►
From YouTube: 2023 Atmosphere Model WWG Day 1 1.31.2023
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
B
A
Think
we
cannot
bring
anything
in
here
other
than
water
over
here,
and
so
the
other
thing
just
quickly
for
people
who
are
here
and
did
not
joined
yesterday.
We
do
have
a
dinner
researcher,
basically
at
Southern,
Sun
tonight
and
and
then
I'm
actually
sure
what
it
was
quarter
to
six
or
five
thirty.
It
is
on
the
registration.
That's
the.
A
B
C
E
Is
you
can
have
your
coffee
and
your
cookies
and
you
just
need
to
go
down
the
hall
to
the
outer
dating
room.
E
A
Okay,
so
I'll
move
up
to
the
other
day
in
this
room
and
then
we'll
come
back
and
bring
some
other
session.
Then
okay
I
think
Let,
Me
Wait
like
just
a
couple
of
well.
You
can
start
experience,
presentation
and
get
started
because
it
always
takes
honor,
always
everything.
That
means
us.
A
Okay
yeah:
do
you
want
to
wait
a
minute
second
I
mean
officially,
we
started
in
eight
at
8,
45,
so
I
think
we
can
start
okay.
So
one
of
my
football
faults:
I'm
the
science
Community
Lane
liaison
industry,
kind
of
working
group
and
I'm,
basically
going
to
give
an
overview
summary
of
can
can
development
and
touch
on
a
couple
of
Diagnostics.
A
Heard
someone
don't
agree
yesterday
and
some
talks,
it's
developments
that
were
going
towards
and
I'll
talk
about,
evaluation
and
Diagnostics
just
a
couple
weeks
later
on
that
and
then
that
will
lead
into
a
discussion
of
user
projects
with
needs
going
forward
from
the
community.
So
on.
The
right
here
is
really
a
schematic.
That's
almost
yesterday
of
the
direction
that
we're
building
we're
going
towards
this
80
kilometer
of
high
top
or
a
course
model
with
93
layers
and
that's
wonderful
worlds.
F
A
A
One
degree
Global
model,
32
layer
and
vertical
levels.
This
is
a
schematic
on
the
left
of
the
vertical
variance
in
a
model
with
one
on
top
of
40
kilometers
and
it's
a
finite
volume,
dynamical
or
default,
and
it
uses
ram4
aerosols
and
we
have
the
opportunity
to
do
specified
Dynamics,
where
we
use
re-analysis
such
as
American
2
work
and
Rich
Britain
onto
the
32
line
available
groups
and
where
we're
going
is
towards
this
cam
Workforce
can
can
within
this
can
Workhorse.
So
we
have
a
58
layer
and
the
93
layer
and.
A
It
uses
it
will
use
mem5
aerosols.
There
will
be
new
upper
boundary
conditions
from
like
X
there'll,
be
Marine,
aerosol
marine
or
vinegar,
aerosol
implemented,
the
dust
module
will
have
emission
updates
and
tvx
will
be
included,
which
is
easier
to
add
more
tolerances
codes
online.
It
has
aerosol
interaction.
A
So
just
talking
about
some
of
the
different
dynamical
causes
we're
changing
to
the
default
from
that
finite
volume,
dynamical
core
to
their
spectral
element,
and
so
the
one
course
will
have
sec
slam,
which
is
a
spectral
element
dynamically
all
with
the
conservative,
stimulagrangian
multi-tracer
transport
and
currently
there's
no
reasonable,
replying
capability
in
that
version.
So
we
will
also
for
music
or
camcam
simulations
currently
need
to
keep
spectral
Elemental,
refined
setup.
A
Like
I,
said
the
finite
volume
we're
moving
away
from
climate
volume
and
towards
the
special
element
dynamical
core
there's
also
an
option
to
get
basically
the.
A
You
can
see
ignition
and
are
not
going,
and
then
we
kind
of
yesterday
also
about
empaths,
which
is
this
laurioni
grid
and
it's
as
Regional
employment
options.
But
it's
currently
in
experimental
phase
being
tested
in
chemistry
and.
E
A
A
And
so
the
responsibilities
of
can
can
into
Workhorse
model
because
we
really
are
going
to
provide
oxidants
to
can
simulations
currently
cgd
has
run
58
and
93
versions,
it's
specified
chemistry,
a
single
SOA
and
updated
metaphor.
We
made
the
prescribed
strain
spring
results
and
Simone
is
running
a
can
chem
on
58
layers,
including
a
topography,
bug
fix
that
was
recently
found,
and
so
the
responsibility
is
to
provide
a
scientifically
validated
configuration
of.
A
Conditions
we
have
for
regional
modeling,
so
there's
a
new
simulation
in
new
community
simulation
that
has
output
from
2001
to
2020
that
can
be
used
as
boundary
conditions
for
regional
simulations,
such
as
wolfchem,
and
so
you
can
request
some
sets
of
this
data
through
the
link
there
and
going
forward.
We'll
continue
those
situations
to
go
further
in
2020..
So
that's
just
I
wanted
everybody
to
be
aware
of
that.
But.
A
And
there's
some
updates
on
musical
we've
heard
a
couple
of
updates
yesterday,
so
we
do
have
a
Wiki
page
for
music
as
well,
and
there
are
three
files
that
are
important
for
people
to
know
about
on
that
page.
So,
there's
a
CNN
bug
of
15
Mega
divisions
and
two
bugs
with
spectral
unlimited
version,
and
you
can
read
more
about
those
on
the
musical
Wiki.
There
are
links
to
those
of
us
on
that
page,
but
it's
important
to
know
about
these
again
film
before
doing
any
simulations
with
musica.
A
We
also
heard
yesterday
about
implementing
geoscam
in
the
music
system,
integrating
model,
independent
chemistry
mechanism
into
models
and
as
being
included
in
the
unified
forecasting
system,
but
Noah,
and
then
we
also
heard
about
the
giant
aerosol
scheme
interface.
That's
been
developed.
A
So
that's
something
you
development,
total
and
now
I've
got
a
couple
of
slides
about
Diagnostics
and
evaluation.
Yesterday
we
had
a
talk
from
everything
about
the
atmospheric
dire
diagnostic
framework.
It's
been
developed
in
Ankara
to
evaluate
kind
of
climate
scale,
simulations
or
longer
scale
simulations
and
he's
also
working
with
us
in
their
ccwg
to
include
chemistry,
analysis,
and
this
is
all
updating
the
previous
Diagnostics
that
were
written
in
the
NCL
code
that
somebody
heaviest
So.
Currently
we
have
our
budget
tables
based
on
beaurus's
work
with
their
spectral
element
brand.
A
There
is
created
some
python
code
and
so
he's
been
working
with
Justin
to
get
those
working
in
ADF
and
Justin
also
has
some
thermal
aerosol
evaluation
and
that
he's
worked
on
David
Fillmore
is
going
to
provide
some
modus
and
Mara
climatology
of
aerosol
Optical
depth
and
I'm
going
to
provide
some
new,
updated
ontology
of
Margaret
column
monoxide
to
compare
with
the
Outlook
and
then
so
that's
what
we're
currently
working
on
and
then
the
next
steps
are
going
to
be
this
ozone
profile,
climatology
updates
and
including
chemistry,
internally
evaluation.
A
A
Written
in
python
as
I'm
watching
the
framework
to
compare
the
moral
results
and
observations
from
an
atmospheric
chemistry
viewpoint.
A
So
there's
a
link
here
to
the
GitHub
and
it's
a
online
to
effort,
Community
effort
of
a
lot
of
people
right
now
and
we're
really
willing
to
have
a
lot
more
people
involved
in
doing
the
developing
the
code
for
the
comparisons
and
on
the
right.
Would
we
have
an
example.
B
A
Ground
stations
of
air
now
compared
against
Wolf
Care
in
terms
of
ozone,
we
also
have
statistical
analysis
as
well.
In
that.
A
Okay
and
then
finally,
we're
moving
on
to
the
discussion
slide.
So
on
the
wiki,
the
chemcam
wiki,
we
have
a
new
page.
That
is
all
about
the
movement
comparisons.
We
have
some
new
model
into
comparisons
coming
up
there
and
things
that
we're
currently
thinking
about,
but
we
really
want
to
get
some
input
from
the
rest
of
the
community
about
what
we
might
have
missed.
Who
is
actually
involved
in
these
into
comparisons?
A
Can
we
love
some
more
information
to
collect
more
information
to
understand
where
people
are
involved
right
now
and
then,
in
terms
of
model
development,
we'd
love
to
discuss
about
what
configurations
the
community
really
requires
like,
for
example,
the
spectral
element
region?
One
client
is
there
a
need
Simone
mentioned
yesterday?
We
really
might
want
to
continue
using
a
32
layer
grid
rental
group
on
your
model.
That
might
be
something
new,
so
I
think
I
will
just
open
it
up
to
discussion.
Yeah.
A
Okay,
yeah,
maybe
we
keep
the
slides
and
when
we
do
this
number
16
minute
discussion,
so
I
think
we
first
go
and
see.
If
there
are
any
questions
in
the
audience
and
online
and
here
and
then
we
can
go
and
and
really
really
talk
about
what
model
configurations
do
we
hear
once
what
do
we
need
to
support
going
forward?
I
A
J
A
On
really
ncesm
to
improve
the
code
to
pull
out
the
aerosol
stuff
that
is
all
over
the
code
and
put
it
in
some
comedy
area.
So
if
anything
code
doesn't
need
to
necessarily
know
what
it
also
called
to.
F
A
You
have
a
place
where
you
can
then
change
aerosol
models
and
it
also
models
in
the
specific.
So
it
will
be
easier
to
actually
Implement
your
layouts
and
speeds.
That's
the
plan,
and
you
know
the
giant
project
and
Alma
I,
don't
know
it's
Alma
is
online,
it's
not
so.
This
is
a
much
bigger
project
like
really
involves
many
groups,
and
people
really
want
to
develop.
Basically.
A
Routines
that
can
be
used
for
various
different
models
that
you
basically
plug
in
in
our
different
schemes,
and
you
know,
our
capabilities
of
making
music
are
flexible,
will
obviously
benefit
from
it,
because
if
other
people
have
routines,
you
can
use
those,
but
that's
a
huge
project
and
that's
a
vision,
paper
I
think
they're
working
on
us,
I
haven't
actually
seen
it
yet,
but
maybe
Louisa
do
you
want
to
add
something.
A
Yeah
so
yeah,
that's
I
think
we
should
actually
have
a
presentation
of
that
at
one
point,
maybe
in
summer,
if
that's
of
Interest
yeah
any
other
questions
and
that's
yeah,
I-
think
the
discussion
on
the
configurations
so
yeah
we
always
have
the
problem
that
cesm,
you
know,
can
the
whole
atmosphere
is
moving
forward
with
developments.
This
is
the
workforce
model
and
they
decide.
It
has
been
decided
that
spectral
element
is
the
dynamical
core
and
we
have
the
high
levels
in
physics
and
everything
will
be
developed
based
on
this.
A
So
usually
chemistry
will
take
that
configuration,
and
then
we
run
with
chemistry,
even
though
we
as
the
chemistry
modelers,
will
then
provide
chemistry
back
to
that
model.
That
doesn't
really
have
the
comprehensive
chemistry
and
describe
oxidants.
So
we
need
to
work.
We
are
working
closely
together
with
them.
The
problem
is,
though,
the
model
is
expensive,
especially.
M
A
Chemistry
and
spectral
element,
it
says,
I
think
it
really
is
not
just
scaling
it.
It
doesn't,
it
seems
to
be
more
expensive
than
with
final
volume
is
what
we
are
experiencing
now
and
if
you
add
chemistry
and
extra
layers,
it
may
be
really
difficult
to
test
things
in
a
configuration
well
running
on
one
degree
and
we
are
running
93
levels.
So
two
in
10
years
of
simulations
to
test
new
schemes
is
very
difficult
and
then
there
were
like.
A
A
A
B
K
K
Yeah
one
thing:
it's
just
it's
expensive
with
school
chemistry
or
their
own
sorts,
yeah.
K
A
Yeah,
the
question
is,
first
of
all
and
I
think
we
have
an
answer.
That
question
yet,
which
will
hopefully
be
answered
in
summer,
is
the
performance
of
the
58
level,
the
same
as
the
performance
of
the
32
level.
I
think
that
will
have
changed
in
the
stratosphere
and
that
we
have
mentioned
that
before
that.
Maybe
we're
not
resolving
the
strategy
as
well
in
the
58
2013
level
model.
H
B
A
Yeah,
so
we
took
the
a
model
that
would
have
a
Stratosphere.
How
much
does
it
save
us
in
computer
class.
K
K
K
We
have
been
a
slate
launching
services,
Instagram.
N
Money
so
then,
maybe
it's
not
expensive
anymore.
A
O
Yes,
actually
it's
it's
like
all.
The
developments
are
in
the
last
few
years
have
been
always
increasing
the
total
number
of
reactions
and
processes
and
to
provide
more
details
and
then,
in
some
cases,
it's
more
important
like
turning
on
or
off
coupling
of
chemical
processes
than
having
them
resolve
with
the
larger
spatial
or
vertical
resolution,
and
and
whenever
you
have
to
make
like
multiple
sensitivities,
to
run
long
simulations,
then
using
a
chip
model
is
always
faster
and
more
easy
to
to
compare
than
the
the
overall
impacts
of
this
new
chemistry.
O
One
is
trying
to
develop.
So
that's
why
that
was
my
the
main
reason
to
to
try
to
think
of
keeping
using
these
32
level
model.
With
a
with
a
wider
resolution.
I
know,
most
of
the
developments
in
income
are
going
in
the
opposite
direction,
going
to
increasing
the
resolution
that
then
for
climate
applications
might
not
be
that
useful
for
initial
development.
That's
what
I
mean.
A
F
A
Yeah
I
think
from
the
software
engineering
standpoint,
It's
probably
hard
to
convince
people
to.
F
A
E
G
P
I
I,
don't
I
mean
as
always
it's
common
right.
It
depends
on
the
the
application
and
what
local,
like
I,
think
you
hear
about
people
running
locally
as
opposed
to
on
the
that
and
car
systems
and
yeah,
so
I
I'm,
not
sure,
but
I
guess
one
of
the
things
that
I
was
thinking
about
is
like,
if
you're
doing
more
of
the
climate
side
of
things
I
mean.
Maybe
two
degree
is
okay
for
some
applications.
But
you
know
if
you're
sort
of
trying
to
think
about
chemistry
on
climate
I
agree
for
air
quality.
P
I
think
you
know
we're
probably
gonna
want
the
more
expensive
higher
resolution
type
simulations,
yeah
I,
don't
have
a
real
answer.
I
just
just
thinking
about
it.
A
Yeah
and
I
wanted
to
I
mean
that's
yeah,
that's
something
we
will
keep
in
mind.
I
think
we
have
one
more
minutes.
One
thing
I
wanted
to
point
out
on
a
different
topic.
We
do
have
readings
music
or
Trucking
meetings
at
Acom
and
for
those
people.
Many
people
already
joined
it
that
are
here
in
the
Scorpion
group
meeting.
A
A
Have
a
whole
bunch
of
developments
that
we
are
pushing
forward
so
when
we're
starting
to
use
the
new
development
versions,
we'll
get
changes
to
the
model
and
they
are
currently
not
scientifically
supported-
just
want
to
make
people
aware
of
this
that
the
model
is
right
now
very
much
in
development,
and
things
may
change
when
you,
when
you
do
runs
similar
nations
with
your
way
out.
For
that,
and
the
second
thing
is
we
moving
over
to
the
ratio
in
a
year
at
least
again,
probably
your
model
ones
that
you
have
done,
you
have
to
be.
A
K
K
E
C
A
Any
more
current
question
or
thoughts
if
this
is
a
short
discussion,
but
I
mean
we
can
always
continue
the
discussions
later,
but
this
is
certainly
to
catch
up
in.
Hopefully
we
have
way
more
results
and
other
things
in
summer
to
show
the
new
world
versions
should
move
on
and
get
ready
to.
The
next
talk
then,.
O
A
It
was
fine
with
the
video
but
yeah.
We
see
it
just
fine,
so.
B
A
So
yeah,
we'll
start
with
our
talks
now
in
the
first
set
of
talks,
will
be
halogens
and
very
short
listed.
Also
so
please
start.
D
Yeah,
okay,
thank
you.
Thank
you
very
much.
First
of
all,
thanks
to
to
Rafa
and
Simone
to
to
organize
the
session
and
managing
nearly
today
might
talk
is
about
the
role
of
iodine
recycling
on
ciso
aerosols
in
the
Global
Marine
boundary
layer.
