►
Description
The 26th Annual CESM Workshop will be a virtual workshop with a modified schedule on its already scheduled date. Specifically, the virtual Workshop will begin with a full-day schedule on 14 June 2021 with presentations on the state of the CESM; by the award recipients; and three invited speakers in the morning, followed by order 15-minute highlight and progress presentations from each of the CESM Working Groups (WG) in the afternoon.
On 15-17 June 2021, working groups and cross working groups have half-day sessions, some with presentations and some that are discussion only.
A
A
B
A
A
A
So
we
hope
that
everyone
will
please
follow
the
guidelines
for
constructive
and
thoughtful
interactions
and
and
throughout
the
presentations
and
discussion,
and
so
we've
for
this
session,
we're
going
to
have
a
series
of
four
or
five
ten
minute
talks
and
then
a
time
to
answer
questions
for
those
previous
talks.
A
So
we
encourage
people
to
type
their
questions
into
the
chat
during
the
talks
and
then
we'll
have
time
for
some
actual
answering
and
discussion
of
them
and
we'll
try
to
take
a
short
break
in
the
middle
of
the
morning
and
end
with
some
discussion.
And
then
we
have
some
networking
time
after
the
end
of
the
session,
we'll
have
some
breakout
rooms
and
which
we've
labeled
with
different
topics.
But
we
hope
everyone
will
stay
around
for
more
informal
discussions
afterwards.
A
C
I
was
trying
to
upload
my
there.
We
go
and
really
see
it.
There
we
go
and
now
it's
work
there
we
go,
and
I
can
just
remove
my
screen
here
anyway.
So
thank
you
so
much
for
the
introduction
lisa.
So
I
didn't
realize
that
I
thank
you
so
much
for
giving
me
the
3
30
pm
talk
which
is
8
a.m
there,
but
I
didn't
realize
that
there
was
a
discussion
at
9
10
and
I
have
to
do
a
school
pick
up
at
4
pm,
so
we'll
try
to
come
back
in
time.
C
C
This
is
based
on
the
university
of
sefiel,
and
what
I'm
going
to
talk
to
you
today
is
some
of
the
implementation
that
we
have
done
at
the
cln5
nitrogen
cycle
and
how
we
use
csm2
to
study
the
use
of
enhanced
road
weathering
with
croplands
to
mitigate
climate
change.
So
I
want
to
say
thank
you.
Some
people
that
have
been
involved
in
this
work
and
also
the
uk
research,
council
and
labor
human
trust
for
for
funding.
C
Okay,
let's
see
oh
there,
we
go
so
just
a
quick
introduction
about
what
is
this
enhanced
broadcast
well
to
to
meet
the
two
degrees
temperature
target
right
2100.
We
need
to
drastically
reduce
greenhouse
gas
emissions
or
fossil
fuel
emissions,
but
we
also
need
to
remove
carbon
dioxide
and
greenhouse
gases
from
the
atmosphere.
C
So
this
past
year
they
had
been
talking
about
cdr
strategies,
carbon
dioxide
strategies,
like
afforestation,
reforestation
bags,
but
enhanced
road
weathering
is
one
of
the
strategies
that
has
been
proposed
and
has
been
recognized
also
by
by
the
united
nations
that
can
help
remove
carbon
dioxide
from
the
atmosphere.
So
how
does
enhanced
real
weathering
work?
C
Well,
a
rain
dissolves
carbon
dioxide
from
the
atmosphere
and
bring
it
back
to
the
soil
and
there
it
reacts
with
the
silicate
that
are
for
a
part
of
the
rocks
like,
for
example,
basal
has
contained
silicates,
which
eventually
are
going
to
form
cations
and
bicarbonate
that
are
going
to
transfer
to
the
ocean,
and
this
carbon
dioxide
is
going
to
be
a
store
there
for
a
thousand
thousand
years.
So
this
is
not
something
new.
C
One
of
the
things
that
this
field
trials
have
been
finding
is
that
the
application
of
basal
treatment
reduces
n2o
emissions,
and
this
is
linked
to
an
increase
in
soil
ph.
So
what
I'm
showing
here
on
the
right
on
the
left
is
how
the
cumulative
and
two
fluxes
decrease
in
the
is
lower
in
the
in
in
the
basal
plot,
compared
to
the
control
plot
and
on
the
right.
C
So
this
is
a
simple
schematic
of
the
nitrogen
cycle
and
what
we?
What
I
want
to
highlight
here
is
that
anything
that
changes
the
soil
ph,
is
going
to
modify
the
rates
in
the
volatilization,
nutrification
and
denitrification
process
and
anything
that
changes.
Those
processes
are
going
to
affect
emissions
of
ammonia
and
nitric
oxide
and
entero
into
the
atmosphere,
and
that
will
have
important
consequences
for
atmospheric
composition
and
climate
so
to
er.
So
what
we
have
done
within
csm2
to
be
able
to.
C
I
study
the
impact
of
and
basal
applications
in
atmospheric
chemistry
and
climate
is.
We
have
implemented
a
few
schemes
within
the
nitrogen
cycle
in
clm5,
so
one
of
the
things
that
we
did
is
we
have
implemented
the
soil,
nitric
oxide,
and
this
is
based
on
descent.
C
We
have
also
implemented
the
soil
ammonium
volatile
volatility
scheme,
and
this
is
based
on
the
ndc
model
and
this
work
was
done
by
camin
phone,
and
this
is
actually
currently
under
review
in
biogeoscience
discussion,
and
we
have
also
modified
the
nitrification
and
denotification
in
cl5
to
link
it
to
soil
ph.
C
This
also
one
thing
that
I
wanted
to
highlight
is
that
we
have
a
bi-directional
coupling,
so
that
means
that
we
can
run
this
skin
with
the
nitrogen
deposition
from
clm
from
kamcam,
but
we
also
can
pass
the
ammonia
and
nitric
oxide
into
or
to
camp
cam
if
needed.
C
So
I
have
a
couple
of
slides
with
validation
of
the
model
I
have
here
on
the
left,
the
annual
and
two
of
fluxes
with
cropland,
and
we
can
see
that
we
have
a
good
special
distribution
of
these
emissions
on
the
right,
I'm
showing
the
seasonality
on
of
the
n2o
emissions
in
croplands,
in
major
crop
regions
and
compared
to
previous
estimates
from
emissions
inventories
and
our
previous
work.
C
We
can
see
that
we
have
a
good
seasonal
cycle
and
we
might
be
under
overestimating
a
slight
a
bit,
the
us
and
europe,
but
in
terms
of
absolute
values
we
are,
I
mean
cm5,
they
capture,
well
the
soil
and
two
emissions
over
crop
regions,
and
here
at
the
bottom
left.
I
have
a
comparison
of
global
and
two
emissions
with
previous
estimates
as
well,
and
we
see
that
we
are
within
a
range
of
previous
observations
and
estimates.
C
So
in
terms
of
soil,
nitric
oxide
emissions.
I
think
we
are
on
the
lower
end
from
previous
estimates.
I
have
compared
here
the
global,
no
emissions
with
previous
estimates,
and
we
have
here
in
red
cln5
and
I'm
comparing
it
with
hemko
and
luis.
I
mentioned
him
call
us
last
monday,
which
is
a
mission
inventory
that
is
going
to
be
available
for
csm2
if
it's
not
available
already,
but
in
terms
of
the
agricultural
you
know,
emissions
I
think
clm5
is
doing
a
pretty
good
job.
C
We
have
in
black
sclm
5
compared
to
other
emissions
inventory
and
it
captures
seasonality
and
pretty
much
of
the
crop
regions.
So
there
is
more
work
going
on
with
the
soil,
nitric
oxide
emissions,
and
this
is
done
by
anthony
wong
and
jeff
guedes
at
university
of
boston,
and
we
will
hear
a
little
bit
more
in
the
next
presentation.
C
C
So
so
what
we
have
done
is
like
this:
I'm
going
to
show
some
preliminary
results
where
I
have
run
a
cam
csm2.
So
I
use
cam4
with
mozart,
chemistry
coupled
with
clm5
and
at
one
degree
resolution,
and
I
have
done
a
10
year,
run
to
look
at
what
what
are
the
implications
of
changing
the
soil
ph
on
the
soil,
nitrogen
emissions
and
how
that
impacts.
Atmospheric
composition.
C
C
But
I
have
here
a
summary
of
the
n2o
in
blue
and
the
you
know
in
in
red,
and
you
see
that
actually
basal
treatment
reduces
soil
into
all
and
you
know
in
all
crop
regions
where
we
have
applied
basal
and
then
these
decreases
actually
are
significant
for
n2o
because
they
respond.
I
mean
they
correspond
to
about
20
percent
of
the
total
into
all
emissions
globally
from
from
kropland.
C
We
have
also
looked
at
what
are
the
impacts
on
summertime
ground
level,
ozone
and
I
have
focused
on
the
mda
8
air
quality
standard,
which
is
the
standard
for
air
quality
for
human
health
and
also
the
aot4t,
which
is
their
quality
standard
for
crop
health,
and
I
have
here
the
the
changes
in
summer
time
also
with
with
their
pasta
treatment.
With
respect
to
control.
I
mean
they
are
not
super
significant
changes,
but
they
can
be
substantial.
C
My
offset
some
of
the
increases
predicted
by
the
increase
in
ocean
projected
by
the
just
by
climate
by
2050
and
a
small
increase
in
aot40
amount
decrease
in
aot40,
my
increase
the
crop
yields
in
corn
and
mace
mace
and
soybean
in
the
u.s
by
two
four
four
percent-
and
this
is
my
my
last
slide.
So
what
we
are
going
to
just
highlight
is
that
we
have
a
flexible
framework
now
to
understand
the
chemistry
climate
feedbacks
resulting
from
enhanced
road
weathering
applications
in
cropland.
C
C
So
I
have
here
two
pictures.
One
is
like
to
show
you
how
simple
is
to
apply
basal
in
the
field,
so
it
follows
like
the
lime,
application
process
techniques,
and
then
I
have
a
picture
here
of
many
members
of
the
liver
hume
center
when
we
could
meet
back
in
2019,
so
we
still
have.
There
is
a
lot
of
people
behind
this
work
and
they
feel
trials
and
lab
experiments
and
great.
This
is
all
what
I
have
to
do.
Thank
you.
A
Thanks
maria,
so
if
you
have
any
questions
for
maria,
please
type
them
in
the
chat
and
discuss
them
later.
C
A
Yeah:
okay,
great
thanks
next
up,
is
anthony
wong.
Who
will
continue
our
discussion
of
soil
no
title.
Can
you
share
your
screen.
A
D
So,
okay,
so,
but
why
do
we
care
about
this
problem?
It
is
pretty.
It
is
because
soil
is
actually
a
very
large
source
of
react.
Atmospheric
reactive,
nitrogen
and
therefore
secondary
air
pollutants.
D
It
has
some
negative
health
impact,
but
arguably,
arguably
more
importantly
when
it
is
a
vital
component
for
atmospheric
for
abstract
photochemical
production
of
ozone,
which
is
another
important
air
pollutants
and
currently
non-agricultural
ammonia
emission
is
set
it's
sitting
at
seven
to
eight
teragram
nitrogen
per
year,
and
you
can
see
that
soil,
soil,
ammonia
emission
is
way
larger
than
that
number,
and
once
the
once
ammonia
get
emitted
into
the
atmosphere
it
will,
it
will
form
it
will
neutralize
the
acidic.
D
So
nox
is
not
it's
not.
The
nox
and
nitrogen
oxides
are
not
the
terminal
products,
but
they
are
the
byproducts
of
knox
of
soil,
nitrification
and
denitrification,
and
also-
and
in
addition,
ammonia
is
also
ammonia.
Emission
is
also
controlled
by
the
volatilization
of
ammonium
ion
because
it
is
highly
soluble
in
water
so
because
it
is
bio
geochemical
in
origin
we
can
already.
D
We
can
already
see
that
there
are
some
factors
that
would
affect
soil
reactive,
nitrogen
emissions,
like
temperature,
soil,
moisture,
ph
salt.
How
much
did
how
much
nitrogen
are
there
in
soil
and
decomposition
rates,
and
things
like
that?
Apart
from
these
factors,
we
can
human
can
also
influence
soil,
reactive,
nitrogens
emissions
by
fertilizer
applications
and
immigrations
and
other
landmark
land
management
practices.
D
D
Would
that
mean
we
will
have
more
soil
reactive,
nitrogen
emissions
in
the
future?
Would
that
mean
some
extra
climate
penalty
on
air
quality
and
also
in
the
future
in
the
future
that
we
have
to
use
more
fertilizer
to
feed
more
people?
Would
that
mean
substantially
larger
soil,
reactive,
nitrogen
emissions
into
the
atmosphere?
D
D
So,
notably,
the
code
base
for
the
code
base
for
the
biogeochemistry
code
base
is
entirely
the
same
as
mariasco
base,
but
I
have
implemented
the
rain
housing
effect
of
of
soil
knox,
and
once
we
finish
once
we
finish
the
once,
we
finish
by
using
clm5
bgc
to
model
the
solar
reactive,
nitrogen
in
the
future.
We
will
apply
them
into
chem
cam,
with
mosaic
to
simulate
both
like
the
changes
in
gas
and
aerosol,
trace,
trace
and
trace
composite
trace
compositions.
D
And
here
is
a
present
day:
clm
soil,
reactive,
nitrogen
emissions.
So
so
the
the
number
is
different
from
maria's,
because
I
have
implemented
the
rain
pulse
and
I
have
I
have
tuned
some.
