July
2004
Final
Report:
December
2003
Peer
Review
of
the
CMAQ
Model
Submitted
to:
Community
Modeling
and
Analysis
System
Center
Carolina
Environmental
Program
The
University
of
North
Carolina
at
Chapel
Hill
Submitted
by:
Praveen
Amar
Northeast
States
for
Coordinated
Air
Use
Management
Robert
Bornstein
San
Jose
State
University
Howard
Feldman
American
Petroleum
Institute
Harvey
Jeffries
The
University
of
North
Carolina
at
Chapel
Hill
Douw
Steyn
The
University
of
British
Columbia
Robert
Yamartino
Consultant,
Integrals
Unlimited
Yang
Zhang
North
Carolina
State
University
2
Introduction
The
external
CMAQ
Model
Program
Peer
Review
Panel
that
reviewed
the
CMAQ
model
in
December
2003
provides
this
report
in
order
to
assess
the
current
state
of
the
model,
to
guide
its
development
in
the
short
term,
and
to
assess
the
appropriateness
of
resources
(
institutional
support,
staffing,
and
operational
funding)
in
the
long
term
to
achieve
desired
advances.
The
report
is
written
from
the
range
of
perspectives
represented
by
the
seven
members
of
the
review
panel:

 
Praveen
Amar
(
Northeast
States
for
Coordinated
Air
Use
Management)

 
Robert
Bornstein
(
San
Jose
State
University)

 
Howard
Feldman
(
American
Petroleum
Institute)

 
Harvey
Jeffries
(
The
University
of
North
Carolina
at
Chapel
Hill)

 
Douw
Steyn
(
The
University
of
British
Columbia)

 
Robert
Yamartino
(
consultant,
Integrals
Unlimited)

 
Yang
Zhang
(
North
Carolina
State
University)

The
panel
members
read
a
considerable
volume
of
material
on
CMAQ
provided
by
EPA,
and
attended
two
days
of
presentations
on
CMAQ
by
EPA
staff.

Federal,
state,
and
local
governments
as
well
as
regional
organizations
have
used
3­
D
deterministic
models
such
as
CMAQ
with
one
fundamental
goal
in
mind:
evaluation
and
the
optimum
selection
of
alternative
control
strategies
to
attain
ambient
standards
for
various
criteria
pollutants
The
main
regulatory
purpose
of
such
models
is
to
understand
how
future
changes
in
emissions
("
delta
E")
would
result
in
future
changes
in
concentrations
of
various
pollutants
("
delta
C").
Ozone
(
1­
h
standard)
has
been
the
primary
focus
so
far,
and
fine
particles
(
those
below
2.5
µ
m
and/
or
10
µ
m
in
diameter)
will
be
the
new
focus
in
addition
to
a
continuing
focus
on
ozone
(
8­
h
standard).
From
the
science,
policy,
and
implementation
perspectives,
particulate
matter
(
PM)
will
be
much
more
challenging
than
ozone,
in
that
PM
has
an
annual
standard
(
in
addition
to
the
24­
h
standard)
that
requires
annual
modeling
(
unlike
ozone,
which
is
usually
modeled
over
periods
of
a
few
days
to
a
few
weeks).
Also,
our
present
understanding
of
PM
chemistry
and
physics
is
considerably
weaker
than
that
for
ozone,
though
the
understanding
of
PM
is
evolving
rapidly.
PM
modeling
has
additional
problems
related
to
the
need
for
acceptable
and
quality­
assured
annual
data
for
emissions,
and
atmospheric
measurements
for
model
evaluation
In
addition,
the
most
sophisticated
treatments
of
aerosol
chemistry
and
physics
will
lead
to
substantial
increases
in
computational
requirements
(
run
time
and
hardware
requirements).

The
CMAQ
model
must
take
into
account
the
fact
that
ozone
plus
PM
modeling
could
involve
considerably
more
than
twice
the
work
of
ozone
modeling
alone.
Also,
in
addition
to
ozone
and
3
PM,
it
is
important
for
CMAQ
to
address
the
issues
of
air
toxics
(
including
mercury)
and
acid
deposition.