D
As
you
can
see,
it's
a
it's
a
it's
a
teamwork
involving
several
scientists
from
from
several
institutions
from
several
countries
and
if
you're
interested
in
in
the
top,
we
actually
accomplished
most
of
the
findings
in
on
GRL
and
yeah,
you
can
have
a
look,
is
the
dividing
into
four
parts,
including
first
some
background
information
on
the
top
spheric,
reactive,
outing
and
then
the
second
one
is
the
the
how
how
we
did
this
simulation,
simulator
design
and
the
third
one
is
the
simulation
results
in
the
Marine
boundary.
D
Okay,
here's
a
rather
complicated
concept
figure
that
I
kind
of
copied
from
Alfonso's
review
about
10
years
ago,
and
it's
still
very
informative
up
to
now,
and
but
basically,
what
I
want
to
show
here
is
that
out
in
chemistry
and
now
in
Associates
are
mostly
from
oceans,
original
regions
and
as
well
as
the
the
polar
regions
and
as
to
turn
down
the
chemistry,
for
it
involves
both
gas
phase
chemistry
and
also
the
air
surface.
D
And
here
we
highlight
the
species
that
I'm
going
to
talk
about
today's
hoi,
both
in
in
the
gas
phrase
and
also
in
the
house
phrase.
So
that's
why
it's
very
critical
species
that
involved
in
many
hydrogen
process.
For
instance,
for
instance,
it
can
be
uptake
on
the
aerosols
if
the
Earth
contains
like
a
broma
in
Corolla,
it
can
activate
this
bromine
chlorine
to
form
IBI
and
Ico
and
which
can
further
fertilize
to
to
release
all
of
the
hydrogen
atoms
like
bromine
chlorine
and
iodine.
So
at
the
at
the
bottom.
D
D
Both
has
proposed
that
theoretically,
the
HRI
could
be
uptake
by
the
SEC
so
also
and
then
to
to
release
this
IBI
and
Ico
molecules
that
without
let's
say,
direct
view
evidence,
then
let's
look
at
the
bottom
left
so,
but
then
in
the
in
the
next
about
10
20
years,
without
directly
with
evidence,
we
we
still
assume
that
depression
could
happen.
So
we
added
that
in
in
the
model
and
then
with
extreme
efficiency,
about
0.01
to
0.06
the
uptake
coefficiency
and
that
about
two
years
ago.
D
Let's
look
at
the
upright
about
two
years
ago
there
has
been
a
paper
reporting,
the
the
direct
view,
evidence
at
one
Coastal
site
in
North
Atlantic.
However,
that
field
study
suggested
that
reaction
is
much
much
faster
about
0.3
update,
coefficient,
much
much
faster
than
previously
assumed.
D
Then
the
question
that
we
asked
our
service:
that
okay,
that's
just
one
side
or
from
one
study,
but
what
about
the
Global
Marine
bundle,
but
it's
it's
it's
Global
phenomenon.
What's
the
global
significance,
whether
you
have?
How
do
you
index
or
not
staying
capacity?
Okay,
then,
that
to
answer
those
questions,
we
need
Global
models
and
we
also
negro
model
type
with
already
with
a
comprehensive
halogen,
chemistry
and
and
Alfonso
school.
That
has
been
developing
hydrogen
chemistry
and
sources
and
added
them
to
can
assess
them
in
the
past
decade.
D
So
I
I
use
this
model
as
a
base
case
than
to
just
through.
Okay.
Here
comes
the
the
simulation
design
just
to
I,
put
it
to
tune
the
optic
coefficient
of
One
processor
and
then
to
see
how
the
the
whole
system
responds
to
it.
D
Okay,
so
here
comes
us
emergency,
so
one
thing
I
didn't
mention
here
is
that
we
actually
we
are
still
using
session,
one
so
I
think
that's
chem,
4
or
chem
five
I
think
rather
not
small
part,
and
we
we
use
specific
Dynamic,
and
we
did
it
for
two
years.
We
simulated
for
two
years
and
then
we
we
discussed
the
first
12
months
as
a
swing
up
and
then
the
resolution
is
and
then
vertically
we
read
it
for
56
layers
and
then
36
30
minute
time
step.
D
We
use
previous
simulation
as
initial
and
videos.
Five
symmetology
condition
and
then
we
like
I,
said
we
here
we
industry,
we
focus
on
the
Marine
boundary
layer,
so
it
was
about
you
know
the
below
900,
100,
pasca
and
the
table
here
shows
the
chemical
simulation
design.
We
have
four
cases
base
case
conventional
so
for
the
base
case.
We
we
allow
the
foi
to
be
uptake
by
the
aerosol
with
a
0106
as
a
coefficient,
but
we
don't
allow
it
to
to
come
back.
D
D
Gamma
0x3
has
the
optical
efficient
and
then
the
same
view
for
IBI
and
ICL,
and
then
we
kind
of
like
oh
okay.
What
will
happen
if
that
we,
the
we
kind
of
go
up
the
to
the
upper
limit
of
the
optical
field?
To
use
0.9?
That's
recall
upper
limit.
So
basically
the
whole
study
is
based
on
one
reaction
here:
Hy
uptake
Pi.
This
is
also
with
bromine
chloride
and
then
to
release
0.5,
ibr
and
0.5
Ico.
Okay,
then,
here
comes
the
simulation
results
in
the
Marine
value
layer.
D
So
the
figure
the
left
shows
the
key
iodine
species
in
the
morning
layer
and
then
we
we
can
see
the
all
of
the
key
iodine
species
at
highest
levels
in
the
in
the
tropics
and
and
also
in
the
sub
shop
regions,
particularly
along
the
coastlines
and
then
hoi
is
the
is
the
dominant
species
which
is
kind
of
like
goes
back
to
to
the
to
the
to.
D
The
purpose
of
the
study
is
that,
okay,
this
is
the
one
of
the
most
important
iodine
species
in
in
the
tropics
tree,
and
then
we
don't
know
much
about
its
fate
in
the
Marine
boundary
lay.
So
we
kind
of
do
this
study
to
kind
of
revisit
the
role,
and
then
the
left
is
ibr
and
Ico,
okay
and
then
on
the
right.
D
We
show
the
iodine
speculation
as
well
as
the
vertical
Profiles
In
from
four
cases,
and
as
we
can
see
that
the
iodine
to
Total
iodine
abundance
decrease
significantly
in
the
lower
one
kilometer,
which
is
the
the
Marine
boundary
layer,
almost
all
of
them,
and
then
it
remained
kind
of
constantly
and
in
the
free
shop
sphere
and
the
difference
between
different
case.
D
Next
year
we
we
show
the
simplify
out
in
chemistry
there,
and
then
we
we
calculate
the
average
concentration
on
missing
ratios
of
these
species,
and
here
the
key
result
here
we
want
to
show
is
that
the
the
form,
the
ibr
level
and
Ico
level,
from
the
from
the
base
case
with
0.01
pptv,
could
go
up
to
like
0.2
or
even
0.3
pptv
with
it's
very
recycling.
D
So
with
this,
among
this
significant
amount
of
precursors
of
the
halogen
atoms,
we
we
kind
of
want
to
know
What
would
happen
to
the
to
the
halogen
atoms.
D
Maybe
maybe
have
those
ibr
and
Ico
formula
number
and
Boundary
layer-
and
here
we
show
the
three
halogen
atom
protection
rate
left
is
outing
and
in
the
middle
is
brought
me
and
on.
The
right
is
the
is
the
it's
the
problem?
Okay,
so
thoughtful
audience
with
the
base
case,
took
and
and
goes
up
to
about,
50,
larger
or
even
even
more,
with
the
when
we
have
one
process
activated
and
for
burning,
it
could
be
even
higher
than
100
larger.
D
So
so,
when
we
have
the
stronger
or
like
slightly
larger
production
rate
of
hydrogen
atoms,
naturally
we
would
the
next
question:
we've
asked:
okay,
we've
maybe
had
more
atoms
in
in
the
system.
What
would
you
mean
to
the
globalization
capacity?
Okay,
so
here
we
show
the
the
similar
levels
of
ozone,
orange
and
electro2
on
the
left.
That's
on
the
left
is
the
simulated
levels
and
then
on
the
right.
D
We
show
the
relative
changes
due
to
this
hoi,
the
Audi
recycling
process,
and-
and
we
can
see
that
the
the
changes
could
be
about
10,
10
percent
and
and
even
even
even
larger,
in
in
some
some
regions
for
ozone
and
oh
and
Azure
too,
and
now
we
we
can
see
that
that's
significant
result
that
we
should
visit
with
with
other
models.
D
Okay,
then,
the
last
one
is
a
very
brief
summary.
D
So,
basically
I
don't
want
to
repeat
the
results.
I'm
gonna
highlighted
some
of
the
implications
that
we
we
found
from
during
this.
It's
kind
of
modern
studies
that,
with
the
development
of
the
measurement
techniques
that
we
we
tend
to
measure
or
or
detect,
more
species
or
or,
let's
say,
lower
levels
of
the
the
the
reactive
species
in
in
the
real
atmosphere.
D
So
when,
whenever
we
have
those
direct
observations
of
this
species,
we
kind
of
as
models
we
kind
of
we
like
to
to
like
revisit
the
things
that
we
thought
was
correct
and
then
we
kind
of
like
we
will
need
to
to
do
discount
simulations
to
to
try
to
be
reconciled
with
the
with
the
observations.
Yeah,
okay,
I,
think
that
will
be
all
thanks.
E
A
H
Nice
talk,
it
seems
like
we,
you
really
do
need
some
observations
for
this
question
to
be
answered
at
some
point.
Is
there
anything
in
the
works
to
measure
hoi
foreign.
L
There
is
one
experiment
going
on
currently
in
Bermuda:
it's
called
boundary
layer,
experiment
on
atmosphere,
chemistry,
the
it's
led
by
Professor
Becky
Alexander
at
the
University
of
Washington
and
she's,
collaborating
with
a
couple
different
institutions
but
I
believe
York
University
is
using
a
Sims
Mass
Spec
to
measure
hoi,
so
stay
tuned.
For
that.
A
Because
you
can
start
sharing,
doesn't
seem
to
be
any
more
questions,
so
the
next
speaker
is
Bruce
understanding
the
distribution
of
clvs
as
and
they
are
interested
transport
based
on
global
model.
Simulations
and
measurement
data.
F
F
N
And
in
this
talk,
I
just
shown
some
of
our
measurement
and
modeling
results
for
the
color
limited,
very
short-lived
substances.
N
So
chimney
talked
about
the
iodine
ones
and
I
would
like
about
the
color
in
the
Box
and
I
specifically
talk
about
the
measurements
that
we
made
during
the
NASA.
N
A
commission
and
I
think
you
all
are
aware
of
atom
but,
as
a
summary
later
was
a
global
Escape
mission
that
had
four
deployments
for
in
in
different
citizens
and
it's
a
unique
data
set
that
can
let
us
investigate
the
global
scales
in
the
remote
regions
for
different
species,
including
bsls
and
I,
will
talk
about
to
a
specific
of
these
species,
dicoloromethane
or
ch
ucl2
and
tetrachloroethane
or
c2cl4,
which
are
majorly
emission
majorly
anthropogenic
sources
and
on
this
map,
I'm
showing
the
distribution
of
dicolor
as
a
example
from
a
recently
developed,
French
Elementary
and
which
shows
very
high
emissions
over
the
Eastern
Asia
and
just
as
a
note
in
in
most
of
my
plots,
I'm
I'm
splitting
the
class
into
different
latitude
beams
and
here
I'm,
showing
different
latitude
means.
N
So
we
have
five
beans
extending
from
the
Arctic
down
to
the
Southern
Ocean,
okay,
so
here
I'm
showing
a
vertical
performance
for
our
diet,
cholero
at
the
best
pollen
and
tetrachlorole
on
the
right
column
in
different
magnitude
beams
as
I
mentioned,
and
these
are
all
the
measurement
data
during
all
four
data
deployments,
and
there
are
many
information
here,
but
I
I
would
like
to
point
out
some
of
them.
So
one
difference
we
are
seeing
between
these
two
columns
is
the
the
values.
N
So
we
are
seeing
very
high
values
up
to
100
or
more
PPT
for
dicoloral
body
where
tetracolor
values
are
very
lips,
but
one
interesting
thing
is:
is
the
inter
hemispheric
differences?
N
So
if
you
look
at
the
Northern
Hemisphere
clouds
we're
seeing
concentrations
are
about
maybe
three
times
higher
than
the
southern
hemisphere
concentrations,
and
also
we
are
seeing
that
the
shape
of
the
vertical
profile
so
in
the
northern
hemisphere,
it's
declining,
the
radiance
or
the
concentrations
are
decreasing
when
we
go
to
the
higher
altitude
units,
but
in
the
southern
hemisphere
we
see
the
inverse
of
that
gradient,
which
is
suggesting
that
the
observed
concentrations
within
the
southern
hemisphere
are
are
because
of
the
interheimer's
transport.
N
So
we
we
did
a
we.
We
tried
to
use
the
updated
halogens
mechanism
in
the
CSA
model
and
I'm
using
the
similar
model,
as
Chini
show,
Mostly,
similar
in
terms
of
mechanism
and
Dynamics,
and
the
blue
lines
are
showing
the
modeling
results
and
we
are
seeing
that
the
model
basically
captured
both
the
vertical
profile,
the
shape
of
the
vertical
profiles
and
where
in
Johannesburg
different
service.
N
But
the
model
is
basically
biased,
low
in
terms
of
Maintenance,
although
there
are
some
overlaps
in
terms
of
the
spreads,
but
the
model
is
quite
slow
and
as
I
mentioned,
these
are
all
the
data
during
all
four
atom
deployments.
But
we
want
to.
We
also
looked
at
the.
E
N
Performance
of
the
model
to
capture
the
seasonality.
So
here,
as
a
reference
on
the
very
bottom
row
I'm
showing
the
flight
profiles,
each
dot
is
showing
the
the
measured
data
during
each
atom
deployment
and
then
on
top
on
each
couple
panel.
The
the
top
panel
is
showing
the
zonal
vertical
profile
in
the
model,
and
the
bottom
is
showing
the
zonal
vertical
profile
in
the
in
the
atom
measurements
during
different
deployments.
N
So
and-
and
this
is
the
results
for
Tetra,
color
and
and
again,
there
are
too
many
information
here,
but
I.
What
in
general
it
shows
it
is
that
the
model
is
capable
of
capturing
the
general
seasonality
between
different
deployments,
but
and
and
and
and
overall,
the
model
is,
is
showing
at
good
results.
N
The
correlation
is
more
than
18
percent,
but
we're
seeing
some
missing
features
in
the
model
that,
for
example,
we're
seeing
that
during,
for
example,
The
Argus,
the
the
modern
values
are
bios,
no
and
also
in
in
the
desert
for
tetrachloro.
We
are
selling
differences,
so
we
think
that
chemistry
plays
an
important
role
in
the
in
the
biases
of
the
model,
and
so
we
look
at
the
removal.
Pathways
of
these
species
and
in
the
model
there
are
three
Pathways
so
dicolor
and
both
of
these
species.
N
They
can
either
fertilized
or
react
with
oh
or
CL
atoms
in
the
model,
and
we
looked
at
the
reaction
rates
within
these
different
Pathways
again,
so
this
column
is
showing
the
reaction
rates
for
with
different
Pathways
in
different
colors
for
dichloral
and
here
I'm,
showing
the
same
results
for
tetracolor
on
the
right.
And
so
what
we
are
seeing
for
dicolor
is
that
the
orange
pathway
is
the
dominant
removal
mechanism
for
Diane
collural.
N
And
if
you
look
at
the
the
scale
of
the
x-axis,
it's
it's
logarithmic.
So
it's
orders
of
magnitude
type
and
we
see
for
tetrachloride
that
there
is
basically
a
competition
between
the
past
the
the
reaction
with
coloring
and
reaction
with,
oh
and
for
both
of
these
species.
N
The
the
blue
line
is
the
photolysis
and
we
say
that
it's
it's
not
basically
a
very
important
password
for
removable
of
things
components,
but
so
for
tetrachloral,
as
I
mentioned,
we
are
seeing
a
competition
between
oh
and
cl
and
its
effects
like
the
the
local
lifetime.
So
here
I'm
showing
The
Book
of
lifetimes
of
of
dicholor
on
the
left
and
but
for
tetrachalurov.
