Page
1
of
49
UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
WASHINGTON,
D.
C.
20460
OFFICE
OF
PREVENTION,
PESTICIDES,
AND
TOXIC
SUBSTANCES
November
12,
2004
MEMORANDUM
SUBJECT:
Transmittal
of
Meeting
Minutes
of
the
FIFRA
Scientific
Advisory
Panel
Meeting
Held
September
9
­
10,
2004
TO:
James
J.
Jones,
Director
Office
of
Pesticide
Programs
FROM:
Joseph
E.
Bailey,
Designated
Federal
Official/
s/
FIFRA
Scientific
Advisory
Panel
Office
of
Science
Coordination
and
Policy
THRU:
Larry
C.
Dorsey,
Executive
Secretary
/
s/
FIFRA
Scientific
Advisory
Panel
Office
of
Science
Coordination
and
Policy
Joseph
J.
Merenda,
Jr.,
Director
/
s/
Office
of
Science
Coordination
and
Policy
Attached,
please
find
the
meeting
minutes
of
the
FIFRA
Scientific
Advisory
Panel
open
meeting
held
in
Arlington,
Virginia
on
September
9
­
10,
2004.
This
report
addresses
a
set
of
scientific
issues
being
considered
by
the
Environmental
Protection
Agency
pertaining
to
the
SOil
Fumigant
Exposure
Assessment
System
(
SOFEA)
using
Telone
as
a
case
study.

Attachment
Page
2
of
49
cc:

Susan
Hazen
Margaret
Schneider
Anne
Lindsay
Janet
Andersen
Debbie
Edwards
Steven
Bradbury
William
Diamond
Arnold
Layne
Tina
Levine
Lois
Rossi
Frank
Sanders
Randolph
Perfetti
George
Herndon
William
Jordan
Douglas
Parsons
Enesta
Jones
Vanessa
Vu
(
SAB)
Jeffrey
Dawson
Michael
Metzger
OPP
Docket
Bruce
Houtman,
Dow
AgroSciences
FIFRA
Scientific
Advisory
Panel
Members
Stephen
M.
Roberts,
Ph.
D.
(
FIFRA
SAP
Chair)
Steven
G.
Heeringa,
Ph.
D.
(
FIFRA
SAP
Session
Chair)
Stuart
Handwerger,
M.
D.

FQPA
Science
Review
Board
Members
S.
Pal
Arya,
Ph.
D.
Paul
W.
Bartlett,
M.
A.
Mark
D.
Cohen,
Ph.
D.
Frank
Gouveia,
C.
C.
M.
Adel
F.
Hanna,
Ph.
D.
Peter
Macdonald,
D.
Phil.
Michael
S.
Majewski,
Ph.
D.
David
R.
Maxwell,
M.
P.
A.,
M.
B.
A.
Li­
Tse
Ou,
Ph.
D.
Thomas
Potter,
Ph.
D.
Frederick
Shokes,
Ph.
D.
Thomas
O.
Spicer,
III,
Ph.
D.
Eric
D.
Winegar,
Ph.
D.
Scott
R.
Yates,
Ph.
D.

SAP
Report
No.
2004­
08
3
of
49
A
Set
of
Scientific
Issues
Being
Considered
by
the
Environmental
Protection
Agency
Regarding:

Fumigant
Bystander
Exposure
Model
Review:
SOil
Fumigant
Exposure
Assessment
System
(
SOFEA
©
)
Using
Telone
as
a
Case
Study
September
9
 
10,
2004
FIFRA
Scientific
Advisory
Panel
Meeting
held
at
the
Holiday
Inn
National
Airport
Arlington,
Virginia
NOTICE
These
meeting
minutes
have
been
written
as
part
of
the
activities
of
the
Federal
Insecticide,
Fungicide
and
Rodenticide
Act
(
FIFRA),
Scientific
Advisory
Panel
(
SAP).
This
report
has
not
been
reviewed
for
approval
by
the
United
States
Environmental
Protection
Agency
(
Agency)
and,
hence,
the
contents
of
this
report
do
not
necessarily
4
of
49
represent
the
views
and
policies
of
the
Agency,
nor
of
other
agencies
in
the
Executive
Branch
of
the
Federal
government,
nor
does
mention
of
trade
names
or
commercial
products
constitute
a
recommendation
of
use.

The
FIFRA
SAP
was
established
under
the
provisions
of
FIFRA,
as
amended
by
the
Food
Quality
Protection
Act
(
FQPA)
of
1996,
to
provide
advice,
information
and
recommendations
to
the
Agency
Administrator
on
pesticides
and
pesticide­
related
issues
regarding
the
impact
of
regulatory
actions
on
health
and
the
environment.
The
Panel
serves
as
the
primary
scientific
peer
review
mechanism
of
the
EPA,
Office
of
Pesticide
Programs
(
OPP)
and
is
structured
to
provide
a
balanced
expert
assessment
of
pesticide
and
pesticide­
related
matters
facing
the
Agency.
Food
Quality
Protection
Act
Science
Review
Board
members
serve
the
FIFRA
SAP
on
an
ad
hoc
basis
to
assist
in
reviews
conducted
by
the
FIFRA
SAP.
Further
information
about
FIFRA
SAP
reports
and
activities
can
be
obtained
from
its
website
at
http://
www.
epa.
gov/
scipoly/
sap/
or
the
OPP
Docket
at
(
703)
305­
5805.
Interested
parties
are
invited
to
contact
Joseph
E.
Bailey,
Designated
Federal
Official,
via
e­
mail
at
bailey.
joseph@
epa.
gov.

In
preparing
these
meeting
minutes,
the
Panel
carefully
considered
all
information
provided
and
presented
by
the
Agency
and
Dow
AgroSciences,
LLC,
as
well
as
information
presented
by
public
commenters.
This
document
addresses
the
information
provided
and
presented
within
the
structure
of
the
charge
by
the
Agency.

TABLE
OF
CONTENTS
Participants                       .     
7
Introduction                       .    ...
9
Public
Commenters                         
10
Summary
of
Panel
Discussion
and
Recommendations           ..
11
5
of
49
Panel
Deliberations
and
Responses
to
Charge          .    ..
16
References                            ..
47
SAP
Report
No.
2004­
08
6
of
49
MEETING
MINUTES
FIFRA
Scientific
Advisory
Panel
Meeting
September
9
 
10,
2004,
held
at
the
Holiday
Inn
National
Airport
Arlington,
VA
A
Set
of
Scientific
Issues
Being
Considered
by
the
Environmental
Protection
Agency
Regarding:

Fumigant
Bystander
Exposure
Model
Review:
SOil
Fumigant
Exposure
Assessment
System
(
SOFEA
©
)
Using
Telone
as
a
Case
Study
Mr.
Joseph
E.
Bailey
Steven
G.
Heeringa,
Ph.
D.
Designated
Federal
Official
FIFRA
SAP
Session
Chair
FIFRA
Scientific
Advisory
Panel
FIFRA
Scientific
Advisory
Panel
Date:
Date:

Federal
Insecticide,
Fungicide
and
Rodenticide
Act
Scientific
Advisory
Panel
Meeting
September
9
 
10,
2004
Fumigant
Bystander
Exposure
Model
Review:
SOil
Fumigant
Exposure
Assessment
System
(
SOFEA
©
)
Using
Telone
as
a
Case
Study
7
of
49
PARTICIPANTS
FIFRA
SAP
Session
Chair
Steven
G.
Heeringa,
Ph.
D.,
Research
Scientist
&
Director
for
Statistical
Design,
Institute
for
Social
Research,
University
of
Michigan,
Ann
Arbor,
MI
FIFRA
Scientific
Advisory
Panel
Members
Stuart
Handwerger,
M.
D.,
Professor
of
Pediatrics,
University
of
Cincinnati,
Children's
Hospital
Medical
Center,
Cincinnati,
OH
FQPA
Science
Advisory
Board
Members
S.
Pal
Arya,
Ph.
D.,
Professor
of
Meteorology,
Department
of
Marine,
Earth
&
Atmospheric
Sciences,
North
Carolina
State
University,
Raleigh,
NC
Paul
W.
Bartlett,
M.
A.,
Research
Associate,
Center
for
the
Biology
of
Natural
Systems,
Queens
College,
City
University
of
New
York,
New
York,
NY
Mark
D.
Cohen,
Ph.
D.,
Physical
Scientist,
Air
Resources
Laboratory
(
R/
ARL),
NOAA,
Silver
Spring,
MD
Frank
Gouveia,
C.
C.
M.,
Meteorologist,
University
of
California,
Lawrence
Livermore
National
Laboratory,
Livermore,
CA
Adel
F.
Hanna,
Ph.
D.,
Research
Professor,
Carolina
Environmental
Program,
University
of
North
Carolina
at
Chapel
Hill,
Chapel
Hill,
NC
Peter
Macdonald,
D.
Phil.,
Professor
of
Mathematics
and
Statistics,
McMaster
University,
Hamilton,
Ontario,
Canada
Michael
S.
Majewski,
Ph.
D.,
Research
Chemist,
US
Geological
Survey,
Sacramento,
CA
David
R.
Maxwell,
M.
P.
A.,
M.
B.
A.,
Environmental
Specialist/
Air
Quality
Monitoring
Specialist,
National
Park
Service,
Denver,
CO
Li­
Tse
Ou,
Ph.
D.,
Scientist,
Soil
&
Water
Science
Department,
University
of
Florida,
Gainesville,
FL
Thomas
Potter,
Ph.
D.,
Research
Chemist,
USDA/
ARS,
Coastal
Plain
Experiment
Station,
Southeast
Watershed
Research
Lab.,
Tifton,
GA
Frederick
Shokes,
Ph.
D.,
Center
Director
and
Professor
of
Plant
Pathology,
Tidewater
Agricultural
Research
and
Extension
Center,
Virginia
Tech
University,
Suffolk,
VA
8
of
49
Thomas
O.
Spicer,
III,
Ph.
D.,
Professor
and
Head,
Department
of
Chemical
Engineering,
University
of
Arkansas,
Fayetteville,
AR
Eric
D.
Winegar,
Ph.
D.,
QEP,
Applied
Measurement
Science,
Fair
Oaks,
CA
Scott
R.
Yates,
Ph.
D.,
Interim
Research
Leader,
USDA/
ARS,
GEBJr.
Salinity
Lab.,
Soil
Physics
&
Pesticides
Research
Unit,
Riverside,
CA
INTRODUCTION
On
August
24­
25,
2004,
August
26­
27,
2004
and
September
9­
10,
2004,
the
Federal
Insecticide,
Fungicide
and
Rodenticide
Act
Scientific
Advisory
Panel
(
FIFRA
SAP)
held
three
separate
meetings
to
consider
and
review
three
fumigant
bystander
9
of
49
exposure
models.
These
meeting
minutes
focus
on
the
FIFRA
SAP
meeting
held
September
9­
10,
2004
to
review
the
SOil
Fumigant
Exposure
Assessment
System
(
SOFEA
©
)
using
Telone
as
a
case
study.
The
FIFRA
SAP
also
met
on
August
24­
25,
2004
to
review
the
Probabilistic
Exposure
and
Risk
model
for
FUMigants
(
PERFUM),
using
iodomethane
as
a
case
study
and
on
August
26­
27,
2004
to
review
the
Fumigant
Exposure
Modeling
System
(
FEMS)
using
metam
sodium
as
a
case
study.
Minutes
from
each
of
these
FIFRA
SAP
meetings
are
available
from
the
FIFRA
SAP
website
at
http://
www.
epa.
gov/
scipoly/
sap/
or
the
OPP
Docket
at
(
703)
305­
5805.

Advance
notice
of
the
September
9­
10,
2004
meeting
was
published
in
the
Federal
Register
on
July
23,
2004.
The
review
was
conducted
in
an
open
Panel
meeting
held
in
Arlington,
Virginia
and
was
chaired
by
Steven
G.
Heeringa,
Ph.
D.
Mr.
Joseph
E.
Bailey
served
as
the
Designated
Federal
Official.
Mr.
Joseph
J.
Merenda,
Jr.
(
Director,
Office
of
Science
Coordination
and
Policy)
and
Randolph
Perfetti,
Ph.
D.
(
Associate
Director,
Health
Effects
Division,
Office
of
Pesticide
Programs)
offered
opening
remarks
at
the
meeting.
Mr.
Jeffrey
Dawson
(
Health
Effects
Division,
Office
of
Pesticide
Programs)
provided
an
introduction
and
highlighted
the
goals
and
objectives
of
the
meeting.
Bruce
Johnson,
Ph.
D.
(
California
Department
of
Pesticide
Regulation)
participated
with
the
EPA
in
this
meeting.
Steven
A.
Cryer,
Ph.
D.
(
Dow
AgroSciences,
LLC)
provided
a
detailed
description
of
SOFEA
©
with
additional
clarifying
comments
being
provided
by
Ian
van
Wesenbeeck,
Ph.
D.
and
Bruce
A.
Houtman,
CIH,
also
with
Dow
AgroSciences,
LLC.

EPA's
Office
of
Pesticide
Programs
is
engaged
in
pesticide
tolerance
reassessment
activities
as
mandated
by
the
Food
Quality
Protection
Act
(
1996).
As
part
of
that
process,
the
Agency
is
currently
involved
in
the
development
of
a
comparative
risk
assessment
for
six
soil
fumigant
pesticides
that
include
chloropicrin,
dazomet,
iodomethane,
methyl
bromide,
metam­
sodium,
and
telone.
Each
of
these
chemicals
has
a
degree
of
volatility
associated
with
it
which
is
a
key
characteristic
needed
to
achieve
a
satisfactory
measure
of
efficacy.
This
volatility,
however,
can
contribute
to
human
exposures
because
these
chemicals
can
travel
to
non­
target
receptors,
such
as
nearby
human
populations.
Commonly
referred
to
as
bystander
exposure,
it
is
considered
by
the
Agency
to
be
the
primary
pathway
through
which
human
exposure
to
fumigants
may
occur.

In
order
to
address
bystander
exposures,
the
Agency
developed
a
method
based
on
a
deterministic
use
of
the
Office
of
Air
model
entitled
Industrial
Source
Complex
Short­
Term
Model
(
ISCST3)
that
is
routinely
used
for
regulatory
decisions.
ISCST3
is
publicly
available
from
the
following
Agency
website:
http://
www.
epa.
gov/
scram001/
tt22.
htm#
isc.
In
this
approach,
the
Agency
uses
chemicalspecific
measures
of
volatility
to
quantify
field
emission
rates
for
modeling
purposes.
Additionally,
the
Agency
uses
standardized
meteorological
conditions
which
represent
a
stable
atmosphere
and
unidirectional
wind
patterns
that
provide
conservative
estimates
of
exposure.
10
of
49
Stakeholders
expressed
concern
that
the
conditions
represented
by
the
current
approach
provide
results
that
are
not
sufficiently
refined
for
regulatory
actions
such
as
risk
mitigation.
In
response,
Dow
AgroSciences,
LLC,
the
registrant
for
Telone
(
Note:
Telone,
or
1,3­
dichloropropene,
will
be
referred
to
as
1,3­
D
throughout
this
report),
has
submitted
the
SOFEA
©
model
for
consideration
as
a
possible
refinement
to
the
Agency's
approach.
The
Agency
believes
that
this
model
also
may
have
the
potential
to
be
used
generically
to
calculate
exposures
for
the
six
soil
fumigants
being
evaluated
in
the
current
risk
assessment.
The
key
differences
between
SOFEA
©
and
the
current
Agency
approach
are
that
it
calculates
fumigant
concentrations
in
air
arising
from
volatility
losses
from
treated
fields
for
entire
agricultural
regions
using
multiple
transient
source
terms
(
e.
g.,
different
treated
fields),
GIS
information,
agronomic
specific
variables,
user
specified
buffer
zones
and
field
re­
entry
intervals.
A
modified
version
of
the
ISCST3
is
used
for
air
dispersion
calculations.

The
purpose
of
this
FIFRA
SAP
meeting
was
to
evaluate
the
approaches
contained
in
SOFEA
©
for
integrating
these
different
factors
into
an
analysis
that
considers
exposures
on
a
regional
level.
Additionally,
the
Agency
sought
a
specific
evaluation
of
the
methods
used
pertaining
to
field
emission
rates,
statistical
approaches
for
data
analysis,
receptor
locations,
modifications
to
ISCST3,
and
defining
the
exposed
populations.
Finally,
the
Agency
sought
a
determination
as
to
the
scientific
validity
of
the
overall
approach
included
in
SOFEA
©
.

In
preparing
these
meeting
minutes,
the
Panel
carefully
considered
all
information
provided
and
presented
by
the
Agency
presenters,
as
well
as
information
presented
by
public
commenters.
This
document
addresses
the
information
provided
and
presented
at
the
meeting,
especially
the
Panel's
response
to
the
Agency
charge.

PUBLIC
COMMENTERS
No
oral
statements
were
made
during
the
meeting.

Written
statements
were
received
from:

California
Rural
Legal
Assistance
Foundation
and
Farmworker
Justice
Fund.

SUMMARY
OF
PANEL
DISCUSSION
AND
RECOMMENDATIONS
11
of
49
The
SOFEA
©
bystander
exposure
model
uses
Excel
for
the
user
interface
and
requires
installation
of
a
proprietary
Excel­
based
software
package,
Crystal
Ball,
to
operate
correctly
and
conduct
the
Monte
Carlo
analysis.
The
model
also
incorporates
the
U.
S.
EPA
air
dispersion
model
the
Industrial
Source
Complex
Short­
Term
Model
(
ISCST3).
The
Panel
indicated
that
since
tables
of
data
are
produced
using
macros
in
the
Excel
framework,
graphical
output
would
also
be
helpful.
Although
the
Panel
thought
that
the
SOFEA
©
User's
and
Installation
Guides
are
generally
clear
and
unambiguous
and
the
Programmer's
Guide
helpful
to
modify
SOFEA
©
,
they
agreed
that
the
documents
would
benefit
from
careful
editing,
some
reorganization,
and
a
more
comprehensive
review
of
the
literature.
A
flow
diagram
and
execution
schematic
that
shows
linkages
and
branching,
and
that
identifies
key
computational
procedures
would
be
helpful.

Regarding
specific
model
components,
the
Panel
identified
the
following
issues:
1)
Excel
spreadsheet
cells
that
the
user
should
not
alter
should
be
locked;
2)
no
information
is
given
on
ISCST3
or
Crystal
Ball;
3)
the
fumigant
flux
is
modeled
crudely
with
scaling
factors,
some
of
which
may
be
difficult
to
obtain;
4)
tarps
and
other
control
technologies
are
important
for
limiting
emissions
especially
for
surface
applications
and
their
effects
should
be
considered;
and
5)
a
more
physically
meaningful
method
of
accounting
for
temporal
differences
is
needed.