I
have
tuned
the
nitrification
rates
over
cropland,
so
it
is
higher
from
maria's
presentation
and
now
it
is
noxian
system
that
is
sitting
at
around
eight
point.
D
Three
telegram
nitrogen
per
year
and
ammonia
is
sitting
at
around
twenty
one
telegram
nitrogen
per
year,
which
is
pretty
close
to
other
estimates,
but
for
the
spatial
distribution
we
still
see
quite
a
lot
of
like
knocks
over
the
high
latitudes
and
maybe
they're
too
little
in
too
little
knocks
somewhere.
There
is
somewhere
there.
There
is
safe.
D
So
maybe
we
are
overestimating
the
the
ammonia
emissions
over
natural
ecosystem
a
bit,
and
certainly
we
will
have
to
do
more
system,
weather
evaluations,
and
these
are
the
results.
These
are.
Some
very
preliminary
runs
on
how
climate
change
may
affect
climate
and
land
use.
Change
may
affect
soil
reactive,
nitrogen
emissions
bear
in
mind
that
our
spin
up
procedures
and
the
code
base
is
a
little
bit
different.
So
the
result,
the
result
is
not
directly
comparable
to
the
last
slide.
D
Here
we
can
see
quite
we
can
actually
see
quite
large
changes
in
soil
in
both
like
soil,
nox
and
ammonia,
emissions
and
and
the
effect
of
land
use,
change
and
climates
are,
are
very
different
for
different
regions.
For
example,
you
can
see
that
climate
has
little
effect
over
the
sahel,
but
land
use
change
has
has
much
larger
effects,
but
over
over
say.
D
Eastern
u.s
climate
has
a
large
effect
on
on
on
knox,
but
the
land
use
is
kind
of
kind
of
offset
that
a
little
bit
so
here
is
because
that's
all
I
can
share
because,
like
the
project
is
still
in
a
very
preliminary
seminary
stage,
but
here
is
the
to-do
list
sort
of
we
want
to
compete.
D
We
want
to
compare
ammonia
and
lox
emissions
like
inventories
in
a
more
in
more
spatial
and
vigorous
manner,
and
we
are
working
with
the
cln
people
to
try
to
diagnose
and
solve
the
spurious
lux
emissions
in
high
latitudes
and
then
like
once
we
get
the
soil-reactive
nitrogen
emissions.
The
task
may
not
be
as
simple
as
just
to
replace
may
not
be
as
simple
as
to
just
replace
the
agricultural
emission
inventories
in
chemcam,
but
we
also
have
to
look
at
the
definitions
of
like
what
is
what
precisely
are
the
agricultural
emissions
in
camcam?
D
That's
it
include
like
natural
emissions
or
or
does
it
include
like
other
sectors
of
of
agriculture,
and
then
we
and
then
perhaps
like
perhaps
another
way
to
look
at
whether
our
soil-reactive
nitrogen
emissions
estimate
makes
sense,
is
to
evaluate
our
camcam
simulations
with
service
concentration
and
satellite
products,
and
we
believe
that
it
is
not.
It
will
not
be
the
first
time
that
we
use
this
framework
use
similar
framework
to
study
the
linkage
between
terrestrial
ecosystem
and
air
quality.
We
believe
that
there
are
a
lot
of
potential
in
this
area.
Thank
you.
A
Please
listeners
type
any
questions
into
the
chat
and
we'll
address
them
in
a
little
bit
and
so
we'll
move
on
to
our
next
speaker,
aditya
nalam
from
iis
s
in
potsdam.
E
Welcome
thanks
lisa.
Can
you
hold
yumi.
F
E
Awesome,
yes,
I
also
started
sharing
my
stream,
so
thanks
for
giving
me
this
opportunity
to
present
my
ongoing
work
on
the
source
attribution
of
tropospheric
ozone
in
cesm.
A
E
So
yeah,
I
will
be
presenting
our
ongoing
work
on
source
attribution
of
tropospheric
ozone
in
csm,
1.2,
cam,
4
cam
for
the
2000
to
2018
period
and
first
I'll
start
off
with
the
source
attribution
scheme
which
has
been
published
in
gmd
in
2018.
E
It's
called
toast
that
is,
preservatives
attribution
of
sources
with
tagging.
So
I
would
just
refer
to
this
as
tagging.
From
now
on,
I
will
present
you
some
important
important
previous
results
using
this
particular
scheme,
and
then
I
will
present
to
you
a
layout
of
the
planned
simulations
for
this
transient
period
between
2000
and
2018,
and
some
preliminary
validation
results
so
far,
all
right.
So
a
brief
schematic
of
what
this
tagging
really
is.
E
E
So
but
when
we
would
like
to
know
which
source
is
responsible
for
a
particular
molecule
of
ozone,
for
example,
then
we
can
label
the
sources
based
on
its
emission
sector
or
emission
region,
and
that
label
is
carried
on
to
the
output
ozone
molecule
and
in
that
way
the
ozone
is
attributed
to
its
source.
E
So,
as
a
simple
illustration,
we
have
a
nox
tagging
simulation
where
the
emitted
no
or
emitted
nox
is
labeled
with
its
sources
and
a
corresponding
ozone
molecule
is
labeled
with
its
source
and
also,
of
course,
stratospheric
influx
and
lighting.
Lightning.
Nox
are
also
correspondingly
labeled,
with
labeled
to
the
particular
ozone
molecule
in
the
simulation.
Similarly,
for
the
voc
tagging,
the
voc
molecules
are
labeled
to
to
their
our
sources
and
their
corresponding
ozone.
E
The
corresponding
resulting
ozone
is
labeled
with
the
with
the
tag
or
the
label
on
the
voc
emitted
voc
molecule,
and
it's
not
just
ozone,
but
also
all
of
its
intermediate
species
that
are
that
are
participating
in
the
reaction.
Sequence
are
carry
the
same
tag
so
for
a
further
illustration
of
this,
I
would
refer
to
for,
for
I'm
going
to
say
for
more
information.
E
E
My
group
has
also
published
another
paper
in
acp
in
2020,
where
I've
used,
2010
edge,
tap
emissions,
two
model
runs
were
performed,
that
is
knox
tag
and
blc
tight
run,
and
so
here,
okay-
and
we
also
use
the
2010
specified
dynamics,
configuration
and
what
what
I
would
like
to
illustrate
here
is
how
this
tagged
resulting
tagged.
Ozone
looks
like
so
here
you
can
see
in
the
not
stacking
simulation.
E
There's
this
total
ozone,
and
there
are
these
tagged
components
of
ozone,
so
ozone
attributed
to
lightning
is
shown
in
this
second
panel
and
ozone
attributed
to
emissions
from
ocean
oceanic
nox
sources
like
shipping
are
shown
here,
so
the
ozone
resulting
from
it
is
predominantly
over
ocean
regions,
but
it's
also
spread
out
across
into
the
land
regions,
which
is
pretty
clearly
illustrated
over
here.
Similarly,
anthropogenic
emissions
from
north
america
anthropogenic
emissions
from
europe
and
anthropogenic
emission
anthropogenic
knox
emissions,
east
asia
are
also
tagged
and
the
resulting
tagged.
E
E
And
these
regions
are
defined
in
the
htap
ensemble
and
these
regions
have
been
tagged
in
our
tagging
system
and
we
can
have
resultant
resulting
outputs
like
this,
for
instance.
Similarly,
for
voc
tagging,
we
have
the
total
surface
ozone,
which
is,
of
course,
the
same
as
the
tool.
E
Surface,
ozone
and
the
most
similar
knots
tag
simulation
and
most
of
the
ozone
from
this
is
can
be
attributed
to
background
methane
and
biogenic
emissions,
whereas
the
anthropogenic
tags
contribute
very
less
because
the
contribution
of
anthropogenic
voc
emissions
is
much
smaller
in
the
global
scale.
So
that's
why
you
can
see
such
a
pattern.
E
So
this
is
just
a
brief
illustration
of
how
this
tagged
tagging
system
works
and
how
it
can
be
applied.
Now
I
will
move
on
to
my
ongoing
work
on
on
simulating
this
tropospheric
ozone,
with
tagging
for
the
historical
period
of
between
20
to
2000
to
2018.
E
Previously
it
was
only
done
for
the
year
2010
and
now
we're
going
to
be
extending
it
to
this
historical
transient
simulation,
a
brief
schematic
of
all
of
the
settings
that
are
used
in
this
study.
So,
firstly,
I
will
be
using
it
two
degree.
Two
degree
resolution
model
56
vertical
levels,
which
is
the
standard
configuration
and
it's
simulated
for
the
year,
2000
to
2010.,
separate
simulations
are
performed
for
knox
and
voc.
This
is
all
this
is
all
done.
E
Also
in
the
previous
papers,
the
transient
emissions
that
is,
emissions
from
the
year
2000
to
2018,
are
derived
from
cams
inventory,
that
is
copenhague's
air
monitoring
service
inventory
and
the
number
of
tanks
that
are
that
are
being
implemented
in
this
study
include
regional
tax.
That
is,
the
source
regions
that
you've
seen
in
the
previous
slide
and
also
tagging
the
emission
control
areas
for
shipping,
which
I
will
elaborate
more
in
the
next
slide.
E
Then
the
stratospheric
boundary
conditions
are
also
relaxed
to
the
annual
values
here,
as
you
can
see
here,
and
these
annual
values
have
been
taken
from
one
of
the
cmf6
ensemble
members,
that
is
cesm2
vacuum
and
we
used
the
mera2d
analysis
data
for
the
2000
to
2018
period
because
it
had
the
consistent
version
of
free
analysis.
Data
set
another
setting
that
we
had
to
use
is
the
relaxation
of
nudging
and
make
it
more
towards
a
free
running
case.
E
This
was
based
on
a
a
suggestion
from
louisa
or
when
we
were
having
problems
with
this,
and
I
will
elaborate
more
in
the
next
in
the
next
slide.
So
what.
E
Okay,
so
what
we've
done
here
is
that
we
compared
the
simulated
and
observed
ozone
with
relaxation
and
without
relaxation
and
we've
seen
that
this
purple
line
that
is
met
metrolax
of
0.2
is
within
the
h,
stack,
ensemble
range,
and
it's
also
closer
to
this
black
line,
which
is
the
observations
I
will
skip
the
next
slide
because
it's
I'm
running
out
of
time.
E
E
As
you
can
see,
these
pink
regions
here,
which
are
the
regions
really
close
to
the
coastline,
and
this
can
give
us
further
insight
about
how
the
near
near
coast
emissions
are
definitely
affecting
the
air
quality
when
compared
to
the
open,
open
ocean,
nox
emissions,
whereas
that's
not
the
case
for
vocs,
because
not
much
voc
emissions
is
happening
directly
in
the
open
ocean.
So,
yes,.
E
Is
this
is
essentially
it
so
this
modified
chemical
mechanism
has
been
implemented,
and
I've
also
illustrated
some
of
these
aspects
and
then
I've
also.
E
I've
also
illustrated
how
the
I
I've
also
summarized
our
settings
for
the
ongoing
simulations
and
also
explained
the
inclusion
of
emission
control
area
attacks.
So,
yes,
this
is
the
end
of
it
and
thanks
a
lot
for
this
opportunity
and
questions
and
comments
and
suggestions
are
welcome.
Thanks
a
lot.
A
Great
thank
you,
nice
to
see
all
this
work
continuing
and
when
those
up
next
improving
fire
representation
in
music
of
version.
Zero
thanks
go.
A
I
Can
you
see
my
screen
and
hear
me
okay,.
I
I'll
just
start
hello,
everyone
today,
I
will
share
some
preliminary
results
from
our
recent
work
on
improving
file
representation
in
the
musical
version.
Zero
model
we
know
plum
dries
and
the
diagonal
cycle
of
fires
are
two
important
features.
They
are
both
important
because
they
impact
the
fire
emission
transport
lifetime
chemistry
and
also
their
impacts.
I
For
example,
we
could
apply
a
single
diagonal
cycle
to
all
files
and
we
could
also
calculate
file
download
cycle
using
satellite
flp
data
and
in
this,
in
this
way,
we
every
file
would
have
its
own
dyno
cycle.
But
in
this
presentation
we
will
only
focus
on
this
one
simple
approach
which,
which
is
applying
the
same
characteristic
dyno
cycle
to
all
files,
and
there
are
also
different
ways
to
apply
chrome
injection
to
file
emissions.
I
For,
for
example,
we
can.
We
can
use
climatology
from
injection
profiles,
as
shown
on
on
this
slide
to
our
files,
and
we
could
also
calculate
plume
rise
within
the
model
online
for
individual
files
and
as
the
first
step
of
this
work
in
this
presentation,
we
will
only
focus
on
applying
this
climatology,
plum,
rice,
plume
injection
profile
to
our
files,
and
I
just
realized.
My
presentation
is
the
first
one
in
this
session
that
involves
music,
although
there
will
be
a
great
ones
afterwards.
I
So
here
is
a
very
brief
introduction
to
them
to
the
model
so
nordica
version
0
is
a
regionally
refined
version
of
chemcam,
and
this
grid
is
with
the
refinement
over
the
corners
region,
and
the
resolution
outside
is
about
1
degree,
and
the
resolution
within
is
about
14
kilometer.
I
I
The
control
room,
which
does
not
have
any
diagonal
cycle
or
plume
rise
and
a
sensitivity
wrong
with
only
a
diagonal
cycle
and
a
sensitivity
wrong
with
only
plume
rice
and
the
last
layer
of
sensitivity
wrong
with
both
polymerase
and
diagonal
cycle
and
to
evaluate
the
model
results.