N
We
calculated
the
local
Lifetime
with
those
reactions
that
I
showed
with
including
the
CL
pathway
on
the
middle
and
excluding
the
CLS
way
on
the
right
panel,
and
we
are
seeing
that
the
cl
is
very
important
and
majorly
affects
what
we
are
seeing
about:
the
bunker
Lifetime
and
and
remove
one
of
petrol.
N
So
one
question
is
that:
do
we
have
that
much
CL
in
in
the
High
Altitude
in
the
models?
And
so
in
the
model?
We
are
seeing
very
high
concentrations
of
Cl
atoms
in
the
cell
atoms
in
the
mid
latitudes
of
metal,
latitudes,
orbital,
troposphere
and
so
but
I
I
do
not
have
the
reaction
rates
of
different
reactions
that
produce
CL
atoms
in
the
model,
at
least
in
in
current
version,
but
based
on
the
different
plots.
N
N
But
we
do
not
have
any
serial
atom
measurements
during
atom,
but
we
have
some
chlorine
molecule
measurements
during
a
term
by
by
Noah
Sims
instrument
and
I
looked
at
them.
N
We
had
that
for
hr3
and
804
and
again
I'm
showing
the
zonal
profiles
here
at
the
model
on
top
the
measurements
on
the
bottom,
and
we
see
that
so
the
model,
as
we
were
expecting
has
high
CO2
concentrations
in
these
regions
and
and
also
the
there
are
many
concentrations
in
the
lower
low
turquoise
sphere
in
the
tropics
that
the
model
that
the
model
is
not
capturing
that
so
this
is
I
think
this
is
an
interesting
finding
that
we
should
work
on
and,
as
a
summary
I
just
showed,
that
the
chlorinated
dsls
have
a
very
large
variability
in
in
terms
of
distribution
and
the
model
is
capable
of
capturing
the
trends.
N
But
this
bias
low
and
we
think
that
the
high
oh
in
the
model
can
be
one
reason
for
the
bias
of
dichola.
As
as
we
said,
there
always
was
the
main
Domino,
and
also
we
saw
too
much
CL
atoms
in
high
high
altitudes
in
the
mid
latitudes,
which
I
think
it
suggests
that
the
inorganic
chlorines
in
in
the
model
should
be
investigated
in
the
in
more
detect.
N
I,
haven't
looked
at
the
for
sale
to
work,
for,
for
the
bsms
we
are
seeing.
Differences
mostly
Pacific
is
is
higher
than
athletic.
It's
because
most
of
the
emissions
are
over
the
Eastern
Asia.
So.
C
C
O
Fun,
no,
there
is
no
talk
about
Sesame
two
vsl's
this
meeting,
but
we
had
some
progress
on,
hopefully
we'll
present
laws
and
everything
will
be
done
for
the
next
meeting
in
the
summer.
O
Yep,
we're
working
on
that
so
can
I
have
a
question
for
this.
Yes,
when-
and
these
sonar
averages
you
are
showing
here
is
this,
for
the
model
is
24-hour
means,
so
this
date
timing.
O
Because
I
I'm
wondering
if
these
chlorine
production
is
declaring
atoms
that
are
too
large,
there
might
be
a
contribution
for
brcl,
because
that
process
was
quite
efficient
in
the
night
because
of
some
recycling
reactions
and
then,
probably,
when
the
day
reached,
it
releases
a
lot
of
chlorine
and
also
bromine
for
sure.
O
So
that
was
just
one
comment
on
on
the
processes
might
be
controlling
the
current
production
and
the
other
one
is.
There
is
a
paper
from
measurements
of
the
hollow
aircraft
from
the
German
Consortium,
which
also
shows
some
nice
hemispheric
differences
on
the
vertical
profiles
of
vibromomethane,
which
might
be
really
interesting
to
compare
with
these
ones.
You
show
for
c8202.
N
K
N
Thank
you
and
thanks
for
the
the
VR
serial
comment,
I
I
think
I
have
saved
the
hourly
data
for
for
some
of
the
species.
I
have
to
look
at
that.
Let's
see
if
we
see
anything
with
the
Knight.
A
Okay,
I
think
we
should
move
forward.
Can
you
stop
sharing
and
the
next
speaker
will
be
Javier
I,
don't
know
what
to
say
the
name
right.
You.
Q
J
K
K
Q
Okay
in
this
presentation
that
will
compare
the
effect
of
natural
allergens
on
propofit
is
awesome
between
pre-industan
and
present
date
and,
in
this
case,
I
use
the
the
camcam
model
and
a
brief
introduction.
Oh,
and
what
are
the
natural
source
of
allergens
and
the
mission
of
allocarbon
very
short
lived,
which
have
a
photochemical
last
times
smaller
than
six
months.
Q
Okay
is
one
of
the
main
source
of
allergen
in
the
in
the
troposphere
they
are
relays
from
the
ocean
via
metabolize
of
marine
organize
such
as
a
phyto,
platoon
and
algae,
and
the
actual
Camp
setup
comes
here.
The
emission
inventory
of
Adonis
by
nine
allocarbons
based
on
clarified
maps
and
in
this
inventory
they
promo
form
and
the
bromometer
are
the
most
abundant
and
avocados
for
iodine
experimental
research.
Q
In
come,
this
route
represent
represents
up
to
75
percent
on
yielding
atmospheric,
but
this
depends
of
the
iodine
flux
on
ozone
is
scheduled
to
to
our
work
and
up
in.
In
fact,
Prados
Romano
indicate
that
increases
in
Ozone
size
bi
for
industrian
amplifies
emissions
of
iodine
in
present
day
via
feedback
and
negative
awesome
iodine.
Q
Q
Q
Okay
in
this
world,
in
this
world,
we
run
CCTV
in
more
specific
Dynamic
for
both
period,
whose
atmospherical
environment
are
simulate
from
the
emissions
corresponding
to
eh
period
and
to
evaluate
around
the
natural
hydrogen
full
and
to
discriminate
their
individual
contribution.
We
prepare
scenario
whose
main
difference
is
related
to
source
of
unlocking
emissions,
foreign.
Q
Q
Change
in
also
in
drive
and
increase
in
the
patching
of
in
in
organic
yielding
be
a
feedback
also
in
urine
more
change
in
Hooks
and
knocks
Drive.
The
change
in
inorganic
iodine
partitioning
increases
the
conversation
for
the
iron
interactive
into
reservoirs,
mainly
in
the
lower
troposphere.
Q
For
the
clothing
and
similar
to
the
urine,
there
is
a
strong
partitioning
into
reservoir
mailing.
The
lower
troposphere
and
the
increases
in
the
solo
Y
is
due
to
that
is
that
the
32
tropospheric
transport
of
air
masses,
with
Rich
in
chlorine,
inorganic
and
enhancement,
the
illumination
on
sea
salt
treatment
by
halogens
for
production
from
for
production
of
chlorine
from
alkate
or
the
nitrogen
on
sea
salt
dry
by
increases
in
the
knocks
near
surface.
Q
Q
In
terms
of
change
on
global
topographic
awesome,
the
natural
hydrogen
induced
a
large
reduction
in
brain
just
in
the
present
day.
Its
effective
is
mainly
governed
by
iodine
and
then
blooming
and
individually.
The
yawning
has
a
role
equal
in
both
period
and
the
bromine
has
a
large
role
in
pi
and
the
opposite.
The
chlorine
is
has
a
role
and
in
present
day,
that's
all
more
impressive.
Q
Q
Q
Yeah
I
suppose
result
of
uptake
of
nitrogen
species.
For
this,
the
the
chlorine
production
is
intensifies
in
the
transition
from
PI
to
to.
Q
In
terms
of
change
in
awesome,
vertical
distributions
and
the
reduction
drive
by
hiring
decreases
from
the
surface
to
three
three
propofia
in
the
transition
frame
paid
to
PD,
mainly,
do
it
the
enhancement
conversion
of
reactivity,
to
pressure
Boris
and
from
free
troposphere
to
Upper
troposphere.
The
role
of
iron
is
prevented
due
to
higher
availability
of
reactive,
yielding
in
present
day
compared
to
Pi.
K
F
A
H
Know
in
your
very
nice
talk
if
your
presentation,
if
you
looked
at
vsli,
vslbr
and
and
via
cell
chlorine-
and
you
sum
those
up
say
for
pi
and
then
you
looked
at
the
DSL
all,
are
they
about
the
same
numbers
I?
Is
there
any
like
non-linear
chemistry
going
on
by
having
all
the
all
the
components
modeled
together.
H
You,
if
you
go
to
your
bar
charts,
just
show
one
example:
your
bar
charts,
yes,.
H
Yeah
yeah,
so
if,
if
I
look
at
the
sum
of
in
the
top
panel
there,
if
you
just
look
at
the
sum
of
p,
I
for
vsli
vsl,
Bromine,
vsl
chlorine,
yes,
and
then
you
compare
that
to
DSL
all
which
is
everything
running
together.
H
Do
they
equal
meaning
that
if
you
sum
every
individual
component
of
the
change
in
Ozone
burden,
does
it
equal
the
total,
when
you
run
them
all
three
together
or
is
there
some
sort
of
cross
Channel
reactions
that
increase
the
overall
loss?
Does
that
make
sense.
Q
H
A
And
then
we
have
we
won.
Who
is
online
and
I
hope
you
can
try
sharing
your
screen.
R
A
And
so
we
talk.
A
S
Okay,
good
morning,
everyone,
it's
my
honor-
to
have
this
chance
to
present
my
work
here
from
UCI
I'm,
a
PhD
student
working
with
Professor
Alex
cancer,
and
so
my
talk
today
will
talk
about
like
how
could
we
model
ISO
Pro
emission,
responds
to
drought
and
Heatwave?
You
make
a
model
and
first
oops,
okay,
first
I
will
give
you
guys
some
background
today.
I
will
talk
about
isoprene,
so
isoprene
is
a
major
organic.
S
We
will
see
coming
from
terrestrial
ecosystem,
it's
a
precursor
of
ozone
and
aerosol,
and
they
can
also
effect
math
things
lifetime,
which
makes
it
very
important
for
air
quality
and
economic
system.
Also
isoprene
has
very
large
emission
amount.
It
accounts
for
50,
but
total
balcony
will
see
globally
and
it
also
it
is
about
three
times
organic
boc
emission
and
so
to
calculate
isoprene
flux
in
general.
S
We'll
use
the
mega
model
developed
developed
by
my
advisor,
and
here
is
a
spatial
distribution
of
azure
printing
plugs
calculated
by
making
in
July
and
here's
the
basic
equation
in
making,
and
the
gamma's
here
represents
the
different
environmental
factors
for
making
they
are:
temperature,
radiation,
Leaf,
age,
CO2,
concentration
and
drought.
So
today
we
will
focus
on
gamma
SM,
which
is
drought,
draws
impact.
S
So
first
we
should
know
like
how
draw
to
effect
astropriation
draft
effect,
as
the
preemption
depends
on
its
severity.
Here's
a
conceptual
model
developed
by
Dr
patasnak.
We
can
see
with
the
drought
developed
so
as
a
first
stage
mailed
or
moderate
drugs
as
a
preemination,
where
it
goes
up,
because
the
leaf
temperature
will
be
changed
by
the
stomach,
conductors
change
induced
by
drought
and
when
the
drug
getting
severe,
the
substrate
of
the
carbon
source
for
isoprene
synthesis
will
be
cut
off
and
the
isopreneur
just
goes
down.
S
So
how
could
we
model
as
the
impact
of
draft
and
Heatwave
on
asoper
emission?
We
need
to
answer
three
questions.
First,
how
to
measure
the
simulative
job.
Second
is
like:
what's
the
threshold
to
activate
the
impact
of
job?
Third
is
like
how
to
quantify
the
impact
of
drought
and
in
the
old
version
making
2.1.
S
We
have
this
algorithm
and
we
use
soy
moisture
to
measure
the
severity
of
drugs
and
we
use
the
melting
point
as
a
threshold
to
activate
the
impact
of
draft
and
we
use
linear
this
e-linear
equation
to
quantify
the
impact
to
draw
and
also
there's
other
drop
algorithms
based
on
soil
moisture.
We
can
see
here
in
this
table
and
also,
if
we
put
them
together,
we
can
see
all
these
soil
moisture
based
algorithms.
They
are
very
similar
to
each
other.
However,
there's
a
limitations
about
the
soil,
moisture
based
drought,
algorithms.
S
First,
there
will
be
uncertainty
of
thresholds.
The
melting
point
in
the
previous
slides,
second,
the
soil,
moisture
value
will
be
increase,
is
could
be
inconsistent
among
different
data
sets
or
different
soil
layers.
Also,
this
algorithms
just
ignore
the
impact
of
atmosphere
or
BP
VPD
here
and
also
the
it
just
cannot
stimulate
the
indirect
impact
of
drugs
as
per
emissions,
to
increase
the
leaf
temperature.
S
So
here's
like
one
example
showing
like
how
unservative
thresholds
affects
the
drug
algorithm
performance,
and
here
we
use
two
different
working
points
and
when
we
use
the
relative
point
0.196
we
can
see
the
drought
algorithms
can
capture
the
SRP
emission
variability
during
the
draft
very
well
and
but
if
we
change
this
melting
point
just
a
little
bit
0.04
about
0.04,
now
we
can
see
the
performance
of
all
these
algorithms.
Just
change
dramatically
now
suggest
research
showing
like
for
the
melting
point.
S
Different
data
sets
will
have
like
diverse
values,
which
is
very
hard
for
us
to
use
this
soil
moisture
based
algorithms.
So
in
that
case
we
may
need
new
ways
to
simulate
the
drought
and
healthy
impact
on
a
supreme.
So
my
colleague
Xiao
Yin
developed
this
algorithm.
We
call
it
unlikely
algorithm
because
it's
hot
it's
coupled
to
the
CRM
and
it
simulates
many
processes
explicitly
so
she
used
the
welding
Factor.
So
what
is
called
like?
Oh,
what
is
sorry
now
it's
the
soil.
S
What's
her
stress,
function,
ability
to
it's
kind
of
like
the
factor
used
in
CRM,
to
evaluate
one
stress
as
like
the
two
evaluate
to
evaluate
the
severity
of
chart
and
also
the
CRM
can
simulate
some
of
the
conductance
and
leave
energy
balance
and
leave
temperature
explicitly.
Then
we
can
have
like
the
Deep
temperature
change
caused
by
job
in
this
Union
and
last
show
and
develop
this
algorithm
using
the
BC
Max
to
represent
the
substrate
Supply
in
this
algorithm.
S
When
the
driving
can
see
where
she,
the
algorithm,
will
be
activated
and
decreased
as
super
emission.
But
this
algorithm
is
highly
connected
to
CRM
and
for
some
like
offline
application,
for
example,
air
quality
simulation
and
there
will
not
be
com
model
over
there.
So
we
need
so.
We
may
need
a
an
offline
way
or
like
easier
way
to
simulate
the
impact
of
chart.
S
So
I
developed
this
another
Alpha,
another
offline
algorithm,
and
in
this
algorithm
we
use
the
evapor
transpiration,
the
ratio
between
the
Revival
transpiration
and
potential
evapor
transformation
as
an
indicator
of
the
drop
to
evaluate
the
simulative
chart.
And
here's
like
the
as
the
algorithms
looks
like
we
have
three
parts
so
gamma
SM,
Max
and
Gamma
sub
and
Gamma
FL,
so
Tema
SM
Max
represents
the
maximum
change
caused
by
the
leaf
temperature
change.
S
Represents
the
impact
of
substrate
Supply
and
leave
temperature
change
respectively.
We
can
see
here
like
when
the
job
developed
from
moderate
stage
to
severe
stage
lift
temperature
response,
will
increase
and
being
stable
and
substrate
Supply
responsible,
just
decrease.
When
you
combine
these
with
two
lines
together,
we
can
get
this
blue
line,
which
is
consistent
with
the
conceptual
model
we
I
mentioned
previously,
and
here
we
did
some
validation
for
the
mode
for
the
model.
Here's
so
the
first
panel
is
for
the
online
model
divided
by
Celine
Dion.
S
The
second
panel
shows
the
offline
model.
We
can
see
the
both
among
so
the
both
two
algorithm
can
capture
the
impact
of
jaw
and
super
emission.
S
So
we
can
say
if
we
don't
consider
the
drops,
the
mega
mode,
we
are
showing
overestimation
during
the
draft,
but
if
we
consider
just,
we
can
capture
the
variety
of
as
something
during
the
draw
very
well
and
here's
the
starter
plot
and
first
two
panels
shows
like
if
we
don't,
if
we
don't
consider
jobs
and
when
we
consider
a
draw,
we
can
capture
the
our
variability
of
isoper
emission
much
better
and
but
also
we
need
a
wider
region.