These
considerations
provided
a
context
for
the
panel's
review
of
CMAQ.
Of
great
interest
and
concern
are
efforts
to
expand
the
range
of
the
model's
scales
of
applicability
into
both
the
micro
scale
(
through
"
urbanization"
of
its
1­
km­
scale
version)
and
the
global
scale.
Such
scale
expansion,
while
consistent
with
a
growing
trend
toward
"
end­
to­
end"
models,
presents
considerable
technical
and
computational
challenges,
made
more
challenging
when
faced
with
limited
resources.

CMAQ
is
a
massive
computer
code
designed
to
model
a
wide
range
of
physical
and
chemical
processes
that
occur
at
particular
scales
in
the
lower
atmosphere.
Some
of
these
processes
are
well
understood
(
for
example,
dispersion
of
pollutants
by
mean
wind
and
turbulent
fluctuations),
some
processes
reasonably
well
understood
(
for
example,
the
production
of
oxidant
substances
by
photochemical
reactions
between
volatile
organic
substances
and
oxides
of
nitrogen),
and
some
processes
only
poorly
understood
(
for
example,
the
production
of
organic
nitrate
aerosols
by
heterogeneous
reactions
between
volatile
organic
substances
and
oxides
of
nitrogen).
This
range
of
knowledge
about
the
processes
being
modeled,
and
the
fact
that
processes
subject
to
uncertainty
are
the
subject
of
active
research
worldwide,
means
that
parts
of
the
model
code
are
well­
established
enough
to
be
considered
fixed,
while
other
parts
of
the
code
are
in
constant
development.
The
result
of
this
is
that
there
must
exist
at
all
times
two
versions
of
the
model:
(
1)
a
currently
active,
reasonably
stable
version
of
CMAQ
that
is
used
in
an
operational
and
regulatory
mode,
and,
in
parallel,
(
2)
a
development
version
of
CMAQ
that
is
continually
being
improved
by
the
CMAQ
development
team.
From
time
to
time,
the
development
version
replaces
the
operational
version,
thus
bringing
recent
advances
in
atmospheric
science
to
the
operational
realm.
Having
a
model
exist
in
both
operational
and
research
realms
is
a
situation
not
uncommon
for
computer
codes.
However,
the
existence
of
these
two
versions
considerably
complicates
the
task
of
this
review
panel.
The
panel
concentrated
its
attention
on
the
operational
version
of
CMAQ,
but
where
appropriate,
has
referred
in
this
report
to
characteristics
of
the
development
version.
Clearly
the
development
version
is
closer
to
"
state
of
the
science"
than
is
the
operational
version,
and
the
differences
are
most
marked
in
areas
that
are
topics
of
current
research.

The
main
substance
of
this
report
is
thus
a
review
and
set
of
recommendations
designed
to
assess
the
current
state
of
CMAQ
and
to
guide
its
development
in
the
short
term.
These
statements
are
primarily
directed
at
the
operational
version
of
CMAQ.
As
requested
by
EPA,
the
panel
concentrated
on
five
areas
of
interest:

1.
What
is
the
overall
quality
of
the
scientific
research
in
the
CMAQ
Modeling
Program?

2.
What
are
the
strengths
and
weaknesses
of
the
science
being
used
within
the
components
of
the
CMAQ
Model
development
program?

3.
What
is
the
quality
and
relevance
of
the
research
applications
and
model
evaluations
being
conducted
as
part
of
the
CMAQ
Modeling
Program?

4.
What
are
your
perceptions
of
the
integration
across
the
elements
of
the
CMAQ
Modeling
Program
(
links
between
model
development,
applications,
evaluation)?
What
is
your
4
perception
of
the
usefulness
of
the
CMAQ
Modeling
Program
to
the
EPA
and
ORD
Agency
mission?