We
use
observations
during
the
5x
eq
field
campaign
that
was
conducted
in
2019
summer
so
far,
execute
provide
comprehensive
observations
to
investigate
the
impact
on
air
quality
and
climate
from
the
wealth
files
and
also
agricultural
files
across
the
continental
uis
and,
as
shown
by
this
figure
on
the
right.
I
And
we
also
looked
at
the
williams
flash
file
as
a
case
study.
This
william
flash
file
was
the
largest
file
sampled
during
the
file
execute.
I
So
if,
if
you
look
at
this
lower
panel,
the
model
grids
within
the
plume
are
shown,
along
with
the
zero
concentrations
and
wounded
field
at
the
motor
surface
layer
and
the
location
of
the
one-minute
merged.
Observations
from
the
dc-8
aircraft
are
also
shown
by
these
black
lines
and
black
dots.
I
I
I
This
slide
shows
the
time
series
of
a
few
modeled
chemical
species,
so
the
modal
concentrations
of
these
species
are
averaged
over
the
whole
region
and
also
averaged
from
the
surface
to
the
500
millibar
and
the
the
black
line
represents
the
control
room
and
the
colored
lines
represents
the
sensitivity
lungs
and
the
dashed
line
is
shows
the
results
from
a
simulation
move.
I
The
classification
with
file
turned
off
just
added
here
as
a
reference,
so,
as
we
can
see,
the
the
simulation
with
only
plumed
rice,
which
is
the
red
line,
is
closer
to
the
control
room
comparing
to
the
other
two
runs
with
diagonal
cycle.
So
this
seemed
to
indicate
that
at
least
for
this
case
file
diagonal
cycle
has
a
stronger
impact
on
the
chemistry
than
the
progress
and
also
as
as
we
know,
atmospheric
background
oxidation
capacity.
I
I
So
even
if
the
daily
total
fail
emissions
are
the
same
for
all
these
simulations,
the
the
simulations
with
diurnal
cycle
has
stronger
emissions
of
primary
species
during
the
daytime,
and
that
corresponds
to
the
peak,
sunlight
and
photolysis
strength,
and
this
is
my
last
slide
and
showing
the
implications
and
the
next
steps.
One
thing
I
would
like
to
say
is
that
plumerize
and
the
diagonal
cycle
should
be
both
considered
together
simultaneously,
because
polymerase
is
strongly
influenced
by
thiodynamocycle
and
in
addition,
dinocycle
and
bronzerites
can
both
interact
with
meteorological
conditions
and
also
I'd.
A
Great,
thank
you
wenbu.
As
usual,
I've
done
a
poor
job
in
keeping
everyone
on
time
went
through.
That
was
fine,
but
so
we
will.
We
can
spend
a
couple
minutes
on
questions.
I
think
most
of
them
have
been
answered
in
the
chat,
but
I
wondered
if
stephanie
and
anthony
had
anything
more.
They
wanted
to
mention
about
the
nox
emissions
in
the
arctic,
because
that
certainly
was
a
striking
feature
of
those
simulations.
A
D
It
is
largely
like
it
is
largely
answered
there,
but
I
just
want
to
emphasize
the
second
point
that,
like
I
think,
like
a
lot
about
your
chemist
have
like
a
lot.
A
lot
about.
Your
chemist
have
have
said
that
like
have
said
that,
like
a
warmer,
a
warmer
climate
in
the
future
will
activate
the
activates
the
biogeochemical
cycle
in
the
arctics.
So
that's
that's
what
that's
why
it
is
even
more
important
for
us
to
get
a
reasonable
value
of
light
present
day
nox
emission
over
the
arctic,
so
that
we
can.
B
A
When
food,
do
you
want
to
say
anything
about
simone's
question
about
the
co,
smaller
co.
I
Yeah,
so
when
I
apply
dino
cycle
to
file
emissions,
I
make
sure
the
the
emissions
for
a
utc
day
is
is
conserved,
but
it's
not
conserved
for
a
local
day.
So
it's
it's
possible.
After
applying
the
diagonal
cycle,
the
emissions
actually
from
the
second
utc
day
was
moved
to
to
from
a
second
local
day
was
moved
to
this
to
this
day,
because
this
williams,
flash
fire,
is
not
a
one
day
fire.
It
continued
for
many
days.
A
Okay,
if
there
are
no
more
questions
at
the
moment,
please
feel
free
to
ask
anything
more
in
the
chat,
and
maybe
we
should
move
on
to
the
next
set
of
talks.
A
A
Five
minutes
on
questions
does
anyone
else
have
any
more
questions,
because
one
of
the
when
I
first
made
the
schedule,
I
had
55
talks
and
there
were,
and
I
didn't
properly
adjust
the
times.
So
people
are.
A
E
E
R2
reanalysis
has
been
a
feature
also
observed
in
cesm2
according
to
one
of
lewis's
emails
and
thanks
to
luisa,
because
she
really
helped
us
out
in
you
know
having
this
idea
about
relaxing
this
nudge
between
prescribed
methodology
and
the
meteorology
that
is
being
used,
and
when
the
nudging
is
relaxed
more
towards
the
free-running
case,
then
the
simulated
surface
ozone
also
reduces-
and
this
overestimation
also
reduces,
but
to
elaborate
more
on
the
physical
and
chemical
processes
that
that
are
responsible
for
this
again,
I'm
not
I'm
not
really
sure
how
this
how
this
works
in
the
model,
but
yeah.
E
E
A
Yeah,
it's
definitely
an
ongoing
issue
and
a
number
of
us
have
been
investigating
the
impacts
of
nudging
and
there
there
are
a
lot
of
options
for
how
to
actually
set
that
up.
So
that's
another
another
aspect
of
this
model
simulations
that
everybody
can
explore
and
try
to
help
understand
the
the
sensitivities
of
it.
A
K
F
F
F
F
So
the
goal
of
this
research
is
to
investigate
the
effects
of
this
grid
resolution
on
gas
and
aerosol
concentrations
and
chemistry,
especially
over
urban
regions,
as
you
already
know,
from
presentations
on
monday
and
yesterday
and
memphis
presentation
just
before
this
talk
musical
version,
zero,
which
was
released
in
the
latest
csm
version.
2.2
has
some
capability
to
use
spectral
elements
with
a
reasonably
climate
grid
like
this
over
corners
or
this
over
the
acting
region.
F
F
So
these
are
my
greece.
I
designed
for
the
course
sake
campaign.
I
made
two
global
greece
on
any
30
and
any
60
and
any
30
corresponds
to
1
degree
or
100
kilometer
resolution
and
any
60
corresponds
to
0.5
degree
or
50
kilometer
resolution,
and
there
are
two
reasonable
refinement.
Grease
correspond
to
14
kilometer
and
seven
kilometer
over
south
korea,
and
this
slide
shows
the
same
grease
for
asia
on
me
and
here
for
south
korea.
So
you
can
see
this
very
fine
scale
structure
here
in
any
30
by
16
green.
F
Let
me
show
you
some
preliminary
results.
I
have
from
this
musical
version
zero
simulation,
so
here
model
dynamics,
physics,
chemistry
and
emissions
are
the
same
for
all
greece,
but
the
grid
resolution
is
the
only
thing
that
they
are
different.
So,
as
you
can
see
from
these
figures
for
black
carbon
concentrations,
simulation
results
are
quite
different.
F
Global
greece
failed
to
resolve
high
concentrations
over
urban
areas
in
china,
and
the
plot
is
still
showing
the
carbon
concentration,
but
over
korea
and
the
black
carbon
here
is
chemically
inert
in
the
model.
So
that
only
follows
the
emission
pattern
and
global
coarse
grids
failed
to
resolve
high
concentrations
over
solar
metropolitan
area.
F
But
these
two
original
refinement
crease
shows
some
high
pollution.
Oversold
area,
and
especially
this
any
30
by
16
model
captures
fine
structure
of
cities
in
korea,
and
this
plot
is
for
ozone
concentration
and
again
the
fine
grit
model.
Simulates.
Some
fine
scale
features
like
some
knox
titration,
so
ozone
concentration
is
very
low,
but
the
course
model
failed
to
simulate
max
titration
effect
and
simulate
high
ozone
concentration
over
urban
area,
and
this
slide
shows
the
model
evaluation
against
aircraft
measurements
during
procedure
campaign.
F
So
you
can
just
see
the
spread
model
values
here,
especially
for
carbon
monoxide
concentrations,
are
very
similar
at
the
surface,
but
they
are
different
in
the
upper
troposphere,
like
a
factor
of
two
differences
between
the
model
grid
settings
and
generally
fine
grade
resolution
shows
higher
co
mixing
ratio
and
for
sulfur
dioxide.
They
are
different
at
the
surface,
but
very
similar
in
the
troposphere.
F
So
the
effect
of
grid
resolution
on
different
chemical
species
are
different,
and
here
I
show
you
the
same
evaluation
plot
against
force,
aq
aircraft
measurement,
but
for
aerosols,
so
for
regaining
aerosol
like
carbon
nitrate
and
surface.
So
you
can
find
that
bridge
resolution
is
important
for
some
species,
but
not
for
all.
So,
for
example,
here
or
regaining
aerosol
concentrations
are
similar
across
different
critical
resolutions
and
I
think
for
organic
aerosol
formation,
because
we
need
more
oxidation
time
in
the
rear
atmosphere
than
other
aerosols
like
black
carbon.
F
F
Actually,
I
think
this
is
a
widely
known
problem
for
chemistry
models,
so
usually
chemistry,
models,
simulate
black
carbon
and
surface
very
well,
but
underestimate
regain
years
of
considerations
and
mothers
overestimate
or
understand
nitrate
aerosols.
F
So,
in
the
next
step
we
are
going
to
investigate
the
region
further.
On
the
estimation
of
organelles
and
nitrate
during
the
coral
eq
campaign,.
F
So,
in
addition
to
spatial
resolution,
I
showed
you
earlier.
Another
thing
we
are
developing
right
now
in
music
is
to
investigate
the
effects
of
temporal
resolution
on
chemistry
by
applying
dyno
cycle
for
emissions.
E
L
F
Okay
thanks,
we
found
that
average
ozone
was
increased
over
the
u.s
europe
and
asia.
Continental
regions,
in
terms
of
mean
ocean
concentration,
and
this
result
is
for
may
june.
2016
error
is
frozen
here
and
the
average
ozone
increase
or
in
the
previous
slide
is
due
to
actually
night
time
was
an
increase
in
the
model.
F
F
So
here
is
the
summary
of
this
token
ongoing
work
plan,
so
we
constructed
new
model
grids
using
musical
version
zero
and
could
investigate
the
effects
of
horizontal
resolution
or
simulations
especially
over
korea,
where
there
are
small
urban
areas
with
high
population
and
I'm
happy
to
take
any
questions,
of
course
in
the
chat.
Thank
you.
K
All
right
thanks
very
much
to
song,
there's
some
questions
in
the
chat
and
if
they
don't
get
answered,
we
can
talk
about
them
in
the
in
the
question
and
discussion
session.
M
Okay,
can
you
guys
see
everything?
Okay,
okay,
great
thanks
dusong
for
the
the
really
nice
talk,
I'm
gonna
talk
to
you
now
about
some
experience.
Experiments
that
myself
and
becky
schwantes,
along
with
all
the
other
co-authors,
have
been
running
again
using
music
aversion,
zero
to
estimate
how
model
assumptions
impact
both
concentrations
of
health,
relevant
pollutants
and
their
health
outcomes,
and
as
kind
of
an
introduction
to
this,
I
know
that
music
has
been
introduced
quite
a
bit.
But
why
specifically
focus
on
health?
M
M
So
musica
version
zero
again
was
released
with
csm
2.2
and
at
finer
resolution.
The
specific
case
that
we're
using
is
a
one
degree
global
grid
with
one
eighth
degree
refinement
over
conus,
and
the
assumption
is
that
we're
going
to
better
resolve
emissions
and
chemistry
will
be
more
accurately
represented
and
we'll
also
be
able
to
calculate
pollutants
at
human
exposure.
Relevant
scales,
which
is
around
the
10
kilometers
or
so.
M
So
the
framework
for
these
experiments
is
that
we
wanted
to
test
two
things
particular
with
this
setup,
so
we
wanted
to
test
the
impact
of
horizontal
scale,
so
we
ran
with
both
a
a
one
degree,
global
grid
or
the
ne30,
and
then
also
with
the
aforementioned
one
degree,
with
one-eighth
degree
magnification
over
conus,
which
is
the
ne0
by
30
by
8
conus
grid.
M
Additionally,
we
wanted
to
test
chemical
complexity
and
see
how
that
impacted
health,
relevant
pollutants
and
their
concentrations.
So
we
ran
with
the
mozart
ts1
mechanism,
which
is
the
base
camcam
chemical
mechanism,
and
additionally,
we
ran
with
the
mozart
ts,
2.1
mechanism,
which
improved
isoprene
and
terpene
chemistry,
in
addition
to
adding
both
a
high
and
low
nox
pathway
for
soa
formation
and
we'll
see
why
that's
important
when
I
get
to
the
results.
M
And
I
will
mention
that
some
of
this
work
was
presented
yesterday,
so
go
back
to
my
talk
yesterday
for
a
little
bit
more
complete
model,
evaluation
or
model
description,
but
I
did
want
to
show
two
things
here.