With
regard
to
the
algorithms,
the
Panel
indicated
that
SOFEA
©
appears
to
generally
perform
the
functions
in
an
appropriate
manner;
however,
an
itemized
list
of
modifications
to
ISCST3
was
recommended.
The
Panel
also
indicated
that
potential
performance
issues
include
the
fact
that
Crystal
Ball
is
not
supported
on
Windows
95
or
Excel
95
and
that
there
may
be
incompatibility
problems
using
SOFEA
©
for
future
editions
of
Windows,
Excel,
Visual
Basic,
or
Crystal
Ball.

Panel
members
were
generally
able
to
successfully
install
and
run
SOFEA
©
,
but
noted
that
Crystal
Ball
is
an
expensive
program
required
to
successfully
use
SOFEA
©
.
Although
the
program
ran
successfully
for
most
Panel
members,
some
had
difficulties
due
to
unexplained
error
messages.
Overall,
initial
evaluation
of
SOFEA
©
by
some
Panel
members
was
found
to
be
more
difficult
due
to
the
use
of
Excel
as
the
user
interface.
After
some
experience
with
SOFEA
©
,
the
Panel
concluded
that
the
advantages
of
Excel
seemed
to
outweigh
the
disadvantages.

Panel
members
concurred
that
the
choice
of
ISCST3
as
the
dispersion
model
in
SOFEA
©
posed
some
problems
related
to
the
weakness
of
ISCST3
to
provide
accurate
estimates
of
chronic
and
acute
exposures
in
comparison
with
other
dispersion
models.
The
Panel
noted
that
SOFEA
©
'
s
methodology
seems
unable
to
relate
townships
to
airsheds,
two
distinctly
different
simulation
domains.
Furthermore,
the
airshed
for
some
chemicals
may
be
larger
than
the
largest
region
considered
by
SOFEA
©
.
The
use
of
the
term
airshed
implies
that
all
significant
source
areas
have
been
included
and
this
is
not
necessarily
the
case
with
SOFEA
©
.

The
Panel
reported
that
the
receptor
grids
used
for
acute
exposures
seem
to
be
adequate.
However,
for
chronic
exposures
the
uniform
grid
assumed
to
be
used
in
SOFEA
©
is
likely
to
underpredict
exposures
because
of
characteristics
of
the
ISCST3
model.
In
the
development
of
receptor
grids
for
chronic
exposure
estimates,
additional
receptors
close
to
treated
fields
should
also
be
considered.
The
Panel
noted
that
no
sensitivity
study
of
the
grid
density
seems
to
have
been
made.
12
of
49
The
aerodynamic
gradient
approach
was
the
field
method
chosen
to
estimate
the
volatilization
flux
of
1,3­
dichloropropene
(
1,3­
D)
and
the
Panel
believed
that
it
is
probably
one
of
the
best
methods
for
determining
volatilization
fluxes
of
pesticides
from
treated
fields.
However,
the
Panel
identified
a
few
limitations
and
drawbacks.
Since
meteorological
input
to
SOFEA
©
is
provided
on
an
hourly
basis,
the
Panel
recommended
that
emission
flux
should
also
be
based
on
hourly
sampled
concentrations.
Several
Panel
members
expressed
concerns
about
the
low
fluxes
determined
during
periods
of
stable
meteorology
at
night.
The
Panel
noted
that
the
aerodynamic
approach
for
flux
calculations
would
benefit
from
the
use
of
improved,
as
well
as
additional
instrumentation.
Several
Panel
members
reiterated
the
importance
of
making
concentration
measurements
with
faster
response
instruments
in
flux
studies
and
in
comparisons
between
modeled
and
predicted
concentrations,
allowing
for
a
more
accurate
representation
of
bystander
exposure.

Regarding
the
probabilistic
scaling
of
fluxes,
SOFEA
©
treats
the
rate
of
application
of
1,3­
D
as
a
stochastic
variable.
The
Panel
indicated
this
to
be
an
unreasonable
choice
because
the
rate
of
application
is
a
quantity
about
which
there
is
the
most
certainty.
Also,
the
Panel
concurred
that
the
probabilistic
varied
flux
rate
for
each
application
based
on
variability
in
field
flux
measurements
is
useful
and
should
be
retained
in
the
model;
however,
a
joint
probability
distribution
is
needed
for
period
flux
rates
and
meteorological
conditions.
The
depth­
of­
injection
scale
factor
needs
to
include
soil
degradation
of
the
applied
fumigant.
One
Panel
member
noted
that
how
soil
structure
(
physics)
influences
diffusive
and/
or
advective
flux
is
the
main
factor
which
influences
surface
flux
relative
to
injection
depth.
The
capability
to
include
probabilistic
flux
inputs
is
valuable
in
assessing
the
overall
sensitivity
of
the
model
to
the
various
parameters.
Since
field
measurements
are
difficult
and
expensive,
this
capability
would
allow
the
investigation
of
various
scenarios
of
field
uncertainty,
thus
giving
a
more
realistic
range
of
the
flux
and
emission
estimates.

The
Panel
agreed
that
using
a
flux/
emission
factor
based
on
a
single
monitoring
study
or
a
few
studies
is
questionable.
One
Panel
member
emphasized
the
importance
of
considering
whether
a
single
flux
profile
is
appropriately
conservative
or
not.
The
emission
flux
behavior
should
be
investigated
for
different
soils
and
soil
moisture
conditions
that
might
exist
in
different
regions.
In
addition,
considering
the
variability
of
emission
fluxes,
a
stability
index
or
parameter
would
be
a
more
appropriate
parameter
to
use
than
air
temperature.
Volatilization
fluxes
depend
on
wind
speed
and
turbulence
whose
effects
may
be
parameterized
through
a
stability
index.
The
use
of
scaling
factors
may
be
reasonable
when
cumulative
fluxes
are
being
considered,
but
they
are
not
appropriate
when
acute,
period
fluxes
and
associated
hourly
downwind
air
concentrations
must
be
estimated.
The
use
of
appropriate
scaling
factors
seems
to
be
based
on
the
user's
judgment,
but
it
would
be
more
appropriate
if
a
mechanistic
approach
was
used
to
develop
them
for
the
various
field
conditions
expected
to
be
encountered.
As
in
most
measurements
and
modeling
applications,
the
question
of
how
representative
the
data
are
of
the
proposed
application
is
critical
to
whether
the
results
are
valid.

The
Panel
commended
the
developers
of
the
SOFEA
©
model
for
attempting
to
consider
multiple,
linked
application
events
in
the
model's
design.
However,
its
reliance
on
the
ISCST3
model
for
description
of
atmospheric
dispersion
limits
the
accuracy
of
the
multiple­
source
simulation.
The
capability
of
considering
multiple
sources
is
very
important,
particularly
when
predicting
chronic
exposures.
More
detailed
land
use
data
may
also
improve
the
model's
capabilities.
13
of
49
The
Panel
noted
that
SOFEA
©
'
s
documentation
does
not
discuss
the
model's
ability
to
address
missing
data.
It
is
implicit
that
all
the
inputs
are
required;
otherwise,
the
model
will
not
run.
In
that
sense,
it
forces
the
user
to
ascertain
whether
the
input
is
complete.
The
Panel
recommended
that
future
documentation
address
how
SOFEA
©
deals
with
missing
data,
particularly
the
use
of
PCRAMMET.

The
Panel
concurred
that
the
addition
of
hourly­
averaged
meteorological
information
and
GIS
data
seems
to
be
a
useful
part
of
the
methodology
implemented
in
SOFEA
©
.
Such
information
seems
to
be
a
step
forward
from
the
present
assumption
of
worst
case
meteorological
conditions
over
the
duration
of
the
release.
For
acute
exposures,
using
meteorological
conditions
from
a
distant
meteorological
station
may
not
accurately
reflect
local
conditions,
especially
effects
such
as
drainage
flows.
For
chronic
exposures,
dispersion
modeling
must
be
done
over
much
larger
distances.
The
use
of
hourly
meteorological
information
from
a
single
monitoring
station
should
not
be
replicated
over
large
areas
(
airsheds)
because
simple
replication
of
the
same
meteorological
data
for
all
sources
will
ignore
terrain
features
which
importantly
effect
conditions.
To
increase
the
reliability
of
meteorological
input
to
a
dispersion
model,
the
Panel
suggested
incorporating
the
Atmospheric
Data
and
Parametrization
Tool
(
ADPT)
or
the
Naval
Research
Laboratory's
Coupled
Ocean/
Atmospheric
Mesoscale
Prediction
Systems
(
COAMPS).

The
Panel
recommended
the
use
of
quality­
assured
meteorological
data
selected
from
the
closest
representative
site,
whether
it
is
a
National
Weather
Service
location,
state
air
quality
or
climatological
site,
or
an
industrial
monitoring
site.
At
least
five
years
of
continuous
data
were
also
recommended,
but
longer
periods
of
data
might
be
necessary
when
evaluating
long­
term
chronic
exposures.
If
on­
site
data
for
a
short
period
are
used,
it
should
be
compared
with
the
nearest
available
long
term
NWS
data.
The
maximum
domain
extent
recommended
for
use
in
SOFEA
©
should
be
influenced
by
the
dispersion
model
it
incorporates.
The
ISCST3
assumption
of
constant
hourly­
averaged
meteorology
and
flat
terrain
over
the
entire
ISCST3
domain
is
questionable
for
larger
model
domains
such
as
might
be
required
for
modeling
an
airshed.
Approaches
to
characterize
the
quality
and
uncertainty
of
meteorological
data
were
described
by
the
developers
of
SOFEA
©
(
e.
g.,
filling
in
missing
data
and
appropriate
data
record
length
for
long
term
exposure
assessments),
but
it
was
not
clear
to
the
Panel
that
these
factors
had
been
fully
addressed.
The
Panel
agreed
that
the
use
of
wind
observations
at
anemometer
heights
used
in
some
data
sources,
other
than
the
standard
height
of
10
m
used
by
ISCST3,
should
be
considered.
For
SOFEA
©
'
s
current
use
of
ISCST3,
the
Panel
believed
that
stability
class
inputs
were
treated
appropriately.
For
a
number
of
reasons,
including
the
coarse
receptor
grid,
the
overflow
algorithm
and
treatment
of
calm
conditions,
the
Panel
concurred
that
SOFEA
©
,
as
currently
configured,
may
not
yield
the
highest
upper­
bound
concentration
estimates.

The
Panel
raised
questions
about
the
accuracy
of
the
use
of
the
PRZM3
model
in
SOFEA
©
to
describe
1,3­
D
flux
from
soil
and
thought
the
CHAIN­
2D
to
be
more
realistic.
They
noted
an
inherent
limitation
of
PRZM3
to
be
the
fact
that
it
produces
results
based
on
daily
(
24­
hour)
time
steps.
In
general,
the
Panel
thought
the
methodology
for
emissions
flux
estimation
in
SOFEA
©
is
too
simplistic.
Meteorological
influences
on
emissions
fluxes
should
be
considered,
as
should
soil
type
and
soil
moisture
and
emission
flux
estimates
should
be
based
on
more
highly
resolved
temporal
measurements
(
e.
g.,
hourly).
In
addition,
the
time
of
application
should
be
considered
as
a
factor.
Flux
estimates
might
be
substantially
off
if
the
time
of
application
is
much
different
from
that
of
the
field
test
used
to
estimate
fluxes.
14
of
49
In
the
ISCST3
model,
estimated
downwind
concentrations
are
inversely
proportional
to
the
windspeed.
When
the
windspeed
goes
to
"
zero",
the
model
cannot
be
used.
In
addition,
its
use
with
low
 
but
non­
zero
 
winds
is
not
recommended.
The
Panel
expressed
concern
about
the
use
of
the
SOFEA
©
model
to
predict
dispersion
estimates
under
calm
conditions.
Several
approaches
were
offered
to
deal
with
SOFEA
©
'
s
shortcomings
associated
with
"
calm"
scenarios
and
wind
reversal
patterns.
They
are:
1)
examine
existing
field
measurement
data
to
determine
if
measurements
were
made
during
calm,
lowwind
or
recirculating
conditions;
2)
consider
abandoning
the
use
of
the
ISCST3
model
as
the
"
engine"
driving
the
dispersion
estimates
in
SOFEA
©
.
(
In
its
place,
a
more
realistic
model
that
does
not
have
as
severe
limitations
under
calm,
low­
wind,
and
recirculation
conditions
could
be
considered
such
as
CALPUFF
or
other
Lagrangian
puff
models.);
and
3)
allow
the
emissions
flux
during
calm
hours
to
build
up,
so
that
the
emissions
in
the
first
hour
after
a
calm
period
would
include
that
hour's
emissions
plus
all
the
emissions
during
the
preceding
calm
period.
This
would
not
address
the
low­
wind
or
recirculation
problem
and
would
only
partially
address
the
calms
problem,
but
it
seems
to
be
better
than
the
present
approach.

While
there
do
not
appear
to
be
major
methodological
problems
with
the
successful
application
of
SOFEA
©
in
settings
other
than
California,
SOFEA
©
developers
reported
on
a
case
study
of
1,3­
D
use
in
central
California
in
which
order­
of­
magnitude
agreement
was
found
between
predicted
and
measured
concentrations.
The
Panel
agreed
that
successful
applications
of
SOFEA
©
to
other
areas
are
hindered
by
data
needed
to
run
the
model
such
as
product
use
data,
flux
estimate
data,
and
weather
and
topographical
data.
One
Panel
member
noted
that
the
development
of
"
scenarios"
that
use
site
specific
data
that
are
more
representative
of
other
regions
of
the
country
may
be
helpful.

The
Panel
concurred
that,
in
many
respects,
the
SOFEA
©
model
does
not
adequately
identify
and
quantify
airborne
concentrations
of
soil
fumigants
that
have
migrated
from
treated
fields
to
sensitive
receptors,
particularly
estimates
of
worst
case,
near
field
exposures.
The
Panel
recommended
the
consideration
of
using
the
CALPUFF/
CALMET
models
in
SOFEA
©
.
Because
of
the
use
of
ISCST3
in
SOFEA
©
,
the
Panel
stated
that
acute
and
chronic
exposures
may
be
underestimated.
Migration
of
soil
fumigants
from
treated
fields
at
large
distances
from
sensitive
receptors
will
be
necessarily
limited
to
about
50
km.
Beyond
this
distance
ISCST3'
s
dispersion
coefficients
should
not
be
used.
Other
more
appropriate
wind
flow
and
dispersion
models
might
be
used
if
long­
term
exposures
from
distant
fields
are
of
interest.

According
to
the
Panel's
assessment,
SOFEA
©
model
results
were
clearly
presented;
however,
concentrations
in
the
upper
ends
of
the
distribution
may
be
underestimated
for
chronic
exposures
at
long
distances,
under
calm
and
low
windspeed
conditions,
and
with
multiple
source
scenarios.
The
ability
of
SOFEA
©
to
predict
worst­
case
concentrations
is
likely
to
be
progressively
worse
for
longer
distances
and
exposure
durations.
Continuous
meteorological
data
over
the
longest
period
of
exposure
may
be
necessary
to
get
concentrations
in
the
upper
ends
of
the
distribution.
No
information
was
found
in
the
SOFEA
©
documentation
about
the
methodology
used
to
calculate
probability
distributions
using
moving
average
concentrations
for
differing
durations
of
exposure.
This
descriptive
documentation
should
be
included.

Ideally,
atmospheric
dispersion
models,
such
as
SOFEA
©
with
ISCST3,
should
be
evaluated
by
comparing
model
predictions
for
a
particular
time
period
at
specific
locations
with
measurements
made
at
the
same
locations
during
the
same
time
periods.
In
carrying
out
such
an
evaluation,
it
would
be
15
of
49
important
to
utilize
the
meteorological
and
emissions
data
for
the
same
period.
The
Panel
recommended
that
this
type
of
model
evaluation
be
carried
out
for
SOFEA
©
.
At
higher
percentiles,
SOFEA
©
underpredicts
monitoring
measurements.
These
results
show
that
further
refinement
of
the
model
is
necessary
to
correct
the
upper
end
of
air
concentration
distribution
estimates
and
to
determine
why
this
underprediction
is
occurring.

The
Panel
suggested
several
specific
factors
to
include
in
the
sensitivity
analysis
of
SOFEA
©
:
background
(
ambient
air)
concentrations,
terrain,
location
(
inland
versus
costal),
and
crop
type.
Model
inputs
that
affect
fumigant
dispersion
and
degradation
need
to
be
included
in
the
analysis
such
as
soil
temperature,
weather
stability,
and
soil
degradation
of
the
applied
fumigant.
Atmospheric
degradation
factors
should
also
be
considered
as
they
affect
the
maximum
volatilization
and
maximum
losses
through
emission
into
the
atmosphere.

Because
SOFEA
©
will
potentially
be
used
for
both
acute
and
chronic
exposure
assessment,
the
Panel
recommended
an
evaluation
of
the
uncertainty
in
both
periodic
and
cumulative
emissions.
Some
Panelists
believed
that
uncertainty
would
be
higher
for
periodic
emissions
than
for
cumulative.
There
was
strong
agreement
for
the
requirement
of
meteorological
record
longer
than
the
five­
year
CIMIS
record
to
ensure
that
some
"
worst
case"
scenarios
would
figure
in
the
sensitivity
analysis.

There
was
general
agreement
among
the
Panel
members
that
the
inputs
required
should
include
the
fumigant
applied,
application
rate,
type
of
application,
application
depth,
tarp
use
or
none,
field
size
(
or
numbers
of
fumigated
fields
for
regional
analysis),
soil
conditions
that
will
affect
fumigant
dispersion
in
the
soil
and
subsequently
into
the
atmosphere,
and
weather
parameters
that
affect
stability.
The
outputs
should
include
flux
rates,
fumigant
concentrations
at
buffer
perimeters
relative
to
toxicity
concentrations,
exceedance
frequency,
distance
from
the
source
at
which
exceedances
occur,
maximum
daily
emission,
and
losses
over
time
through
emission
into
the
atmosphere.

The
Panel
noted
that
SOFEA
©
,
like
any
other
model
at
this
stage
of
development,
will
need
a
line­
by­
line
code
audit
by
an
independent
programmer
to
ensure
that
the
code
does
what
it
is
supposed
to
do.
Documentation
and
testing
of
the
random
number
generator
in
Crystal
Ball
is
needed,
and
if
it
proves
to
be
deficient,
a
better
random
number
generator
must
be
used.

The
Panel
was
concerned
that
"
calms"
could
be
very
important
and
had
not
been
adequately
incorporated
into
the
model.
Perhaps
the
ISCST3
model
is
not
conservative
enough
in
this
regard.
It
is
a
limitation
of
ISCST3
that
no
stability
categories
are
applicable
for
nighttime
calm
and
near
calm
conditions.