S
Validation,
so
I
use
for
model
height
from
satellite
to
to
validate
the
draft
algorithm.
So
I
use
the
only
formaldehyde
column
concentration
during
year,
2012,
which
is
a
year
there's
a
big
drop
happening
in
the
center
of
the
U.S
and
also
I
run
the
cam
cap
model
with
different
treatments.
So
three
columns
represents
the
different
treatments.
The
first
one
is
no
drug
stress
and
second
and
third
one
represents
online
and
offline
algorithms.
We
can
see
if
we
consider
the
drugs
we.
S
If
we
don't
consider
drought,
we
can
see
in
center
of
your
U.S.
The
model
can
overestimate
the
formaldehive
of
all
column
concentration
and
when
we
consider
a
drug
that
we
can
diminish
this
overestimation,
but
also
for
the
average
semi-arid
region
like
Texas
and
the
model
will
only
estimates.
S
The
common
height
column
concentration
but
considered
drugs
will
make
it
worse
because,
like
com
mode
cannot
simulate
the
drought
very
accurately-
and
here
is
like
comparison
between
the
mostly
Omi
formaldehyde
and
camcam
from
under
high
volume
concentration
in
the
colors
region,
as
a
color
of
the
points
represents
the
severity
of
water
stress
we
can
see
here.
If
we
we
don't
consider
jobs,
there
will
be
the
overestimation
form
of
formaldehyde
I'm
going
to
draw
this
degree
and
then,
when
we
consider
draw,
we
can
diminish
this
overestimation
and
we
can
decrease
mbme
and
rmse.
S
Both
parts
are
online
and
offline
algorithms.
Here
we
can.
We
wanted
to
evaluate
the
capacity
of
CRM
to
simulate
Jaws.
So
we
compare
the
surface
soil
moisture
simulated
by
crm5
and
with
the
esa
CCI
soil,
moisture
data
set
with
the
satellite-based
sorry
moisture
data
side.
We
can
see
the
model
on
the
estimate
story,
moisture,
which
means
the
model
will
exaggerate
the
drought.
Similarity
in
that
case,
if
the
CRM
cannot
simulate
the
drug
ability
accurately,
then
we
cannot
act
means
the
drop
impact
and
supreme
very
accurately.
S
So
if
we
can
improve
the
cell
crm's
capacity
to
simulate
draw,
the
isopro
emission
can
also
benefit
from
this
process,
but
also
our
parameter
rise.
Algorithm
also
have
weakness,
for
example.
Here,
when
the
chart
happens,
the
algorithms
can
only
release
isoprignition
by
27,
but
in
reality,
when
the
chart
happened
here
we
can
see.
This
example
is
like
the
it's
an
experiment
conducting
the
lab
and
when
they
give
the
chart
condition
to
the
plants,
we
can
see
the
leaf
temperature
can
change
a
lot.
S
It
could
be
one
to
five
degrees,
but
we
cannot
have
this
in
this
parameter
like
algorithm.
So
in
that
case,
we
can
now
capture
this
High
values
during
the
draft.
So
if
we
can
find
a
better
way
to
represent
the
connections
between
some
other
conductance
definitely
balance,
and
if
temperature
we
can
better
simulate
these
high
values
better.
S
So,
including
in
this
study,
we
have
a
better
way
to
represent
the
water
stress
in
a
super
emission
model,
and
we
have
a
very
simple
model
framework
for
simulating
the
impact
of
water
stress
answers
for
animation,
but
also
the
inaccurate
drug
stimulation
can
directly
Factor
modeling
of
drought
impact
Oracle
range,
and
this
work
has
been
published
on
gems
recently.
If
you
want
to
know
more
details,
you
can
check
this
paper
now.
Also
I
will
take
any
questions
from
here.
Thank
you.
A
Yeah
very
nice
talk,
I,
have
a
question,
so
you
look
like
I
suffering
emissions.
They
they
do
change
with
your
implementation
right.
You
might
have
shown
that.
Could
you
repeat
that
and
my
second
question
is
this
going
to
be
implemented
in
then
the
new
version
of
the
model
is
this
part
of
Megan's
free.
S
So
yeah
sorry
I
think
globally,
the
other.
If
we
have
like
Trot
in
the
current
cim,
it
will
decrease
as
a
pre-mission
by
11
percent
and
both
two
average
rhythms
showing
the
similar
results
is
that
your
question
yeah.
A
G
Then,
oh
sure,
so
we
know
that
can
isoprene
region
has
the
temperature
dependence.
Is
this
child
effect
that
you
show
here
utters
the
temperature
dependence
shape
that
we
normally
understand
for
ISO
praying.
S
No,
oh,
that's
a
good
question,
so
our
understanding
about
like
drought
and
hit
or
Draw
10
temperatures
like
so
there
are
like
two
different
mechanisms
under
this.
So
we
didn't
change
the
temperature
response
curve.
We
just
changed
like
how,
like
the
draft
change,
the
substance
price
and
for
the
temperature
part
will
mainly
induced
by
the
drought.
I'm.
Sorry,
so,
basically
like
the
drought,
change,
the
leaf
temperature
and
leave
temperature
will
increase
as
a
precision
and
following
the
current
Leaf
temperature
response
curve,
you
make
it
foreign.
M
K
M
I
think
there's
a
recent
paper
by
Wells
at
all,
looking
at
isoprene
columns
using
the
Chris
satellite
and
they
seem
to
indicate
in
the
Southeast
U.S
that
the
models
underestimate.
M
M
If
the
model
takes
takes
into
account
the
drought,
the
model
isoprene
will
be
even
lower
than.
S
Oh
I
see
I
think
it
depends
on
like
the
which
gears
it
looks
at
I.
Think
the
things
like
so
chart
didn't
happen
like
every
year,
also
like
what
as
a
like
the
likes.
S
What
do
we
see
like
I
said
sorry,
we
could
go
back
to
the
box
of
these
slides
so,
for
example,
as
a
mild
and
moderate
stage
and
the
throttle
will
increase
the
answer
pretty
Mission,
but
so
in
that
case,
if
the
draft
is
always
in
like
the
mail
another
message,
it
can
increase
it
as
a
privilege,
but
then
just
getting
into
the
severe
stage
it
will
decrease
as
a
prenation.
S
So
when
we
do
the
global
estimation,
so
basically
in
most
of
cases
we
see
like
the
severe
chart,
and
so
that's
why
it's
globally
will
decrease,
but
for
some
specific
is
for
some
specific
region
for
the
corners
region.
Maybe
if
we
always
like
suffering
from
like
some
else,
male
drought,
there
will
be
like
like
higher
as
a
pretty
mission.
F
A
A
A
S
I
think
so,
I
didn't
really
look
at
monitoring,
but
I
I
go
over
some
literature,
so
so
the
drought
in
general
will
decrease
the
monitoring
emissions.
A
Any
more
questions:
hey
thanks
for
all
that
work
also
and
moving
it
towards
the
new
model
version.
That's
great
okay,
so
we're
a
little
behind
and
I
think
we'll
still
want
15
minute
break,
so
maybe
we'll
meet
five
minutes
later
than
the
program
says:
we'll
have
a
break
till
10
35
and
then
we'll
be
back
here
to
start
right
on
time.
A
Here
it
was
a
great
coffee,
is
in
the
outer
day,
member
and
I
think
there's
some
of
us
actually
going
on
so
I'm
able
to
just
come
back
here
we
are
supposed
to.
We
can
bring
actually
food
entrance
in
this
room,
and
so
maybe
we'll
just
get
through
copying
of
that
here.
I,
don't
think
it'd.
F
B
B
B
B
B
K
K
R
D
R
F
K
So,
oh
and
I'm
out.
K
R
J
K
K
R
K
R
F
K
F
F
F
A
F
A
A
Okay
thanks
everybody
well
starting
up
again
after
our
break
I.
A
And
so
now
we
continue
with
some
work,
ecology
and
emissions,
and
then
we
move
forward
with.
F
A
I
All
right,
thanks
I,
don't
actually
use
these
models
at
all,
but
I
was
talking
to
people
from
the
community
and
they
said
that
they
might
be
some
interesting,
actually
I'm,
not
using
the
models,
but
actually
I
am
now
using
them
for
validations
yeah.
I
Just
a
2d
gun
model
competitions.
I
So
the
background
why
I'm
doing
this
to
stay
down
every
four
years,
this
yoga
system
operation
involved
in
that
Community
looking
at
I,
mean
long-lived
Hello
carbons,
things
like
that,
and
so
then
we
were
expected
to
want
to
fly
Global
emissions.
But
things
like
this
additional
scientific
papers
and
more
in-depth
studies.
So
for.
I
Was
an
assessment
for
the
ones
coming
out.
I
This
has
to
be
done
routinely.
It's
got
a
lot
of
work
to
do,
there's
two
different
networks,
students
as
well,
so
that's
fine
submissions
using
measurements;
and
so
so,
basically,
what
we
do.
We
take
measurements,
work
backwards
between
and
there's
two
different
Global
networks
to
the
use
of
Noah,
a
gauge
which
is
a
Consortium
of
different
organizations,
and
it's
probably
I
say
about
three
people
globally.
Who
do
this
between
three
properly
odds,
mostly
small,
two.
I
And
computation
needs
a
lot
of
work.
It
wouldn't
be
so
we
need
to
sort
of
things
quick
wasted,
trying
to
get
this
done
routinely
So.
Currently
we
just
use
a
box.
It's
oh!
That's
for
the
state
of
the
art.
To
do
completely
is
a
12
Box
model
downstairs,
so
that's
little
schematic
level
the
place
for
the
Box.
What
it's
doing
is
transferring
information
boxes.
I
We
use
surface
measurements,
well,
there's
probably
10
20
sites
that
we
could
use.
So
it's
on
a
huge
social
network,
they're
very
long-term
measurements,
but
they're
not
particularly
right,
spread
out
around
the
world.
I
Features
in
it
course
resolution,
of
course
it's
Box
model,
so
yeah.
It
has
quite
a
big
rough
technician
in
space
and
time,
but
there's
been
a
lot
of
things
recently
talking
about
effects
like
Qbo
I
mean
how
even
would
be
very
long-lived
substances
stuff,
like
Qbo
Excellence
concentrations,
which
is
not
capturing
these
models
and
a
lot
of
these
Services,
which
are
controller
of
the
laundry
protocol,
the
emissions
of
produced
production
scheme
and
to
say
COC.
Since
2010.
I
people
are
really
interested
in
quite
small
deviations
from
a
long-term
Trend,
and
these
deviations
of
people
becoming
interested
in
are
about
the
same
size
as
what
the
error
caused
by
not
including
histological
Dynamics.
I
I
I
We've
seen
you
for
say,
50
different
substances,
two
computations
and
just
wasn't
just
a
Manpower
that
he's
going
to
set
up
these
models.
It's
quite
issues
so,
for
example,
something
to
have
50-year
records.
I
I
K
I
A
2d
model
given
by
Arabian
analysis
fields
allowing
translation
to
vary
monthly,
so
it's
not
at
the
best
solution
there,
transport
that
it
is
because
still
we
should
be
able
to
help
you
capture.
Some
of
these
features
2D
models
a
little
bit
funny,
even
though
you
have
to
parametersize
transported
using
any
tests
represent.
I
Through
Perfection
diffusion,
which
is
not
true
of
excellent
diffusion,
the
way
you
use
it,
but
it
is
parameterization
and
that's
going
to
be
much
progress
with
them.
They'll
use
a
lot.
I
can
say,
like
you
said
about
the
90s,
but
then
once
computational
power
allowing
the
context
between
one
of
those
people
really
description,
specific
cases,
but
it's
something
about
the
ocean
literature.
They
still
use
them
for
a
lot
so
to
diagnose
things
like
any
transport.
So
there's
a
bit
more.
I
There
has
been
some
progress,
maybe
not
in
the
answer
experiences
as
much,
and
then
this
is
at
the
moment
in
what
I've
been
doing.
I
I
Tracer
setups
lots
of
different
configurations
to
try
and
diagnose
it
in
those
philosophies.
I
I
You
have
diagonal
Direction
and
that's
actually
very
hard
to
do
in
some
final
difference
between
the
Machine
model
and
basically
like
a
little
passage
to
it's
introduced.
So
that's
it
prefer
to
welcome
that
physics.
Community,
don't
speak,
we're
getting
around
that
problem
so
implemented
that
and
a
lot
of
these
described
so
lots
of
stuff
describes
this
Spirit.
How
well
so
this.
S
I
I
I
I
So
yeah
certain
about
this
making
this
in
terms
so
measurements
service
sites,
surface
sea
level,
one
service
to
altitude.
So
this
just.
I
I
Scratch
that's
different
state
mounted,
so
it's
quite
another
piece
measurement
sites
a
bit
away
from
population
standards
on
top
of
things
like
even
amounts
of
jobs,
so
already
we're
using
2D
modeling
slightly
better
still
for
the
course
of
a
solution.
I
only
have
30
levels,
they're
uniform
as
well,
so
it's
still
quite
close
to
resolution
even
with
that
would
be
much
better
expensive
and
more
fractions
measurement
size.
J
I
Models
particularly
but
the
surface
is
a
bit
of
an
ambiguous
thinking,
particularly
and
they've
generally
look
used
to
live
in
Midlands
middle
atmosphere.
F
I
And
really
what
we
want
to
do
is
Spotify
missions
with
it,
so
yeah
I've
done
just
10
years
to
quantify
our
missions.
This
University.
H
I
So
run
this
of
six
to
ten
years,
keeping
emissions
constant
to
go
out
there
well
fractions
and
then
Recon
spot
emissions,
and
you
also
have
some
a
to
our
estimates
so
you're
using
basic
statistics.
You
try
probability
what
you
guessed.
What
do
you
think
those
missions
are
so
basically
I
just
randomly
generated
some
big
priority
emissions
by
what
I
thought
that
might
be
interested
line,
the
true
emissions
are
there
orange
and,
and
then
foreign.
I
I
And
the
big
question
is
actually
the
representation
of
a
Qbo
I'm
trying
to
Qbo
is
is
only
varying
wins
and
I'm,
using
his
only
average
model
to
try
and
through
digging
cycle
Qbo,
so
I
think
it's
something
like
seriously
possible
pressure
normally
against
data
with
a
described
Builders
field.
There
is
some
signal
which
type
of
qbl,
but
oscillation
is.
I
It's
going
to
be
analysis
States
anyway,
smoothing
this
does
propagate
all
the
way
down
to
the
surface,
which
is
see
you
know,
it's
not
great.
It
is
not
it's
not
going
to
solve
all
our
problems,
but,
but
actually
that
there's
something
there
and
it's
it's
better
than
what
we've
had
with
the
Box
models,
actually,
which
is.
I
Described
sing-saw
or
maybe
have
to
generate
some
of
those
as
users
to
drive
missions
and
uses
Michigan
sites.
Things
like
comparing
some
of
the
other
atmospheric
concentrations
to
certified
issues
as
well
see
how
well
we're
doing
not
just
at
the
surface,
but
it's
fair
and
then
hopefully,
look
at
some
applications.
So
I.
I
Lot
of
benefits
to
having
sort
of
almost
12
models,
other
applications,
too
yeah
playing
around.
R
R
Have
lost
processes
that
yeah
at
the
top
of
the
model,
and
you
have
the
circulation,
bringing
it
gas
and
high
latitudes.
I
Then,
yes,
that's
how
you
have
a
lot.
So
that's
so
that's
something
that
you
think
about
whether
you
can
just
you
know
how
to
exaggerate
your
loss
at
the
hundred.
So
what
is
your.
I
I
need
to
sort
of
see
how
much
that
impacts,
because
really
at
the
moment
it
is
useful
service
concentration.
So
how
much
does
that
impact
service?
It
doesn't
really
impact
service
concentrations.
Then
it's
a
matter.
Otherwise.
You
have
to
think
about
doing
something
about
having
some
sort
of
yeah
boundary
conditional,
significant
boxes,
yeah.
A
Okay,
well,
we
should
move
on
to
the
next
part
nice
to
have
you
today
and
we
see
your
group
there.
That's
nice
I
hope
you
can
see
us
all
guess
what
yes,
okay,.
P
Can
you
hear
us
okay,
too
yeah
great,
let's
see
yeah.
P
P
Cool
okay,
great
yeah,
so
it's
it's
great
to
at
least
kind
of
be
Hybrid
part
of
this
meeting,
thanks
for
including
us
and
so
first
I
want
to
start
by
acknowledging
my
co-authors
on
this
paper
that
just
came
out
near
the
very
end
of
2022
and
this
new
IOP
Journal
Environmental
Research
climate
and
I'd
like
to
just
point
out.