M
One
is
that
we
did
evaluate
the
model,
both
with
respect
to
flight
campaigns
and
with
respect
to
surface
observations
and
for
the
work
presented
here.
We
did
make
one
change
specifically
on
how
we're
evaluating
the
model
or
how
we're
using
the
model
output
for
health
relevant
metrics,
but
this
is
becky's
work.
Looking
at
the
the
five
field
campaigns
over
the
us
and
as
you
can
see,
the
the
the
best
performing
model
is
the
regional
refined,
ts,
2.1
case
in
most
pollutants.
M
So
right
here,
we're
showing
no,
but
she's
done
this
analysis
for
most
ozone
precursors-
and
this
is
the
the
update
that
I
wanted
to
talk
about
with
the
results
that
we
had
from
this.
M
So,
what's
being
shown
here
is
the
model
first
level
ozone,
and
this
is
a
correlation
plot
with
all
epa
aqs
sites
and
the
one
on
the
left
is
showing
model
first
level
and
the
second
one
is
showing
a
surface
corrected
ozone
which
takes
in
the
the
surface,
roughness
and
deposition
velocity
to
as
a
correction
factor
to
reduce
the
ozone
concentrations
at
the
surface
versus
the
first
level,
and
this
is
kind
of
a
spatial
map
showing
this
is
for
june
july
august
average
concentrations.
M
So
here
I've
picked
out
two
of
the
epa
regions
and
looked
at
kind
of
the
daily
mda
concentrations,
both
with
the
correction
and
without
the
correction,
and
here
you
can
see
we're
falling
much
closer
to
the
observed
mda8,
which
is
the
blue
line
on
the
bottom
and
the
green
is
the
surface
corrected
and
then
the
top
musical
line
is
the
first
level
ozone.
So
this
surface
correction
has
really
brought
us
closer
to
the
observations
which
I
think
is
important
when
evaluating
this
as
a
health
relevant
metric
and
for
reference.
M
The
the
red
line
shown
here,
that's
what's
called
the
the
counter,
factual
and
I'll
get
into
that
in
the
next
slide,
but
below
that
level
is
the
point
that
we
can
say
that
we
do
not
have
a
statistical
impact
of
ozone
concentrations
on
human
health.
So
that's
that
level
is
at
26.7
for
some
of
the
calculations.
M
It
varies
depending
on
what
metric
you
use,
but
I
just
wanted
that
on
there
as
a
reference
to
show
that,
particularly
in
epa
region
8,
which
is
the
mountain
west
where
we
are,
we
are
consistently
above
the
the
health
relevant
metric
for
mda8
ozone.
M
So
the
methods
of
how
we
are
doing
these
health
mk
impact
calculations
is
using
what's
called
a
beta
function,
and
this
is
the
general
form
is
shown
here
where
you
have.
M
Some
change
in
mortality
is
a
function
of
some
incidence
rate
minus
a
beta
function
or
the
the
inverse
of
the
beta
function,
where
these
are
basically
calculated
as
the
the
hazard
ratio
or
the
kind
of
percent
increase,
or
increased
likelihood
that
you
will
have
of
a
specific
disease
outcome
based
on
studies
that
have
looked
at
hospitalizations
and
death
counts
and
then
created
the
statistical
relationship.
M
For
this
case,
we're
evaluating
both
ambient
pm,
2.5
and
ozone,
and
for
pm
2.5
we're
using
the
global
exposure
mortality
model
or
gem,
which
is
adapted
from
brunette
all
2018,
and
this
is
the
most
up-to-date
cohort
study-
includes
a
number
of
cohorts
not
just
in
the
u.s
and
europe,
but
also
in
china
and
for
ozone.
We're
using
a
beta
function
from
the
acs,
cps2
study
and
I'll
go
over
this
quickly.
M
But
the
the
baseline
mortality
rates
that
we're
getting
are
from
ben
map,
which
is
a
epa
model
used
for
health
impact
analysis
and
I've.
Also
looked
at
them
from
the
cdc
wonder
and
compared
how
the
different
formulations
of
those
work
and
then
adjust
the
the
final
two
three
slides
here,
I'll
try
and
go
through
quickly
for
the
sake
of
time,
but
this
is
kind
of
the
nuts
and
bolts
of
the
findings
that
we
have
here
and
probably
to
me,
and
I
think
to
the
other
authors
on
the
paper.
M
One
of
the
most
novel
findings
of
this
work
is
that
both
resolution
and
chemical
complexity
are
of
similar
importance
when
you're
looking
at
how
they
impact
human
health.
So
what
we
have
here
in
these
plots
is
the
attributable
desk
for
each
region.
Green
is
for
the
regional
refined,
blue
is
for
the
one
degree
model
and
the
hatched
is
for
the
more
complex
chemistry
case.
M
So
here
we
have.
Basically,
if
you
look
at
this,
this
difference
plot,
I
think
this
is
really
highlighting
what
I'm
trying
to
show.
Basically,
this
is
now
showing
the
shifts
between
those
and
green
and
blue
are
kind
of
of
equal
magnitude
here,
especially
when
you're
in
populated
regions
such
as
epa
region,
two,
which
is
new
york,
new
jersey
and
epa
region.
Eight
or
sorry
region,
nine,
which
is
includes
california.
So
that's
where
we're
seeing
the
largest
impacts
of
both
resolution
and
chemistry.
M
Ozone-
I
did
want
to
put
this
in
here
for
the
sake
of
being
complete,
we're
updating
this
analysis
based
on
some
discussions.
I've
had
with
the
health
community
in
general,
and
but
the
kind
of
the
the
one
takeaway
here
is
that
resolution
seems
to
be
more
important
when
it
comes
to
ozone
impacts,
not
as
much
chemistry,
and
you
can
kind
of
see
that,
if
you're
looking
at
concentrations,
the
largest
concentration
changes
are
due
to
ozone
and
not
as
much
with
chemistry
and
then.
M
Finally,
these
are
kind
of
the
the
total
number
of
attributable
deaths
for
each
of
the
different
runs,
and
here
you
can
see
for
pm.
2.5
we're
really
getting
similar
values
around
20
000
for
a
change
either
in
resolution
or
in
chemistry.
So
that's
really
highlights
how
similar
they
are
and
in
ozone
you're,
seeing
that
most
of
the
results
are
from
from
the
change
in
resolution.
M
K
N
N
So
csm2
has
historically
been
relying
on
the
mozart
family
of
chemical
mechanisms
to
perform
atmospheric
chemistry.
So
here
we
propose
to
add
a
new
chemistry
option
using
the
geoscam
chemistry
mechanism
that
stands
alongside
mozart.
N
So
in
terms
of
the
technical
description,
csm2gc
features
a
coupling
with
man4,
where
I
also
have
overlap
between
the
geoscan
chemistry
mechanism
and
mam,
such
as
black,
carbon
and
organic
matter,
are
essentially
passed
from
man
to
geoscale.
N
Geoscan
performs
its
chemistry
and
then
the
tendency
tendencies
are
then
passed
back
onto
map
for
other
houses
that
are
not
that,
but
do
not
overlap
between
the
two
models
they
they're
just
handed
off
to
geoscale,
and
here
I
want
to
point
out
that
we
use
the
geoscan
and,
with
the
complex
sorry
option,
to
facilitate
the
mapping
between
the
s-waste
pcs
in
geoscan
and
the
vbs
scheme
in
map
for
dry
deposition.
N
We
have
two
options
that
are
currently
implemented
and
sent
to
gc
the
first
one
is
the
same
as
what
mozart
is
doing
where
we,
where
it
uses
the
velocity
with
velocities
computed
by
clm
and
pass
through
the
copper
to.
Can
the
the
other
option
is
to
directly
pass
the
land
tabs
and
the
leaf
higher
industries
from
clm
to
cam,
and
then
to
use
the
geoscale
android
depletion
routines
to
compute
the
dry
deposition
velocities
for
weight
deposition.
We
rely
on
the
new
scheme
to
compute
the
washer
of
gaseous
species
and
geoscan
ourselves
for
photolysis.
N
So,
first
of
all,
whenever
possible,
I
choose
emissions
through
hamco
and
all
three
models
unless
we
have
a
better
option
in
any
simulation,
so,
for
instance,
in
csm2,
there's
an
online
option
to
compute
lightning
max
which
there's
not
in
the
geoscan
ctm,
so
that
in
that
case,
we
and
the
two
in
the
first
destinations
we're
only
using
online
lightning
lost
missions
or
biogenic
emissions,
for
instance,
and
all
three
simulations
also
use
specified
dynamics
using
the
r2.
The
resolution
is
approximately
identical
in
all
three
simulations
and
whenever
possible,
we
also
use
online
land
types.
N
I
compare
individual
processes,
so
I
provided
individual
processes
such
as
I
draw,
and
what
deposition
fluxes
of
ozone,
nitrogen
and
sulfur
that
I
compare
against
the
literature
and
also
compare
full
model
results
by
looking
at
surface
composition
of
ozone
against
crown
station
measurements
over
the
us
and
over
europe
also
compare
vertical
profiles
of
ozone
and
other
species
against
ozone,
climatology
and
aircraft
company
measurements.
And
finally,
I
also
look
at
common
totals
of
ozone
by
comparing
two
and
the
observations
of
paramount
mls
and
also
a
seal
column
with
the
dataset
from
mobile.
N
So
now,
let's
move
on
to
the
compare
to
the
intercomparison.
So
I
start
by
comparing
the
the
word
depression
tendencies
of
nitrogen,
as
of
as
through
three
the
three
models
so
from
top
to
bottom
csm2gc
camcam
and
the
gsm
ctm,
and
I
compared
those
results
to
observations
for
the
year.
2005
from
vettetal
and
all
models
provide
consistent
results
with
a
correction
coefficient
of
0.66,
but
we
know
that
this
is
a
slight
negative
bias
at
like
higher
what
deposition
fluxes.
N
Let's
say,
a
value
is
greater
than
five,
where
all
three
models
have
kind
of
negative
parts,
but
we
know
that
the
slope
from
the
regression
line
is
closer
to
the
parity
line
in
the
case
of
csm2
gc
for
sulfur.
For
the
software
with
the
position,
we
do
the
same
comparison
where
we
compare
all
three
models
to
the
same
rainwater
composition,
measurements
for
the
year
2005..
N
So
what
we
first
see
is
that
all
three
models
kind
of
predict,
the
same
trend,
but
that
this
trend
is
far
away
from
the
party
line,
and
so
the
model
results
were
obtained
for
the
year
2016,
but
whereas
the
measurements
were
done
for
2005-
and
I
think
this
is
this
discrepancy
is
explained
by
the
fact
that
there's
been
a
decrease.
N
That's
been
observed
in
literature
and
the
deposition
of
sulfur
at
a
rate
of
approximately
minus
one
kilogram
software
per
hectare
per
year
since
the
early
2000s
up
until
now,
and
that's
been
observed
in
a
number
of
studies
that
have
been
recently
published
when
I
look
at
column
column
totals
here.
We
compare
the
the
observations
from
omni
mls
for
total
column
to
the
biases
that
we
observe
in
csm2,
gc
and
chemcam.
N
So
I
find
that
there's
an
annual
bias
in
calm
ozone
of
approximately
minus
eight
thousand
units
for
csm2g
series.
This,
the
same
simulation
in
cancum
yields
a
bias
of
plus
6.5
dots
and
units
when
you
average
annually,
but
I
find
there's
also
strong
seasonal
variations
in
the
torque
on
calm
ozone,
where,
for
instance,
for
system
gc
values
between
minus
16
to
minus
six
thousand
units,
whereas
in
cancun
expanded
between
minus
three
and
plus
seven
thousand
units.
N
And
when
you
look
at
the
tropospheres
like
ozone
column
independently,
I
find
that
csmgc
underestimates
both
of
those
by
approximately
minus
two
and
minus
six
stops
in
units
respectively.
Whereas
camcam
is
doing,
has
a
larger
bias
in
magnitude
for
the
tropospheric
ozone
column.
But
does
slightly
better
in
the
stratosphere.
N
Now
I
compare
the
vertical
profiles
of
ozone
by
comparing
the
the
results
from
the
csm2
gc
and
chem
configurations
to
ozone
songs.
N
So
if
you
look,
for
instance,
at
the
first
column
in
the
tropics,
you
see
that
both
cam
and
geoscan
the
results
from
both
chemcam
and
gs,
camera
relatively
close
and
they
estimate
roughly
similar
ozone,
mixing
ratios
and
the
middle
column
and
the
mid-latitudes
in
the
troposphere.
You
see
that
geoscan
is
kind
of
lying,
underneath
the
ozone
climatology
and
underneath
the
resource
of
chemical
as
well,
but
it
does
a
bit.
N
N
I
can
also
compare
the
same
results
to
aircraft
campaigns
to
kind
of
have
an
idea
of
the
vertical
profiles
of
different
species
other
than
puzzle.
N
And
we
see
once
again
that
csmgc
in
red
underestimates
the
ozone
mixing
ratios
at
different
latitude
and
in
and
seasons
and
whereas
is
closer
to
the
observations
but
for
other
species.
And
we
see
that
both
models
are
relatively
consistent
and
quite
close
to
the
observations.
N
I
look
at
when
I
look
at
the
muppets
carbon
monoxide
comparison.
I
see,
for
instance,
in
csmtc
there's
a
smaller
nearly
average
ozone
bias.