5.
Are
there
modeling
research
areas
that
are
not
being
addressed
or
are
given
insufficient
attention
within
the
CMAQ
Modeling
Program?
Are
there
current
areas
of
research
emphasis
that
might
be
given
lower
priority
or
eliminated?
For
the
resources
available
to
the
CMAQ
Modeling
Program,
are
they
being
used
in
an
effective
manner
in
terms
of
the
choice
and
quality
of
research
being
conducted?
5
Panel
Report
This
panel
report
is
presented
as
a
set
of
recommendations.
These
are
not
ranked
in
order
of
priority,
but
are
grouped
merely
for
convenience.

Overall
Recommendations
 
The
CMAQ
effort
should
remain
focused
on
its
main
mission.
From
the
perspective
of
its
major
clients,
this
is
urban/
regional
modeling
of
PM
and
ozone
for
State
Implementation
Plan
(
SIP)
purposes.
The
main
regulatory
purpose
of
CMAQ
and
similar
models
is
to
provide
an
understanding
of
how
future
changes
in
emissions
would
result
in
future
changes
in
concentrations
of
various
pollutants.

 
The
core
research
effort
of
the
CMAQ
modeling
program
should
focus
on
model
improvements
and
urban/
regional
applications.
The
research
effort
in
air
quality
modeling
at
a
fine
scale
and
up
to
a
global
scale
to
study
the
linkages
among
air
quality,
health
impacts,
and
global
change
is
important,
but
should
not
distract
from
the
main
effort
of
developing
CMAQ
as
an
urban­
to
regional­
scale
model.

 
We
believe
that
air
quality
forecasting
is
an
important
and
worthwhile
goal
that
could
be
accomplished
with
CMAQ,
supplemented
with
other
less
intensive
tools
such
as
Autoregression
Integrated
Moving
Average
(
ARIMA)
or
Artificial
Neural
Net
(
ANN)
statistical
models.
Since
long­
term
air
quality
forecasting
(>
1
month)
may
provide
potentially
important
information
for
SIPs,
EPA
should
coordinate
their
efforts
and
resources
effectively
with
the
lead
organizations
charged
with
achieving
this
capability,
such
as
the
National
Oceanic
and
Atmospheric
Administration
(
NOAA)
and
National
Center
for
Atmospheric
Research
(
NCAR).

Core
Model
Capability
and
Applicability
 
We
recommend
that
enhancing
the
chemical
and
dynamic
aspects
of
PM
modeling
in
CMAQ
become
a
top
priority.
We
note
that
the
primary
objective
of
the
Atmospheric
Modeling
Division's
(
AMD's)
CMAQ
model
research
program
is
to
develop
operational
air
quality
models
for
use
by
EPA
and
the
states
in
making
emission
management
decisions.
A
continuing
focus
of
scientific
improvement
should
be
on
CMAQ's
aerosol
treatments.
Much
effort
has
gone
into
improving
these
treatments
over
the
past
several
years,
but
there
remains
room
for
improvement.
It
is
also
important
to
note
that
while
CMAQ
is
a
leader
in
this
area
among
other
operational
air
quality
models,
all
such
models
lag
noticeably
behind
worldwide
research
efforts.
6
 
We
recommend
that
EPA
maintain
its
position
in
the
forefront
of
model
evaluation
for
the
types
of
applications
for
which
CMAQ
is
used.
Operational
evaluation
should
be
expanded
into
more
insightful
diagnostic/
mechanistic/
probabilistic
evaluation.
A
guideline/
protocol
for
PM
evaluation
should
be
published
by
EPA.

 
We
recommend
that
the
CMAQ
team
perform
an
in­
house
evaluation
of
emissions
modeling
by
use
of
techniques
such
as
"
inverse
modeling."

Fine­
Scale
and
Global­
Scale
Model
Applicability
 
We
recommend
that
EPA
and
others
undertake
an
investigation
of
the
range
of
scales
(
temporal
and
spatial)
over
which
the
model
can
legitimately
be
applied.
Results
of
this
investigation
should
be
publicized
to
all
users.