The
Panel
recommends
running
a
series
of
simulations
to
determine
whether
the
shape
of
fumigant
flux
profiles
impact
acute
or
chronic
exposure
estimates
generated
by
the
model.
Profiles
can
be
developed
from
published
or
unpublished
studies
or
they
could
be
simulated.
Simulations
should
be
run
under
a
variety
of
worst­
case
conditions
to
determine
the
extent
to
which
"
extreme"
conditions
(
high
or
low
temperature,
wind,
stability
etc.)
may
influence
results.

Because
of
the
wide
range
of
expertise
on
the
Panel,
there
were
many
suggestions
for
enhancing
the
model,
and
some
of
these
may
make
a
significant
difference
in
model
output
under
some
scenarios.
In
summary,
the
Panel
recommended
incorporating
those
proposed
enhancements
that
look
most
16
of
49
promising
and
doing
more
validations
(
or
pseudo­
validations)
in
comparison
to
field
data,
looking
particularly
for
agreement
in
upper
percentiles
and
under
both
typical
and
extreme
scenarios.

PANEL
DELIBERATIONS
AND
RESPONSE
TO
THE
CHARGE
The
specific
issues
to
be
addressed
by
the
Panel
are
keyed
to
the
Agency's
background
documents,
references
and
Agency's
charge
questions.

Agency
Charge
Critical
Element
1:
Documentation
Question
1:
The
background
information
presented
to
the
SAP
Panel
by
the
SOFEA
©
developers
provides
both
user
guidance,
a
technical
overview
of
the
system,
and
a
series
of
case
studies.

(
1A)
Please
comment
on
the
detail
and
clarity
of
these
documents.

Panel
Response
SOFEA
©
uses
Excel
for
the
user
interface.
The
spreadsheet
contains
17
worksheets
for
input
and
output.
A
proprietary
Excel­
based
software
package,
Crystal
Ball,
is
used
to
conduct
the
Monte
Carlo
analysis.
This
program
must
be
installed
before
SOFEA
©
will
operate
correctly.
The
model
also
incorporates
the
U.
S.
EPA
air
dispersion
model
the
Industrial
Source
Complex
Short­
Term
Model
(
ISCST3).
One
worksheet
is
used
to
define
the
input
probability
density
functions
(
PDFs)
and
other
model
parameters,
several
worksheets
are
used
to
include
spatial
and
temporal
information,
and
four
worksheets
provide
the
primary
output.
Optional
Geographical
Information
System
(
GIS)
input
of
crop
cover,
elevation,
and
population
is
available
in
SOFEA
©
.
None
of
the
output
is
presented
graphically.
Since
tables
of
data
are
produced
using
macros
in
the
Excel
framework,
graphical
output
should
be
able
to
be
automatically
generated
and
would
be
helpful
for
interpretation
and
quality
assessment
of
the
model
predictions.

Regarding
the
documentation,
the
SOFEA
©
User's
and
Installation
Guides
were
generally
clear
and
unambiguous.
A
Programmer's
Guide
was
included
that
would
help
a
user
modify
SOFEA
©
(
even
though
such
changes
might
complicate
model
standardization).
The
documents
would
benefit
from
careful
editing,
some
reorganization,
and
a
more
comprehensive
review
of
the
literature.
The
oral
presentation
to
the
Panel
provided
additional
explanation
that
would
be
useful
to
include
in
the
written
documentation.

In
the
SOFEA
©
User's
Guide,
it
can
be
difficult
to
identify
information
on
certain
topics
(
e.
g.
PDFs
that
describe
agronomic
practices).
There
were
also
many
incorrect
citations
(
e.
g.,
all
appendices)
and
in
some
cases
failure
to
properly
support
statements
such
as
"
Research
has
shown".
A
flow
diagram
and
execution
schematic
that
shows
linkages
and
branching,
and
that
identifies
key
computational
procedures
would
be
very
17
of
49
helpful.

Stylistically,
the
User's
Guide
is
a
blend
of
a
"
user's
manual"
and
a
"
case
study".
Users
who
are
not
familiar
with
California
regulatory
programs
could
be
confused
by
references
to
terms
such
as
the
"
maximum
total
mass
that
can
be
applied
to
a
township",
which
is
unique
to
California.
"
Side­
bar"
discussions
on
this
and
other
topics
(
e.
g.
buffers)
could
help
avoid
confusion.
A
number
of
other
issues
identified
by
the
Panel
are
as
follows:

 
There
is
no
information
in
the
User's
Guide
for
the
`
Forecast
Worksheet'.
Space
appears
to
be
available
for
text,
so
maybe
this
was
a
printing
error.
Also,
#
REF
was
found
in
the
spreadsheet
indicating
an
addressing
error.
 
Comment
fields
are
used
to
help
describe
the
cellular
data,
for
example
a
PDF
cell.
It
is
not
clear
if
Crystal
Ball
or
the
authors
of
SOFEA
©
generated
the
comment
statements
in
the
spreadsheet.
 
For
the
Town_
Mass_
Wt
worksheet,
it
seems
that
more
than
the
township
mass
could
be
entered
without
the
user
knowing
this
happened.
It
is
possible
to
enter
data
in
two
locations
causing
the
data
to
be
incorrectly
added.
Warning
messages
are
given
in
comment
statements,
but
there
should
be
some
form
of
error
checking
to
make
sure
this
is
not
allowed.
 
There
is
substantial
documentation
in
the
Excel
spreadsheet,
but
since
all
data
are
available
to
the
user
in
the
spreadsheet
format,
it
is
easy
to
be
overwhelmed
while
learning
SOFEA
©
.
It
can
be
difficult
to
know
where
to
make
changes
in
the
file.
Although
this
may
not
be
a
problem
for
an
experienced
user,
it
might
be
helpful
to
summarize
input
and
critical
output
on
a
single
worksheet.
 
The
section
that
describes
how
fields
are
handled
during
overflow
conditions
is
somewhat
confusing.
Rewriting
this
section
would
be
helpful.
 
A
brief
description
should
be
provided
on
how
Crystal
Ball
is
used
in
Excel
(
i.
e.,
the
basic
manipulations
required
to
incorporate
Crystal
Ball
into
Excel).
This
should
include
some
information
about
how
to
create
a
PDF
cell,
how
to
change
the
Monte
Carlo
parameters,
and
guidance
on
assigning
probability
distribution
parameters.

One
Panel
member
found
that
the
Programmer's
Guide
contained
much
useful
information
that
was
not
necessarily
related
to
programming.
For
example,
the
flow
diagram
in
Figure
5
(
page
15)
helped
explain
the
flux
calculations.
It
is
recommended
that
much
of
the
information
contained
in
the
first
section
of
the
Programmer's
Guide
be
included
in
the
User's
Guide.
Also
regarding
the
Programmer's
Guide,
Figure
11
on
page
22
of
the
guide
is
confusing
because
it
is
not
clear
how
the
three
tables
denoted
by
a),
b)
and
c)
that
contain
numerical
figures
relate
to
the
three
graphs
below
the
tables.
18
of
49
(
1B)
Are
the
descriptions
of
the
specific
model
components
accurate?

Panel
Response
With
regard
to
the
descriptions
of
the
specific
model
components,
the
following
data
integrity
issues
were
identified:

 
Most
(
if
not
all)
data
cells
can
presently
be
changed
in
the
Excel
spreadsheet.
It
might
be
a
good
idea
to
lock
all
the
spreadsheet
cells
that
the
user
should
not
alter.
For
experienced
users,
a
button
could
be
added
to
unlock
selected
cells.
If
input
data
are
deleted
or
altered
without
the
user's
intention,
the
output
could
be
in
error
without
the
user's
knowledge.
In
some
worksheets,
rudimentary
error
checking
is
performed
(
i.
e.,
calculating
sums
that
should
theoretically
equal
100%,
etc).
Further
error
checking
would
improve
SOFEA
©
.
 
SOFEA
©
utilizes
ISCST3
but
no
information
is
given
on
ISCST3
(
which
can
be
obtained
from
the
EPA
web
site).
 
No
information
or
source
of
information
is
provided
for
Crystal
Ball.
Some
Panel
members
found
it
easy
to
change
the
PDF
parameters,
but
it
was
not
so
clear
how
to
use
Crystal
Ball's
more
sophisticated
features.
 
The
fumigant
flux
is
modeled
crudely
with
scaling
factors
reflecting
the
type
and
depth
of
application
as
well
as
seasonal
effects
based
on
temperature,
and
scale
factors
are
applied
to
measured
fumigant
flux
from
a
particular
field
study
with
known
application
rate.
This
scale
factor
would
be
location
and
application
specific
and
could
be
difficult
to
obtain.
The
method
does
not
account
for
the
effect
of
soil
degradation
of
the
applied
fumigant,
which
can
be
important
for
surface
applications
using
a
tarp
(
especially
with
virtually
impermeable
films).
 
Tarps
are
an
important
control
measure
for
limiting
emissions
especially
for
surface
applications,
but
their
efficacy
is
highly
dependent
on
temperature
and
fumigant
used.
Using
PRZM3,
the
presence
of
a
tarp
was
predicted
to
reduce
emissions
to
64%
of
those
in
the
absence
of
a
tarp.
For
high
density
polyethylene
(
HDPE),
1,3­
D
is
highly
permeable
and
for
a
soil
with
low
degradation
(
i.
e.,
0.06
d­
1),
cumulative
emission
would
be
approximately
91%
of
the
applied
1,3­
D.
Increasing
the
degradation
rate
by
a
factor
of
10
(
i.
e.,
0.6
d­
1)
gives
a
cumulative
emission
of
76%.
 
Temporal
scaling
is
based
on
a
personal
communication
with
California
Department
of
Pesticide
Regulation
(
CDPR)
and
utilizes
a
ratio
of
summer
to
winter
flux
of
1.6.
Although
one
would
expect
a
summer
flux
to
be
greater
than
in
the
winter,
a
simple
ratio
is
not
likely
to
be
generally
correct.
A
more
physically
meaningful
method
of
accounting
for
temporal
differences
is
needed.
 
A
detailed
list
of
subroutines
was
provided
and
would
be
very
helpful
if
someone
wished
to
modify
the
program.
This
list
would
also
be
helpful
for
error
checking
or
debugging.
19
of
49
(
1C)
Do
the
algorithms
in
the
annotated
code
perform
the
functions
as
defined
in
this
document?

Panel
Response
With
regard
to
the
algorithms,
SOFEA
©
appears
to
generally
perform
the
functions
in
an
appropriate
manner.
Error
messages
seen
during
model
execution
caused
concern
about
how
well
the
algorithms
perform
and
about
potential
effects
on
the
results
of
an
assessment
(
discussed
below).
An
itemized
list
of
modifications
to
ISCST3
is
needed.
Also,
some
demonstration
should
be
provided
showing
that
the
core
ISCST3
algorithm
has
not
been
affected
by
the
modifications.
Some
potential
performance
issues
include
the
following:

 
Crystal
Ball
is
not
supported
on
Windows
95
or
Excel
95,
but
this
should
not
be
a
serious
problem
since
this
operating
system
and
program
are
not
likely
to
be
used
in
the
future.
 
There
may
be
incompatibility
problems
using
SOFEA
©
for
future
editions
of
Windows,
Excel,
Visual
Basic,
and
Crystal
Ball.
Any
significant
change
in
these
programs
may
cause
the
system
to
fail
to
run
successfully.

(
1D)
Please
discuss
any
difficulties
encountered
with
respect
to
loading
the
software
and
evaluating
the
system
including
the
presented
case
study.

Panel
Response
Panel
members
were
generally
able
to
successfully
install
and
run
SOFEA
©
.
Crystal
Ball
is
an
expensive
program
($
2500
for
Professional
version)
that
is
required
to
use
SOFEA
©
.
A
trial
version
of
Crystal
Ball
was
obtained
to
test
SOFEA
©
.
Many
of
the
buttons
in
the
Excel
spreadsheet
do
not
function
properly
unless
Crystal
Ball
has
been
installed.
Some
of
these
buttons
do
not
appear
to
have
any
direct
relationship
with
the
Monte
Carlo
analysis,
so
the
reason
for
their
lack
of
operation
is
puzzling.
However,
the
documentation
is
clear
about
the
required
use
of
Crystal
Ball.
Problems
reported
by
Panelists
included
the
following:

 
During
a
demonstration
simulation
of
SOFEA
©
,
the
"
Run
Township
Simulation"
button
on
the
"
PDF
parameters"
page
was
pressed
prior
to
installation
of
Crystal
Ball.
The
simulation
started
and
completed
in
15
minutes.
It
is
difficult
to
know
what
the
results
represent
since
the
User's
Guide
states
that
Crystal
Ball
is
a
mandatory
requirement
for
SOFEA
©
.
If
this
is
the
case,
then
SOFEA
©
should
generate
an
error
message
and/
or
not
allow
execution
of
the
program
if
Crystal
Ball
is
not
actively
running.
 
One
Panel
member
experienced
problems
running
the
program,
possibly
due
to
using
an
older
computer.
Once
installed,
Crystal
Ball
returned
three
error
messages
during
each
loop:
"
Unable
to
Complete
the
Operation
Due
to
an
Unexpected
Error".
The
cause
of
these
error
messages
was
never
determined.
After
Crystal
Ball
was
installed,
SOFEA
©
reached
50%
completion
before
total
20
of
49
failure.
It
seems
that
the
ISCST3
weather
file
for
the
current
(
failed)
loop
was
for
1999,
but
the
input
file
listed
1996.
Trying
to
run
ISCST3
in
stand­
alone
mode
produced
an
error.
Editing
ISCST3.
INP
and
changing
the
dates
to
match
would
allow
ISCST3
to
run.
However,
the
error
in
SOFEA
©
occurs
at
a
point
where
restart
is
not
possible
(
a
panel
member
was
not
able
to
change
ISCST3.
INP,
run
ISCST3,
then
get
SOFEA
©
to
restart).
The
problem
may
be
related
to
the
poor
performance
characteristics
of
the
test
computer
(
i.
e.,
Dell
Inspiron
8000,
800
MHz,
256MB).
SOFEA
©
was
run
to
completion
by
restricting
the
PDF
for
YEAR
to
1996­
1997
(
i.
e.,
only
allow
1996).
This
forced
all
the
ISCST3
output
files
to
have
the
correct
information.
However,
three
Crystal
Ball
error
messages
were
again
received
and
the
cause
was
never
determined.
 
The
program
ran
correctly
for
other
Panel
members.

Initial
evaluation
of
SOFEA
©
by
some
Panel
members
was
found
to
be
more
difficult
due
to
the
use
of
Excel
as
the
user
interface.
This
is
a
somewhat
unusual
approach
to
create
a
user
interface.
While
this
has
many
advantages
(
ease
in
creating
"
what
if"
scenarios,
near
universal
availability,
no
or
low
cost
(
for
Excel),
everything
in
one
place,
programming
flexibility,
etc.)
some
are
offset
by
the
high
cost
of
Crystal
Ball,
potential
for
SOFEA
©
to
be
incompatible
with
Excel,
potential
to
change
something
inadvertently,
and
the
potential
to
get
"
lost"
in
the
pages
and
pages
of
numbers.
The
importance
of
this
latter
point
could
be
reduced
if
the
columnar
output
was
captured
in
some
figures.
After
some
experience
with
SOFEA
©
,
the
Excel
interface
advantages
seemed
to
outweigh
the
disadvantages.
One
Panel
member
did
feel
that
the
Excel
interface
made
the
program
user
friendly
and
facilitated
its
installation
and
use,
but
recommended
that
a
"
standalone"
version
that
does
not
use
Crystal
Ball
be
developed.
The
Panel
member
indicated
that
given
the
limited
use
of
stochastic
sampling,
a
version
of
SOFEA
©
that
operates
on
set
scenarios
may
be
equally
or
more
effective
in
developing
potential
exposure
profiles.

Critical
Element
2:
System
Design/
Inputs
Question
2:
In
the
background
documents,
a
series
of
detailed
individual
processes
and
components
included
in
SOFEA
©
are
presented.
The
key
processes
include
(
1)
incorporation
of
ISCST3
into
SOFEA
©
,
(
2)
probabilistic
scaling
of
flux
rates;
(
3)
defining
source
placement
within
an
airshed;
(
4)
development
of
receptor
grids
within
airsheds;
and
(
5)
generation
of
probability
distribution
functions
based
on
use
patterns
and
application
parameters.

(
2A)
Please
comment
on
these
proposed
processes,
the
nature
of
the
components
included
in
SOFEA
©
and
the
data
needed
to
generate
an
analysis
using
SOFEA
©
.

Panel
Response
The
choice
of
ISCST3
as
the
dispersion
model
incorporated
in
SOFEA
©
poses
some
problems
as
follows:
21
of
49
 
For
1,3­
D,
chronic
exposures
must
be
estimated
over
long
distances.
Consequently,
dispersion
estimates
must
be
made
over
distances
that
are
much
greater
than
are
considered
appropriate
when
using
ISCST3.
There
was
concern
expressed
that
the
original
intent
of
estimating
the
acute
exposure
to
a
bystander
is
fundamentally
a
different
task
from
estimating
chronic
exposure
to
the
public
(
because
of
the
very
different
time
and
length
scales
for
these
two
problems).
Furthermore,
the
inherent
weaknesses
of
the
ISCST3
model
when
applied
to
estimating
chronic
exposures
over
long
distances
severely
limits
the
applicability
of
SOFEA
©
to
the
long
term
health
issues
surrounding
the
use
of
1,3­
D.

 
For
acute
exposure,
concentrations
must
be
estimated
over
near
field
distances
(
less
than
100
m).
Over
such
short
distances,
there
is
some
indication
that
ISCST3
may
underpredict
concentrations
in
comparison
with
other
dispersion
models.
Isakov
et
al.
(
2004)
report
that
ISCST3
may
substantially
underestimate
pollutant
concentrations
in
the
vicinity
of
the
source.
Coulter
and
Eckhoff
(
1998)
report
that
ISCST3
consistently
predicts
concentrations
that
are
lower
than
CALPUFF
when
variable
meteorological
conditions
are
used.

The
model,
in
the
form
that
was
presented,
computes
flux
estimates
using
a
single
flux
profile
developed
during
a
registrant
study
in
California.
In
SOFEA
©
,
the
proposed
flux
profile
is
adjusted
by
stochastically
sampling
PDFs
representing
agronomic
factors
such
as
depth
of
application,
application
rate,
and
timing.
The
variation
in
flux
associated
with
differences
in
soil
properties,
moisture
content,
temperature,
and
other
factors
is
not
presently
taken
into
account
with
a
PDF.