That's
a
carbon
monoxide
bias,
but
both
models
still
understand
carbon
monoxide
and
northern
hemisphere,
but
this
magnitude
is
smaller
in
csm2gc
and
finally,
looking
at
surface
ozone.
N
Here
we
compared
the
resource
of
system
to
gc
camp
cam
and
the
gscan
ctm
to
ground
station
measurements
from
the
epa
and
the
european
environmental
agency,
and
I
find
that
generally,
a
system
to
gc
has
a
lower
estimate
of
surface
ozone.
Mixing
ratio
compared
to
the
other
two
models,
but
generally
agrees
quite
well
with
the
observations.
N
So,
for
instance,
if
you
look
at
the
correction
coefficient
in
north
america,
csm2gc
predicts
a
caution
condition
of
like
almost
0.5.
The
gsm
ctm
does
a
bit
better,
but
whatever
you're
obsessing
to
gc
performs
better
so
to
wrap
up
this
presentation.
We
included
the
geoscan
mechanism
within
csm2
and
we
provide
a
fair
comparison
with
identical
emissions
and
identical
meteorology
between
kamcam
and
the
g
and
geo
scan
both
embedded
within
csm2.
N
I
find
that
csm2
gc
predicts
a
lower
tropospheric
occlusal
mixing
ratio
than
camcam
towards
the
surface
and
also
allows
for
a
reduced
column.
Carbon
monoxide
bias
compared
to
camkin
and
finally,
ozone
and
no2
surface
mixing
ratios
from
csm2
gc
agree
reasonably
with
gram
versions.
So
if
you
have
any
questions
regarding
the
methodology
or
the
results,
just
let
me
go
during
the
discussion
session
or
just
type
it
in
the
chat.
Thank
you.
K
Thanks
tibo
yeah
there's
a
few
questions
in
the
chat
right
now
and
if
anyone
else
has
any
questions
feel
free
to
type
them
in
our
next
speakers,
ben
gober,
from
ncar
who's
going
to
talk
about
aerosol
data
assimilation
and
cesm.
B
O
Full
screen
now,
yes,
all
right,
I'm
gonna
show
some
very
preliminary
work
on
estimating
aod
to
constrain
aerosol
concentrations
and
emissions.
O
So,
basically,
when
we
do
regular
scenes.
O
O
So
basically
get
a
fire,
then
modis
is
able
to
detect
fire
counts
of
burnt
area
and
those
are
translated
into
emissions.
For
instance,
finn
is
using
the
burn
area
from
bodies
and
beers
and
provide
emissions
of
according
to
emission.
O
You
you
get
the
binary
and
convert
that
into
fluxes
for
the
different
chemicals
so
here
for
co,
and
you
can
see
very
strong
correlation
between
cobc
and
npom,
but
also
other
vocs,
and
so
the
basic
idea
is
then,
when
you
run
camcam,
you
get
aerosol
mass
concentrations
that
translate
into
aod,
and
then
you
compare
you
interpolate
the
aod
to
the
modi's
location
and
compare
with
this
observations
with
simulation
and
correct
the
concentrations,
but
also
emissions
related
to
a
permanent
bc.
O
So
I'm
running
the
csm
2.2
with
nudging
to
the
dot-com.
We
analyze
this
for
my
spin-up
and
then
I'm
using
the
regular
csm2.2
options
with
ts1
chemistry
and
mem4
and
vbs.
O
O
O
So
energy
parameters
are
the
the
strength
of
the
nudging
to
uv
and
t
so
we'll
be
15
hours
on
average,
more
or
less
as
you
that
15
so
a
couple
of
nine
hours
to
25
hours
nudging,
so
that
allows
to
first
perturbation
in
the
physical
parameters
but
then
later
on,
I'm
starting
the
assimilation
of
those
weather
parameters.
So
then
I
don't
need
to
nudge
anymore,
but
I'm
applying
perturbation
for
the
dust
and
sea
salt,
biomass,
burning
and
anthropogenic
that
are
completely
different.
O
So
I'm
purchasing
the
global
dust
factor,
global
sea
salt
factor
to
have
a
perturbation
in
the
respective
aerosols
and
all
of
the
gases
and
aerosol
direct
emissions
from
fenian
camps,
which
would
be
independent
from
the
biogenic
as
well,
so
that
hopefully,
the
ensemble
covariances
between
all
of
those
parameters
will
be
correlated
when
it's
needed
where
transport
errors
are
important
for
all
of
them,
but
then
independent
when
the
sources
are
different
to
be
able
to
differentiate
the
different
error
subtypes.
O
Because
the
problem
is
that
you
have
one
colon
observations
that
is
used
to
update
27
state
variables
actually
and
I'm
adding
two
emission
fluxes
to
that.
So
it's
largely
under
constrained.
So
you
want
to
be
able
to
have
an
ensemble
that
guides
you,
the
differences
between
sulfate
dust
sea
salt
and
the
different
sources
of
black
carbon
and
primary
organic
and
secondary
organic
as
well,
so
and
and
the
respective
numbers.
O
O
This
is
in
observation
space.
So
I'm
the
white
is
where
there
is
no
observations
over
the
10
days
and
you
can
see
at
this
point.
There
is
fires
in
oregon
that
shows
higher
aud.
They
were
predicted
by
finn
and
there
is
actually
a
larger
plume
shown
by
modis.
O
That
seems
to
be
due
to
the
same
fires
but
lower
aod
over
the
source
region,
at
least
seen
by
modis.
O
So
when
you
compare
the
two
so
observation
minus
model,
we
can
see
that
the
actual
source
region
was
overestimated
and
the
room
was
underestimated
and
after
simulation
with
the
beginning
here,
starting
from
the
same
initial
conditions,
some
of
that
biases
has
been
reduced
by
increasing
the
emissions
or
equating
the
reversals
close
to
the
source.
O
But
then
what
is
with
camcam?
You
can
look
at
aud
according
to
the
different
sources,
and
you
can
see
that
those
biomassbanding
sources
are
mainly
in
the
form
of
pc
bomb
and
soa.
That
contributes
a
lot
to
the
signal
of
the
total
aod.
O
While
the
subways
are
important
in
the
south
and
east,
together
with
the
soa
from
the
biogenic
sources
in
august.
O
So
now,
if
we
look
at
the
change
between
the
modis
assimilation
run
and
the
control
run,
you
can
see
that
those
differences
are
going
in
the
right
direction
where
bc
and
pom
are
updated,
but
in
in
one
direction
and
the
sulfates
are
digging
in
another
direction
and
dissolved
in
another
direction
too.
So
you
don't
increase
artificially
the
cell
fades
in
oregon
because
you're
seeing
a
fire.
O
The
next
10
days
of
simulation
is
getting
closer
to
the
fires,
we're
interesting
in
washington
state
and,
interestingly,
we
are
seeing
similar
behavior,
where
the
observations
over
the
fires
suggest
another
estimation
there,
but
then,
over
a
reagan,
the
bias
have
has
then
completely
disappeared,
which
is
probably
due
to
the
emissions
update,
which
is
a
good
good
thing,
and
then,
for
that
same
period
we
are
kind
of
saying
the
same
thing
where
the
the
plume
is
is
not
very
well
constrained
until
it
reaches
the
east
coast
in
this
case
and
an
overestimation
of
the
washington
source
that
is
developing
at
this
point.
O
So
first
modis
assimilation
with
very
standard
setup.
No
tuning
at
all
is
performing
reasonably
well
so
far,
and
the
mission
of
that
seems
to
be
an
important
way
to
improve
the
forecast
and
reducing
the
biases
over
the
source
region.
O
O
So
that's
something
we
need
to
take
into
account
and
make
sure
that
if
the
sources
are
not
very
well
defined
by
modis,
we
should
be
able
to
correct
that
downwind
because
otherwise
we're
gonna
our
emissions.
Estimation
is
gonna,
be
underestimated,
so
that
we're
also
gonna
add
the
muppet
ceo
to
be
to
be
able
to
constrain
bc
and
pom
further.
O
O
So
if
you
have
any
questions,
feel
free
to
email,
many
I'll
be
sticking
around
a
little
bit.
K
Thanks
all
right,
thank
you
ben.
Our
final
speaker
for
the
session
is
simone
who's,
going
to
talk
about
the
implementation
of
karma
and
csm2.
B
J
All
right,
thank
you,
and
so
I'm
going
to
talk
about
our
ongoing
work
on
the
integration
of
the
karma
model,
a
slice
resolving
aerosol
model
into
csm2,
and
this
work
is
funded
by
the
noaa
program
of
the
earth
radiation
budget,
and
I
want
to
acknowledge
my
co-authors,
mike
miltz,
when
she
andrew
princess
witch
and
chuck
benin
and
we're
working
with
ping
feu
as
well
on
this.
J
J
This
is
much
more
detailed
way
of
representing
aerosols
and,
and
so
that
will
be
an
opportunity
also
for
the
community
to
develop
further
other
models
in
this
version,
and
so
we're
hoping
that
this
work
will
eventually
be
released,
hopefully
getting
into
the
trunk
in
this
year
and
also
being
released
in
the
next
version.
J
So
another
thing
is
that
we
supporting
the
music
effort
and
integrating
different
aerosol
schemes
and
also
supporting
eventually
different
resolutions
and
dynamical
cores,
which
is
not
the
case
at
the
moment.
So
what
is
the
difference
between
a
modal
and
a
sectional
aerosol
model?
We
are
mostly
using
the
mam
scheme
in
csm.
You
can
see
that
on
the
left.
J
We
have
four
modes
currently
and
we'll
see
a
later
talk
of
fifth
mode
that
can
be
added,
but
it's
basically
mixed
aerosol
particles
in
each
of
these
modes,
and
since
it
has
the
mode
description,
it
is
cheaper
and
has
been
used
now,
essentially
for
climate
simulations,
but
it
is
has
shortcomings,
it
has
constraints
fixed
modes
and
especially
for
situations
like
high
errors
or
loads
or
other
things.
J
It
is
constrained
in
these
limits
and
it
may
not
be
realistic,
then,
for
the
sectional
model
on
the
right.
What
we
are
using
here
is
a
model
that,
as
I
said,
is
a
full
aerosol
description,
so
it
describes
two
groups.
J
One
is
a
mixed
aerosol
group,
which
contains
the
organic
and
black
carbon
organic
carbon,
black
carbon,
so
a
sulfate
salt
and
dust,
while
the
other
group
is
a
pure
sulfate
group
that
is
separate
and
both
of
these
groups
have
20
bins
with
a
slightly
different
size,
distribution
and
so
having
that
model
will
allow
us
a
better
representation
movement
between
the
bins
of
the
aerosols,
and
there
are
various
observations
that
can
be
used
to
compare
those
size
distributions.
J
So
we
we
expect
that
this
is
a
benchmark
model
also
in
terms
of
radiation
and
heterogeneous
chemistry
may
be
improved,
and
so
what
needs
to
be
done?
What
has
been
done
in
the
implementation?
J
So,
since
csm
aerosols
are
coupled
to
clouds
radiation
and
so
on,
we
have
to
do
this
also
for
karma,
and
we
try
to
stay
very
similar
to
the
framework
like
what
mem
is
using.
So
we're
trying
to
do
a
similar
approach
in
implementing
that
that
will
make
it
easier
later
to
combine
those
schemes
in
the
musical
version.
In
a
more
you
know,
more
modular
version
of
of
the
code,
so
we're
going
through
radiation
and
optics
karma
has
a
course.
Shell
approach
assumption
with
currently
lock-up
tables.
J
We
include
aerosol
cloud
interactions,
ice,
nucleation
cloud,
aerosol
interactions
in
terms
of
interstitial
to
cloud-borne
movement.
J
I
want
to
show
some
preliminary
results
because
right
now
we
are
already
emphasized
for
the
stratosphere,
and
what
you
can
see
here
is
the
simulation
of
the
mount
pinatubo
eruption,
which
is
a
really
nice
example,
because
you
go
from
very
low,
aerosol
volts
to
very
high
loading,
and
so
you
can
nicely
use
it
as
a
case
study
and
there
are
many
lines.
But
if
you
first
look
at
the
dashed
lines,
this
is
basically
the
injection
of
the
so2
burden.
That
is
the
injection
right.
J
After
the
mount
peanut
tubule,
you
see
that
strong
peak
and
then
it
is
declining
because
then
so2
is
forming
sulfate
in
the
stratosphere
and
so
say
well
we're
looking
at
the
blue
line,
which
is
the
main
model,
which
is
the
dashed
line
underneath
those
others.
They
have
the
same
amount,
it's
a
10-0,
sulfur
injection
and
then
the
blue
line.
The
main
model
shows
this
following
nicely:
the
observations
of
a
peak
and
then
a
decline,
and
these
are
like
high
resolution.
J
I
think,
greater
observations
of
sulfates
after
mount
pinatubo.
However,
if
we
run
karma,
the
first
attempt
was
the
same
10
teragrams.
We
see
much
lower
peak
and
then
we
realized
that
actually
earlier
model
studies
by
jason,
english,
he
has
excessively
used
karma
for
for
those
evaluation
of
mount
pinatubo.
J
Is
that
actually
the
more
realistic
or
the
more
measured
amount
of
injection
of
sulfur
after
mount
pineal
tubule
was
like
18
teragrams
of
sulfur
or
something
around
that?
So
we
we
increase
the
amount,
and
then
you
see
a
much
better
representation
at
least
of
the
tail
of
this,
and
there
are
always
large
uncertainties
in
those
observations
as
well.