Sarah
Hancock,
who
is
a
Columbia
undergraduate
with
me,
is
now
a
second
year
graduate
student
at
Harvard
and
she
was
a
big
player
in
helping
us
get
this
project.
P
The
analysis
off
the
ground,
and
actually
she
has
a
first
author
paper
that
should
be
coming
out
or
maybe
just
came
out
this
week.
I'm
not
sure
if
it's
published
yet
in
the
same
new
Journal,
using
the
same
simulations
that
all
of
us
from
here
at
MIT
will
be
presenting
today.
Her
Sarah's
paper
is
not
part
of
our
Trilogy
today,
but
she's
worked
on
aerosols
and
meteorology
from
1950
to
2014.
P
P
So
just
I'm
not
going
to
speak
through
all
the
co-authors,
but
also
acknowledge
the
folks
who
have
helped
us
with
processing
observations
and
making
it
super
easy
to
do
comparisons
with
a
global
chemistry,
climate
models.
So,
on
the
right
hand,
side
what
I'm,
showing
you
is
the
annual
main
show
of
the
Sierra
ozone
Garden.
We
see
from
1950
to
2014
that
there's
been
about
an
80
teragram
increase
and
each
of
these
individual
spaghetti
lines
is
an
initial
condition.
P
P
All
of
a
lot
of
the
spots
I'll
show
the
the
thick
line
will
be
The,
Ensemble
meme,
which
should
be
our
best
estimate
of
the
force
signal,
meaning
the
response
to
the
anthropogenic
emissions
in
greenhouse
gases,
which
are
identical
in
each
Ensemble
member
and
then
the
difference
across
numbers,
of
course,
is
the
role
of
internal
climate
variability.
P
P
So
obviously,
we
have
lots
of
observational
data
sets,
but
they
tend
to
be
sparse
in
in
space
and
there
are
limited
records
that
go
back
multiple
decades
and
so
a
lot
of
what
we
understand
also
relies
on
models.
I
want
to
acknowledge
that
tropospheric
ozone
assessment
report,
which
has
been
important
in
terms
of
not
only
synthesizing
our
current
understanding,
but
also
bringing
together
observational
data,
sets
and
processing
them
in
ways
that
make
that
comparisons
with
global
models
most
meaningfully.
P
In
the
paper
that
I
mentioned
on
the
the
first
slide,
which
I'll
pull
results
from
for
the
rest
of
the
talk,
we
did
look
at
six
ground-based
long-term
multi-decatal
monitoring
sites
around
the
globe.
We
also
use
the
Omi
MLS
as
well
as
tons
satellite
products
and
but
in
the
interest
of
time
today,
I'm
just
going
to
focus
on
the
comparison
with
the
iago's
data
set,
which
are
basically
an
ozone
instrument
sitting
Award
commercial
aircraft
and
I'll.
P
Tell
you
a
little
bit
more
about
those
comparisons
in
a
bit
so
I'm,
not
given
the
the
workshop
I.
Don't
think
I
need
to
step
through,
what's
in
a
chemistry
climate
model
for
this
Cloud,
but
mostly
just
wanted
to
acknowledge
all
of
the
the
work
that
that
we've
been
able
to
to
build
upon
by
by
having
this
model
available.
That
has
fully
coupled
stratospheric
and
tropospheric
chemistry
with
interactive
aerosols
and
also
to
to
try
to
make
that
make
a
case
today.
P
That
I
think
we
can
learn
a
lot
about
not
just
tropospheric
ozone
Trends
but
also
other
atmospheric
chemic
chemistry
species
of
in
the
troposphere,
as
well
as
the
stratosphere.
Through
these
initial
condition.
Ensembles
and
that
they
kind
of
allow
us
this
new
window
into
the
role
of
climate
variability
and
what
role
that
might
play
in
in
The
observed
records
that
we
have,
and
so
we
have
a
12-member
initial
conditional
Ensemble,
which
I'll
combine
in
the
rest
of
the
slides
with
three
send
up
six
members.
P
I
will
mention
that
this
12
member
is
actually
now
a
13
member,
but
in
the
paper
that
I'm
talking
about
today,
we
only
had
the
use
the
the
first
12
members,
but
I
think
you
and
chindanju,
who
will
be
talking
in
a
little
bit.
I
will
use
the
13
member
on
top
okay.
So
how
do
we
know
if
the
model
is
you
know
fit
for
purpose
or
appropriate
for
the
the
application
that
we're
using
it
for
and
so
with
chemistry,
transport
models
or
specified?
P
Dynamics
runs,
of
course,
we're
driving
the
chemistry
with
the
observed
meteorology.
So
we
certainly
expect
that
the
model
should
be
able
to
capture
the
observations,
and
in
this
case
our
focus
is
observed.
Long-Term
trends.
However,
when
we
have
a
chemistry
appointment
model,
that's
generating
its
own
weather.
Of
course,
we
we
kind
of
need
to
reframe
the
question
a
bit,
and
so
the
question
then,
is:
do
the
observed
long-term
trends
that
one
realization
that
we
have
from
The
Real
World?
P
Does
it
fall
in
the
range
of
possible
trends
that
are
simulated
by
the
individual,
Ensemble
members
in
the
chemistry
climate
model
and
so
to
we'll
we'll
look
at
this
now
with
the
iagos
data
and
huge
thanks
to
Audrey
godell?
P
Let's
say
this
is
a
figure
taken
from
her
2020
paper,
but
she
did
an
enormous
amount
of
work
to
carefully
sample
the
data
and
process
it
and
construct
tropospheric
ozone
columns
and
just
the
free
tropospheric
ozone
column
over
these
11
regions,
where
the
commercial
aircraft,
with
the
the
Ozark
instruments
on
them,
are
routinely
descending
and
ascending
from
airports,
and
so
in
her
paper.
She
reports
Trends
in
each
of
these
regions
and
those
are
the
black
dots
on
this
figure.
P
So
what
we're
looking
at
is
the
ozone
Trends
here
and
then
organized
by
latitude
so
that
the
most
northern
region
Europe
is
to
the
top
of
the
plot.
Okay
and
what's
superimposed
on
here
is
the
csm2
Wacom
6.
This
is
the
range
of
those
15
members,
the
ensemble
mean
is
shown
as
the
Square
here,
and
so
there
are
several
points
I'd
like
to
make.
P
The
first
is
that
that
range
crosses
zero
for
this
European
region
and
indicates
that
we
might
have
observed
a
negative
Trend
with
the
exact
same
anthropogenic
emissions
in
greenhouse
gas,
forcing
just
because
the
climate
varied
in
a
certain
way.
However,
the
ensemble
mean
over
all
of
these
regions
is
always
positive,
suggesting
that
the
force
signal
is
a
continual
increase
at
free
tropospheric
ozone
over
all
of
these
regions.
P
Okay
and
then,
if
we
look
the
difference
between
these
two
panels
now
are
just
the
free
tropospheric
column
versus
the
full
tropospheric
column.
So
this
includes
down
to
950
hectare
pascals
and
a
couple
points
here:
I
guess
I'll,
just
zoom
in
one
thing,
is
that
we
notice
that
there's
some
places
where
even
the
15-member
range
of
Trends
does
not
overlap.
P
What's
observed,
and
so
there's
kind
of
two
ways
we
can
think
about
that,
or
at
least
two,
but
one
is
that
it
may
be
that
our
our
15-member
Ensemble
is
just
too
small
to
sample
the
climate.
P
Variability
that
would
actually
happen
in
the
real
world,
but
it
could
also
indicate
a
problem
in
the
model
and
in
this
case
perhaps
given
that
we
see
more
of
an
issue
when
we
look
at
the
full
tropospheric
column,
one
one
hypothesis
is
that
the
emission
growth
rate
over
some
of
these
regions
over
the
period
when
these
iago's
observations
are
available,
which
I
should
have
said,
go
back
roughly
to
the
mid
to
late
90s
these,
depending
on
the
region,
that
that
emission
growth
rate
may
be
underestimated
in
the
sense
inventory
that
underlies
the
cmf6
simulation
and
so
to
dig
into
that.
P
A
little
bit
further
Sarah
Hancock
went
and
got
us
the
steam
of
six
models
that
have
interactive
tropospheric
chemistry,
and
so
those
are
listed
here.
So
we
basically
have
five
additional
models
along
with
csf2
whack
up
six,
and
it
turned
out
that
some
of
those
have
a
different
configurations
and
multiple
Ensemble
members
and
so
taken
together.
The
csm2
Wacom
6
Plus.
These
other
five
models
gives
us
73
Ensemble
numbers
to
compare
with
the
iago's
data,
and
we
see
that
in
many
cases
now
that
broader
cmx
6
range
does
overlap.
P
P
P
And
what
I'd
like
to
point
out
is
that,
just
by
picking
two
different
Ensemble
members,
we
can
see
if
we
focus
over
the
U.S,
that,
depending
on
that
initial
condition
back
in
1950,
we
might
have
simulated
zero
to
slightly
positive
trend
or
negative
Trend
over
the
US
and
Europe,
and
the
sign
of
the
trend
can
also
vary
in
other
regions,
but
also
the
magnitude
of
the
trend
depends
on
on
that
sort
of
imprint
of
climate
variability
as
well.
P
Okay,
focusing
still
on
these
last
two
decades
of
95
to
2014.
What
we're
doing
here
is
now
just
dividing
the
world
into
10
degree
latitude
bands
staying
with
that
lower
tropospheric
ozone
burden,
but
now
separating
by
Seasons,
so
June
July
August
is
green.
December
January
February
is
that
orange
color
and
really
some
of
the
takeaways
here
are
that
the
ozone
Trends
in
the
lower
tropis
here
tend
to
be
stronger
in
the
winter
season.
The.
P
Hemisphere
is
in
the
winter
season,
and
that
makes
sense
if
we
think
about
the
lifetime
of
ozone
is
just
a
lot
longer
in
the
winter.
I
also
want
to
call
your
attention
to
the
trends
overlapping,
zero
in
sort
of
40,
North
and
and
forward
here
and
in
summertime,
and
so
this
is
presumably
in
response
to
the
emission
controls
that
have
happened,
for
instance
in
North,
America
and
Europe
over
this
period
to
try
to
improve
ground
level
ozone.
P
So
let's
contrast
that
with
what's
going
on
in
the
upper
troposphere
in
the
next
plot-
and
here
we
see
that
even
in
summertime
in
the
northern
hemisphere,
pretty
much
all
across
the
Northern
Hemisphere,
we
actually
have
a
stronger
increases
in
Upper
tropospheric
ozone
than
we
do
in
winter
time,
and
so
this
kind
of
decoupling
of
this
upper
tropospheric
ozone
trans
from
the
lower
tropospheric
ozone
Trends
in
summer
here
suggests
that
ozone's
role
as
a
greenhouse
gas
is
continuing
to
grow,
even
as
efforts
have
been
made
to
reduce
lower
troposphere
goes
on
for
air
quality
and
health
reasons.
P
Okay,
so
the
other
point
to
make
here
is
just
that.
We
see
that
upper
tropospheric
ozone
is
increasing
over
this
two
decade
period
throughout
the
the
globe,
pretty
much
other
than
the
polls,
and
then
the
other
point
is
that
we
do
have
places
where
the
for
instance,
if
we
look
in
kind
of
here,
we
don't
have
any
overlap
in
the
seasonal
ranges.
P
Across
The
Ensemble
members,
suggesting
that
there's
a
detectable
difference
between
summer
and
winter
and
in
some
of
the
observational
work,
including
work
that
Audrey
godell
has
done
with
the
algo's
data.
This
seems
to
also
be
detected
in
the
data
in
the
observations.
P
Okay,
so
I'm
just
going
to
sum
up
here
and
say
that
I
hope
that
you'll
also
think
that
that's
initial
conditioning
Ensemble
might
be
interesting
to
use
for
some
of
your
analyzes
we're
hoping
that
it
will
be
broadly
useful
to
the
community
we're
finding
differences
in
magnitude
and
in
sign
in
some
cases
of
20-year
lower
tropospheric
ozone
Trends.
P
Okay,
I
think
in
the
interest
of
time.
I
won't
read
through
all
of
this,
but
just
to
say
that
we
think
we're
finding
some
discrepancies
in
the
emissions,
the
growth
rate
over
the
last
two
decades,
over
Malaysia
in
the
Gulf
of
Guinea
and
then
finally,
we
find
strong,
Regional
and
seasonal
variability
in
trans
and
tropospheric
ozone
in
the
lower
troposphere
extra
Tropics.
P
The
strongest
increases
tend
to
occur
in
Winter,
whereas
in
the
upper
troposphere,
ranging
from
30
South
to
40
now
at
40,
North
is
where
we
see
the
largest
Trends
and
those
increase
most
in
summer.
So
again,
there
seems
to
be
this
decoupling
that
overall
tropospheric
ozone
is
still
Rising,
even
though
in
some
regions
in
the
lower
troposphere,
there's
been
great
strides
made
to
at
least
stem
the
growth
of
ozone
and
actually
reduce
the
highest
levels
of
ozone
in
summertime.
P
H
Yeah
really
nice
talk
as.
H
Admission
issue:
have
you
thought
about
running
with
specified
Dynamics,
where
you
kind
of
take
out
the
very
building
and
just
to.
P
See
if
you're
getting
the
right
answer
in
the
knee
yeah,
that
is
I
seem
to
have
lost
my
Google
meet
window,
but
but
anyways.
Yes,
so
that's
a
great
question
and
sure
we
could
do
that
it
turns
out.
P
Actually
there
were
two
papers
that
had
been
published
in
the
last
year
that
were
I,
don't
think
any
of
them
were
with
cesm
I
think
they
may
have
both
been
geoscam,
although
I'm
not
remembering
exactly
but
looking
at
troposphere,
goes
on
trans
with,
basically
with
specified
Dynamics
and
those
also
found
similar
issues.
So
it
does
seem
like
there's
something.
A
Maybe
quick
other
question:
I
have
a
change
seems
to
play
an
important
role
too
right
because
it
seems
like
we
get
an
increase
in
the
upper
tropos
here.
He
is
in
that
place
an
important
role.
P
P
I
suppose
not
something
you
know
we
could
try
to
separate
out
at
some
point.
My
thinking
was
that
the
role
of
stratrobe
exchange
would
be
more
contributing
to.
A
P
Okay,
great
so
I'm
just
gonna
share
my
screen
and
then
let
I
lost
my
Google
screen,
but
we're
good.
Now,
okay,.
A
C
Hi,
everyone
I'm
Michelle,
a
pleasure.
This
student
working
with
Arlene
Dory
and
today
I'm,
going
to
share
some
findings
based
on
csm2.com6,
which
is
about
characterizing
impacts
of
external
forces
and
internal
controverability
or
enter
a
new
apples
like
also
vibrations,
and
in
this
talk,
I
will
use
vot
to
represent
arbitrust
for
you,
okay,
so,
first
of
all,
a
robot
track
also
has
increased
the
load
since
pre-industrial
time
and
observed
the
ozone
Trends
driven
by
both
human
and
natural
forces,
could
could
have
been
Amplified
or
masked
by
noise
in
response
to
Natural
internal
criminal
variability.
C
Here
we
will
isolate
the
law
of
internal
contourability
and
external
forces
based
on
13
member
initial
condition,
also
about
historical
simulations
of
csm2
economics
and
is
externally
post
the
trends
of
iot
ozone.
We
will
call
the
signal
which
is
estimated
by
the
anxiety
and
the
globality.
Also
signal
is
shown
as
the
black
line
in
this
graph
and
the
noise
remember.
Internal
climate
variability
is
qualified
as
the
differences
between
each
individual
Ensemble
member
and
The
Ensemble,
meaning
by
characterizing
the
signal
and
the
noise
patterns
in
annual
30
ozone
operations.
C
And
then
we
can
say
that
this.
These
are
the
noise.
You
have
one
and
noise
pc1
for
noise.
You
have
one
we
can
find
out
that
there
is
a
dominant
like
the
type
of
features
in
the
tropics,
and
this
is
mainly
driven
by
internal
climuliability,
and
this
pattern
is
obsessed
in
the
middle
attitudes
and
then
we
will
follow
a
message,
a
fingerprint
method,
that
by
centers
work
we
calculate
the
signal
to
noise
ratios
to
explore.