Regarding
the
probabilistic
scaling
of
fluxes,
SOFEA
©
treats
the
rate
of
application
of
1,3­
D
as
a
stochastic
variable.
This
seems
to
be
an
unreasonable
choice
because
the
rate
of
application
is
a
quantity
about
which
there
is
the
most
certainty
(
farmers
are
going
to
choose
how
much
to
apply
and
carefully
make
that
application,
especially
in
California
where
strict
reporting
is
required).
Since
the
total
amount
of
1,3­
D
applied
in
a
township
is
limited
by
regulation
(
California
is
the
only
state
where
township
caps
apply)
and
most
townships
use
the
maximum
amount
allowed,
stochastic
variation
will
mean
that
the
model
will
be
making
predictions
as
if
some
townships
(
arbitrarily)
use
less
and
some
use
more
­­
a
modeling
technique
which
will
incorrectly
predict
the
tails
of
the
distribution.
Other
parameters
associated
with
the
flux
estimates
would
seem
to
be
more
appropriately
modeled
as
stochastic
variables.
For
flux
estimates
based
on
field
experiments,
variability
arises
from
factors
such
as
meteorological
conditions
that
reflect
the
differences
between
field
measurements
and
modeled
application
days.
In
contrast,
deterministic
models
for
flux
can
be
used
with
stochastically
varied
meteorological
conditions
to
predict
the
variability
in
a
given
application.

Regarding
source
placement
within
an
airshed,
SOFEA
©
uses
the
definition
of
a
township
as
the
basic
simulation
domain
because
the
maximum
amount
of
1,3­
D
that
can
be
applied
is
specified
by
township.
For
the
purpose
of
this
response,
an
airshed
will
be
defined
by
the
source,
transport,
and
fate
of
the
chemical
in
question,
but
airshed
boundaries
have
no
relationship
with
townships,
and
SOFEA
©
'
s
methodology
seems
22
of
49
unable
to
relate
the
two.
Furthermore,
the
airshed
for
some
chemicals
(
such
as
1,3­
D
for
which
chronic
exposure
is
important)
may
be
larger
than
the
largest
region
considered
by
SOFEA
©
(
23
by
23
township
region
or
222
km
by
222
km).
As
discussed
above,
ISCST3
model
predictions
should
not
be
used
at
such
long
distances.

SOFEA
©
places
treated
fields
randomly
in
a
township.
In
practice,
treated
fields
may
be
clustered
and
even
juxtaposed,
although
treatments
may
not
occur
at
the
same
time.
The
average
concentrations
downwind
from
such
a
cluster
might
be
larger
than
if
the
fields
were
distributed
randomly.

For
townships
with
high
application
rates,
source
placement
can
be
dictated
by
the
overflow
algorithm,
which
is
invoked
when
the
random
placement
of
fields
restricts
the
placement
of
additional
fields
in
the
same
township
section.
Consequently,
the
overflow
algorithm
spreads
sources
over
a
larger
geographic
area
than
happens
in
practice.
Since
the
same
amount
of
1,3­
D
is
modeled
by
the
overflow
algorithm
as
a
release
over
a
larger
area,
the
modeled
exposure
will
be
less
than
would
be
predicted
if
the
1,3­
D
application
were
modeled
over
the
correct
area.
Consequently,
the
overflow
algorithm
reduces
the
predicted
exposure
in
areas
where
1,3­
D
application
is
the
highest.

The
receptor
grids
used
for
acute
exposures
seem
to
be
adequate.
However,
the
receptor
grid
for
chronic
exposures
is
assumed
to
be
uniform,
and
such
a
uniform
grid
is
likely
to
underpredict
exposures
unless
it
is
fairly
dense,
particularly
because
of
the
characteristics
of
the
ISCST3
model
used
in
SOFEA
©
;
ISCST3
predicts
a
plume
trajectory
to
follow
the
wind
trajectory
from
a
source
to
downwind
distances
of
greater
than
50
km
based
on
the
hourly
wind
direction.
Put
simply,
the
trajectory
of
the
maximum
concentrations
for
a
plume
may
miss
all
grid
receptors
if
the
grid
is
sparse,
and
while
the
Gaussian
plume
models
maximum
concentrations
at
a
given
downwind
distance
reasonably
well,
it
does
a
notoriously
poor
job
of
predicting
the
locations
of
such
maxima.
No
sensitivity
study
of
the
grid
density
seems
to
have
been
made.
In
the
development
of
receptor
grids
for
chronic
exposure
estimates,
additional
receptors
close
to
the
treated
fields
should
also
be
considered.
Spatial
averages
of
these
near­
field
receptors
should
be
given,
in
addition
to
those
for
the
uniformly­
spaced
receptors.
Nearfield
receptors
are
likely
to
be
exposed
to
much
higher
concentrations
on
both
short
term
and
long
term
bases.
However,
computer
resource
requirements
can
increase
to
impractical
levels
for
dense
grids.
This
problem
only
appears
to
be
discussed
very
briefly
in
the
documentation,
and
clearer
warnings
and
more
specific
guidance
should
be
provided.
Also,
it
would
be
helpful
if
a
warning
message
would
be
generated
if
the
user
selects
a
grid
size
too
large
to
provide
meaningful
answers
to
the
questions
under
consideration.

(
2B)
Are
there
any
other
potential
critical
sources
of
data
or
methodologies
that
should
be
considered?

Panel
Response
The
preliminary
efforts
at
including
GIS
information
and
land
use
data,
such
as
the
Pesticide
Use
Records
(
PUR),
look
as
though
such
efforts
may
be
effective
at
providing
realistic
exposure
estimates.
However,
there
were
some
concerns
expressed
as
23
of
49
to
the
reliability
of
PUR
especially
as
it
relates
to
missing
data.
A
member
of
CDPR
admitted
that
this
was
a
concern
and
that
work
was
being
done
to
address
this
issue.

Question
3:
The
determination
of
appropriate
flux/
emission
rates
is
critical
to
the
proper
use
of
the
SOFEA
©
model
as
these
values
define
the
source
of
fumigants
in
the
air
that
can
lead
to
exposures.
Upon
its
review
of
how
flux
rates
can
be
calculated,
the
Agency
has
identified
a
number
of
questions
it
would
like
the
Panel
to
consider.
In
SOFEA
©
,
measured
flux
rates
specific
to
the
conditions
at
the
time
of
the
monitoring
studies
used
are
adjusted
based
upon
incorporation
depth
and
seasonal
differences
to
account
for
varying
application
conditions.
Emissions
of
1,3­
dichloropropene
are
sensitive
to
soil
temperature
and
incorporation
depth.
Incorporation
depth
is
addressed
using
the
EPA
model
PRZM3
and
also
the
USDA
model
CHAIN­
2D.
Scaling
factors
were
used
to
address
temperature
differences.

(
3A)
What,
if
any,
refinements
are
needed
for
this
process
including
the
manner
in
which
flux
values
were
directly
monitored
and
calculated
using
the
aerodynamic
flux
approach?

Panel
Response
The
aerodynamic
gradient
approach
was
the
field
method
chosen
to
estimate
the
volatilization
flux
of
1,3­
D.
This
method
has
been
widely
used
and
is
well
documented
in
the
literature.
It
is
probably
one
of
the
best
methods
for
determining
volatilization
fluxes
of
pesticides
from
treated
fields,
although
it
does
have
a
few
limitations
and
drawbacks
that
will
be
discussed
below.
The
primary
usage
regions
for
1,3­
D
were
reported
to
be
Washington,
California,
North
Carolina,
Georgia,
and
Florida.
To
date,
field
studies
have
only
been
conducted
in
Florida
(
single
study)
and
California
(
four
studies)
using
a
variety
of
application
methods.

The
single
study
conducted
in
Salinas,
California
was
selected
as
the
"
worst
case"
or
most
representative
of
shank
application
conditions
because
it
resulted
in
the
highest
mass
loss
and,
therefore,
the
most
conservative
example.
The
resulting
14­
day
flux
profile
was
used
as
the
representative
source
emission
input
to
the
model
in
all
test
cases.
For
acute
exposure
assessment,
however,
use
of
this
high
mass­
loss
flux
profile
does
not
guarantee
the
"
worst­
case"
because
factors
such
as
application
timing
and
meteorological
conditions
were
not
considered.
Long
sampling
periods
used
in
the
field
study
(
two
6
hour
samples
followed
by
a
12
hour
sample
overnight)
cannot
reflect
the
variability
in
the
emission
fluxes.
The
environmental
conditions
that
affect
volatilization
are
continuously
changing,
and
very
dramatic
and
important
changes
can
occur
over
much
shorter
time
periods.
The
empirical
flux­
gradient
relations,
called
stability
correction
factors
by
the
presenters,
are
generally
based
on
30
to
60
minute­
averaged
data
and
may
not
be
appropriate
for
6­
12
hour
averaging
or
sampling
times
over
which
stability
conditions,
turbulent
eddy
diffusivities,
and
fluxes
can
vary
considerably.
Using
flux
estimates
averaged
over
several
hours
to
estimate
hourly
exposure
concentrations
will
underpredict
the
actual
downwind
air
concentrations
at
some
times
and
overpredict
the
concentrations
at
others.
Since
meteorological
input
to
SOFEA
©
is
provided
on
an
hourly
basis,
the
24
of
49
Panel
recommended
that
emission
flux
should
also
be
based
on
hourly
sampled
concentrations.
The
Panel
recognized
that
this
will
add
considerable
expense
to
field
studies
to
collect
hourly
gradient
samples.
It
is
possible
to
automate
hourly
air
concentration
sampling
using
multiple
cartridges
and
a
solenoid
switching
system
at
each
height,
similar
to
the
one
used
in
the
chamber
studies
described
in
the
presentation
(
slide
39).

As
described
in
the
oral
presentation,
concentration,
wind
speed,
and
air
temperature
measurements
were
made
at
33,
50,
90,
and
150
cm
heights
above
ground
level;
concentration
was
also
measured
at
15
cm.
By
convention
from
some
previous
work,
only
the
data
collected
at
two
heights
(
33
and
90
cm)
were
used
to
estimate
the
flux
densities
with
the
aerodynamic
gradient
approach.
It
would
be
more
appropriate
to
use
data
from
all
height
levels
and
reduce
the
error
in
the
calculated
flux
by
averaging
the
estimated
values
at
several
heights
based
on
gradient
estimates
between
adjacent
heights.
For
gradient
estimates,
the
logarithmic
finite­
difference
approximation
is
considered
superior
to
the
linear
finite­
difference
approximation
and
is
often
used
in
micrometeorology
(
Arya,
2001).
The
same
approach
should
be
used
for
the
estimation
of
the
gradient
Richardson
number.
The
logarithmic
approximation
can
be
used
between
adjacent
heights
even
when
the
whole
profile
may
not
be
logarithmic,
as
the
log
law
is
strictly
valid
only
under
neutral
stability
conditions.

Several
Panel
members
expressed
concerns
about
the
low
fluxes
determined
during
periods
of
stable
meteorology
at
night.
The
measured
values
at
or
near
zero
do
not
seem
real.
This
is
likely
due
to
the
relatively
high
threshold
levels
for
the
wind
speed
sensors
utilized
in
the
study.
The
following
suggestions
about
the
use
of
improved
and
additional
instrumentation,
would
benefit
the
aerodynamic
approach
for
flux
calculations:

 
A
potential
inaccuracy
in
the
measurement
of
the
near­
surface
temperature
arises
from
the
use
of
naturally
aspirated
temperature­
sensor
housings.
These
housings
were
used
appropriately
in
the
presented
field
study
and
are
well
accepted
for
similar
studies.
There
is
a
problem,
however,
with
using
these
sensor
housings
because
the
housing
may
artificially
cool
the
sensor
at
night
when
the
wind
is
less
than
2
m/
s
and
the
sky
is
clear.
A
1996
study
compared
measured
air
temperatures
from
naturally
ventilated
and
forced
ventilated
temperature
housings
(
NUMUG,
1996).
Although
the
cited
study
used
less
efficient
housings
than
the
Dow
field
study,
a
similar
effect
would
be
present.
It
is
conceivable
that
such
temperature
measurement
errors
could
lead
to
lower
flux
estimates.
Furthermore,
the
Richardson
number
is
dependent
on
the
average
air
temperature.
This
may
help
explain
the
differences
between
emission
rates
calculated
by
the
aerodynamic
gradient
approach
and
back­
calculation
flux
methods
made
at
night.
 
The
incorporation
of
completely
correlated
wind
and
chemical
measurements
would
be
of
value
in
reducing
the
flux
measurement
uncertainty.
Hourly
wind
averages
along
with
hourly
air
temperature
and
concentration
measurements
would
allow
a
better
understanding
of
the
dynamics
over
time.
 
Meteorological
sensors
using
sonic
anemometry
could
be
used
to
collect
very
low
threshold
values.
Vaisala,
Inc.
(
www.
vaisala.
com)
has
a
2­
D
sonic
sensor
with
a
25
of
49
"
virtually
zero
starting
threshold."
Other
companies
have
3­
D
research
grade
instruments
that
allow
the
measurement
of
vertical
components
(
www.
apptech.
com)
as
well
as
a
higher
frequency
for
gathering
information
on
turbulence.
A
fast
3­
D
anemometer/
thermocouple
would
make
direct
measurements
of
momentum
and
heat
flux.
The
use
of
sonic
anemometers
will
also
lower
the
wind
speed
measurement
threshold
thereby
enabling
fluxes
to
be
estimated
during
very
stable/
low
wind
conditions.
With
new
technologies
on
the
market,
such
sensors
are
now
inexpensive
and
appropriate
for
use
in
field
studies.
 
Another
sensor
that
may
be
useful
is
an
IR
thermocouple
to
measure
soil
surface
temperature.
This
parameter
would
be
helpful
when
comparing
results
from
different
times
of
the
day,
and
even
different
field
studies.
An
IR
thermocouple
is
inexpensive,
easy
to
incorporate
into
existing
field
loggers,
and
is
very
stable
and
accurate.
 
Although
it
is
understood
that
open­
path
FTIR
was
used
with
poor
results
in
one
of
the
field
trials,
another
possible
open­
path
technology
is
tunable
diode
laser
(
www.
boreal­
laser.
com,
www.
unisearch.
com).
This
technology
is
potentially
more
sensitive
than
FTIR
due
to
beam­
throughput
issues
and
is
well­
suited
for
a
single
species
measurement
scenario.
 
Conventional
air
monitoring
equipment
can
easily
meet
the
need
for
short­
term
air
samples
with
high
sensitivity.
Using
EPA
Method
TO­
15
(
www.
epa.
gov/
ttn/
amtic/
airtox.
html),
one
can
collect
samples
for
periods
as
short
as
approximately
one
minute
and
up
to
24
hours
or
more
(
www.
entechinst.
com).
There
are
at
least
20
commercial
analytical
laboratories
that
provide
detection
limits
as
low
as
0.1
part
per
billion
by
volume
(
ppbv)
on
a
routine
basis
(
e.
g.,
www.
airtoxics.
com,
www.
easlab.
com).
 
Another
instrument
that
may
be
useful
is
a
fast­
response
closed­
path
IR
spectrometer.
Although
it
may
not
be
fast
enough
for
direct
measurement
of
chemical
flux,
this
instrument
may
be
appropriate
for
measurement
of
chemical
concentrations
with
averaging
times
between
several
seconds
and
a
few
minutes.
 
A
commercially
available
portable
gas
chromatograph/
mass
spectrometer
with
an
on­
board
pre­
concentrator
module
can
provide
for
sub­
ppbv
measurements
with
10
minutes
of
sample
collection
followed
by
a
10
minute
or
less
analysis
time,
providing
MS­
quality
data
consistent
with
laboratory
measurements
(
www.
hapsite.
com).
This
instrument
has
been
evaluated
by
EPA's
Environmental
Technology
Verification
Program
(
www.
epa.
gov/
etv/
verifications/
verification­
index.
html).
Other
field­
portable
instruments
such
as
gas
chromatographs
are
available
(
e.
g.,
www.
photovac.
com),
although
it's
not
certain
all
would
meet
the
main
requirement
of
ppbv
sensitivity.
 
One
panel
member
noted
that
a
gas
chromatograph
with
an
electron
capture
detector
and
a
pre­
concentrator
module
would
be
the
best
choice
for
field
portable
instrumentation
because
it
would
likely
be
more
stable
in
the
field
than
a
GC­
MS
instrument
and
provide
enhanced
sensitivity
for
1,3­
D
over
gas
chromatography
with
photoionization
detection
instrumentation.
26
of
49
Commercially
available
technology
exists
for
collecting
emissions
data
using
the
relaxed
eddy
accumulation
method,
a
gradient
method
similar
to
the
aerodynamic
method
(
www.
apptech.
com,
www.
hamptontechnologies.
com).

Several
Panel
members
reiterated
the
importance
of
making
concentration
measurements
with
faster
response
instruments
in
flux
studies
and
in
comparisons
between
modeled
and
predicted
concentrations.
Faster
response
concentration
measurements
will
allow
for
a
more
accurate
representation
of
bystander
exposure.

(
3B)
SOFEA
©
can
easily
be
modified
to
probabilistically
vary
flux
rate
for
each
application
based
on
variability
in
field
flux
measurements
(
e.
g.,
application
method
or
temperature)
or
model
generated
flux.
Please
comment
on
this
potential
modification.

Panel
Response
The
probabilistically
varied
flux
rate
for
each
application
based
on
variability
in
field
flux
measurements
is
useful
and
should
be
retained
in
the
model.
To
obtain
the
probabilistic
flux
values,
however,
a
joint
probability
distribution
is
needed
for
period
flux
rates
and
meteorological
conditions.
The
emission
fluxes
are
dependent
on
temperature,
atmospheric
stability,
and
precipitation.
The
stochastic
selection
of
a
flux
value
should
also
depend
on
these
processes.

The
depth­
of­
injection
scale
factor
needs
to
include
soil
degradation
of
the
applied
fumigant.
In
the
absence
of
soil
degradation,
cumulative
emissions
will
be
100%.
Soil
degradation
is
the
controlling
process
affecting
cumulative
emissions.
Depth
of
injection
affects
emissions
by
changing
the
soil
residence
time
(
i.
e.,
the
amount
of
time
over
which
soil
degradation
occurs).
The
depth­
of­
injection
scale
factor
is
valid
only
for
soils
with
the
"
calibrated"
degradation
rate
from
the
reference
field
study.

The
capability
to
include
probabilistic
flux
inputs
is
valuable
in
assessing
the
overall
sensitivity
of
the
model
to
the
various
parameters.
Since
field
measurements
are
difficult
and
expensive,
this
capability
would
allow
the
investigation
of
various
scenarios
of
field
uncertainty,
thus
giving
a
more
realistic
range
of
the
flux
and
emission
estimates.