 
Recognizing
pressure
from
users
and
regulators
to
provide
a
nested
modeling
capability
down
to
the
scale
of
urban
canyons,
we
recommend
EPA
move
cautiously
in
this
logical
next
step,
given
the
significant
challenges
on
the
meteorological
side
of
developing
a
linkage
of
the
Pennsylvania
State
University/
NCAR
Mesoscale
Model
(
version
5)
(
MM5)
to
computational
fluid
dynamics
(
CFD)
and/
or
large
eddy
simulation
(
LES)
codes.

Air
Quality
Forecasting
 
We
recommend
that
EPA
resist
pressures
to
implement
nesting
within
the
Weather
Research
and
Forecasting
model
(
WRF),
as
we
believe
that
this
task
should
be
undertaken
by
NCAR.
We
note
also
that
WRF
will
have
to
incorporate
nudging
capability
for
it
to
be
an
appropriate
driver
for
CMAQ,
but
such
data
assimilation
capabilities
are
also
demanded
by
weather
forecasters.
Thus,
EPA
should
also
not
be
responsible
for
implementing
nudging,
but
should
work
to
ensure
that
this
capability
is
included
into
WRF
in
a
manner
well­
suited
to
both
weather
forecasters
and
air
quality
modelers.

 
We
recommend
that
EPA
coordinate
efforts
and
resources
effectively
with
other
organizations
(
e.
g.,
NOAA,
NCAR)
to
develop
a
version
of
CMAQ
for
real­
time
operational
forecasting
that
leads
to
development
of
a
nationwide
air
quality
forecasting
program.

CMAQ
User
and
Developer
Communities
 
Recent
key
staff
retirements
in
the
ORD
CMAQ
group
raise
concern
about
continuation
of
the
PM
modeling
effort.
Given
the
importance
of
PM
modeling
to
EPA's
mandate
in
the
PM2.5­
PM10
areas
and
given
the
complexity
of
aerosol
physics
and
chemistry,
we
recommend
that
the
modeling
group
work
actively
to
ensure
they
have
sufficient
research­
level
staff
to
support
needed
PM
modeling
efforts.

 
We
recommend
that
the
CMAQ
model
development
team
increase
its
number
of
postdoctoral
researchers,
as
these
personnel
will
be
most
likely
to
introduce
continuing
improvements
into
CMAQ.
7
 
We
recommend
that
EPA,
or
others,
develop
a
group
of
beta
testers
who
will
test
research
and
operational
(
pre­
released)
versions
of
CMAQ,
while
the
current
released
version
is
in
active
use.

 
We
recommend
that
EPA
widely
and
publicly
acknowledge
the
contributions
of
non­
EPA
developers
and
beta
testers
of
CMAQ.
This
should
include
those
providing
code
extensions,
modifications,
corrections,
and
bug
fixes.
A
mechanism
to
implement
bug
fixes
discovered
in
this
way
should
be
established.

 
We
recommend
that
EPA
staff
be
given
time
and
resources
to
build
collaborative
links
to
non­
EPA
researchers,
so
as
to
ensure
that
CMAQ
reflects
the
state
of
the
science
in
all
its
aspects.
This
could
be
achieved
through
focused
workshops
or
conferences
on
topics
central
to
the
development
of
CMAQ.

 
We
recommend
the
creation
of
a
mechanism
for
the
periodic
review
of
CMAQ
to
ensure
that
it
remains
state
of
the
science.
We
strongly
recommend
that
the
scientific
review
of
CMAQ
be
made
an
ongoing
activity
and
be
undertaken
about
every
two
years.

 
We
recommend
that
the
CMAQ
team
carefully
consider
efforts
underway
to
improve
MM5,
so
as
to
avoid
working
in
problem
areas
that
parallel
the
efforts
of
the
meteorological
community.
This
will
free
staff
to
work
on
problem
areas
that
are
"
CMAQspecific

 
We
support
EPA's
current
and
quite
successful
efforts
in
its
outreach
to
scientific
and
regulatory
communities
through
CMAS.
It
is
extremely
important
that
CMAS
have
sufficient
public
and
private
resources
to
offer
focused
training
classes
and
workshops
relating
to
various
components
of
CMAQ.