J
Many
other
models
actually
show
a
very
similar
distribution,
so
so
karma
is
able
to
get
the
same,
sulfide
burn
but
needs
much
more
injection
than
what
meme
uses
and
that's
because
of
the
difference
in
how
meme
is
constrained
in
their
range
of
stigma
and
forming
larger
aerosols.
So
they,
the
aerosols,
don't
get
as
large
and
they
don't
fall
out
as
quickly.
So
this
is
an
aod
plot,
and
that
is
the
same
model
simulations
again.
J
The
blue
line
is
the
main
simulation
and
you
see
the
comparison
to
lidar
and
sage,
which
is
very
nice.
If
we
do
look
at
our
csn
camera
one
with
the
10
teragram
sulfur,
we
don't
reach
those
led
values.
However,
if
we
do
double
or
go
to
18
teragrams,
then
we
get
a
much
better
representation
and
there
is
also
the
study
from
jason
english,
as
I
mentioned
before,
that
had
a
very
similar
distribution
than
than
we
have
with
the
18
teragrams.
So
there's
the
the
point.
This
is
really
uncertain.
J
Actually
how
much
injection
you
need
to
get
the
same
result
for
after
mountain
peanut
tuba,
which
is
critical.
If
you
think
about
that,
for
instance,
for
geoengineering
studies,
we
want
to
estimate
how
much
sulfur
is
needed.
So
that's
one
thing
we're
going
to
look
in
the
future.
We
also.
E
J
For
two
more
minutes
or
yeah,
we
look
more
at
evaluation
in
terms
of
size
distribution.
There
are
some
nice
balloon
data
of
alarmy
when
shannon
made
this
plot,
comparing
a
size
distribution
before
the
eruption,
shortly
after
and
then
a
year
after
so
not
quite,
but
both
men,
which
is
the
lines
and
so
the
dots
of
the
balloon
observation.
This
is
number
concentration
with
regard
to
the
radius
and
the
sulfates.
J
The
pure
sulfates
are
in
blue
and
the
mixed
particles
are
in
red
both
actually
share
very
nice
agreement
with
the
balloon
data.
So
that's
one
way
to
look
at
these
data.
Another
way
is:
will
will
compare
to
atom
observation
for
other
periods
for
lower
loading,
aerosol
cases
and
so
on.
So
this
is
work
in
progress
and,
in
summary,
yeah.
We
are
implementing
this
this
version.
J
We
have
some
initial
results
and
still
have
to
evaluate
already
seeing
some
very
interesting
results
with
regard
to
mount
pinatuba
simulations,
and
one
thing
is
that
we
continue
to
adopt
in
the
future
a
new
transport
scheme,
a
new
dynamic,
of
course,
that
hopefully,
we
can
also
couple
to
the
musica
version,
zero
options
of
regional
refined
grids,
and
with
that
I
stopped
thank
you
for
the
attention.
K
Great
thanks
simone,
so
we've
got
some
time
for
questions
and
discussion
now
until
10
25.
So
if
you
want
to
stay
on
time,
maybe
five
six
minutes.
A
Yeah
and
then
we
can
take
a
10
minute
break
so
sorry
about
my
last
minute
changes
to
the
agenda.
I
left
in
some
typos.
That's
fine.
K
No,
I
didn't
catch
them
from
when
you
posted,
so
it's
everybody's
fault,
wow.
The
chat
has
a
lot
of
questions.
I'm
not
sure
how
many
have
or
have
not
been
answered.
K
Does
anyone
have
anything
that
they
want
to
discuss
from
the
chat
that
sort
of
need
some
more
open
discussion?
We
haven't
got
a
lot
on
multi-scale
modeling
here,
which
is
certainly
one
important
topic
for
the
week.
A
A
A
So
it
may
be
technically
possible
to
run
the
regionally
refined
grid
down
to
a
16th
of
a
degree
in
some
regions
that
not
all
the
physics
will
probably
work
well
and
so
we're
looking
forward
to
using
cam
with
empaths
to
handle
those
much
finer
resolutions.
K
This
is
somewhat
related,
so
I
know
the
numbers
for
wacom.
This
is
certainly
because
we've
got
questions
about
computational
expense
right
now,
but
what's
the
core
hour
cost
per
year
for
a
cam
when
you're
running
with
regional
refinement,.
M
M
But
I
think
that
for
a
couple
proposals,
we've
calculated
it
as
a
kind
of
a
per
grid
cell
indicator,
and
that
seems
to
scale
really
well
between
the
different
grids.
So
that
may
be
a
good
a
number
to
publish
on
the
wiki,
so
that
for
people
that
are
doing,
the
grids
can
have
a
rough
idea
of
what
the
computational
cost
is.
Yeah.
K
I
was
going
to
say
that
it's
the
it's
the
chemical
complexity
and
then
also
just
you
know
it's
not
just
going
to
the
finer
resolution
in
terms
of
spatial
scale.
You
really
have
to
drop
the
time
time
step
quite
a
bit,
so
there's
it's
very
hard
to
keep
track,
sometimes
of
all
these
different
configurations
and
what
they
cost
and
I'm
guessing.
If
you
do
like
a
quarter
degree,
you
don't
have
to
do
such
a
big
time
step
cut.
So
I
just
it's
good
to
get
those
numbers
out
there.
M
Hi,
nick
just
so
that
the
number
that
I
have
for
conus
with
just
cam,
not
cam
chem,
is
is
about
55
or
60.,
and
so
that
indicates
that
the
chemistry
is
is
doubling
the.
N
K
Yeah
and
definitely
to
echo
what
forest
said
if
you
save
a
lot
of
output
that
really
like
I
found
it
can
almost
like,
if
you
save
like
every
time
step,
because
you're
really
interested
in
that
kind
of
stuff,
could
almost
double
your
run
time
at
these
kind
of
resolutions.
So
it's
a
big
big
swing
in
core
hours
with
all
these
different
choices.
A
J
Yeah,
and
especially,
we
also
have
the
land
spin
up
files
for
that,
at
least
it
doesn't
have
to
be
consistent
for
the
land.
It
can
be
a
slightly
different
year,
at
least
that's
what
we
usually
do,
so
that
will
also
help
to
to
have
that,
so
the
users
only
have
to
use
an
interpolation
program
to
produce
their
initial
condition
files,
and
those
runs
are
by
the
way
done.
They
just
have
to
be
announced
and
still
put
on
the
website.
But
if
you
already
want
to
know
more
about
it,
you
can
email.
J
A
J
But
it
I
did
currently,
it
seems
to
be
about
three
times
as
much,
even
though
yeah
we're
having
120
more
affected
tracers
that
is
based
on
a
two-degree
kmk
model
version.
I'm
not
sure
how
that
changes
when
we
have
like
running
webcam
or
something
like
that,
but
I
don't
expect
that
it
would
be
different,
we'll
see.
J
Looking
at
that
yeah,
there
are
more
questions
so
karma
in
terms
of
prescribed
partition
and
fraction
to
different
size
spins
for
each
specie,
so
yeah
we
we
start
with
a
certain
mass
and
then
it
will
be
fractionally
increase
with
the
bin.
So
the
the
mass
bins
are
the
same.
It's
not
the
size
spins
that
that
we're
using,
but
it's
kind
of
mass
bins
that
describes
the
different
20
bins
and
maybe,
when
channel
might
can
clarify
your
chuck.
P
Yeah,
so
for
for
the
different
emissions,
we
use
the
error
net
observations
to
to
assume
the
size
distribution
and
then
we
use
that
size
distribution
to
to
do
the
prescribed
partitioning
of
the
of
the
emissions
into
different
size.
P
A
Good,
well
maybe
we
should
take
a
break,
let
everybody
get
up
and
walk
around
and
we'll
restart
at
10
35
and,
let's
see
ming.
Let
me
I
made
a
corrected
agenda
and
hopefully,
z
main,
can
finish
condensing
his
20-minute
presentation
in
ten
minutes.
A
A
H
H
H
Cool
okay,
sorry.
A
Q
I
can
share
this
session
great
thanks,
so
it's
10
35
here
in
colorado.
So
why
don't
we
restart?
And
the
first
speaker
of
this
final
section
of
talks
is
simeon
key
from
texas,
a
m
university
talking
about
the
impact
of
volcanic
eruptions
on
the
ice
nucleation
and
its
sensitivity
to
emission
scale,
and
you
have
10
minutes
with
some
buffer.
B
H
Okay,
thank
me
so
thank
you.
Thank
you
for
this
great
opportunity
to
share
my
recent
study
about
the
impact
impact
of
the
volcanic
eruptions
on
the
servers
cloud.
This
work
has
been
done
by
dr
shahoon
news
research
group,
and
we
want
to
thank
the
help
and
communication
from
simon,
luisa
and
daniel,
and
also
thank
the
supporters
from
the
doe
ngd
project.
H
H
H
H
H
So
the
map5
is
designed
to
solve
this
dust
of
bios.
The
reason
that
generators
are
extra
dust
in
the
stratosphere
is
due
to
the
remaining
process
between
the
accumulation
and
the
coarse
mode.
This
is
a
design
designed
to
promote
the
lifetime
of
the
surface
in
the
stratosphere,
particularly
after
the
volcano
eruptions.
H
However,
this
process
also
includes
the
dust
and
the
sea
salt
that
does
lifetime
enhanced
by
this
process
and
the
accumulation
of
the
dust
are
converted
into
the
crossing
of
the
dust
which
can
participate
in
the
surf
circular
service
cloud
ice.
Nucleation,
the
man5
added
another
mode
to
isolate
isolated
the
stratosphere
profit
to
treat
the
repeat
of
the
stratosphere
sulfate
in
the
separately
under
it
no
longer
exchange
the
dust
and
the
season
between
the
accumulation
and
the
coarse
mode
in
the
atmosphere.
H
Here
we
show
how
the
aerosol
would
impact
the
ice
nucleation
in
the
serous
cloud
in
current
6
eyes,
nucleation
skin
the
ice
nucleation
skin
is
based
on
you
and
the
panel
2005
and
the
due
at
o2
7..
This
parameterization
means
ice
nucleation
with
the
aerosol
number
temperature
and
the
updraft
velocity.
The
ak
mode
lumber
is
used
to
represent
the
sulfate.
H
It
can
move
the
lumbar
to
participate
in
the
homogeneous
location
and
the
cause
of
the
dust
number
is
used
to
present
as
a
dust
and
participate
in
the
heterogeneous
ice
nucleation,
also
the
accumulation
mode
and,
of
course,
with
the
surface
lumber
directly
added
into
the
active
ice
number
in
the
stratosphere.
Based
on
the
update
in
the
2017.
H
And
here
we
design
for
simulations,
so
the
band
4v
is
the
control
room
using
the
current
default.
Welcome
configuration
the
mem4
envy
is
the
c
as
same
as
memphis,
but
without
a
volcanic
emission.
The
man
5-way
is
using
man5
with
changes
in
the
ice
location
scheme.
The
mam5
envy
is
the
same
as
the
man
5-way,
but
without
a
mechanical
emission.
H
All
the
simulation
are
larger
to
mirror
2
uvt
and
conducted
by
the
wacom
6,
and
the
simulation
time
is
from
1990
to
the
1994.
The
biggest
volcano
eruption
during
this
period
is
a
mountain,
pinatable
eruption.
Mountain
pinatobo
is
located
at
the
philippines
and
erupted
on
the
june
15th
1991
and
injected
the
10
telogram
so2
into
the
stratosphere.
H
H
The
red
means
the
month
5
results,
while
the
blue
chooses
the
amount
4
results.
The
dashed
line
are
the
wrong
without
volcano
emission
under
the
sodium
line
of
the
wrong
ways:
volcano
emissions:
let's
look
at
the
dust
results.
The
first
rule.
First,
the
man
5
ways
and
without
volcano
emission
shows.
H
H
So
with
the
death
stars,
the
bias
in
the
lower
stratosphere
and
upper
troposphere,
we
can
evaluate
the
impact
over
the
volcano
user
options
on
the
servers
cloud
by
comparing
the
results
between
bank
5v
and
the
95
envy.
Okay.
Here
we
see
the
results
from
the
heterogeneous
ice
nucleation
first,
the
left
panel
shows
the
relative
difference
of
the
core
starts
the
longer
concentration
in
1992,
that
is,
between
band
5v
and
the
9.5
envy.
H
1992
is
the
year
after
pinot
eruption,
the
red
and
the
yellow
boxes,
means
mean
are
shown
in
the
middle
and
the
right
panel.
There
is
no
difference
between
the
two
sig
two
simulations
as
we
expected
it.
I
mean,
though
significant
difference,
and
then
therefore
we
see
no
clear
difference
in
the
heterogeneous
isolation
between
these
two
simulations,
as
there
is
no
does
the
difference.
H
H
And
we
also
see
the
time
series:
this
is
the
lower
stratosphere,
and
this
is
our
troposphere.
We
see
this.
I
can
move
the
lumber
for
the
wrong
ways
of
a
calorie
emissions
that
I
can
do
the
number
increase
initially,
but
several
months
later
it
decreased
sharply
and
the
lower
than
the
wrong
without
volcano
emissions,
and
this
this
decrease
will
last
for
years
unless
another
volcano
erupts
in
at
the
end
of
1994..
H
H
So
here
we
see
the
homogeneous
ice
lumber
relative
difference.
The
left
panel
displays
the
volcano
emission
enhanced
the
homogeneous
nucleation
in
1992
the
middle
panel,
and
the
right
panel
shows
the
enhanced
enhancement
last
four
years.