1,3­
D,
as
commercially
marketed,
consists
of
two
isomers,
cis­
and
trans­.
The
two
isomers
have
different
physicochemical,
biological,
and
toxicological
properties.
These
should
be
considered
as
two
different
chemicals.
The
degradation
rates
of
cis­
and
trans­
1,3­
D
in
Florida
sandy
soils
have
been
found
to
be
essentially
the
same
as
that
in
sterile
soils
with
both
isomers
having
the
same
degradation
rate.
Soils
with
a
history
of
repeated
application
of
1,3­
D
were
found
to
exhibit
more
rapid
degradation
in
live
soil
than
in
sterile
soil,
and
the
degradation
of
trans­
1,3­
D
was
more
rapid
than
cis­
1,3­
D
(
Chung
et
al.,
1999;
Ou
et
al.,
1995).
This
phenomenon
has
been
termed
differential
enhanced
degradation.
The
cis­
and
trans­
1,3­
D
isomers
in
non­
enhanced
soils
were
principally
degraded
by
chemical
hydrolysis
to
the
corresponding
cis­
3­
chloroallyl
alcohol
(
IUPAC
name:
3­
chloropropene­
1­
ol)
and
trans­
3­
chloroallyl
alcohol.
Since
the
27
of
49
hydrolysis
rates
of
the
two
isomers
in
water
are
the
same,
the
degradation
rates
for
the
two
isomers
in
non­
enhanced
soil
should
be
the
same.
The
two
chloroallyl
alcohols
were
then
degraded
microbially
to
corresponding
cis­
and
trans­
3­
chloroacrylic
acid,
and
eventually
to
CO2
and
H2O
(
Ou,
1998).
A
bacterial
degrader
has
been
isolated
from
the
enhanced
soil
that
also
exhibited
differential
enhanced
degradation
between
cis­
and
trans­
1,3­
D.

Since
cis­
and
trans­
1,3­
D
are
first
degraded
in
soil
to
cis­
and
trans­
3­
chloroallyl
alcohol
by
chemical
hydrolysis
(
non­
enhanced
soil)
or
chemical
and
biological
hydrolysis
(
enhanced
soil),
chemical
hydrolysis
rates
of
the
two
isomers
depend
on
soil
or
water
temperature.
For
example,
hydrolysis
half­
lives
of
cis­
and
trans­
1,3­
D
in
water
at
20
and
30

C
are
11.3
and
3.1
days,
respectively
(
McCall,
1987).
The
average
half­
life
values
for
both
cis­
and
trans­
1,3­
D
in
non­
enhanced
soils
and
sterile
soils
at
24­
25

C
are
about
8
days.
Beside
biodegradation
rates
of
cis­
and
trans­
3­
chloroallyl
alcohol,
no
information
on
the
physicochemical
and
toxicological
properties
of
these
two
alcohols
has
been
found
in
the
literature.
Depending
on
the
test
duration
and
the
body
temperature
of
the
laboratory
animal,
the
toxicity
of
cis­
and
trans­
1,3­
D
may
actually
come
from
the
two
alcohols,
or
a
combination
of
1,3­
D
and
the
alcohols
because
cis­
and
trans­
1,3­
D
may
completely
or
partially
hydrolyze
to
corresponding
cis­
and
trans­
3­
chloroallyl
alcohol.
The
cis­
1,3­
D
isomer
is
more
toxic
to
soil
nematodes
than
the
trans­
isomer
(
McKenry
and
Thomason,
1974;
Shoemaker
and
Been,
1999).

Cis­
1,3­
D
has
a
higher
vapor
pressure
than
trans­
1,3­
D,
34.3
and
23
mm
Hg
(
Hornsby
et
al.,
1995),
respectively.
After
application
of
a
commercial
product
of
1,3­
D
to
field
plots,
cis­
1,3­
D
was
always
found
to
be
the
first
chemical
to
volatilize
from
a
field
plot
surface
in
Florida
sandy
soil,
usually
1
to
5
hours
after
application
and
followed
by
trans­
1,3­
D
1
to
3
hours
later,
depending
on
soil
temperature.
It
was
also
found
that
during
the
first
48
hours
after
application,
the
flux
rate
for
cis­
1,3­
D
was
about
1.5
to
>
3
times
greater
than
trans­
1,3­
D.
The
ratio
between
the
two
isomers
gradually
declined.
Since
large
amounts
of
cis­
and
trans­
chloroallyl
alcohol
may
form
in
soil,
the
two
alcohols
may
volatilize
into
the
atmosphere.
A
similar
chemical,
2­
chloropropene­
1­
ol,
has
a
boiling
point
of
about
133

C.
Therefore,
the
two
alcohols
could
be
less
volatile
than
cis­
and
trans­
1,3­
D.

Because
of
toxicity
difference
and
higher
volatility
of
cis­
1,3­
D,
one
possible
approach
would
be
to
take
into
account
the
individual
toxicity
and
emission
flux
for
the
two
isomers
for
the
establishment
of
buffer
zones
and
threshold
concentrations.

(
3C)
How
appropriate
is
it
to
use
a
flux/
emission
factor
from
a
single
monitoring
study
(
or
small
number
of
studies)
and
apply
it
to
different
situations
such
as
for
the
same
crop
in
a
different
region
of
the
country?

Panel
Response
Using
a
flux/
emission
factor
based
on
a
single
monitoring
study
or
a
few
studies
is
questionable.
One
Panel
member
emphasized
the
importance
of
considering
whether
a
28
of
49
single
flux
profile
is
appropriately
conservative
or
not.
The
emission
flux
behavior
should
be
investigated
for
different
soils,
soil
moisture
conditions
and
environmental
factors
that
might
exist
in
different
regions.
In
addition,
considering
the
variability
of
emission
fluxes,
a
stability
index
or
parameter,
such
as
the
bulk
Richardson
number,
would
be
a
more
appropriate
parameter
to
use
than
air
temperature.
Volatilization
fluxes
depend
on
wind
speed
and
turbulence
whose
effects
may
be
parameterized
through
a
stability
index,
such
as
Pasquill's.
Both
published
and
unpublished
studies
describing
volatilization
losses
of
1,3­
D
and
other
fumigants
should
be
carefully
reviewed.
At
a
minimum,
the
model
developers
should
use
data
from
all
of
their
own
flux
field
studies
(
described
during
their
presentation)
to
develop
a
PDF
describing
volatilization
losses
of
1,3­
D.
The
work
of
Gan
et
al.
(
1998),
Kim
et
al.
(
2003),
Thomas
et
al.
(
2004)
and
Schneider
et
al.
(
1995)
should
be
consulted.

The
plot
of
the
field
study
summary
(
slide
49
in
the
presentation)
shows
the
cumulative
volatilization
losses
as
a
percent
of
the
applied
material.
The
results
from
the
four
studies
show
very
different
cumulative
volatilization
behavior.
Both
the
Georgia
(
drip)
and
Florida
(
shank­
bed)
studies
show
very
steep
(
high)
initial
losses
for
about
four
or
five
days
followed
by
a
leveling
off
of
the
emissions.
The
cumulative
loss
profiles
for
the
two
California
studies
(
both
shank)
are
not
only
different
from
the
GA
and
FL
studies,
but
they
also
show
a
very
different
cumulative
loss
pattern.
The
presenters
stated
that
the
Imperial,
CA
data
were
discounted
because
the
study
only
lasted
eight
days.
Unless
there
was
some
technical
reason
that
the
Imperial
data
cannot
be
used
(
none
was
stated),
then
it
is
reasonable
to
compare
the
results
for
the
two
studies.
The
loss
pattern
from
the
Salinas
study
shows
almost
a
2­
day
delay
before
the
onset
of
significant
losses,
while
the
Imperial
study
shows
a
very
gradual
loss
pattern
with
time.
These
are
very
different
results.
The
slope
of
the
curve
is
significantly
different,
with
the
values
on
day
2
(
according
to
the
flux
profile
plot)
being
the
highest.
But
the
difference
between
Imperial
and
Salinas
shows
1%
vs.
15%
of
the
applied
fumigant
volatilized.
It
is
hard
to
see
how
a
scaling
factor
based
on
the
Salinas
data
could
be
accurately
applied
to
the
Imperial
data.
The
data
presented
in
slide
49
is
probably
the
best
example
of
why
the
results
of
only
one
study
should
not
be
used
as
the
basis
of
all
source
emission
data
input
for
the
model.

The
use
of
scaling
factors
may
be
reasonable
when
cumulative
fluxes
are
being
considered,
but
they
are
not
appropriate
when
acute,
period
fluxes
and
associated
hourly
downwind
air
concentrations
must
be
estimated.

Although
the
use
of
the
scaling
factors
may
be
reasonable
for
accounting
for
some
application
factors,
there
is
concern
that
some
scale
factors
may
not
be
realistic.
For
example,
the
factor
of
1.6
between
summer
and
winter
temperatures
appears
too
simplistic.
In
the
summertime
alone
there
is
a
probable
temperature
differential
of
close
to
that
between
the
hot
inland
valleys
such
as
around
Kern
County,
California
and
the
cooler,
coastal
areas
around
Santa
Cruz
and
Monterey
Counties,
California.
29
of
49
The
use
of
appropriate
scaling
factors
seems
to
be
based
on
the
user's
judgment,
but
it
would
be
more
appropriate
if
a
mechanistic
approach
was
used
to
develop
them
for
the
various
field
conditions
expected
to
be
encountered.
That
is,
what
are
the
most
influential
factors
or
processes
that
are
driving
the
observed
fluxes
and
how
can
these
be
input
to
the
process.
The
model
documentation
should
include
a
table
of
appropriate
scaling
factors
recommended
by
the
authors.

The
use
of
local
meteorological
data
is
also
suggested
because
weather
patterns
are
significantly
different
between
different
regions
such
as
the
coastal
areas
and
the
valley
areas
of
California.
Other
inputs
into
ISCST3
such
as
mixing
height
are
regionally
dependent.

(
3D)
Please
comment
on
SOFEA
©
'
s
capability
to
adequately
consider
multiple,
linked
application
events
on
an
airshed
basis
as
well
as
single
source
scenarios.

Panel
Response
The
model
developers
can
be
commended
for
attempting
to
consider
multiple,
linked
application
events.
The
regulatory
agencies
are
urged
to
consider
the
effects
of
multiple,
linked
application
events
of
a
particular
compound,
and
indeed,
of
other
compounds.
In
the
real
world,
such
events
are
happening
and
people
are
exposed
to
fumigant
emissions
and
other
types
of
emissions
in
combination.
However,
as
discussed
in
relation
to
other
questions,
SOFEA
©
'
s
reliance
on
the
ISCST3
model
for
description
of
atmospheric
dispersion
limits
the
accuracy
of
the
multiple­
source
simulation.
Field
locations
cannot
presently
be
specified
as
source
area
inputs
to
the
model,
but
certain
areas
in
the
quadrant
can
be
assigned
a
higher
weighting
factor.
This
was
viewed
by
some
Panelists
to
be
an
acceptable
compromise,
but
other
Panelists
pointed
out
that
fields
can
be
abutted
in
practice
(
which
would
increase
impact
over
that
predicted
by
the
model),
and
the
overflow
algorithm
effectively
spreads
application
of
the
fumigant
over
a
larger
area
(
which
would
decrease
predicted
impact).
The
capability
of
considering
multiple
sources
is
very
important,
particularly
when
predicting
chronic
exposures.
More
detailed
land
use
data
may
improve
the
model's
capabilities.

(
3E)
Does
SOFEA
©
appropriately
address
situations
where
data
are
missing?

Panel
Response
The
ability
of
SOFEA
©
to
address
missing
data
was
absent
from
the
presentation
and
the
provided
documentation.
It
is
implicit
that
all
the
inputs
are
required;
otherwise,
the
model
will
not
run.
In
that
sense,
it
forces
the
user
to
ascertain
whether
the
input
is
complete,
which
can
be
a
beneficial
process.
The
documentation
does
not
address
how
it
deals
with
missing
data,
however,
and
this
aspect
should
be
included
in
the
future,
particularly
the
use
of
PCRAMMET.
30
of
49
Question
4:
The
integration
of
meteorological
data
into
ISCST3
is
one
of
the
key
components
that
separates
the
SOFEA
©
methodology
from
that
being
employed
by
the
Agency
in
its
current
assessment.
This
information,
coupled
with
GIS
(
Geographical
Information
Systems)
data
such
as
the
amount
of
ag­
capable
land
cover,
elevation,
and
population
densities
are
optional
inputs
for
SOFEA
©
.

(
4A)
Can
the
Panel
comment
on
the
value
of
adding
this
information
for
conducting
spatially
realistic
simulations?

Panel
Response
The
addition
of
hourly­
averaged
meteorological
information
and
GIS
data
seems
to
be
a
useful
part
of
the
methodology
implemented
in
SOFEA
©
.
Such
information
seems
to
be
a
step
forward
from
the
present
assumption
of
worst
case
meteorological
conditions
over
the
duration
of
the
release.
As
discussed
previously,
the
case
study
for
SOFEA
©
was
based
on
1,3­
D
for
which
acute
and
chronic
exposures
must
be
considered.

For
acute
exposures,
using
meteorological
conditions
from
a
distant
meteorological
station
may
not
accurately
reflect
local
conditions,
especially
effects
such
as
drainage
flows.

For
chronic
exposures,
dispersion
modeling
must
be
done
over
much
larger
distances.
The
use
of
hourly
meteorological
information
from
a
single
monitoring
station
should
not
be
replicated
over
large
areas
(
airsheds)
because
simple
replication
of
the
same
meteorological
data
for
all
sources
will
ignore
terrain
features
which
importantly
effect
conditions.
(
Although
SOFEA
©
provides
for
gridded
inputs
of
elevation
and
land
cover,
it
does
not
utilize
them
to
determine
wind
vectors
or
surface
roughness.)
A
massconsistent
wind
flow
model,
such
as
the
National
Atmospheric
Release
Advisory
Center's
(
NARAC)
Atmospheric
Data
and
Parametrization
Tool
(
ADPT)
or
the
Naval
Research
Laboratory's
Coupled
Ocean/
Atmospheric
Mesoscale
Prediction
Systems
(
COAMPS),
could
be
incorporated
into
SOFEA
©
.
Such
a
flow
model
takes
a
few
observations
of
wind
and
temperature
from
surface
stations
and
profilers,
and
creates
a
grid
of
surface
wind
vectors.
This
increase
in
sophistication
of
wind
flow
comes
at
a
heavy
cost
but
would
substantially
increase
the
reliability
of
meteorological
input
to
a
dispersion
model.
Alternatively,
mesoscale
models
such
as
the
National
Center
for
Atmospheric
Research
(
NCAR)
Mesoscale
Model
version
5
(
MM5),
or
the
Colorado
State
Regional
Atmospheric
Modeling
System
(
RAMS),
can
be
employed
to
yield
terrain­
influenced
wind
vectors.

Using
a
constant
value
of
mixing
height
may
be
adequate.
SOFEA
©
'
s
suggested
value
of
320
m
appears
to
be
on
the
conservative
side,
but
local
air
regulators
should
be
consulted.
In
reality,
daytime
mixing
heights
have
a
wide
range
and
show
strong
diurnal
and
seasonal
variations.
They
also
depend
on
the
land
use
and
proximity
to
coastline.
Changes
to
the
mixing
height
will
only
affect
predicted
concentrations
several
kilometers
from
the
source.
31
of
49
(
4B)
There
are
several
potential
sources
of
meteorological
and
GIS
data
(
e.
g.,
National
Weather
Service
and
California
Irrigation
Management
Information
System
or
CIMIS).
Please
comment
on
the
methods
used
to
select
these
data
including
locations
for
meteorological
stations.

Panel
Response
Major
sources
of
meteorological
data
are
the
NWS,
CIMIS
for
California,
and
other
local
and
state
climate,
agricultural,
and
industrial
meteorological
stations.
Measurement
height
for
surface
data
may
vary
from
2
to
10
m.
Quality­
assured
data
should
be
selected
from
the
closest
representative
site,
whether
it
is
a
NWS
location,
state
air
quality
or
climatological
site,
or
an
industrial
monitoring
site.
At
least
five
years
of
continuous
data
are
recommended,
but
longer
periods
of
data
might
be
necessary
when
evaluating
long­
term
chronic
exposures.
If
on­
site
data
for
a
short
period
are
used,
it
should
be
compared
with
the
nearest
available
long
term
NWS
data.
The
local
air
quality
regulators
should
be
consulted
for
the
selection
and
use
of
appropriate
meteorological
data.

(
4C)
What
criteria
should
be
used
to
identify
airsheds
for
analysis
and
how
should
data
be
selected
to
address
each
airshed?
Please
comment
on
the
manner
in
which
these
data
are
processed.

Panel
Response
An
airshed
is
not
defined
by
the
particular
domain
of
the
SOFEA
©
model,
i.
e.,
23
X
23
townships
or
smaller,
but
rather,
by
the
relevant
meteorology
and
transport
of
the
chemical
of
interest.
For
some
chemicals,
the
significant
airshed
(
i.
e.,
source
areas
over
which
a
given
receptor
may
experience
significant
exposure
levels)
may
be
larger
than
the
largest
domain
allowed
in
the
current
version
of
SOFEA
©
.
The
use
of
the
term
airshed
implies
that
all
significant
source
areas
have
been
included.
This
is
not
necessarily
the
case
with
SOFEA
©
.

The
maximum
domain
extent
recommended
for
use
in
SOFEA
©
should
be
influenced
by
the
dispersion
model
it
incorporates.
With
the
choice
of
ISCST3,
the
domain
should
be
limited
such
that
the
largest
distance
between
any
source
and
receptor
of
interest
is
less
than
about
50
km.
Over
larger
distances,
Gaussian
dispersion
parameters
cannot
be
specified
because
the
Pasquill­
Gifford
dispersion
coefficients
used
in
ISCST3
were
originally
based
on
experimental
data
taken
within
about
30
km
from
the
source.
(
ISCST3
seems
to
arbitrarily
limit
the
extent
of
a
plume
to
80
km.)
The
ISCST3
assumption
of
constant
hourly­
averaged
meteorology
and
flat
terrain
over
the
entire
ISCST3
domain
is
questionable
for
larger
model
domains
such
as
might
be
required
for
modeling
an
airshed.
32
of
49
(
4D)
Data
quality
and
uncertainty
associated
with
these
data
vary
with
the
source.
Does
the
Panel
agree
with
the
approaches
used
to
characterize
these
factors?