C
Time
series
and
the
results
are
shown
in
the
first
panel
and
the
only
other
noise
will
project
it
onto
the
same
fingerprint
and
then
painted
with
no
overlapping
earlier
Trends
The
Noise
series
is
estimated
as
a
standard
deviation
of
the
area,
Trend
distribution
and
the
results
are
shown
in
the
second
panel.
The
actual
ratio
is
Lu
signal
in
first
panel,
divided
by
the
respective
values.
In
the
second
panel
from
those
two
in
ratios,
we
can
find
that
the
fingerprint
of
your
closing
noise
in
response
to
external
forces
is
significantly
identifiable
in
each
75
years.
C
We
use
the
satellite
data
from
the
only
balance
measurements
and
we
use
the
same
method,
and
the
difference
is
that
the
signal
time
series
will
calculated
by
projecting
the
slide
data
to
the
model
being
complete
instead
of
the
ensembled
members,
and
the
nice
theories
was
also
calculated
using
the
universal
noise
and
quantify
the
best
of
model
simulations
same
as
previous
slides
that
are
model
observed,
afterward
ratios
were
Quantified
and
we
can
finalize
the
model.
Fingerprint
is
also
significantly
identifiable
in
17
years.
I
like
the
records
and
the
detection.
C
Time
is
12
years,
so
here
we
we
show
the
potential
for
applying
a
climate
detection,
the
attribution
method
to
the
chemistry
to
the
atmospheric
chemistry,
and
also
here
you
can
find
that
globality
of
in
long-term
observation
data.
There
are
still
needs
for
us
to
understanding
the
characteristics
of
iot
ozone
noise,
which
can
help
us
understand
the
variations
in
observed.
Truth
work
also.
So
in
The
Following
part,
we
will
focus
on
the
noise.
C
What
I
wanted
to
mention
here
is
that
our
variables
we
will
talk
about
were
calculated
as
the
differences
between
individual
Ensemble
member
and
the
answer
for
me
and
the
previous
Studies
have
assured
that
the
strong,
also
signal
is
formed.
Introsperical
column
ozone.
So
here
we
want
to
explore
that
for
the
uto
zone,
algorical
Partners
also
like
dipole
features,
are
from
annotation
noise.
You
have
one,
and
there
is
a
correlation
coefficient
of
0.94
between
minerals
and
wonderful
Rich,
SSC
anomalies
and
the
pc1
of
your
closing
noise.
C
Then
we
apply
the
correlation,
allows
analysis,
Toyota,
also
noise
and
the
needle
support
access
your
anomalies,
which
shows
a
single
pattern
as
a
noise.
You
have
one,
so
we
can
find
that
we
can
conclude.
There
is
also
a
strong
signal
in
the
utility
ozone
noise.
So,
according
to
previous
studies,
there
are
a
number
of
mechanisms
so
which
also
could
impact
the
tropical
column
also
to
explore
that
for
our
tools,
also,
we
will
discussed
in
two
aspects.
C
So
first
one
is
the
relationship
between
all
utils
and
noise
and
the
changes
of
ozone
production
which
is
associated
with
also
in
terms
of
lightning
production.
Progress
study
has
shown
that
vanilla
needs
to
increase
the
Deep
convection
in
the
tropical
Pacific,
which
increase
the
lightning
produced
at
home,
and
here
we
show
the
overall
relationship
in
the
tropical
Pacific
Beach.
C
Illinois
they
actually
in
the
Scatter
Plots,
which
indicates
that
the
decreases
in
lightning
and
low
production
Theory
are
Nino
correspond
to
Deep
accuracies
in
utl
and
UT.
Also,
however,
when
we
look
into
the
special
patterns
of
the
correlation
coefficient
in
the
near
quarter
region
during
Arduino,
the
Latino
production
increases,
which
is
contrast
to
utl
and
UT.
C
Whether
you
also
I
mean
it
would
I
mean
aperture
whatever
Electronics,
to
represent
the
topic
of
welding
rate
and
the
fact
that
UT
ozone,
noise
pc1
is
positively
correlated
with
tropical
uprounding,
which
also
means
the
increases
of
the
deep
convection
in
the
tropical
Pacific
will
decrease
the
LG
also
to
have
a
better
understanding
of
the
influence
of
anomalous
transport
or
your
ozone
noise
or
plot
the
sensitivity
of
both
the
ozone
noise
and
the
mass
stream
function
to
answer
related
assets.
Variability,
and
here
is
the
meridian
values
near
the
quarter.
C
The
background
color
refers
to
the
regression
coefficients
of
ozone
against
the
needles
to
Quantified
anxiety,
anomalies
and
the
overlay.
The
lines
are
the
results
of
the
workers
circulation
to
the
Arduino,
which
is
shown
by
the
regression
coefficients
from
the
from
regressing.
The
zombie
mastering
function
against
an
industrial
point
of
anomalies,
and
the
third
line
means
the
positive
value
and
the
Dachshund
line
means
the
negative
value
compare
compared
with
the
normal
circulation.
C
We
can
find
that
the
cellular
line
means
the
anomalous
anomalous
clockwise
circulation
and
then
the
dash
line
means
the
anomalous
con
anti-clock
alarmized,
circulations
and
there's
only
also
when
it's
shown
here,
which
is
above
the
30
about
above
the
380
Pascal
and
between
and
under
the
sugar
pulse,
the
response
in
the
ozone
in
lower
middle
lower
and
the
middle
troposphere
to
also
are
similar
to
that
in
the
atmosphere.
C
C
So
for
the
Eastern
region
there
are
stronger
upwelling
motion
and
which
we
should
put
the
leader
to
the
decreased
in
ut,
also
the
decreasing
immunity
ozone,
and
it
is
worth
noting
that
in
the
Eastern
regions,
the
positive,
also
normally
in
the
middle
of
fields
May
because
of
that
in
Korea,
is
the
stratosphere
to
terms
of
the
exchange
also,
but
we
should
try
to
do
more
work
to
prove
it
and
so
for
the
western
region.
The
situations
is
eventually
the
difference.
C
C
A
Between
other
climate
node
index,
variability.
P
There's
so
I
think
that
the
answer
as
well
yeah
yeah.
T
Hi,
that
was
really
nice.
I
had
a
question
about
the
relationship
with
lightning
knocks
I
guess
I
I
might
have
misunderstood
what
you
said,
but
I
think
my
interpretation
of
what
you
said
was
that
it's
expected
that
in
La
Nina
you
get
increases
in
deep
convection
in
the
tropical
Pacific
which
increased
lightning
knocks,
but
actually
you
were
finding
El
Nino.
It
was
happening
during
oh
no.
Now
now
that
you
showed
the
slide,
you
show,
you
said
during
El,
Nino,
there's
less
lightning,
nuts,
okay,.
C
Oh
sorry,
I
should
share
the
screen.
C
Okay,
so
for
this
part,
yeah
yeah,
so
overall
in
tropical
Pacific,
we
find
that
when
there
is
lamina,
Latino
will
increase.
But
when
we
look
into
like
the
special
patterns
in
the
near
quota
regions,
there
are
Nina,
let
me
know,
will
increase
so.
R
Ahead,
I'm
just
curious,
this
eof
one
that
you
derived
from
the
65-year
time
series.
R
Is
it
good
to
assume
it's
constant
or
if
you
split
your
data
set
into
two
32-year
act,
you
know
I'm
wondering
if
it's
if
this
pattern
is
constant
over
the
climate
record
or
whether
you
we'd
actually
get
a
trend
in
that
pattern,
but
do
you
get
the
same
need
less?
If
you
split
into
232-year
Time
series.
P
That's
a
great
question
yeah,
so
he's
he's
asking
if
you
split
your
data,
we
should
do
that
and
into
two
two
32-year
periods
right,
because
there's
reasons
with
the
admissions
changing.
We
might
expect
that
the
pattern
that
the
signal
is
actually
changing
this
time.
P
A
Okay,
qualifications
move
on
to
the
next
book
by
Nori,
Norwich
West.
B
U
Just
because
there's
some
background,
yeah
there's
some
background
noise
in
the
area
that
I'm
at
right
now
give
me
one.
Second,
okay,.
A
Yeah,
that's
great
if
you
can
show
that
and
Nori
will
talk
about
evaluation
of
model,
simulated
ozone
and
it's
precursor
using
high
resolution
model
simulations
here
in
the
Michigan
Ontario
also
on
Source
experiment,
moves
or
teacher.
Well,
there's
we
don't
have
a
password,
which
means
that
we
have
to
find
this
master
password
available.
A
We
can
also
share
it
if
I
find
your
recording
here,
I
can.
A
With
the
book
I
do
it
from
here,
then
I
guess
I.
K
U
Yourself,
let
me
mute
myself,
really
quick
I,
don't
know
if
I.
A
U
A
True,
okay,
I
need
to.
A
F
F
U
Yeah,
if
it,
if
it
doesn't
work
I
can
I
can
try.
Let
me
let
me
let
me
just
see
if
I
can
move
into
this
room
over
here.
U
I
can
go
ahead
and
share
and
and
present
it
here.
Just
excuse
me
if
there's
any
any
noise
in
the
background.
A
Are
you
trying
to
yeah
share
your
presentation.
U
U
Okay,
you
can
see
it
right,
awesome,
hello,
everyone,
my
name
is
Nori
Beth
mariskell
and
I'm,
a
PhD
student
at
Wayne,
State
University
in
Detroit
Michigan,
today,
I'm
going
to
just
be
discussing
a
project
that
I've
been
working
on,
titled
evaluation
of
model,
simulated
ozone
and
its
precursors
using
high
resolution
model
simulations
during
the
Moose
campaign,
and
you
know
this
will
essentially
become
part
of
my
PhD
dissertation.
So
I'd
like
to
thank
the
co-authors
that
are
listed
here
for
for
their
support.
U
So
the
main
motivation
of
this
project
is
essentially
so
because
in
Southeast,
Michigan
ozone
levels
have
continued
to
exceed
National
ambient
air
quality
standards.
Despite
there
having
been
reductions
in
its
precursor
emissions,
nitrogen
oxides
and
volatile
organ
compounds,
many
factors
could
be
Associated
to
this
ozone
exceedance,
which
include
you
know
anthropogenic
emission
sources.
U
I've
listed
some
examples
here
with
like,
for
example,
in
Monroe
Michigan,
which
is
the
bottom
most
magenta
star,
there's
a
power
plant,
Detroit
Michigan
in
the
center
there's
the
Marathon
Oil
Refinery
and
then
just
across
Port
Huron
Michigan
into
Canada,
which
is
Sarnia
Ontario,
also
known
as
the
chemical
Valley.
There
is
a
a
mix
of
industrial
emission
sources
as
well
that
could
be
contributing
to
this
ozone
non-attainment
in
southeast
Michigan,
due
to
factors
such
as
long-range
transport
and
the
land
Clique
interactions.
U
Southeast
Michigan
is
bordered
by
two
of
the
five
Great
Lakes
Lake
Huron
up
here,
and
then
we
have
Lake
Erie
to
the
east
of
Southeast
Michigan.
So
because
of
this
more,
you
know,
although
there
have
been
previous
campaigns
studying
ozone,
exceedances
in
Michigan,
more
detailed
and
Innovative
measurements
and
modeling
studies
are
needed
for
understanding
ozone,
production
and
loss,
particularly
in
southeast
Michigan
in
the
present
day.
U
So
you
know
with
this
being
said,
I
have
you
know
two
main
objectives
throughout
this
project
which
are
to
investigate
the
sensitivity
of
model
stimulated
ozone
and
its
precursors
using
model
horizontal
grid
resolutions
in
Musica,
through
the
creation
of
an
approximate
seven
kilometer
grid
over
the
state
of
Michigan
and
implementing
and
evaluating
this
new
grid
with
a
recent
field
campaign
that
was
conducted
in
the
area
in
2021
and
2022.
U
U
So
you
know
one
of
the
an
additional
like
main
motivation
of
this
project
was
you
know,
through
the
Michigan
Ontario
ozone
Source
experiment,
which
was
led
by
NASA,
the
Michigan
Department
of
environment,
Great,
Lakes
and
energy
environment,
climate
change
Canada
as
well.
As
you
know,
participation
from
various
universities.
This
included
the
participation
of
you
know
our
research
group
as
well,
and
you
know,
the
goal
of
this
campaign
was
essentially
to
Define
potential
attainment
strategies
in
southeast
Michigan
to
better
understand
what
contributes
to
Ozone,
and
you
know,
determine
these
attainment
strategies.
U
It
was
conducted
in
two
separate
phases
in
the
early
Summers
of
2021
and
2022
and
included
a
mix
of
you
know:
high
resolution
aircraft,
mobile
lab
and
in-situ
measurements
as
well.
The
images
that
I
have
shown
here.
So
there
are
two
sites
that
our
research
group
worked
heavily
at,
so
the
Southwest
Detroit
site.
It's
located,
you
know
in
in
the
Detroit
area,
amidst
it's
down
the
street
from
the
Marathon
Oil
Refinery
near
an
International
High,
a
very
large
Highway,
and
then
the
international
border
crossing
to
Canada.
U
It's
also
located
right
next
to
the
large
Construction
Construction
Site
for
the
new
bridge,
that's
being
built
to
Canada
called
the
Gordy
house.
So
it's
a
it's.
A
great
space
to
you
know,
learn
about
the
anthropogenic
emission
sources
that
may
be
contributing
to
Ozone
and
then
separately
at
the
New
Haven
site.
This
is
located
in
a
more
suburban
area
on
the
coast
of
Lake
Saint
Claire,
which
is
essentially
this
Lake
connects
two
of
the
Great
Lakes
that
I
mentioned
in
the
previous
slide.
U
So
it's
a
it's
a
great
site
to
look
at
the
landslake
interaction
aspect
of
this
project.
U
So
for
this
project,
as
you
know,
many
of
you
on
the
call
already
are
familiar
with
I
used
musica
version
zero,
which
is
the
first
implementation
of
the
musica
framework.
As
a
configuration
of
the
community
atmosphere
model
with
chemistry.
It
uses
the
spectral
element
dynamical
core,
which
enables
Regional
refinement
and
has
a
Default
Resolution
of
approximately
14
kilometers
over
the
contiguous
United
States.
U
For
this
project,
I
decided
to
take
advantage
of
the
regional
refinement,
capabilities
and
Define
a
grid
over
Michigan
of
approximately
seven
kilometers,
which
can
be
seen
down
here
through
the
use
of
the
community
mesh
generation
toolkit
and
various
input
processing
tools
available
on
the
musica
Wiki.
So
after
creating
and
you
know,
testing
the
initial
runs
for
this.
U
U
In
this
phase
and
Incorporated
the
you
know,
Megan
bug
fix
as
well
as
the
addiction
book
better
that
have
been
mentioned
in
previous
talks
throughout
the
working
group
meeting,
I
used,
meritu
meteorological
fields
and
then
anthropogenic
emissions
from
cams
and
aircraft
emissions
from
cams
as
well,
and
then
biomass,
burning
emissions
from
Q,
fed
and
fin
after
running
the
initial
simulations,
I
took
a
look
at
the
model.
U
Cost
between
the
two-
so
you
know,
running
musica
with
the
14
kilometer
grid
is,
is
relatively
computationally
expensive,
but
it
can
become
even
more
expensive
when
using
the
TS
to
chemical
mechanism.
So
you
know,
when
you
know,
refining
this
grid
over
the
the
state
of
Michigan.
It
did
see
of
an
approximate
36
percent
reduction
in
terms
of
model
cost
which,
in
my
case,
because
in
in
the
future
I,
would
like
to
run
some
sensitivity
simulations.
U
Could
you
know
help
with
that
computational
cost
and
make
the
the
models
run?
You
know
faster
and
smoother
as
well.
So
you
know
that
was
one
of
the
benefits
there
have
also
been.
You
know.
Various
campaigns
in
the
state
that
you
know
might
be
able
to
take
advantage
of
this
Michigan
grid
for
for
future
uses
on
you
know,
projects
other
than
than
mine
here
and
I
here.
I
have
just
a
moving
image
of
carbon
monoxide
over
Michigan
for
May,
5th,
2022,
I'm,
sorry,
May,
5th,
May,
22nd
of
2021.
U
Sorry
about
that
and
with
figure
a
showing
the
Default
Resolution
and
then
figure
B,
showing
the
more
the
seven
kilometer
resolution
that
I
made
over
Michigan
and
as
you
can
see
from
this
plot,
it's
just
essentially
showing
that
the
concentrations
are
definitely
you
know
more
I
guess
the
point
sources
can
be
better
identified
with
this
new
resolution,
which
is
definitely
needed
for
a
project.
That's
looking
at
such
a.