This
means
the
increment
of
the
sophie
aerosol
number
in
the
accumulation
and
the
course
mode
overcomes
the
lumbar
reduction
in
the
acre
mode
as
the
course
and
the
accumulation
mode
longer
have
a
higher
potential
to
participate
in
the
ice
nucleation.
H
Here
we
show
the
tidal
forcing
so
first.
First
we
show
the
table
is
the
mang5
v
by
the
summer,
five
envy.
That
is
a
long
wave
cloud
of
forcing
in
five
years,
and
here
we
should
use
the
minor
five
memphis
virus
among
four
envy,
that
also
long
wave
cloud
of
forcing
in
the
same
five
years
as
we
expected
in
1990.
H
So
here's
my
conclusions:
man5
improves
the
duster
bias
in
the
lower
stratosphere
and
upper
troposphere.
The
mountain
pinatable
eruption
increase
the
accumulation
and
cost
of
the
sophie
aerosol
number,
but
decreases
it
can
move
the
number
for
years
of
further
eruption
through
enhance
enhancing
the
homogeneous
ice
nucleation.
The
mountain
climate
eruption
promoted
serous
cloud
ice,
lumber
concentration
and
increased
the
long
wave
cloud
forcing
by
0.3
watt
per
square
meter
globally
for
two
to
three
years.
H
Q
All
right,
yeah,
thank
you
jimmy
and
thanks
for
last
minute.
Shortening
of
your
your
talk.
I
appreciate
that.
So,
let's
move
on
to
the
next
presentation,
which
will
be
given
by
zhu
from
ncar,
and
it's
persisting:
volcanic
ash
particles,
impact
stratospheric,
so2,
lifetime
and
aerosol
optical
properties.
P
Hello
everyone
today,
I
will
talk
about
the
impact
of
the
volcanic
eruptions
on
the
stratus
for
chemistry
and
aerosols.
I
changed
the
title
slightly,
but
the
content
is
the
same.
P
Okay,
so
volcanic
ash
is
usually
neglected
in
the
climate
simulation
because
ash
are
considered
large
high
density
particles
and
fall
fast
during
the
in
a
few.
In
a
few
days
or
weeks,.
P
P
So
those
drawers
those
motivate
us
to
look
into
some
volcanic
eruptions
with
ash
and
we
picked
the
2014
mount
kalut
eruption
of
the
this.
The
calypso
reported
by
the
veneer
2016
sees
that
the
volcanic
aerosol
layers
persist
more
than
a
month,
and
we
you
you
can
say
that
maybe
those
are
sulfate
particles,
because
sulfate
usually
persists
pretty
long
after
the
eruption.
P
However,
from
a
depolarization
channel
from
calypso,
it
sees
those
particles
are
actually
odd,
shaped,
not
round
like
sulfate,
so
those
are
actually
persistent
ash
particles.
Then
we
may
wonder
how
could
those
ash
particles
persist
so
long?
Is
it
because
they
are
very
small,
like
submicron
sized
sulfate,
and
we
have
some
very
valuable
data
from
aatrax
that
observed
per
super
micron?
Sized
particles
from
the
month
can
loot
eruption
a
month
after
the
eruption,
so
this
figure
shows
a
track.
P
However,
we
still
miss
a
lot
of
particles,
that's
larger,
that's
larger
than
five
microns,
and
then
we
switch
the
density
of
ash
particle
from
2.3
to
0.5,
and
actually
we
reproduce
the
size.
Distribution
very
well
and
those
particles
are
not
volcanic
glass
and
it's
like
pumice
with
parallel
structures.
P
So
to
answer
the
first
question,
so
those
persistent
ash
particles
are
actually
micro
sized
with
a
density
close
to
pumice
and
then
another
interesting
point
of
our
research
is
about
the
lifetime
of
so2.
So
our
the
the
modeling
community,
or
at
least
me,
we,
I
always
think
the
lifetime
of
so2,
is
about
a
month.
P
So
we
look
into
the
kalut
case
and
then
to
compare
the
observed
so2
lifetime
without
with
the
model.
The
black
line
here
shows
the
wacom
case
that
the
so2
can
react
with.
Oh
and
oh,
is
get
depleted
very
rapidly
in
the
plume,
because
lots
of
it
to
get
injected
the
lifetime
is
26
days.
However,
the
observation
says
that
lifetime
is
only
seven
days.
Why
is
they
are
so
different?
P
So
we
we
subtract
the
so2,
that's
below
the
omit
detection
limit
and
it
actually
reduce
the
observed,
no
no
reduce
the
model,
so2
lifetime
to
13
days
getting
closer,
but
still,
and
we
tried
many
other
methods
to
reduce
this
lifetime,
but
nothing
worked
really
well
and
finally,
we
find
some
lab
experiment
that
says
a
heterogeneous
reaction
of
so2
on
ash
particles
happening
and
when
we
apply
this
reaction.
In
the
model,
it's
actually
explaining
the
remaining
description
descriptions
of
pencil,
so
this
reaction-
I
have
a
very
brief
chart
here.
P
P
So
first
we
have
the
so2
and
we
have
this
mineral
particle
and
through
a
process
called
absorption
and
the
so2
can
quickly
stick
on
the
surface
of
the
mineral
ash,
with
no
barrier.
So,
unlike
the
nucleation
take
quite
quite
some
energy
and
then
those
so2
get
oxidized
into
the
into
the
sulfate.
P
So
in
conclusion,
here
that
none
of
the
observations
in
the
model
gets
the
so2
correct,
so
the
actual
so2
lifetime
is
17
days
and
at
the
end,
I'm
this
figure
shows
that
the
ash
particle
can
impact
the
s
impact
to
the
sulfur
sulfate
burden
in
the
stratosphere.
P
Interact
with
the
sulfate
through
the
heterogeneous
reaction,
and
then
we
have
the
case
in
red
that
also
adding
the
so2,
heterogeneous
reaction
on
ash
and
adding
this
reaction
results
in
50,
more
percent
of
sulfate
removal
within
60
days,
okay,
so
in
summary,
well,
this
will
work
is
conducted
in
wacom,
4
karma
model.
P
So
in
summary,
we
we
have
to
modify
the
diagram
of
physics,
scan
chemical
process
in
the
volcanic
clouds,
so
in
a
typical
model
model
that
we,
the
volcanic
eruptions,
erupted
gas,
so2
and
ash
in
the
in
the
model
and,
however,
the
ash
can
sediment
very
quickly,
so
it
is
usually
neglected
in
a
climate
simulation.
P
So
the
simulation
only
consider
the
so2
react
with
oh
and
form
the
sulfate
aerosol.
However,
our
study
says
we
can
also
consider
the
ash
and
react
with
it.
The
so2
and
form
the
mixed
aerosol
and
those
aerosols
can
also
react
with
the
pure
sulfate,
the
large
ash
sediments.
When
we
sulfate
take
part
of
the
sulfate
away
from
the
stratosphere
small
ash
micro,
sized
particles
with
sulfate
can
persist
longer
and
impact
impact,
the
climate
in
a
longer
term.
P
Yeah.
Thank
you,
and
this
is
an
article
we
published
and
you're-
welcome,
to
take
a
look.
Q
R
Wacom,
okay,
thank
you.
Thank
you.
Nick
I'm,
a
postdoc
in
susan
solomon's
group
and,
first
of
all,
I
want
to
thank
all
the
co-authors
for
that
enable
this
work
and
we
are
interested
in
the
in
the
in
on
how
small
scale
gravity
waves
impact
the
chemistry
in
vacuum,
and
we
know
from
oh
sorry
it
might
be,
and
we
know
from
observations
that
sorry
I
have
some.
R
We
know
from
observations
that
the
the
that
small
scale
gravity
waves
can
lead
to
temperature
perturbations
up
to
15
kelvin
in
the
in
the
stratosphere,
but
we
also
know
that
the
chemistry
not
only
linearly
depends
on
the
temperature.
For
example,
polar
strategic
clouds
exist
only
below
a
specific
temperature
threshold
and
reaction
rate
constants
themselves.
R
But
of
course,
these
small
scale
gravity
base
are
not
resolved
by
the
model,
but
we
have
the
gravity
parametry
safe
in
the
model
that
that
treats
this
small
scale
gravity
base,
and
but
they
are
currently
not
coupled
with
the
chemistry
in
the
model,
and
this
is
exactly
where
the
idea
of
our
project
comes
in,
because
we
can
get
vertical
displacements
of
the
cra
of
the
gravity
base
or
or
of
pulse
air
puzzles
in
the
in
the
gravity
wave
and
the
vertical
gradient
of
of
the
temperature
transparent
to
the
chemistry
and
then
get
a
temperature
perturbation
due
to
the
gravity
base.
R
What
we
then
did
as
first
test
simulations,
we
used
vacuum
sd,
the
the
sd
version
of
wacom,
relaxed
to
merit2,
we
simulated
time
period
from
2010
to
2011,
and
we
conducted
two
simulations
one
without
the
sub
cycling
approach
and
the
second
one
with
this
gravity-based
parametrization
included,
and
currently
we
implemented
it
only
for
orographic
gravity
waves,
because
this
is
kind
of
easier
to
implement
and
on
the
right
hand,
side.
R
You
can
see
the
distribution
of
these
deprived
values
and
you
can
actually
see
the
distribution
of
of
the
mountains
in
the
model
quite
well,
and
especially
at
hot
spots
like
the
andes.
They
are
quite
high
values
of
of
t,
prime
and
also
at
the
antarctic
peninsula,
and
you
can
also
see
that
the
the
magnitude
of
these
t,
prime
values
is,
is
comparable
to
what
we
expect
from
measurements,
which
is
quite
promising.
R
But,
of
course
we
have
to
evaluate
our
t,
prime
values
first
and
here's.
Here's
the
first
comparison
from
katherine
wilker's
phd
thesis
who,
who
graduated
this
year.
R
We
we
co
here
we
compared
the
the
vacuum
distribution
of
t,
prime
values,
with
the
cosmic
t,
prime
values,
which
looks
quite
good
in
the
summer
months
here
is
shown
for
for
the
end,
is
for
january.
R
We
further
looked
into
in
the
comparisons
of
the
chemical
substances,
with
with
observations
and
here's
this
here,
it's
shown
for
for
mls.
You
can
see
here
time
series
for
a
specific
point
where
you
can
see
that
that
these
peaks
in
clo,
which
is
an
indicator
of
chlorine
activation,
is
quite
good,
represented
in
both
simulations,
but
with
in
the
simulation
with
gravity
waves,
which
is
the
blue
one.
R
You
can
actually
see
that
these
peaks
are
actually
improved
due
to
our
new
parametrization
of
including
p
prime,
which
is
a
quite
promising
result,
but
we
have
to
further
look
into
this.
Of
course,
here
you
can
see
the
relative
differences
between
the
two
simulations,
the
simulation
with
the
sub-cycling
approach
and
the
simulation,
the
reference
simulations
for
the
tropics
and
the
high
latitudes,
and
you
can
see
that
there
are
quite
large
relative
differences
in
the
dropper
tropical
tropopause
region,
where
we
expect
cirrus
clouds
to
exist.
R
So
this
is
a
region
where
we
expect
that
that
heterogeneous
chemistry
is
active
and
where
we
could
expect
that
also
the
the
this
difference
can
could
come
from.
Also,
of
course,
we
see
differences
in
the
certain
high
latitudes.
R
Here
it's
shown
for
september
2011.,
you
can
see
that
at
the
polar
at
the
edge
of
the
polar
vortex,
there
are
some
large
relative
differences,
but
you
might
know
that
that
the
chemistry
is
one
of
the
most
expensive
parts
of
the
model,
and
actually,
if
we
use
this
10
times
sub-cycling
of
the
chemistry,
it
actually
uses
twice
the
time
we
usually
use
in
wacom.
R
So
we
are
currently
also
thinking
about
approaches
how
to
do
how
to
get
similar
results
with
fewer
sub-cycles,
and
we
came
to
this
result
that
that
if
we
use
sine
wave
distributed
random
t
primes,
we
can
actually
use
only
three
sub-cycles
to
get
the
same
risk
actually,
the
same
result
as
for
for
the
time
sub
sampling
approach
I
mentioned-
and
this
is
basically
shown
here
for
clo
on
the
x-axis,
the
the
differences
of
the
10
times
sub-cycling
to
the
reference
is
shown
and
on
the
y-axis.
R
The
same
is
shown
for
the
random
approach
here
and
you
can
see
that
they
are
quite
similar
in
especially
these
large
differences.
Here.
Color-Coded
is
the
pressure,
and
this
is
exactly
the
pressure
range
where
the
heterogeneous
chemistry
is
active.
So
this
is
a
really
promising
result
that
you
can
actually
use
only
three
sub-cycles
instead
of
10
to
get
similar
results
using
our
t,
prime
parametrization.
R
So
to
summarize,
we
implemented
the
math
method
currently
for
oral
graphic
gravity
phase.
Only
we,
of
course,
are
planning
also
to
include
non-horographic
gravity
waves.
We
did
first
comparisons
to
measurements,
but
we
have
to
look
into
that
a
bit
further
and
I
showed
that
using
sine
wave
distributed.
Q
All
right,
thank
you.
There
is
a
quick
question
in
the
chat.
Maybe
you
can
answer
there
or
we
can
answer
at
the
the
end
of
the
q
a
period
so
we'll
move
on
to
the
next
presentation,
which
is
given
by
yoga
richter
from
ncar,
and
the
title
is
small
ensemble
of
climate
intervention
simulations
using
stratospheric
aerosols.