Panel
Response
Approaches
to
characterize
the
quality
and
uncertainty
of
meteorological
data
were
described
by
the
developers
of
SOFEA
©
,
but
it
is
not
clear
that
these
factors
have
been
fully
addressed.
A
procedure
for
filling
in
missing
data
was
discussed.
The
SOFEA
©
model
applies
meteorological
data
from
a
previous
or
future
year
for
the
date
and
hour(
s)
of
the
missing
data.
A
better
alternative
would
be
to
use
the
meteorological
data
from
the
previous
valid
hour
or
valid
hour
after
the
missing
period,
thereby
using
the
consistency
method.

A
critical
factor
to
consider
is
the
length
of
the
meteorological
record
used
in
the
model.
In
the
case
study
reported,
the
CIMIS
meteorological
data
of
5
consecutive
years
was
used
in
a
much
longer­
term
exposure
analysis.
These
data
were
sampled
stochastically
during
a
simulation
assuming
a
uniform
distribution
with
each
weather
year
assigned
an
equal
probability.
The
5­
year
length
of
weather
record
in
this
case
study
appears
inappropriately
short
for
the
treatment
of
long
(>
20
years)
term
exposures.
Criteria
should
be
developed
for
the
appropriate
data
record
length
for
long
term
exposure
assessments
that
will
ensure
that
extreme
events
which
most
likely
contribute
the
highest
exposures
will
be
taken
into
account.
This
is
the
standard
practice
when
conducting
exposure
assessments
to
estimate
the
magnitude
of
human
and
ecological
risks
associated
with
pesticide
use.

(
4E)
Anemometer
sampling
height
has
been
identified
as
a
concern
by
the
Agency
in
preparation
for
this
meeting.
What
are
the
potential
impacts
of
using
data
collected
with
different
anemometer
heights
in
an
analysis
of
this
nature?

Panel
Response
In
SOFEA
©
,
the
ISCST3
model
uses
the
mean
wind
speed
and
wind
direction
at
a
standard
height
of
10
m
to
estimate
the
mean
transport
wind
conditions
in
the
Gaussian
plume
formulas
for
surface
sources.
Some
data
sources
have
wind
measurements
at
a
lower
elevation
(
e.
g.,
2
m
in
CIMIS).
The
exact
details
of
how
the
ISCST3
model
uses
windspeed
and
elevation
data
for
area
sources
as
used
in
SOFEA
©
needs
to
be
considered.

The
use
of
wind
observations
at
2
m
elevation
may
be
appropriate
for
evaluating
short
term,
acute
exposures
because
the
lower
measurement
level
may
be
more
reflective
of
conditions
relevant
to
near­
field
1.5
m
high
receptors.
The
difference
between
the
wind
measured
at
2
m
and
10
m
is,
in
general,
subject
to
the
stability
conditions
of
the
atmosphere
and
the
type
of
terrain.
(
It
should
be
noted
that
the
Panel
reviewing
the
Fumigant
Exposure
Modeling
System
(
FEMS)
using
metam
sodium
as
a
case
study,
stated
it
is
preferable
to
have
vertically
resolved
air
concentrations
and
to
have
33
of
49
meteorological
data
for
1.5­
2
m
and
10
m
during
the
testing
period.
See
SAP
Report
No.
2004­
07
dated
November
9,
2004.)

(
4F)
Does
SOFEA
©
treat
meteorological
stability
class
inputs
appropriately?

Panel
Response
SOFEA
©
treats
stability
class
inputs
appropriately
for
its
current
use
of
ISCST3.
Dispersion
coefficients
are
specified
as
a
function
of
the
Pasquill
stability
classes
(
which
are
a
discrete
measure
of
stability).
Other
dispersion
models,
such
as
AERMOD,
can
treat
stability
as
a
continuous
variable.

(
4G)
Does
SOFEA
©
appropriately
calculate
bounding
air
concentration
estimates?

Panel
Response
The
current
approach
used
in
SOFEA
©
may
not
yield
the
highest
upper­
bound
concentrations.

 
The
coarse
receptor
grid
used
for
chronic
exposures
may
be
a
serious
limitation.
For
determining
highest
upper­
bound
concentrations,
additional
receptors
should
be
considered
at
short
distances
(
just
outside
the
buffer
zones)
from
the
treated
fields.
 
As
pointed
out
in
the
response
to
Question
2,
multiple
treated
fields
can
be
abutted
in
practice
(
which
would
increase
impact
over
that
predicted
by
the
model),
and
the
overflow
algorithm
effectively
spreads
application
of
the
fumigant
over
a
larger
area
(
which
would
decrease
predicted
impact).
 
The
emission
flux
based
on
the
single
field
test
and
treated
in
a
stochastic
manner
may
not
yield
the
highest
concentration,
especially
when
considering
the
tails
of
the
exposure
distribution
and
the
potentially
high
degradation
rates
in
the
selected
field
test.
 
The
highest
concentrations
from
surface
sources,
such
as
treated
fields,
are
likely
to
occur
during
very
stable,
low­
wind
or
calm
conditions.
A
Gaussian
dispersion
model,
such
as
ISCST3
used
in
SOFEA
©
,
is
not
particularly
applicable
under
such
conditions.
For
such
conditions,
more
sophisticated
models,
such
as
CALPUFF,
should
be
considered.
Drainage
flows
would
likely
impact
acute
exposures,
but
these
are
not
considered
in
ISCST3.
 
A
recent
study
has
shown
that
ISCST3
may
significantly
overestimate
vertical
dispersion,
i.
e.,
"
sigma
z"
(
Minnick
et
al,
2002).
Such
an
overestimation
of
vertical
dispersion
will
result
in
an
underestimation
of
concentration.
Such
underestimates
may
greatly
influence
the
peak
concentrations
observed
in
the
near
field.

Question
5:
The
Agency
model,
ISCST3
is
a
critical
component
of
the
SOFEA
©
approach.
This
model
has
been
peer
reviewed
and
is
commonly
used
for
regulatory
34
of
49
purposes
by
the
Agency.
SOFEA
©
also
uses
other
Agency
systems
such
as
PCRAMMET
and
PRZM3
as
well
as
the
USDA
model
CHAIN­
2D.

(
5A)
Please
recommend
any
parameters
that
should
be
altered
to
optimize
the
manner
that
they
are
used
in
SOFEA
©
.

Panel
Response
Questions
were
raised
about
the
accuracy
of
the
PRZM3
model.
CHAIN­
2D
was
considered
to
be
more
realistic
than
PRZM3.
However,
it
was
recognized
that
the
computational
resources
needed
to
run
the
CHAIN­
2D
model
were
much
greater
than
that
needed
for
PRZM3.

SOFEA
©
developers
used
PRZM3
simulations
to
describe
1,3­
D
flux
from
soil.
Order
of
magnitude
agreement
between
measured
1,3­
D
losses
and
simulated
results
(
cumulative
loss)
were
reported
(
Cryer
et
al.,
2003).
These
results
suggest
that
PRZM3
may
provide
realistic
flux
estimates;
however,
considerably
more
data
are
needed
before
a
meaningful
(
i.
e.,
statistically
valid)
conclusion
can
be
reached.
An
inherent
limitation
of
PRZM3
is
that
it
produces
results
based
on
daily
(
24­
hour)
time
steps.
This
presents
a
problem
in
linkage
of
model
outputs
to
ISCST3,
which
simulates
dispersion
based
on
hourly
time
steps.
Since
meteorology
can
vary
greatly
from
hour
to
hour,
the
use
of
emissions
flux
estimates
averaged
over
much
longer
time
periods
may
introduce
significant
errors
into
the
estimates.
Cryer
et
al.
(
2003)
addressed
this
(
in
their
application
of
PRZM3)
by
converting
the
PRZM3
daily
flux
to
weighted
hourly
flux
estimates.
A
number
of
assumptions
were
required,
and
it
is
unknown
to
what
extent
such
assumptions
affected
estimates
because
sensitivity
analyses
were
not
reported.
No
flux
data
were
collected
on
hourly
time
scales
in
any
of
the
field
studies
described
so
that
data
are
not
available
to
evaluate
the
validity
of
the
assumptions.

In
general,
the
methodology
for
emissions
flux
estimation
in
SOFEA
©
is
thought
to
be
too
simplistic.
Meteorological
influences
on
emissions
fluxes
should
be
considered,
as
should
soil
type
and
soil
moisture.
Emission
flux
estimates
should
be
based
on
more
highly
resolved
temporal
measurements
(
e.
g.,
hourly).
In
addition,
the
time
of
application
should
be
considered
as
a
factor.
If
the
time
of
application
is
much
different
from
that
of
the
field
test
used
to
estimate
fluxes,
then
the
estimates
might
be
substantially
off.

The
algorithms
used
in
the
PRZM3
model
are
believed
to
be
very
similar
to
those
used
in
the
PEARL
and
PELMO
models,
described
in
Wolters
et
al.,
2003.
In
this
paper,
the
latter
two
models
were
evaluated
by
comparing
their
predictions
with
experimental
measurements.
It
was
found
that
" 
model
predictions
deviated
markedly
from
measured
volatilization
rates
and
showed
limitations
of
current
volatilization
models ."
The
deviations
were
particularly
pronounced
in
the
initial
emissions
stages,
and
sometimes
the
measured
flux
was
significantly
higher
than
the
simulated
flux.
35
of
49
(
5B)
ISCST3,
as
integrated
into
SOFEA
©
,
was
run
in
regulatory
mode
which
includes
the
use
of
the
"
calms"
processing
routine.
Does
the
Panel
concur
with
this
approach?
If
not,
please
suggest
a
suitable
alternative?

Panel
Response
In
the
ISCST3
model,
estimated
downwind
concentrations
are
inversely
proportional
to
the
windspeed.
When
the
windspeed
goes
to
"
zero",
the
model
cannot
be
used.
In
addition,
its
use
with
low
 
but
non­
zero
 
winds
is
not
recommended.
Thus,
methodologies
have
been
developed
for
the
use
of
ISCST3
model
in
these
conditions.

There
is
some
uncertainty
as
to
exactly
what
methodologies
were
followed
in
these
conditions
within
the
SOFEA
©
modeling
system.
The
model
developers
present
at
the
meeting
were
not
able
to
describe
the
methodologies
used.
And,
unfortunately,
there
seems
to
be
some
ambiguity
in
various
documents
purporting
to
describe
these
methodologies
(
e.
g.,
ISCST3
source
code
and
User's
Guide;
EPA
web
site
material;
40
CFR
Ch
I.;
PCRAMMET
source
code
and
user's
guide).

In
the
following,
a
particular
set
of
methodologies
is
assumed
to
have
been
followed.
The
Panel
assumed
that
in
the
regulatory­
mode
application
of
the
ISCST3
model,
if
the
meteorological
data
specifies
"
calm"
or
a
wind
speed
of
0
m/
sec,
all
downwind
concentrations
are
set
to
zero.
A
slight
correction
is
made
by
not
counting
that
particular
hour
in
estimating
the
average
concentration.
That
is,
if
one
is
averaging
over
24
hours,
and
2
of
the
hours
are
"
calm",
then
one
takes
the
average
concentration
just
for
the
22
hours
that
were
non­
calm.
However,
even
this
slight
correction
only
goes
so
far.
For
example,
in
regulatory
mode,
if
there
are
less
than
18
non­
calm
hours
in
a
24­
hour
period,
then
the
24­
hour
average
is
estimated
by
dividing
the
sum
of
the
non­
calm
concentrations
by
18.
This
has
the
effect
of
reducing
the
estimated
average
concentration.

If
the
meteorological
data
specifies
a
wind
speed
greater
than
zero
but
less
than
1
m/
sec,
then
the
wind
speed
is
arbitrarily
increased
to
1
m/
sec
in
the
use
of
ISCST3
in
regulatory
mode.
This
procedure
also
has
the
effect
of
reducing
the
estimated
concentrations.

The
methodologies
are
described
in
40
CFR
Ch.
I.
Unfortunately
these
methodologies
have
the
potential
to
allow
the
highest
actual
concentrations
to
be
underestimated
or
even
set
to
zero.
This
is
perhaps
the
most
critical
of
all
the
ways
in
which
the
SOFEA
©
model
may
underestimate
concentrations
in
high
exposure
situations.

Not
all
models
or
methodologies
have
the
limitations
that
the
ISCST3
model
has
at
zero
and
low
wind
speeds.
In
one
study
(
Coulter
and
Eckhoff,
1998),
the
CALPUFF
model
was
compared
with
ISCST3,
and
because
of
the
method
of
handling
calms,
lowwind
speed
situations,
and
wind
reversals
in
ISCST3,
the
ISCST3
model
tended
to
underestimate
concentrations
relative
to
the
CALPUFF
model.

The
"
recirculation"
problem
deserves
special
mention
here.
Consider
the
following
set
of
circumstances.
In
the
first
hour,
the
wind
is
blowing
in
a
particular
direction
at
a
fairly
low
speed.
As
a
result,
there
are
relatively
high
concentrations
in
the
near
field
downwind
of
the
source.
In
the
next
hour,
imagine
that
the
wind
"
reverses"
36
of
49
direction
but
still
remains
relatively
low.
In
reality,
the
pollutant
dispersed
the
previous
hour
will
be
blown
back
to
the
same
near­
field
receptors
"
hit"
in
that
previous
hour,
increasing
their
exposure.
Since
these
are
the
conditions
of
highest
exposure
(
i.
e.,
low
wind
speed),
the
"
extra"
exposure
due
to
this
wind
reversal
may
be
significant.
In
the
ISCST3
model,
when
the
wind
reverses
that
second
hour,
all
upwind
concentrations
are
set
to
zero,
and
the
"
extra"
exposure
due
to
recirculation
is
not
counted.
In
estimating
average
exposure
averaged
over
all
wind
directions
for
long
periods,
the
effect
of
this
unrealistic
ISCST3
methodology
may
not
be
overwhelmingly
significant.
However,
in
the
estimation
of
peak
concentrations
and
exposure,
the
effect
of
this
problem
may
be
very
significant.
This
is
another
example
of
the
way
in
which
the
current
version
of
the
SOFEA
©
model
is
vulnerable
to
underestimating
peak
exposures.

There
are
several
approaches
that
might
be
investigated
to
attempt
to
deal
with
the
various
shortcomings
discussed
above.

 
Existing
field
measurement
data
should
be
examined
to
determine
if
measurements
were
made
during
calm,
low­
wind,
or
recirculating
conditions.
If
such
data
exist,
they
can
be
used
to
(
a)
characterize
the
degree
of
"
error"
in
the
ISCST3
simulation
and
(
b)
serve
as
a
basis
for
developing
an
empirical
correction
to
model
predictions.
If
insufficient
data
are
available
from
existing
studies,
current
and
future
field
studies
could
be
modified
to
introduce
measurements
under
calm,
low­
wind,
and/
or
recirculation
conditions.
 
Consider
abandoning
the
use
of
the
ISCST3
model
as
the
"
engine"
driving
the
dispersion
estimates
in
SOFEA
©
.
In
its
place,
a
more
realistic
model
that
does
not
have
as
severe
limitations
under
calm,
low­
wind,
and
recirculation
conditions
could
be
considered
such
as
CALPUFF
or
other
Lagrangian
puff
models.
 
A
simple
approach
that
might
offer
some
improvement
would
be
to
allow
the
emissions
flux
during
calm
hours
to
build
up,
so
that
the
emissions
in
the
first
hour
after
a
calm
period
would
include
that
hour's
emissions
plus
all
the
emissions
during
the
preceding
calm
period.
This
would
not
address
the
lowwind
or
recirculation
problem
and
would
only
partially
address
the
calms
problem,
but
it
seems
to
be
better
than
the
present
approach.
Other
ad
hoc
approaches
may
be
possible,
but
all
such
approaches
should
be
carefully
considered.

Critical
Element
3:
Results
Question
6:
Soil
fumigants
can
be
used
in
different
regions
of
the
country
under
different
conditions
and
they
can
be
applied
with
a
variety
of
equipment.

(
6A)
Please
comment
on
to
what
extent
the
methodologies
in
SOFEA
©
can
be
applied
generically
in
order
to
assess
a
wide
variety
of
fumigant
uses?
What
considerations
with
regard
to
data
needs
and
model
inputs
should
be
considered
for
such
an
effort?
37
of
49
Panel
Response
The
oral
presentation
at
the
meeting
described
a
case
study
of
1,3­
D
use
in
central
California
(
CA)
and
found
order­
of­
magnitude
agreement
between
predicted
and
measured
concentrations.
However,
results
indicated
that
SOFEA
©
may
under­
predict
both
chronic
and
peak
exposures
at
the
high­
end
of
exposure
distributions
(>
90%).
It
is
unknown
whether
this
is
a
characteristic
of
the
model.
Additional
study
may
provide
insight.
Several
recommendations
were
made
by
SAP
Panel
members,
which
may
help
to
guide
future
efforts.

While
there
do
not
appear
to
be
major
methodological
problems
with
the
successful
application
of
SOFEA
©
in
settings
other
than
the
CA
Central
Valley,
successful
applications
are
hindered
by
data
needed
to
run
the
model.
There
are
four
principal
areas
of
concern:
product
use,
flux
estimates,
weather
and
topography.
One
Panel
member
noted
that
the
development
of
"
scenarios"
that
use
site
specific
data
that
are
more
representative
of
other
regions
of
the
country
may
be
helpful.

Product
Use
For
the
CA
1,3­
D
case
study,
the
registrant
hired
a
contractor
to
"
mine"
use
data
from
the
CA
Pesticide
Use
Records
(
PUR)
database
to
obtain
critical
information
needed
to
run
SOFEA
©
.
PUR
includes
information
on
application
locations
(
Township,
Range,
and
Section),
application
date,
rate,
depth,
field
size,
crop
type,
and
total
pounds
of
fumigant
used.
One
SAP
Panel
member
expressed
misgivings
about
the
quality
of
data
available
in
the
PUR
database.
The
California
Department
of
Pesticide
Regulation
(
CDPR)
representative
present
at
the
meeting
acknowledged
this
and
indicated
that
there
were
some
efforts
to
"
correct"
extreme
and
or
missing
values.
While
uncertainties
remain,
PUR
currently
represents
the
best
available
data
and
is
the
only
data
gathering
effort
of
this
type
in
the
country.
In
the
absence
of
data
of
this
type,
pesticide
use
data
at
a
watershed
and/
or
airshed
scales
are
estimated
using
"
census
of
agriculture"
data,
farmgate
reports
(#
acres
in
production
by
county)
or
other
estimates
of
land
use
and
USDANational
Agricultural
Statistics
Service
pesticide
use
profiles
for
various
crops.
The
general
approach
is
described
by
Thelin
and
Gianessi
(
2000).
The
result
provides
a
"
best­
guess"
estimate
of
pesticide
use
by
active
ingredient
type
and
amount.
Uncertainty
bounds
on
estimates
are
unknown
and
are
likely
large
in
some
cases.
Agricultural
census
data
are
collected
and
reported
nationally
on
5
year
cycles.
Where
population
pressures
result
in
rapid
conversion
of
land
to
non­
agricultural
uses,
the
census
may
not
be
representative
of
current
conditions.
This
is
also
the
case
with
cropping
patterns
where
changes
in
crop
distributions
may
be
quite
rapid
due
to
factors
such
as
pest
pressure
or
economics.