U
A
small
area
compared
to
the
rest
of
the
state
and
some
preliminary
results
that
I
have
shown
here
so
I
compared
the
initial
musica
runs
for
ozone,
nitrogen
oxide
and
formaldehyde
against
the
Moose
aerodyne
mobile
lab
measurements,
where
the
magenta
line
here
represents
the
default
resolution
in
Musica.
The
green
line
represents
the
the
finer
resolution
and
then
the
X's
represent
the
observations
for
each
of
these
species
for
the
entirety
of
the
mobile
lab
data
sets,
which
is
from
May
22nd
of
2021
to
June
30th
of
2021..
U
The
overall
amine
bias
performed
significantly
well,
when
you
know
taking
an
initial
look
at
the
at
the
comparison.
So
with
for
ozone,
the
mean
bias
between
the
the
new
resolution
and
the
observations
was
1.3
parts
per
billion
compared
to
you
know
negative
2.9
parts
per
billion.
So
there
are,
you
know
significant
differences
there
and
then
for
nitrogen
oxides
and
formaldehyde.
The
mean
bias
compared
very
well
compared
to
the
observations
when
looking
at
this
overall
mean
bias
of
the
data
set.
U
But
when
looking
closer
at
these
plots-
and
you
know
more
at
the
AML
flight
tracks,
that
I
have
you
know
listed
to
the
right
on
this
image
here,
you
know
it
can
be
seen
that
you
know
there
are
these
spikes
in
each
of
these
species,
as
well
as
various
other
plots
that
I've
I've
had
the
chance
to
take
a
look
at.
U
You
know
that,
where
you
know
neither
model
captures
which
could
be
you
know
like
could
be
due
to
you
know
the
path
that
the
AML
track.
You
know
took
that
day
or
maybe
there
was
a
you
know,
spike
in
temperature,
something
like
that.
So
that
is
something
that
I'm
still
looking
at
a
little
bit
more
in
depth.
U
So,
looking
at
the
day
by
day,
measurements
versus
just
an
overall
mean
bias
is
important
in
this
case,
to
identify
what
is
you
know
leading
to
these
increases
in
in
the
southeast
Michigan
region?
U
So,
like
I
mentioned,
you
know,
I'm
continuing
the
evaluation
of
the
models
with
the
observational
constraints
for
moose
on
a
day
by
day
basis,
as
well.
As
you
know,
taking
a
closer
look
at
the
meteorology
on
the
plot
that
I
have
here
to
the
right,
it
just
shows
temperature
over
Michigan
for
a
particular
day
in
June,
and
you
know,
although
the
models
ran
with
identical
configurations,
the
temp,
there
are
some
temperature
discrepancies
when
you
know
looking
at
it
compared
to
observations
as
well.
U
As
you
know
this,
these
just
initial
heat
maps
of
of
temperature,
so
where
for
the
Default
Resolution,
it
tends
to
be
hotter
in
the
southwest
part
of
the
of
the
plot
it
tends
in
the
new
resolution.
It
tends
to
be
hotter
in
the
Northwest,
so
I'm
kind
of
looking
at
you
know
why
these
changes
are
happening
is
also
you
know
on
my
list
of
future
ongoing
and
future
work.
U
Additionally,
like
I
mentioned,
I
would
like
to
run
sensitivity,
experiments
to
understand
the
impact
of
the
emission
sectors
on
ozone
in
southeast
Michigan,
so
I
will
be
doing
that
once
I
finish
up
the
evaluation
and
then
in
the
future,
I
hope
to
combine
the
musica
model
output
with
an
exposure
model
to
study
the
impacts
of
non-attainment
on
human
health
in
in
the
area.
U
It's
an
area
where,
where
I'm
you
know
currently
located
at
and
I,
have
been
living
here,
my
whole
life
so
I
think
it's
it's
important
and
to
look
at
these
human
health
impacts,
and
you
know
I
think
musica
provides
a
great
framework
for
that.
U
So
yeah
and
I'd
like
to
just
acknowledge
that
a
lot
of
this
work
is
based
on
you
know,
funding
supported
and
supports
from
NSF
ncar,
as
well
as
the
NASA
Michigan
spacecraft,
Consortium
and
I'd,
also
like
to
thank
Tara,
Brian
and
Francesca
from
aerodyne
research
and
the
rest
of
the
Moose
science
teams
for
their
the
campaign.
Data
sets
that
have
been
used
in
the
study.
U
Yeah
so
I
I
created
a
nudging
window
over
over
the
like
the
the
area
of
the
grid,
essentially
for
both
of
the
of
the
simulations.
A
You
nudge
inside
the
refined
region,
or
do
you
exclude
the
latching
inside
the
refined
region.
U
Yeah
definitely
something
to
consider
foreign.
N
U
So
I'm
currently
processing
the
when
data
right
now
and
hope
to
have
some
further
results
with
that
in
the
future.
So
definitely
because
you
know
a
lot
of
the
not
only
with
temperature,
but
I've
noticed
a
lot
with
a
lot
of
the
meteorological
species.
There
are,
you
know,
discrepancies
between
each
grid,
even
though
they've
been
you
know,
run
with
very
with
identical
configuration,
so
could
just
be
the
way
that
the
model
is
analyzed
or
you
know
running.
A
A
M
You,
okay,
excellent,
so
hello,
everyone!
Thank
you
for
having
me,
my
name
is
Ben
Lee
I.
L
M
A
research
scientist
here
at
the
University
of
Washington
in
Seattle
I
work
primarily
with
Professor
Joel,
Thornton
building
and
deploying
instrumentation
to
study
the
chemistry
occurring
in
the
atmosphere
for
this
project.
I've
had
the
pleasure
of
working
with
Professor
Abby
Swan,
also
here
at
the
UW
Abby
specializes,
in
looking
at
the
interaction
between
ecosystems
and
climate.
M
So
this
project
this
is
her
first
foray
into
chemistry.
This
is
my
first
Venture
into
modeling
period,
so
as
you'll
see
by
the
end
of
this
dog,
well,
I'm
sure
we'll
have
more
questions
for
you
than
we
have
results
to
share.
So
here
we
go.
M
Oh
one
thing
to
mention,
so
Abby
will
be
at
ncar
next
week
for
a
different
project
that
she's
working
on.
So
if
anyone
has
any
questions,
you
could
ask
me
here
at
the
end
of
this
talk,
or
also
talk
to
Abby
next
week
at
encore,
so
when
Abby
first
sort
of
started
this
project,
the
main
goal
for
it
was
to
try
to
identify
any
kind
of
observables
or
Trends
resulting
from
Regional
perturbations,
namely
deforestation,
and
when
I
was
tasked
with
this
project.
M
Of
course,
given
my
background,
I,
naturally
gravitated
towards
atmospheric
chemistry,
so
the
question,
the
more
specific
question
became,
can
Regional
deforestation.
Events
affect
atmosphere,
chemistry,
not
just
over
this
region,
but
sort
of
on
a
wider
scale,
possibly
a
global
scale.
So
the
model
that
we
used
was
cam
6
chem
the
land
was
in
BGC
mode
with
fires,
the
sea
surface
temperature
was
prescribed
one
of
the
the
runs
that
we
ran
was
a
historical
transient
simulation
from
1970
to
2015.
M
comset
FC
fire
hist.
For
those
of
you
who.
J
M
Interested
so
as
we
started
to
dive
into
the
output,
the
results,
one
of
the
first
things
that
we
noticed
was
that
the
episode
that
the
isoprene
levels
were
too
high.
M
So
what
you're
looking
at
here
on
the
figure
on
the
left
is,
is
over
most
of
South
America,
focusing
on
the
Amazon
you're,
looking
at
color
Contours
of
isoprene
mixing
ratios
in
the
atmosphere
in
the
level
of
the
atmosphere,
that's
closest
to
the
surface,
so
you're,
looking
at
anywhere
from
30
upwards
of
40
parts
per
billion,
which
is
clearly
too
high.
M
The
figure
on
the
bottom
right
is
from
way
at
all
2018
and
they
are
showing
isoprene
measurements
during
the
go
Amazon
campaign
as
a
function
of
time
of
day,
as
you
go
from
left
to
right,
you're
looking
at
June,
September,
December
and
January
and
I
believe
go
Amazon
was
stationed
just
outside
of
manasse
in
Brazil,
so
you
can
see
that
there's
a
lot
of
variability,
but
typically
you're,
looking
at
isoprine
levels
near
the
surface,
anywhere
from
four
upwards
of
12
parts
per
billion.
M
So
now
you're.
Looking
at
over
the
same
domain,
oh
in
the
atmosphere
near
the
surface
and
where
isoprene
is
really
abundant,
you're
you're,
seeing
that
the
oh
is
too
depressed.
So
you're,
looking
at
in
these
regions
anywhere
from
one
and
a
half
to
upwards
of
three
and
a
half
times
10
to
the
four
molecules
per
centimeter
cubed,
which
is
really.
J
M
It's
probably
Low
by
effect
of
at
least
10
upwards
of
20
25.
So
that's
quite
low
and
others
have
seen
this
as
well.
This
is
a
figure
from
Wells
at
all
in
2020,
and
this
is
in
January,
I,
believe
of
the
Year
2013.
The
figure
on
the
left
shows
isoprene
column
as
measured
by
the
cross-track
infrared,
Sounder,
satellite
or
Chris,
and
the
panel
on
the
right
shows
isoprene
column
as
measure
as
sorry
modeled
by
geoscam
or
the
isoprene.
M
J
M
Underestimation
of
oh,
which
is
caused
by
an
underestimation
of
no
emitted
by
soils,
so
I
should
also
mention
the
sort
of
really
high
isoprene
in
the
boundary.
B
M
Or
not,
we're
not
the
first
to
have
reported
this
I
believe
Joe
at
all
in
2021
and
ACP
also
made
a
note
of
this
as
well
so
moving
forward
so
yeah.
So
what
does
soil?
M
No
emissions
look
like
in
cesn,
so
this
is
a
Time
series
from
from
1970
to
about
2015
you're,
looking
at
cesm
soil,
anal
fluxes
over
the
Amazon
and
units
of
micrograms
per
meter
squared
per
hour,
and
you
see
the
seasonal
variability
as
you
should,
and
the
range
is
roughly
between
four
and
eight
micrograms
per
meter
Square
per
hour.
M
So
how
does
that
compare
to
what
has
been
observed?
So
this
is
a
figure
that
I
quickly
put
together.
This
is
a
soil,
no
fluxes.
The
gray
shaded
region
is
sort
of
roughly
the
range
that
we
encounter
in
the
cesm
model
and
these
markers
represent
measurements
from
field
sites,
a
unique
site
or
or
time
periods.
This
is
just
in
an
unperturbed
force
in
the
Amazon,
so
it
does
not
include
pasture
land
or
agricultural
Fields,
no
logging
sites.
M
So
this
is
just
undisturbed,
foresights
and-
and
you
can
see
that
that
what
is
measured
or
I
should
say,
what's
prescribed
in
CSM
lies
on
the
lower
end
of
what
has
been
observed
out
in
the
field
so
out
in
the
field
how
these
measurements
are
done.
There
are
these
big,
Chambers
or
domes
that
go
over
a
known
area
of
the
soil
and
that
don't
that
chamber
stays
there
for
a
known
amount
of
time
and
instruments
measure
how
much
nitric,
oxide
or
nox
accumulates
in
that
air
space.
M
So
so
you
can
see
that
what
we
have
modeled
into
esm
seems
to
be
low,
so
as
a
first
cut,
quote-unquote
six,
which
it's
not
really
a
fix,
but
it's
our
first
sort
of
attempt
at
trying
to
address
what
we
think
is
an
underestimation
of
soil.
I
know
emissions,
we
multiplied
the
soil
and
no
flux
by
a
factor
of
10.,
so
we
go
from
between
four
and
eight
to
about
40
to
80.
and
that's
what
you're
seeing
here
is
our
sort
of
first
cut
attempt
at
trying
to
address
this
shortcoming.
M
We
multiply
Justice
soil,
no
flux,
so
we
didn't
touch
the
anthrogenic,
no
emissions
or
lightning
or
fire
just
the
soil,
no
by
a
factor
of
10,
and
when
we
do
that,
we
see
that
the
oh
increases.
So
this
is
you're.
Looking
at
surface,
oh
mixing
ratio,
it's
a
fractional
change
in
the
surface.
Oh
mixing
ratio,
as
you
increase
the
no
by
a
factor
of
10
keeping
everything
else,
constant
and
oh
increases
roughly
by
a
factor
of
four
and
the
chemistry
I'm
sure
you
guys
are
familiar
with.
When
you
have
KNOX
in
particular
nitrogen
dioxide.
M
It
photochemically
produces
ozone
which
photochemically
produces
hydroxyl
radicals,
nitric
oxide
no
in
the
atmosphere
can
also
react
with
peroxy
radicals
stemming
from
the
oxidation
of
voc,
such
as
ice
cream
and
monoterpenes,
and
in
that
reaction
you
typically
you.
Well,
you
reform
the
NO2,
which
can
further
go
on
to
measure
that
form
ozone
which
forms
more
oh,
but
you
also
form
these
carbonyls
and
HO2,
which
is
the
the
partner
to
oh.
Together
they
make
up
what
we
call
hops.
M
The
h02
itself
can
react
with
no
to
give
you
o
h
again.
It
gives
you
back
the
NO2
and
some
of
these
carbonyls
on
the
order
of
several
minutes
to
hours
can
fertilize
to
give
you
more
HO2,
so
long
story,
short
more
knocks,
particularly
in
these
pristine
environments,
gives
you
more
Hucks,
especially
when
you
have
a
lot
of
voc
around.
M
So
what
does
the
isoprene
look
like?
So
this
is
you're
looking
at
now
the
absolute
change
in
isoprene
as
you
increase
the
soil
and
no
emission
by
a
factor
of
10,
and
you
see
that
the
isoprene
drops
on
the
order
of
10
to
15
parts
per
billion,
so
we're
sort
of
approaching
a
more
sort
of
reasonable
value.
So
a
value
that's
closer
to
what's
been
observed
in
the
field
and
the
implications
are
this
sort
of
Correction
for
no
is
correct.
M
That
has
a
large
impact
on,
oh,
which
then
has
implications
on
the
lifetime
of
reactive
species
such
as
methane
and
isoprene.
It
also
impacts
the
distribution
of
these
precursors
and
their
oxidation
products
so
stuff
like
secondary
organic
aerosol
ozone,
nitric
acid,
all
those
good
things
so
just
to
put
that
into
context.
This
is
a
figure
from
Turner
at
all
2019,
it's
a
perspective
in
pnas
and
they're.
Looking
at
methane,
a.
J
M
Global
background
methane
mixing
ratios,
so
this
is
a
Time
series
from
about
1985
to
2016
2017..
On
top
of
this
figure,
they're
they're
showing
methane
mixing
ratios
in
both
the
northern
hemisphere
and
southern
hemisphere
and
they're
focused
on
this.
What
they
call
stabilization
period
between
the
years
2000
and
2007.
You
can
see
that
in
the
face
of
sort.
M
Emissions
background
methane
levels,
sort
of
stalled
out,
so
there's
been
a
lot
of
studies
focused
on
why
this
has
occurred
and
Turner
at
all
and
Rigby
and
all
and
others
have
pointed
out.
So
if
you
look
at
the
bottom
part
of
this
figure
that
the
stabilization
of
methane
is
due
to
the
sort
of
uptick
in
oh
during
this
period,
which
state
were
able
to
determine
by
using
I,
believe
halocarbons
and
methyl
chloroform
I
think
measurements
from
the
a
gauge
sites.
M
M
So
then,
so,
since
we
sort
of
had
this
quote-unquote
fix
or
it's
more
like
a
patch
for
what
we
think
is
the
true
soil
and
Knox
emissions,
we
pivoted
to
our
original
question,
which
is
what
happens
to
your
atmospheric
chemistry
when
we
chop
down
a
fork.
So
we
simulated
deforestation
over
the
Amazon
by
altering
the
land
cover
which
basically
took
down
the
leaf
area
index
from
about
six
two.
M
Essentially,
nothing
and
that's
what
you're
seeing
on
this
figure
here
on
the
left
is
the
change,
the
absolute
change
in
the
leaf
area
index
over
the
same
domain.
So
we
go
from
a
forest
to
basically
a
pestry
land,
so
that
also
means
that
the
VOC
emissions
emissions
of
things
like
ice
cream
and
model
terpenes
essentially
became
negligible.
So
when
you
do
that,
the
change
in
oh
is
huge.
So
because
you
took
away
this
the
sink
of
a
wage.
