparison
to
SCI­
GROW
model
predictions
as
presented
in
the
NRDC
objection
(
Imidacloprid
Order
Denying
Objections
to
Issuance
of
Tolerance,
Federal
Register
Notice
OPP­
2004­
152;
FRL
7355­
7;
hereafter
referred
to
as
the
Imidacloprid
FR)
and
the
trifluralin
TRED.
In
the
TRED,
SCI­
GROW
model
predictions
are
lower
than
the
maximum
reported
trifluralin
concentration
in
the
USGS
NAWQA
groundwater
monitoring
data.
In
contrast,
the
SCI­
GROW
model
performance
assessment
in
the
Imidacloprid
FR
indicates
the
trifluralin
concentration
in
groundwater,
as
predicted
by
SCI­
GROW
,
is
greater
than
the
99.8
percentile
concentration
from
the
NAWQA
groundwater
monitoring
data.
The
relative
difference
of
the
SCI­
GROW
estimates
and
its
comparison
to
monitoring
data
can
be
explained
by
the
use
of
different
percentile
concentrations
as
well
as
use
of
different
trifluralin
application
rates.
SCI­
GROW
modeling
in
the
TRED
was
conducted
using
an
application
rate
of
4
lbs
ai/
A
(
maximum
application
rate
for
sugarcane
on
the
CORNBELT
LABEL,
EPA
Reg.
No.
11773­
17),
median
Koc
of
7384
L
kg
oc­
1
(
MRID
40673501),
average
aerobic
soil
half­
life
of
168.7
days
(
MRID
41240501),
and
application
interval
of
180
days.
In
contrast,
the
model
performance
assessment
of
SCI­
GROW
was
conducted
using
similar
environmental
fate
parameters
and
an
application
rate
of
only
2
lbs
ai/
A.
Predicted
groundwater
concentrations
of
trifluralin
were
0.0346
ug/
L
and
~
0.015
ug/
L
for
the
TRED
and
model
performance
assessment,
respectively.
These
differences
are
not
unexpected
because
the
SCI­
GROW
model
predictions
are
linearly
dependent
on
application
rate
when
all
other
fate
data
are
similar.

The
USGS
NAWQA
ground
water
monitoring
in
the
TRED
were
obtained
from
the
NAWQA
data
warehouse
on
August
12,
2003
(
http://
water.
usgs.
gov/
nawqa/
data).
These
data
represent
NAWQA
groundwater
monitoring
data
as
of
September
30,
2002.
In
contrast,
the
SCI­
GROW
model
performance
assessment
was
conducted
using
June
11,
2001
data
(
covering
samples
collected
up
to
about
one
to
two
years
to
the
date
the
analytical
results
were
compiled).
It
is
important
to
note
the
NAWQA
data
warehouse
is
not
a
static
data
set
because
it
is
being
updated
at
regular
intervals.
Therefore,
the
data
set
available
for
any
compound
is
dependent
on
the
time
of
data
retrieval.

Additional
characterization
of
the
USGS
NAWQA
ground
water
monitoring
data
in
the
TRED
is
required
for
equal
comparison
with
the
SCI­
GROW
model
performance
assessment.
In
the
TRED,
the
SCI­
GROW
estimate
is
compared
to
the
maximum
value
(
100
percentile)
in
the
monitoring
data.
Since
no
risk
concerns
were
identified
even
with
exposure
to
the
maximum
observed
concentration
of
trifluralin,
no
additional
comparisons
to
the
monitoring
data
was
conducted,
including
the
examination
of
the
frequency
distribution
of
trifluralin
residues
in
ground
water.

Unlike
the
trifluralin
TRED
drinking
water
assessment,
the
SCI­
GROW
model
performance
assessment
for
the
Imidacloprid
FR
was
conducted
using
the
99.8
percentile
concentration.
This
percentile
was
selected
because:

1.
Due
to
sample
size
limitations
of
some
studies,
it
is
the
highest
percentile
concentration
used
to
develop
the
SCI­
GROW
regression
model.
2.
Some
individual
observation
wells
may
be
positioned
in
a
way
that
leaching
from
normal
field
application
of
pesticides
may
result
in
higher
levels
of
contamination
than
occur
in
the
ground
water
actually
used
for
drinking
water
in
the
study
area.

Further
characterization
of
the
USGS
NAWQA
groundwater
monitoring
data
used
in
the
TRED
indicates
the
99.8
percentile
concentration
is
0.012
ug/
L.
The
distribution
is
highly
skewed
because
99.5%
of
the
data
(
10,034
samples
from
10,083
samples)
were
non­
detections.
These
data
show
the
SCI­
GROW
estimate
is
greater
than
the
99.8
percentile
concentration
in
the
USGS
NAWQA
groundwater
monitoring
data.
A
similar
observation
was
made
with
the
earlier
NAWQA
data
set
used
in
the
SCI­
GROW
model
performance
assessment
supporting
the
imidacloprid
FR.