As
pointed
out
by
SOFEA
©
developers,
obtaining
land­
use
data
that
delineates
whether
land
is
agriculturally
capable
is
relatively
straightforward.
LANDSAT
imagery
and
other
data
sources
are
readily
available
and
can
be
used
for
this
purpose.
The
problems
are
in
identifying
how
much
land
is
associated
with
a
given
crop
and
its
38
of
49
pesticide
treatment
history
for
a
given
year.
In
the
absence
of
these
data,
the
"
actual
use"
approach
described
for
the
1,3­
D
case
study
does
not
appear
feasible
in
many
locations.

An
alternative
is
to
create
crop
use
scenarios.
This
approach
is
well­
established
in
FQPA
drinking
water
risk
assessments.
Scenarios
for
a
variety
of
crops
have
been
developed
for
the
PRZM3
model
to
examine
the
potential
for
pesticide
runoff
impacts
on
surface
water
quality.
In
these
scenarios,
pesticide
applications
are
assumed
to
be
at
the
maximum
label
rate
with
adjustments
for
the
percent
crop
area
in
a
watershed
(
USEPA,
1999).
This
is
in
keeping
with
the
need
to
assess
potential
versus
actual
exposures.
This
approach
could
be
extended
to
fumigant
exposure
risks
where
product
use
data
are
not
available.

Flux
Estimates
Issues
surrounding
use
of
a
single
flux
profile
developed
in
a
CA
study
to
compute
flux
estimates
are
described
under
Question
3
in
this
report.
There
was
agreement
among
the
Panel
that
this
approach
has
significant
limitations
even
for
use
within
the
region
where
measurements
were
made.
This
is
not
to
say
that
the
approach
is
without
merit
for
regulatory
purposes,
provided
there
is
agreement
on
what
constitutes
an
appropriately
conservative
flux
profile.

In
a
generic
sense,
to
use
the
model
in
other
settings
or
for
other
fumigants,
region
and
fumigant
specific
profiles
need
to
be
determined
experimentally.
This
should
take
into
account
factors
such
soil
type
and
properties
such
as
bulk
density
and
organic
matter
levels,
water
content,
the
potential
for
enhanced
biodegradation,
local
weather,
and
other
conditions
which
likely
influence
flux
rates.
It
is
important
to
obtain
data
for
new
and
improved
fumigation
practices,
especially
those
that
include
emission
reduction
methods.
An
example
is
the
use
of
tarps
described
in
the
presented
1,3­
D
case
study.
Use
of
traditional
films
(
i.
e.,
high­
density
polyethylene)
may
not
be
very
effective
in
controlling
1,3­
D
or
other
fumigant
emissions.
Surface
water
sealing
may
be
useful
as
a
costeffective
emission­
reduction
strategy.
Virtually
impermeable
films
have
been
shown
to
significantly
reduce
emission
in
small­
scale
studies.
More
information
is
needed
at
agronomic
scales.
If
emissions
can
be
reduced,
lower
application
levels
should
provide
equivalent
control.

When
studies
are
conducted
to
obtain
flux
estimates,
the
Panel
agreed
that
the
aerodynamic
method
is
the
best
approach
(
see
responses
to
Question
3).
Some
studies
using
1,3­
D
were
described
at
the
meeting,
and
several
Panel
members
noted
that
there
is
a
large
body
of
soil
flux
data
for
fumigants
such
as
methyl
bromide.
Compilation
and
comparison
of
results
across
regions
may
prove
useful
in
the
identification
of
generic
profiles
that
are
suitable
for
use
as
a
SOFEA
©
input.
The
same
approach
may
prove
useful
in
identifying
ways
to
scale
flux
loss
by
the
time
of
year
in
which
the
fumigant
is
applied.
The
case
study
approach
used
a
single
value
to
represent
summer
and
winter
conditions.
Other
metrics
such
as
soil
temperature
on
application
date
may
be
a
more
effective
method
of
seasonal
adjustment
in
other
regions.
In
the
humid
southeast,
where
rainfall
is
120
to
150
cm
per
year,
some
consideration
should
be
given
to
conduct
of
39
of
49
experiments
that
quantify
impacts
of
rain
and
or
irrigation
on
fumigant
flux.
Data
reported
by
Gan
et
al.
(
1998)
indicated
that
water
sealing
(
which
may
occur
after
a
treated
field
receives
rain
or
is
irrigated)
may
suppress
cumulative
emissions
by
as
much
as
two­
fold.
Not
taking
this
into
account,
will
tend
to
make
flux
estimates
(
and
the
modeled
concentrations)
more
conservative.

In
the
absence
of
measured
data,
it
was
suggested
that
models
may
be
used
to
simulate
flux
profiles.
Use
of
the
USEPA
model
PRZM3
for
this
purpose
was
reported
in
one
published
study
(
Cryer
et
al.,
2003).
Order­
of­
magnitude
agreement
between
measured
and
simulated
flux
profiles
was
reported
in
two
cases.
However,
the
utility
of
PRZM3
to
predict
fumigant
flux
in
other
settings
is
unknown.
Given
this
and
other
concerns
discussed
in
Question
5,
it
appears
that
other
models
need
to
be
identified
and
evaluated.
Cryer
et
al.
(
2003)
showed
that
simulated
results
obtained
with
the
USDA
model,
CHAIN­
2D,
provided
a
reasonable
fit
for
measured
values.
While
potentially
useful,
it
appears
that
data
requirements
may
limit
CHAIN­
2D
applications
in
some
cases.

Weather
and
Topography
Proximal
weather
data
in
sufficient
detail
and
quality
(
observations
and
length
of
record)
may
not
be
available
to
conduct
SOFEA
©
simulations.
Thus
the
best
available
data
from
other
stations
(
in
close
proximity)
are
used.
There
are
inherent
uncertainties
in
this
approach
that
are
difficult
to
quantify.
Meteorological
situations
including
urban,
complex
terrain,
fields
in
densely
wooded
areas
that
may
need
special
treatment
should
be
identified.
Mesoscale
models
such
as
the
National
Center
for
Atmospheric
Research
(
NCAR),
Mesoscale
Model
version
5
(
MM5),
and
the
Colorado
State
Regional
Atmospheric
Modeling
System
(
RAMS)
can
be
used
to
generate
more
sophisticated
gridded
micrometeorological
wind
vector
data
for
these
situations.
In
addition,
when
used
for
regional
or
statewide
assessments,
emission
data
should
accurately
characterize
the
average
behavior
across
the
region
for
each
time
period.
For
accuracy,
this
information
needs
to
be
appropriate
for
a
given
locale,
time
of
year,
and
fumigation
type.
This
could
be
aided
by
development
of
a
series
of
cropping
and
"
worst­
case"
weather
scenarios
that
would
serve
to
provide
a
template
for
SOFEA
©
applications.

Question
7A:
Please
comment
on
whether
SOFEA
©
adequately
identifies
and
quantifies
airborne
concentrations
of
soil
fumigants
that
have
migrated
from
treated
fields
to
sensitive
receptors.

Panel
Response
In
many
respects,
the
SOFEA
©
model
does
not
adequately
identify
and
quantify
airborne
concentrations
of
soil
fumigants
that
have
migrated
from
treated
fields
to
sensitive
receptors.
In
particular,
estimates
of
worst
case,
near
field
exposures
do
not
appear
to
be
adequately
identified
or
quantified
by
the
SOFEA
©
model.
Some
of
the
reasons
for
this
deficiency
are:
40
of
49
 
Poor
or
nonexistent
treatment
of
calms
and
weak
winds
(<
2
m/
s)
with
highly
variable
wind
direction
(
see
Question
5);
 
Cutoff
of
emission
events
at
14
days,
when
significant
emissions
might
occur
after
that
period;
 
Inadequate
consideration
of
worst­
case
conditions
of
application
location,
receptor
location,
emissions,
meteorology,
etc.,
that
might
result
in
extreme
exposures;
and
 
Actual
times
and
conditions
of
application
are
different
from
those
of
the
field
measurements
used
to
derive
emission
flux
profile
(
see
Question
5).

SOFEA
©
assumes
a
receptor
height
of
1.5
meters,
to
simulate
an
adult's
inhalation
exposure.
Children
are
also
potentially
exposed,
and
lower
receptor
heights
should
be
considered
especially
for
near­
field
locations
where
the
concentrations
are
highest.
Because
children
are
more
vulnerable
to
exposure
(
due
to
their
lower
body
weight
and
potential
interferences
with
developmental
processes),
any
underestimate
of
their
exposure
should
be
avoided.

The
CALPUFF
model
follows
the
trajectory
of
previously
emitted
air
pollutants
when
calculating
hourly
concentrations
which
ISCST3
cannot
do.
The
use
of
CALPUFF/
CALMET
models
in
SOFEA
©
should
be
considered.
Acute
and
chronic
exposures
may
be
underestimated
because
of
the
use
of
ISCST3
in
SOFEA
©
Migration
of
soil
fumigants
from
treated
fields
at
large
distances
from
sensitive
receptors
will
be
necessarily
limited
to
about
50
km
beyond
which
ISCST3'
s
dispersion
coefficients
should
not
be
used.
Other
more
appropriate
wind
flow
and
dispersion
models
might
be
used
if
long­
term
exposures
from
far
fields
are
of
interest.

(
7B)
The
Agency
is
particularly
concerned
about
air
concentrations
in
the
upper
ends
of
the
distribution.
Are
these
results
presented
in
a
clear
and
concise
manner
that
would
allow
for
appropriate
characterization
of
exposures
that
could
occur
at
such
levels?

Panel
Response
SOFEA
©
model
results
seem
clearly
presented.
However,
concentrations
in
the
upper
ends
of
the
distribution
may
be
underestimated
for
several
reasons
previously
discussed:
chronic
exposures
at
long
distances
(
see
Question
4G),
calm
and
low
windspeed
conditions
(
see
Question
5),
and
multiple
sources
(
see
Question
3).
Empirical
evidence
for
the
inability
of
the
model
to
estimate
exposures
at
the
upper
ends
of
the
distribution
was
provided
by
slides
93
and
112
of
the
oral
presentation
at
the
SAP
review
meeting.

The
ability
of
SOFEA
©
to
predict
worst­
case
concentrations
is
likely
to
be
progressively
worse
for
longer
distances
and
exposure
durations.
Continuous
meteorological
data
over
the
longest
period
of
exposure
may
be
necessary
to
get
concentrations
in
the
upper
ends
of
the
distribution.
SOFEA
©
predicts
receptor
exposure
41
of
49
with
ISCST3,
which
assumes
steady­
state
plume
existence
in
the
wind
direction
for
each
hour
of
meteorological
data.
ISCST3
seems
to
arbitrarily
limit
the
extent
of
a
plume
to
80
km
(
see
responses
to
Questions
2A
and
4C).
Put
another
way,
SOFEA
©
could
predict
no
exposure
beyond
ISCST3'
s
artificial
limit
even
for
very
long
periods
of
time.

There
was
considerable
discussion
by
the
Panel
about
the
placement
of
near­
field
receptors
for
chronic
exposures.
A
comparative
study
of
peak
concentrations
and
spatial
distribution
of
peak­
to­
mean
ratios
might
be
useful.
It
is
possible
that
the
highest
calculated
concentrations
at
uniformly
spaced
grid
locations
may
be
multiplied
by
a
defensible
factor
to
estimate
the
potentially
higher
concentrations
in
the
gaps
between
modeled
receptor
locations.
Two
sets
of
receptors
may
be
used:
(
1)
a
uniform
grid
for
far­
field
exposures;
and
(
2)
receptors
in
close
proximity
(
minimum
distance
allowable
in
the
model)
to
treated
fields.
Although
it
would
increase
the
computation
time,
the
uniform
grid
spacing
should
be
tested
by
reducing
it
to
one­
half
of
that
used
in
the
current
application
and
comparing
the
results.

(
7C)
Please
comment
on
SOFEA
©
'
s
approach
for
calculating
and
presenting
probability
distributions
of
moving
average
concentrations
for
differing
durations
of
exposure.

Panel
Response
No
information
is
found
in
the
documentation
about
the
methodology
used
to
calculate
probability
distributions
using
moving
averages.
It
should
be
described
in
greater
detail
as
applied
to
the
SOFEA
©
model
results.

Hourly
observations
of
winds
should
be
sufficient
to
characterize
dispersion
in
every
case
except
when
sub­
hour
exposure
is
important
to
risk
assessment.
To
estimate
short­
term
acute
exposure,
a
temporal
peak­
to­
mean
ratio
could
be
employed.

(
7D)
Please
comment
on
the
types
of
monitoring
data
that
would
be
required
to
define
the
accuracy
of
simulations
made
with
SOFEA
©
for
differing
durations
of
exposure.

Panel
Response
Ideally,
atmospheric
dispersion
models,
such
as
SOFEA
©
with
ISCST3,
should
be
evaluated
by
comparing
model
predictions
for
a
particular
time
period
at
specific
locations
with
measurements
made
at
the
same
locations
during
the
same
time
periods.
In
carrying
out
such
an
evaluation,
it
would
be
important
to
utilize
the
meteorological
and
emissions
data
for
the
same
period.
This
type
of
model
evaluation
could
be
carried
out
for
SOFEA
©
and
is
needed,
but
does
not
appear
to
have
been
done.
Data
sets
with
fast
response
measurements
could
be
time
averaged
with
different
time
scales
to
compare
with
model
predictions
of
the
same.

Data
were
presented
for
a
comparison
of
simulated
and
measured
concentrations
for
selected
field
experiments
(
slide
33
of
the
oral
presentation).
While
the
initial
visual
42
of
49
comparison
of
the
simulated
and
measured
concentrations
looks
reasonable,
on
closer
inspection
one
finds
significant
under­
predictions
of
concentrations
during
certain
periods
of
high
concentrations.
Also,
the
results
of
only
two
out
of
eight
sampler
locations
were
presented
for
one
field
experiment.
It
is
unknown
whether
this
comparison
is
representative
of
the
other
experiments
and
measurements.
In
the
pseudoevaluation
shown
on
slide
93,
the
results
for
2001
were
presented,
but
what
were
the
results
for
the
year
2000?
Such
comparisons
should
be
reported
completely.

The
best
way
to
evaluate
a
model
is
through
comparison
with
actual
field
monitoring.
This
was
done
in
the
Kern
County
exercise
when
the
model
concentration
estimates
at
selected
downwind
receptor
points
were
compared
to
California
Air
Resources
Board
(
CARB)
ambient
monitoring
results
for
1,3­
D.
The
10­
year
simulated
concentrations
versus
exceedance
percentiles
(
slide
93
of
the
oral
presentation)
seem
to
have
good
agreement
with
the
CARB
ambient
air
monitoring
data
up
to
about
the
90th
percentile.
At
higher
percentiles,
SOFEA
©
underpredicts
monitoring
measurements.
These
results
show
that
further
refinement
of
the
model
is
necessary
to
correct
the
upper
end
of
air
concentration
distribution
estimates
and
to
determine
why
this
underprediction
is
occurring.
Depending
on
how
much
1,3­
D­
targeted
monitoring
information
is
available
from
the
CARB,
or
if
1,3­
D
is
included
in
the
toxics
air
monitoring
network,
all
available
ambient
air
concentration
data
should
be
used
to
evaluate
the
model
output,
including
the
concurrent
environmental
conditions,
such
as
meteorology,
field
characteristics
(
soil
type,
moisture,
etc.),
etc.,
to
determine
the
most
influential
processes
affecting
the
results
at
either
end
of
the
concentration
distribution.
These
may
be
different
than
the
original
set
of
factors
used.

SOFEA
©
results
are
strongly
dependent
upon
the
quality
of
the
source
emission
flux.
Using
the
"
worst­
case"
field
data
and
the
appropriate
scaling
factors
may
be
a
rational
approach
for
estimating
the
flux.
The
coarse
concentration
averaging
periods
(
6,
6,
and
12
hr
sampling
durations)
influenced
the
estimated
flux
values.
The
accuracy
of
exposure
estimates
can
be
assessed
by
taking
field
data
at
sufficient
detail
to
capture
the
effects
of
important
parameters
(
such
as
meteorological
conditions).
Such
data
can
then
be
averaged
over
different
time
scales,
and
SOFEA
©
simulations
can
be
compared
for
various
durations
of
exposure.

Question
8A:
What
types
of
sensitivity/
uncertainty
analyses
of
SOFEA
©
are
recommended
by
the
Panel
to
be
the
most
useful
in
making
scientifically
sound,
regulatory
decisions?

Panel
Response
In
the
initial
stages
of
model
development,
it
is
enough
to
run
selected
scenarios
and
interpret
the
results
one
scenario
at
a
time.
SOFEA
©
is
now
ready
for
more
than
that.
There
are
good
discussions
of
experimental
design
for
sensitivity
analysis
in
SAP
Minutes
2004­
01
"
Refined
(
Level
II)
Terrestrial
and
Aquatic
Models
Probabilistic
Ecological
Assessments
for
Pesticides:
Level
II
Aquatic
Model
Session"
(
see
the
response
to
charge
3d)
and
2004­
03
"
Refined
(
Level
II)
Terrestrial
and
Aquatic
Models
43
of
49
Probabilistic
Ecological
Assessments
for
Pesticides:
Terrestrial"
(
in
the
General
Comments).
In
the
Level
II
Aquatic
Model
Session
(
2004­
01),
the
work
of
Kleijnen
(
2004)
was
cited.
This
approach
uses
principles
of
experimental
design,
fractional
factorials
in
particular,
and
response
surface
methodology,
to
determine
which
assumptions
in
the
model
are
critical
and
which
factors
drive
the
simulation.
The
Panel
again
advocates
that
the
Agency
try
these
methods.

Panelists
had
a
number
of
specific
suggestions
for
factors
to
include
in
the
sensitivity
analysis:
background
(
ambient
air)
concentrations,
terrain,
location
(
inland
versus
coastal),
and
crop
type.
Inputs
into
the
model
that
affect
fumigant
dispersion
and
degradation
need
to
be
included
in
the
analysis:
soil
temperature,
weather
stability,
and
soil
degradation
of
the
applied
fumigant.
Atmospheric
degradation
factors
should
be
considered
as
they
affect
the
maximum
volatilization
and
maximum
losses
through
emission
into
the
atmosphere.