All
these
things
like
ice
cream
and
model
terpenes
that
react
with.
M
M
On
the
other
hand,
HO2
the
partner
to
oh
and
making
up
Hawks
decreases
by
a
lot
because
you've
taken
away,
isoprene
and
monoterpenes
and
other
VOC,
so
you've
taken
away
ro2.
But
you
still
have
no
and,
as
you
recall,
the
the
reaction
between
r2no
can
generate
HO2,
so
you've
taken
away
that
Source
term.
So
HO2
goes
down
and
the
amount
that
HO2
decreases
far
outweighs
that
the
amount
that
oh
increases.
M
So
since
we're
running
low
on
time,
I'll
sort
of
blow
past
this
pretty
quickly,
but
the
point
that
I
want
to
make
here
is
that
so,
when
you
change
both
boc
boc
and
Knox,
the
change
in
oh
is
about
a
factor
of
50
150.
So
it's
a
lot
higher
than
the
product
of
the
effect
from
the
two
previous
scenarios.
That
I
showed-
and
this
is
sort
of
a
long
way
of
saying,
is
that
that
CSM
chem
is
a
for.
M
M
Other
thing
to
note
is
that
so
for
this
talk,
I've
just
showed
results
from
the
region
where
we
induce
these
perturbations,
but
we
see
changes
in
the
Hawks
away
from
those
regions
but
and
we're
still
sort
of
investigating,
whether
that's
an
actual
sort
of
chemically
driven
change
or
if
this
is
sort
of
just
some
noise
that
we
see
from
model
runs
model
runs.
So
that's
still
under
investigation.
M
So
I'll
finish
with
this.
So
this
is
a
figure
from
Davidson
in
1991.
I
think
this
is
the
first
paper
that
described
this
hole
in
a
pipe
model
that
describes
nitrification
and
denitrification,
which
are
responsible
for
emissions
of
no
and
n2o
from
soils.
You're
looking
at
fluxes
of
these
compounds
as
a
function
on
the
x-axis
of
water
filled.
M
I'm,
sorry,
okay,
so
oh
yeah!
So
so
so
in
the
two
in
the
two
sort
of
simulations
that
we
did
with
the
force
intact
and
deforested,
there's
clearly
a
difference
in
soil
temperature
and
moisture,
but
what
we
see
the
resulting
soil,
anal
fluxes.
We
don't
see
a
change,
so
we'd
like
to
investigate
further
what
how
soil
and
Emissions
are
prescribed
in
the
market.
So
with
that
I'll
stop
and
take
your
questions.
Thank
you
for
your
attention.
Thank.
F
A
Your
hands
up
and
then
maybe
we
move
on,
but
there's
certainly
a
lot
to
talk
about
and
follow
up.
Thank
you.
T
Yeah
I
just
had
a
quick
question
about
how
you're
thinking
about
the
canopy
reduction
Factor,
because
chamber
soil,
Knox,
watches,
probably
shouldn't
agree
with
top
of
the
canopy
Knox
luxes,
because
the
Amazon
canopy
is
so
tall
and
there's
a
lot
of
chemistry
and
stomatic
uptake
of
of
NO2.
That
happens.
M
T
T
B
C
T
M
Right
right,
yeah,
so
I
need
to
learn
more
about
within
the
Cannabis
scale.
How
that
chemistry
is
done,
but
I
guess
what
I'm,
what
we're
trying
to
get
right
is
yeah.
How
much
yeah
is
it
right
to
compare
what's
coming
out
of
the
soil,
to
what's
I
guess
what's
modeled
at
the
surface
level,
yeah
but
I
mean
clearly
what
we
see
is
that
yeah
sorry
God,
no.
A
A
It's
great
there's
already
a
group
working
on
putting
in
interactive
soil
emissions
into
CLM.
A
Sheffield
and
Jeff
Getty
at
Boston
University,
so
we
should
get
in
touch
with
them.
Yeah.
F
A
Okay,
that's
sorry:
we
need
to
move
to
the
last.
G
A
Please
try
to
share
your
screen
and
she's
talking
about
characterizing
Continental,
scalable,
age,
Trends
and
csm2
webcam
6
climate.
One.
G
Okay,
I
will
try
to
make
a
shot
because
I
know
till
the
session
is
already
over
time.
Hi
everyone,
my
name,
is
Jin
Andrew
I
am
a
NOAA
climate
and
Global
change.
Post-Op
fellow
working
with
Arlene
at
the
MIT
and
my
talk
is,
is
again
always
related
talk,
but
I
focus
on
the
csm2
climate
model
instead
of
another
model
that
I've
been
mentioned.
G
The
measurements
of
the
Lonely
species,
such
as
special
Coral
foam,
on
the
other
hand,
there's
another
large
amount
of
of
the
oil
studies
that
focus
on
the
process
level
and
digging
to
the
very
integrated
detail
of
the
oh
chemistry.
In
order
to
get
a
better
understanding
of
the
chemical
mechanism
behind
our
bed,
but
for
these
two
groups
of
studies,
there's
apparently
a
big
gap
in
the
spatial
scales.
G
That's
why,
during
my
PhD
I,
moved
to
the
urban
scale
and
look
at
the
urban
relations
in
one
decade
in
49,
North,
American
cities
and
in
this
talk
during
my
postdoc
I,
moved
to
the
Continental
original
scale
and
want
to
see
how
the
Continental
steel,
oh
Trends,
look
like
and
what's
the
difference
of
that
seem
to
Arlington
serious
talk.
I
also
use
the
13-member
initial
sample
model
simulation
spending
between
1950
to
2014
generated
with
the
cesm2
climate
model.
Configured
with
working
six
chemical
mechanism
and
I
also
select
the
11
regions.
G
Spending
between
the
charcoal
region
to
the
northern
latitude
cities-
and
these
are
these
regions,
always
are
Overland
and
with
virulence,
population
very
dense
population
and
the
figure
on
the
top
left
just
shows
the
original
chance
from
the
cesm2
model
simulations
across
these
11
regions
and
the
solid
line
represents
the
Osama
mean,
and
the
change
represents
a
range
of
the
some
different
assembled
members
also
to
constitutely
describe
characterize
these
always
trends
that
also
calculates
the
iav
IV
here
denotes
the
inter-annual
variability
and
from
the
left
to
the
right.
G
I
ordered
these
regions
by
the
latitude.
So
you
can
see
that
there's
a
very
large
scattering
inside
IAB
across
these
elaborations,
the
highest
iub
of
about
14
percent,
is
found
in
the
charcoal
region
in
Malaysia,
but
regions
like
golf
or
guinea
show
a
very
low
IB,
which
is
about
like
two
percent.
G
Given
this
pattern,
we
really
want
to
identify
what
the
dominant
drivers
will
be
simulated
always
chance,
and
what
caused
this
Regional
variation,
and
also
on
top
of
that,
we
want
to
investigate
how
represent
how
representative
of
these
simulated
awaiting
the
csmg
model,
comparing
to
comparing
against
the
reality.
And
how
can
we
use
the
model
simulation
to
mimics?
The
react,
the
always
in
the
reality
to
address
these
two
questions.
I
use.
G
Machine
learning
model
to
include
interface
is
always
chance,
specifically
I
use
one
of
the
data
set
from
one
model
example,
as
the
training
data
set
to
train
a
machine
learning
model
called
gradient,
boost,
C3
model
and
use
the
remaining
12
Ensemble
members,
as
the
testing
data
set.
The
target
of
this
machine
learning
model
is
to
predict
this
monthly
mean
Mass
weighted
transfer
mean
oh
in
order
to
get
a
better
performance
of
this
machine
model.
G
I
carefully,
select
the
input,
features
and
I
end
up
having
input
features
listed
below
with
three
groups,
including
the
meteorological
drivers,
including
a
water
vapor
temperature
lighting
factories
and
cloud
cover
radiation
driver,
including
drill,
single
D
and
chemical
driver,
including
ozone,
Co
O2,
colon
and
format
High
column,
and
this
selection
of
those
input
features
are
based
on
our
understanding
of
the
oh
chemistry
that
I
show
again
on
on
the
right.
G
The
figure
here
just
shows
the
performance
of
this
machine
model
by
showing
the
comparison
between
machine
learning
predicted
always
versus
the
csm2oh.
From
the
testing
data
set,
and
we
see
that
this
Machinery
model
were
able
to
reproduce
the
orange
from
the
csm2
model,
with
a
relatively
relative
rooming
Square
Arrow
of
about
five
percent.
G
So
now
we
have
a
good
tool
which
is
machine
learning
to
mimics
the
CSM
to
model
simulations
of
the
oh.
How
can
we
interpret
the
oh
and
address
our
first
question
so
here
I
picked
two
regions,
as
example.
The
solid
line
represents
the
eastern
North
America
Origins,
and
the
dash
line
represent
the
oil
Trans
in
the
southeast
Asia,
and
these
two
lines
were
from
the
machine
learning
predictions
used
using
the
original
outputs
from
the
cdsm2
models.
G
So
it's
pretty
much
seeing
a
good
similarity
to
the
direct
cdsm
to
OS
simulations,
and
we
know
that
there's
a
bit.
There
is
already
a
big
radio
narration
across
these
two
regions
and
then
I
designed
the
first
scenario.
I
call
it
the
meteorological
driver,
fixed
scenario.
What
what
I
did
is
to
remove
the
inter-annual
variability
of
these
meteorological
drivers,
including
these
four
different
meteorological
parameters
that
I
put
in
the
input
drivers
construct.
G
This
input
features
with
this
fixed
meteorological
drivers,
plus
the
chemical
drivers
in
radiated
drivers
same
as
the
csm2
model
output,
get
a
new
OS
prediction
from
the
Machinery
model,
showing
the
RS
line,
comparing
between
the
blue
line
and
orange
line.
We
know
that
when
we
remove
the
internal
ability
of
the
meteorological
drivers
in
the
eastern
North
America,
it
was
able
to
reduce
the
increase
in
the
oh
Trends
after
1990.
G
However,
in
the
southeast
Asia,
the
difference
is
relatively
small
for
the
second
scenario,
I
keep
the
meteorological
drivers
the
same
as
the
csm2
model
output,
but
remove
the
inter-annual
variability
of
all
the
chemical
drivers.
I
call
it
chemical
driver,
face
scenario
and
construct
the
input
driver
input,
features
using
the
cat
fixed
chemical
drivers.
Again
guys
are
always
predictions
from
machine
learning
model
showing
the
green
line.
Comparing
group
between
blue
line
and
green
line.
G
We
can
pretty
much
see
that
when
we
fix
the
chemical
drivers,
the
inter-annual
variability
of
the
oil
Trends
were
gone
and
the
oil
change
at
these
two
regions
seem
to
converge,
and
this
observed
pattern
can
also
be
reflected
by
calculating
the
iav
here
and
here
I'm,
showing
the
IV
calculator
from
these
three
different
scenarios
over
these
two
regions
and
for
the
eastern
North
America.
We
know
that
for
either
fixing
the
meteorological
or
chemical
drivers
will
reduce
the
imv
by
by
50.
G
However,
in
the
southeast
Asia
Only
reduces
the
chemical
drivers
will
reduce
the
IAB
substantially
expanding
these
analysis
to
full
elabor
regions.
Here
are
our
takeaway.
The
first
takeaway
will
be
the
patterns
of
the
oh
Trends
in
tropical
regions
such
as
Malaysia
are
pretty
much
dominant
by
the
chemical
drivers,
but
in
regions
like
eastern
North
America.
We
know
that
the
oil
change
is
sensitive
to
both
meteorological
drivers
and
the
chemical
drivers.
G
But
if
we
look
at
the
green
bar
alone,
which
is
the
machine
prediction
with
chemical
drivers,
we
know
that
the
inter-annual
variability
of
the
chemical
drivers
pretty
much
dominates
the
original
variation
of
the
ohns.
That's
that
if
we
remove
the
internal
variability
of
the
chemical
drivers,
we
know
that
the
oh
Trans,
as
these
11
regions
will
be
very
similar
to
each
other.
G
Given
the
importance
of
the
chemical
drivers,
they
can
also
post
the
potential
for
us
to
estimates
the
always
chance
in
reality,
because
OS
measurements
can
be
very
scarce,
but
the
observations
of
these
chemical
drivers
are
available
in
recent
years.
For
instance,
we
have
the
satellite
observations
for
the
O2
colon
and
for
math
High
column
here
I'm,
showing
the
comparison
between
only
retrieved
or
O2,
colon
and
format
column
versus
the
csm2
simulations.
G
The
blue
box
represents
the
only
two
column,
Orange
Box
representative
format,
High
column,
the
largest
difference
are
observed
for
the
only
two
column
in
this
Northern
mid-latitude
regions,
with
the
high
bias
by
a
factor
of
1.5
for
the
format
High
column.
We
also
know
the
high
bias,
but
it's
relatively
smaller,
weighs
about
15
reason
wise.
We
know
that
larger
differences
found
in
the
northern
main
latitudes,
but
in
the
charcoal
regions
the
agreement
seems
to
be
better
for
another
two
chemical
drivers-
oh
and
seal.
G
We
can
obtain
the
measurements
from
the
yeah
ghost
air
aircraft
measurements
and
these
measurements
are
available
after
1995,
so
again,
I'm
comparing
the
CEO
and
ozone
between
these
eye
goes
observations
and
the
cdsm2
simulations.
The
green
box
represents
the
ozone
comparison
and
Redbox
represent
the
seal
comparison
so
compared
to
the
I
want
to
call
it
a
formatted
column.
We
know
that
the
rapid
difference
between
observations
and
models
for
sale
and
also
are
much
more
smaller,
but
also
windows
for
most
operations.
This
is
a
low
bias
by
about
20,
but
for
Co.
G
We
also
a
high
bias
High
bias
about
less
than
between
0
to
40
percent.
Here
is
a
very
quick
overview
of
how
the
cdsmg
model
is
representing
the
chemical
drivers.
Now
we
want
to
investigate
how
these
buyers
in
the
chemical
drivers
would
affect
the
oh
Trends
so
to
address
this
question.
I
get
this
observation
constraint
or
always
predictions.
It's
extensions.
G
First
of
all,
we
note
replacing
the
chemical
drivers
from
the
csmg
model
with
observations.
The
machine
learning
prediction
is
overall
lower
than
the
cdsm2
simulated.
Oh
in
most
regions
in
Europe,
North,
America,
Asia,
partially
Asia.
The
low
bias
will
be
about
five
percent
to
20
percent.
That
means
the
csm2
is
over
predicting
the
Oh
by
less
than
10,
by
about
10
percent
and
for
our
observation
constraint,
always
we
only
know
one
region
which
is
southeast
Asia
that
the
predicted
oh,
is
higher
than
the
cesm2.
By
about
five
percent.
G
I
want
to
wrap
up
my
talk
by
showing
all
these
conclusion,
as,
on
the
other
hand,
since
this
project
is
ongoing,
I
also
want
to
show
some
of
my
future
plans.
First
of
all,
we
want
to
look
into
the
contribution
of
the
individual
drivers
to
the
oh
Trans
in
each
region
and
for
the
eastern
North
America,
where
we
see
a
very
high
sensitivity
to
the
meteorological
drivers
of
for
the
ohms.
We
want
really
want
to
figure
out
why?
G
Where
does
this
High
sensitivity
come
from
and
it
will
potentially
incur
or
introduce
a
larger
influence
of
the
climate
variability
to
the
oh
Trans
in
this
region
and
at
last
we
only
show
the
observation
constrained,
oh
in
the
absolute
values
and
in
the
future.
We
also
want
to
investigate
how
the
only
chance,
always
chance
concerned
by
the
observations,
differ
from
the
cesm2
model
simulations.
A
Little
short
in
time,
we
have
to
go
back
but
yeah
this
one
here
again
really.
R
R
G
R
G
A
And
or
NO2
or
whatever,
and
so
the
question
is,
is
that
something
this
approach
could
maybe
you
know
we
could
use
this
approach
to
adopt
a
diagonal,
Cyclone
or
something.
G
It
could
be
possible,
but
in
these
projects,
specifically
where
the
target
of
this
machine
learning
model
is
the
monthly
mean?
Oh
so
there's
no,
there
are
no
circle
in
it
because
we
only
see
the
monthly,
oh,
but
I
agree
that
it
can
can
have
more
like
applications
of
this
tool.
A
Yeah,
certainly
something
this
further
discuss.
I
do
I
think
we
should
now
have
people
here
go
to
lunch,
because
we
have
to
raise
them
again
at
one
of
our
place,
everybody
and
all
the
speakers
one
more
time.