The
horizontal
(
Sigma­
Y)
and
vertical
(
Sigma­
Z)
dispersion
coefficients
in
the
ISCST3
model
could
be
examined
probabilistically
by
selecting
a
random
multiplier
to
Sigma­
Y
and
Sigma­
X
in
the
ISCST3
model
code.
The
distribution
of
the
random
multiplier
would
be
based
on
the
cumulative
distribution
function
observed
in
field
experiments.

Because
SOFEA
©
will
be
used
for
both
acute
and
chronic
exposure
assessment,
there
is
a
need
to
evaluate
the
uncertainty
in
both
periodic
and
cumulative
emissions.
Some
Panelists
believed
that
uncertainty
would
be
higher
for
periodic
emissions
than
for
cumulative.

While
the
efforts
taken
to
produce
reliable
emission
inputs
should
reduce
uncertainty
of
the
experimental
conditions,
information
about
sources
of
error
and
uncertainty
in
the
flux
estimation
should
be
provided.

Sensitivity
analysis
could
be
done
on
the
individual
components
of
the
model,
ISCST3,
PRZM3
and
CHAIN­
2D.
These
results
might
already
be
available
in
the
literature.

There
was
strong
agreement
that
a
meteorological
record
longer
than
five­
year
CIMIS
record
was
required,
to
ensure
that
some
"
worst
case"
scenarios
would
figure
in
the
sensitivity
analysis.

(
8B)
What
should
be
routinely
reported
as
part
of
a
SOFEA
©
assessment
with
respect
to
inputs
and
outputs?
Are
there
certain
tables
and
graphs
that
should
be
reported?
44
of
49
Panel
Response
Since
Excel
is
the
user
interface,
all
information
concerning
input
(
system
parameters,
probability
distributions,
spatial
and
temporal
inputs)
and
basic
numeric
output
are
in
an
Excel
workbook.
Users
should
have
no
difficulty
adding
graphs
and
tables
to
a
worksheet
and
these
will
be
updated
automatically
each
time
SOFEA
©
is
run.
A
strength
of
SOFEA
©
is
the
possibility
of
"
what
if"
scenarios.
The
output
figures
and
tables
will
be
dependent
on
the
problem
being
studied.

SOFEA
©
returns
tables
giving
exposure
at
many
locations
at
a
sequence
of
times.
In
the
first
stages
of
testing,
all
sorts
of
plots
will
be
needed
to
look
for
aberrant
values
and
generally
help
decide
if
the
results
make
sense.
Time
series
plots,
box
and
whisker
plots,
and
scatter
plots
will
be
useful
here
for
as
many
variables
and
combinations
of
variables
as
one
can
think
of.
Note
that
while
box
and
whisker
plots
are
very
useful
for
exploratory
data
analysis,
they
are
very
clumsy
to
create
in
Excel.
Further
down
the
line,
end­
users
will
appreciate
geographical
contour
plots
for
median
and
upper
percentiles
of
acute
and
chronic
exposure.

The
results
for
upper
percentiles
will
only
be
meaningful
if
the
model
captures
all
sources
of
variation
and
enough
simulations
are
run
under
each
scenario.

The
plot
of
concentration
versus
exceedance
percentile
for
the
pseudo­
validation
shown
in
the
Agency
presentation
(
handout
page
47)
shows
concentration
on
a
log
scale.
Even
though
statisticians
like
log
scales
because
the
plots
look
neater,
a
linear
scale
would
in
this
case
de­
emphasize
the
good
agreement
at
low
concentrations
and
exaggerate
the
poor
agreement
at
high
concentrations,
giving
a
very
different
impression.
If
we
accept
that
it
is
more
important
for
models
to
be
accurate
at
upper
percentiles,
diagnostic
plots
should
be
on
linear
scales.

Not
all
Panelists
were
able
to
try
running
SOFEA
©
because
most
did
not
have
access
to
Crystal
Ball,
and
hence
only
a
few
Panelists
were
able
to
make
specific
suggestions
concerning
routine
reporting
of
inputs
and
outputs.

There
was
general
agreement
that
the
inputs
required
should
include
the
fumigant
applied,
application
rate,
type
of
application,
application
depth,
tarp
use
or
none,
field
size
(
or
numbers
of
fumigated
fields
for
regional
analysis),
soil
conditions
that
will
affect
fumigant
dispersion
in
the
soil
and
subsequently
into
the
atmosphere
and
weather
parameters
that
affect
stability.
The
outputs
should
include
flux
rates,
fumigant
concentrations
at
buffer
perimeters
relative
to
toxicity
concentrations,
exceedance
frequency,
distance
from
the
source
at
which
exceedances
occur,
maximum
daily
emission,
and
losses
over
time
through
emission
into
the
atmosphere.

Another
recommendation
is
to
add
graphics
showing
the
distribution
of
statistical
parameters
at
a
selected
number
of
receptors,
say,
100
 
200
receptors
at
various
distances
from
the
source.
The
graphics
can
show
the
median
at
each
receptor
based
on
the
Monte
45
of
49
Carlo
runs,
as
well
as
the
coefficient
of
variation
(
standard
deviation/
mean),
and
variability/
uncertainty
range:

(
2.5th
percentile
­
mean)/
mean;
(
97.5th
percentile
­
mean)/
mean.

8(
C)
Does
the
Panel
recommend
any
further
steps
to
evaluate
SOFEA
©
and
if
so,
what?

Panel
Response
SOFEA
©
,
like
any
other
model
at
this
stage
of
development,
will
need
a
line­
byline
code
audit
by
an
independent
programmer
to
ensure
that
the
code
does
what
it
is
supposed
to
do.
The
hardest
programming
errors
to
detect
are
those
that
deliver
results
that
look
correct
but
in
fact
are
wrong.
A
code
audit
should
pick
up
any
errors
of
this
kind.
The
FORTRAN
code
in
particular
needs
to
be
audited
because
it
is
so
detailed.
SOFEA
©
relies
on
code
within
Crystal
Ball
and
Excel.
A
number
of
the
statistical
functions
in
Excel
are
known
to
be
deficient.
Serious
problems
with
the
Excel
random
number
generator
were
identified
in
SAP
Minutes
2000­
01
"
Session
III:
Dietary
Exposure
Evaluation
Model
(
DEEM)",
citing
McCullough
and
Wilson
(
1999).
We
need
documentation
and
testing
of
the
random
number
generator
in
Crystal
Ball
and
if
it
too
proves
to
be
deficient,
a
better
random
number
generator
must
be
used.

The
Panel
was
concerned
that
"
calms"
could
be
very
important
and
had
not
been
adequately
incorporated
into
the
model.
Perhaps
the
ISCST3
model
is
not
conservative
enough
in
this
regard.
It
is
a
limitation
of
ISCST3
that
no
stability
categories
are
applicable
for
nighttime
calm
and
near
calm
conditions.

The
Panel
recommends
running
a
series
of
simulations
to
determine
whether
the
shape
of
fumigant
flux
profiles
impact
acute
or
chronic
exposure
estimates
generated
by
the
model.
Profiles
can
be
developed
from
published
or
unpublished
studies
or
they
could
be
simulated.
Cumulative
losses
would
be
held
constant.
A
factorial
experiment
where
cumulative
losses
are
also
varied
may
prove
insightful
and
help
guide
decisions
on
the
conduct
of
additional
flux
field
experiments
or
in
the
selection
of
an
"
appropriately"
conservative
flux
profile
for
use
in
exposure
assessments.
Another
area
that
should
be
explored
is
the
impact
of
weather.
Simulations
should
be
run
under
a
variety
of
worstcase
conditions
to
determine
the
extent
to
which
"
extreme"
conditions
(
high
or
low
temperature,
wind,
stability
etc.)
may
influence
results.

Further
evaluation
of
SOFEA
©
'
s
ability
to
simulate
acute
and
chronic
exposure
at
receptor
points
would
be
helpful,
but
this
will
be
difficult
without
the
availability
of
extensive
data
sets.
It
may
be
possible
to
use
available
methyl
bromide
data,
but
evaluation
of
SOFEA
©
with
this
chemical
may
not
be
of
interest
to
the
developers.
This
may
be
an
appropriate
activity
for
EPA
or
CDPR.
This
might
also
be
a
good
first
step
in
promoting
SOFEA
©
to
the
user
community.
46
of
49
Despite
these
difficulties,
SOFEA
©
should
be
evaluated
with
other
fumigants
under
various
conditions.
There
is
a
need
to
evaluate
the
model
with
at
least
two
other
different
types
of
fumigants
using
real
data
from
different
areas
where
it
could
prove
useful.
Since
a
specific
fumigant
may
be
widely
used
in
certain
areas
and
use
rates
are
frequently
high,
knowledge
of
chronic
exposures
could
be
a
useful
regulatory
tool
for
risk
assessment
and
management.

The
broader
question
of
determining
whether
the
model
is
good
enough
is
much
more
difficult
to
address.
Because
of
the
wide
range
of
expertise
on
the
Panel,
there
were
many
suggestions
for
enhancing
the
model,
and
some
of
these
may
make
a
significant
difference
in
model
output
under
some
scenarios.
Because
we
could
go
on
forever
improving
the
model,
the
question
is
not
so
much
whether
the
model
is
completely
realistic,
but
rather,
is
it
complete
enough
for
regulatory
purposes?
At
this
stage,
the
Panel
recommends
incorporating
those
proposed
enhancements
that
look
most
promising
and
doing
more
validations
(
or
pseudo­
validations)
in
comparison
to
field
data,
looking
particularly
for
agreement
in
upper
percentiles
and
under
both
typical
and
extreme
scenarios.
It
is
encouraging
that
comparison
with
observed
field
data
seems
to
be
very
feasible
in
these
applications.

(
8D)
SOFEA
©
uses
a
Monte
Carlo
based
approach
based
on
varied
random
number
streams
for
each
simulation.
Can
the
Panel
comment
on
the
appropriate
statistical
techniques
that
should
be
used
to
define
differences
between
outputs
for
different
scenarios?

Panel
Response
This
is
the
correct
way
to
run
simulations.
In
the
exploratory
stage
of
development,
scenarios
should
be
run
several
times
with
independent
random
number
streams.
It
is
important
to
run
enough
replications
to
see
stability,
allowing
for
variability
in
field
sampling
data.

If
the
results
are
presented
as
cumulative
probability
distributions,
however,
it
will
be
difficult
to
analyze
them,
as
statistical
tests
such
as
the
Kolmogorov
Smirnov
test
to
compare
distributions
are
too
powerful
and
cannot
easily
be
extended
to
compare
the
results
of
many
different
scenarios.
The
best
approach
would
be
to
summarize
each
run
by
an
upper
quantile,
making
the
response
univariate.
The
variability
in
the
results
can
then
be
displayed
simply
with
box
and
whisker
plots
or
superimposed
time
series.

When
the
developers
proceed
to
a
more
formal
sensitivity
analysis
using
the
methods
advocated
in
(
8A),
the
variability
between
simulations
due
to
independent
random
number
streams
will
be
taken
into
account
in
the
factorial
analysis
of
variance.
47
of
49
REFERENCES
Arya,
S.
P.
2001.
Introduction
to
Micrometeorology,
Second
Edition,
Academic
Press,
2001,
pp.
228­
232.

Chung,
K.
 
Y.,
D.
W.
Dickson,
and
L.­
T.
Ou.
1999.
Differential
enhanced
degradation
of
cis­
and
trans­
1,3­
D
in
soil
with
a
history
of
repeated
applications
of
1,3­
D.
J.
Environ.
Sci.
Health
B34:
749­
768.

Coulter,
C.
T.,
and
P.
A.
Eckhoff.
"
A
Comparison
of
CALPUFF
with
ISC3",
EPA
report
EPA­
45/
R­
98­
020,
December
1998.

Cryer,
S.
A.,
I.
J.
van
Wesenbeck
and
J.
A.
Knuteson.
2003.
Predicting
regional
emissions
and
near­
field
air
concentrations
of
soil
fumigants
using
modest
numerical
algorithms:
a
case
study
using
1,3­
dichloropropene.
J.
Agric.
Food.
Chem.
51,
3401­
3409.

El
Hadiri,
N.,
M.
Ammati,
M,
Chgoura,
and
K.
Mounir.
2003.
Behavior
of
1,3­
dichloropropene
and
methyl
isothiocyanate
in
undisturbed
soil
columns.
Chemosphere
52:
893­
899.

Gan,
J.,
S.
R.
Yates,
D.
Crowley,
and
J.
O.
Becker,
1998.
Acceleration
of
1,3­
dichloropropene
degradation
by
organic
amendments
and
potential
application
emissions
reduction.
J.
Environ.
Qual.
27:
408­
414.

Gan,
J.,
S.
R.
Yates,
D.
Wang,
and
F.
F.
Ernst,
1998.
Effect
of
application
methods
on
1,3­
dichloropropene
volatilization
from
soil
under
controlled
conditions.
J.
Environ.
Qual.
27:
432­
438.

Hornsby,
A.
G.,
R.
D.
Wauchope,
and
A.
E.
Herner.
1995.
Pesticide
properties
in
the
environment.
Springer­
Verlag.
New
York.

Isakov,
A.,
T.
Sax,
A
Venkatram,
D.
Pankratz,
J.
Heumann,
and
D
Fitz.
2004
(
April)
Near­
field
dispersion
modeling
for
regulatory
applications.
J.
Air
&
Waste
Manage.
Assoc.
54:
473­
482.

Kim,
J.
H.,
S.
K.
Papiernik,
W.
J.
Farmer,
J.
Gan,
and
S.
R.
Yates.
2003.
Effect
of
formulation
on
the
behavior
of
1,3­
dichloropropene
in
soil.
J.
Environ.
Qual.
32:
2223­
2229.

Kleijnen,
J.
P.
C.
2004.
An
overview
of
the
design
and
analysis
of
simulation
experiments
for
sensitivity
analysis.
European
Journal
of
Operational
Research,
in
press.
(
Available
online
at
http://
center.
uvt.
nl/
staff/
kleijnen/
ejor_
review_
proof.
pdf)

McCall,
P.
J.
1987.
Hydrolysis
of
1,3­
dichloropropene
in
dilute
aqueous
solution.
Pestic.
Sci.
19:
235­
242.
48
of
49
McCullough,
B.
E.
and
B.
Wilson.
1999.
On
the
accuracy
of
statistical
procedures
in
Microsoft
Excel
97.
Computational
Statistics
and
Data
Analysis
31:
27­
37.

McKenry,
M.
V.,
and
I.
J.
Thomason.
1974.
Part
II.
Organism­
dosage­
response
studies
in
the
laboratory
with
several
nematode
species.
Hilgaria.
42:
422­
437.

Minnich,
T.
R.,
R.
L.
Scotto,
J.
J.
Brady,
Jr.,
H.
J.
Kohlmann,
P.
O'Connor,
S.
Mallik,
and
K.
Parekh.
2002.
Use
of
Open­
Path
FTIR
Spectroscopy
to
Support
Development
of
Refined
Estimates
of
H2S
Emissions
from
a
New
York
City
Municipal
Wastewater
Treatment
Plant.
Presented
at
the
Air
&
Waste
Management
Association's
95th
Annual
Conference
and
Exhibition,
June
23­
27,
2002,
in
Baltimore
Maryland,
available
at
http://
www.
msiair.
net/
nycwtp.
html
Nuclear
Utility
Meteorological
Data
Users
Group
(
NUMUG),
San
Francisco,
CA,
April
1996
available
as
the
report
UCRL­
JC­
123717.

Ou,
L.­
T.
1998.
Enhanced
degradation
of
the
volatile
fumigant­
nematicides
1,3­
D
and
methyl
bromide
in
soil.
J.
Nematol.
30:
56­
64.

Ou,
L.­
T.,
K.­
Y,
Chung,
J.
E.
Thomas,
T.
A.
Obreza,
and
D.
W.
Dickson,
1995.
Degradation
of
1,3­
dichloropropene
(
1,3­
D)
in
soils
with
different
histories
of
field
applications
of
1,3­
D.
J.
Nematol.
27:
249­
257.

Schneider,
R.
C.;
B.
S.
Sipes,
C.
H.
Oda,
R.
E.
Green,
and
D.
P.
Schmitt.
1995.
Management
of
1,3­
dichloropropene
in
pineapple
for
efficacy
and
reduced
volatile
losses,
Crop
Protect.
14:
611­
618.

Shoemaker,
C.
H.,
and
T.
H.
Been.
1999.
Compound
models
describing
the
relationship
between
dosage
of
(
Z)­
and
(
E)­
isomers
of
1,3­
dichloropropene
and
hatching
behavior
of
Globodera
rostochiensis.
Nematology.
1:
19­
29.

Thelin,
G.
P.
and
L.
P.
Gianessi.
2000.
Method
for
Estimating
Pesticide
Use
for
County
Areas
of
the
Conterminous
United
States.
U.
S.
Geological
Survey
Open
File
Report
00­
250.
USGS,
Sacramento,
CA.
(
available
on­
line
at
http://
ca.
water.
usgs.
gov/
pnsp/
rep/
ofr00250)
(
verified
September
15,
2004)

Thomas,
J.
E.,
L.
H.
Allen,
Jr.,
L.
A.
McCormack,
J.
C.
Vu,
D.
W.
Dickson,
and
L­
T.
Ou.
2004.
Diffusion
and
emissions
of
1,3­
dichloropropene
in
Florida
sandy
soil
in
microplots
affected
by
soil
moisture,
organic
matter
and
plastic
film.
Pest
Manag..
Sci.
60:
390­
398.

USEPA.
1999.
Estimating
the
Drinking
Water
Component
of
a
Dietary
Exposure
Assessment.
U.
S.
Environmental
Protection
Agency,
Office
of
Pesticide
Programs,
Washington,
D.
C.
(
available
on­
line
at
http://
www.
epa.
gov/
fedrgstr/
EPA­
PEST/
1999/
November/
Day­
10/
6044.
pdf)
(
verified
September
15,
2004)
49
of
49
USEPA.
2001.
General
Principles
for
Performing
Aggregate
Exposure
and
Risk
Assessments.
U.
S.
Environmental
Protection
Agency,
Office
of
Pesticide
Programs,
Washington,
D.
C.
(
available
on­
line
at
http://
www.
epa.
gov/
pesticides/
trac/
science/
aggregate.
pdf)
(
verified
September
15,
2004)

Wolters,
A.,
V.
Linnemann,
M.
Herbst,
M.
Klein,
A.
Schaffer,
and
H.
Vereecken
(
2003).
Pesticide
Volatilization
from
Soil:
Lysimeter
Measurements
versus
Predictions
of
European
Registration
Models.
J.
Environ.
Qual.
32:
1183
 
1193.