L
Good
morning
my
name
is
yaga
richter
and
I'm
going
to
present
work.
That's
collaborative
effort
with
severals
from
cornell
university,
danielle,
vigiani,
doug,
mcmartin
walker,
lee
sally
woodhouse,
and
several
people
in
cg
do
cgd,
who
have
helped
either
setting
up
simulations
or
giving
input
to
their
design.
L
So
yeah,
as
nick
mentioned
I'll
talk
about
a
small
ensemble
of
climate
intervention
simulations
using
stratospheric
aerosols
and
the
reason
we
do.
This
research
is
that
climate
intervention
could
potentially
reduce
the
worst
consequences
of
climate
change.
The
stratospheric,
aerosol
injection
or
sai
seems
to
be
the
most
promising
or
can
say
most
studied
strategy.
Marine
cloud
brightening
is
another
strategy,
but
there
have
not
been
very
many
recent
global
studies
of
that,
so
sai
looks
most
promising
for
now.
However,
there
still
exists
very
large
uncertainties
in
the
impacts
of
sai,
so
back
in
2016
and
17.
L
The
north-south
gradient
and
the
multiple
gradient
will
remain
roughly
at
2020
levels,
so
we're
trying
to
get
the
climate
as
close
to
present
day
as
possible
and
a
network
on
glenn's.
We
use
csm1,
wacom
5
and
only
with
middle
atmospheric
chemistry.
So
there
was
limited
potential
for
evaluating
tropospheric
chemistry,
impacts
that
which
is
not
available
and
glenn's
was
run
with
a
very
high
emission
scenario.
L
So
rcp
8.5
to
2100
with
a
start
in
2020
and
injections,
were
also
at
a
relatively
high
altitude
of
23
25
kilometer,
which
is
really
hard
for
the
current
technologies
to
reach
and
simona
has
done
additional,
similar
simulations
to
similar
to
blends
with
csm
to
wacom
6
and
they
had
very
similar
setup.
Except
now
she
was
using
ssps
585
and
the
overshoot
scenario,
but
also
started
in
2020
and
injections
at
these
high
altitudes.
L
L
This
is
on
average,
where
most
models
show
that
we
go
over
the
one
and
a
half
degree
threshold,
and
that
is
kind
of
normalizing
for
where
we
are
right
now.
So
if
you
normalize
to
the
wacom's
pre-industrial
temperatures,
we
would
reach
one
and
a
half
degrees
in
2025,
which
we
know
it's
not
true,
because
we're
already
at
2021-
and
we
got
a
little
bit
of
ways
to
go
so
most
ipcc
models
show
it
going
over
roughly
around
2030
a
little
bit
later,
and
we
start
the
intervention
a
little
bit
after
that.
L
We're
also
focusing
on
a
moderate
emission
scenario,
hoping
that
there
will
be
more
efforts
to
mitigate
emissions
to
so
we
wouldn't
have
to
offset
such
a
huge
warming
and
we
lower
the
injection
altitude
to
21
kilometers,
because
there
is
research
done
on
how
high
sulfate
kirby
carried
into
the
stratosphere,
and
that
is
that
currently
could
be
achieved
with
the
aircrafts
that
exist
or
could
be
easily
built.
L
And
we
also
cap
the
simulations
at
2069
again
with
past
2069.
It
is
so
hard
to
tell
what
the
emissions
will
be
and
we
want
to
just
keep
focus
on
the
shorter
term
impacts,
which
would
give
us
a
lower
signal
to
noise,
which
you
may
say.
Well,
it's
bad
for
doing
research.
L
L
Okay,
so
finally,
some
figures
on
these
in
comparison
to
glenn's
and
the
runs
done
by
simona.
So
in
our
new
simulations,
the
injections
are
primarily
going
at
15
south
and
a
little
bit
at
15,
north
and
other
locations,
but
there
is
a
huge
dominance
of
15
south
injections.
This
is
in
contrast
to
glands,
in
which
the
algorithm
was
picking
injections,
primarily
at
30,
north
and
30
south,
and
in
simona's
simulations.
The
injections
were
going
primarily
in
30
south,
and
this
is
where
the
injections
go
is
really
not
fully
understood.
L
It
depends
obviously
on
how
much
the
surface
temperatures
depart
from
the
goal,
and
then
the
algorithm
is
picking
the
best
injection
to
meet
that
goal.
But
it's
not
a
something
that
we
fully
understand,
just
because
of
the
complexities
of
the
system
and
I'll
talk
a
little
bit
about
it
and
there's
differences
between
csm1
and
csm2,
like
in
the
amoc
they're
contributing
to
the
model,
picking
more
southern
hemisphere.
But
it's
not
exactly
clear
why
it
would
pick
more
at
15,
south
versus
30
south,
for
example,
between
these
two
sets
of
runs
that
are
quite
similar.
L
Okay,
so
here
are
the
zonal,
mean
temperature
changes
when
you
inject
these
large
and
quite
large
amounts
of
aerosols
on
these
bottom
panels.
I'm
showing
the
period
of
2050
to
2069
relative
to
our
control
of
2020
to
2039,
and
you
get
about
four
and
a
half
degrees
of
stratospheric
warming,
and
it
is
kind
of
comparable
to
what
we
had
in
glenn's
in
an
earlier
time
period
of
2040
to
2059
again
by
the
end
of
the
century.
L
In
glenn's,
our
stratospheric
heating
was
twice
of
what
we
have
right
now
at
the
end
of
our
simulations,
and
if
you
notice
there
is
a
little
bit
of
warming
in
the
southern
hemisphere,
which
will
become
important
for
when
we
look
at
the
surface
impacts.
L
So
here
is
a
similar
figure,
but
looking
at
zonal
mean
wind.
So
if
you
change
the
temperature
structure
near
the
equator
in
the
stratosphere
by
thermal
wind
balance,
you
expect
the
winds
to
be
changed
as
well.
So
specifically
in
djf,
we
anticipated
strengthening
of
the
northern
hemisphere
polar
vortex,
and
we
saw
this
in
glenn's
as
well
and,
however,
in
the
tropics,
the
responses
are
a
little
bit
more
complicated.
L
L
Actually,
you
see
the
speeding
up
of
the
winds
a
little
bit
close
to
30
south
and
30
north,
and
why
is
that
different?
And
again
it
comes
down
to
where
the
injections
are
going.
So
in
this
plot
on
the
right,
I
showed
just
single
point:
injection
results
from
our
very
old
simulations
with
csm1,
which
show
that
if
you
inject
just
in
15
south,
you
do
not
necessarily
get
the
decrease
of
winds
across
the
entire
tropical
region.
The
response
is
a
little
bit
more
mixed.
L
L
However,
this
didn't
happen
in
glenns,
because
the
injections
were
going
in
30,
south
and
30
north,
so
even
by
the
end
of
the
century,
in
glenns,
the
qba
remained
with
a
period
of
about
20
months
and
in
simona
simulations
it's
about
at
16
months.
Whereas
in
these
new
simulations
we
go
to
a
cubicle,
that's
super
fast.
L
So
what
I
have
here
on
the
right
hand,
side
is
glenn,
so
this
is
first
rcp
8.5
changes
so
warming
everywhere
and
if
you
look
at
rg
engineering
simulations
relative
to
control
that
the
dominant
feature
was
this
winter
warming
over
northern
eurasia,
and
so
we
thought
that
there's
pretty
a
persistent
feature
of
this
geoengineering
strategy
when
you
get
this
large
tropical
heating.
However,
that
is
not
the
case
in
the
new
simulations
on
this
plot.
L
Here
I'm
showing
the
2050
to
2069
temperatures
relative
our
control,
and
you
see
the
key
feature
is
we
can
call
it
a
warming
hall
or
a
cooling
hall?
In
any
case,
the
primary
feature
is
that
you
get
a
lot
of
cooling
over
the
ocean
here
over
the
labyrinth
or
sea,
which
is
largely
attributed
to
the
changes
in
amok
between
csm1
csm2
and
you're,
seeing
a
little
bit
of
cooling
over
the
northern
continents.
L
But
you
are
not
seeing
this
warming,
and
so
here
is
a
djf
just
for
the
same
comparison,
so
you
see
the
patterns
and
glands
and
you
see
the
patterns
in
csm2.
They
are
not
the
same,
and
here
briefly
I'm
just
showing
you
the
amok.
So
in
csm1
this
is
a
figure
from
aishu
who,
if
you
look
at
all
the
rcp
scenarios
and
all
of
them,
the
emac
is
rapidly
slowing
down,
especially
in
the
rcp
8.5
scenario.
However,
for
the
more
moderate
scenarios
the
amok
does
not
decline
as
fast.
L
However,
in
csm2,
if
you
look
at
all
the
ssps,
the
amok
is
really
decreasing
fast
for
all
of
them.
It's
really
nose,
diving
and
in
glenns
we
were
able
to
recover
the
amok
with
geoengineering
to
this
blue
line
here
compared
to
rcp
8.5,
and
it
is
just
not
happening
in
the
new
simulations
and
simona
has
shown
that
also
in
her
simulations,
there
is
the
recovery
of
a
mic
and
glenns,
and
there
is
no
recovery
of
maybe
somewhat
of
a
recovery
in
the
simulations
that
she
has
done.
L
But
there
is
a
huge
difference
between
csm1
and
csm2
so
anyway,
in
short,
we're
carrying
out
the
small
ensemble
of
climate
intervention,
simulations
that
are
set
up
to
design
to
be
realistic,
but
really
they
need
to
be
just
taken.
We
need
to
look
at
these
simulations
and
simone
simulations
and
glands
together
and
just
see
what
the
impacts
are
and
why
they're
happening
and
really
understand
the
key
features
of
the
where
the
injections
are
going
and
what
is
driving
the
impacts
and
since
many
impacts
already
are
showing
differences
from
glands.
L
Q
Yeah,
thank
you
yoga.
We
move
on
to
the
the
next
talk
which
will
be
given
by
doug
kennison
from
incar,
and
it
is
evaluation
of
the
ccmi
wmo,
ref
d1
wacom
6
simulation.
S
Can
you
see
my
simulation
or
my
my
powerpoint?
Yes,
we.
S
Okay:
okay,
today,
I'm
going
to
talk
about
the
valuation
of
the
ccmi,
wmo,
reft-1
and
refd2
simulations
and
I'll
go
into
some
details
on
that
in
a
minute.
But
first
I
really
want
to
thank
everybody.
That's
been
involved
in
developing
csm2
wacom
6,
and
it
really
does
take
a
village
to
get
all
this
stuff
done
and
there
there
are.
S
There
are
a
lot
of
people
that
have
contributed:
let's
speed,
okay,
so
these
new
simulations,
these
refd
ones,
and
these
ref
d2s
are
aimed
at
supporting
the
wmo
scientific
assessment
of
ozone
depletion.
S
S
But
there's
also,
these
new
simulations
that
spark
ccmi
has
put
together
for
the
wmo
and
one
is
a
hindcast
simulation
from
1960
to
2018
and
it's
designated
as
the
refd1,
and
this
simulation
will
use
an
hqbo
and
it
will
also
be
used
to
grade
the
model
processes,
the
dynamics,
transport
and
chemistry,
and
then
the
second
simulation
refd2
is
a
new
forecast
simulation
going
from
2015
to
2100,
and
this
will
be
based
on
the
ssp
245.
S
The
overall
goal-
and
I
don't
know
if
we're
going
to
be
successful
in
the
time
frame
we
have
left-
is
that
papers
will
be
written
on
their
fd1
and
results
will
be
used
from
that
to
weight.
The
reft2
models
results
to
say,
for
example,
an
ozone
recovery
and
only
the
models
to
do
a
better
job
with
the
dynamics,
transport
and
chemistry
will
be
part
of
the
multi-model
mean
now.
For
this
talk,
I'm
going
to
give
talk
about
five
different
simulations
I'll
show
specified
dynamics
with
using
mara.
S
Essentially,
what
we're
doing
here
is
we're
going
from
most
constrained
to
most
interactive,
the
refti
one,
which
is
got
an
observed
ocean
interactive
dynamics,
but
has
a
nudge
qbo
and
it
uses
wmo
halogens.
At
the
end
of
the
assessment
there
around
2015
to
2018.
the
amip
simulations
were
in
the
cement
six
assessment
and
they
have
an
interactive
qbo
that
yaga
and
rolando
worked
on
and
have
published.
S
The
coupled
simulation
is
the
same
as
the
amiibo
has
a
a
deep
ocean,
interactive
ocean
and
then
finally,
the
ref
d2,
which
is
similar
to
the
coupled
simulation,
except
it
uses
the
different
halogen
recovery.
So
I'm
going
to
go
quickly
through
some
some
results
and
then
and
I'll
be
done
here.
So
if
you
first
want
to
look
at
just
the
qbo
on
the
left
is
what
was
published
in
the
gentleman
all
paper.
S
S
So
this
is
one
reason
that
we
wanted
for
for
spark
ccmi
and
for
wmo
to
use
the
nudge
qbos.
We
know
that
we
could
do
a
better
job
representing
the
qbo,
and
you
see
that
on
the
right
again
compared
to
the
era,
data
at
the
westerly
zoo,
extend
down,
and
also
if
you
want
to
compare
the
satellite
data,
you
can
align
the
qbo's
with
with
the
data
we
also
took
a
look.
This
is
work.
S
Rolando,
garcia
did
to
look
at
the
refti
wants
to
make
sure
that
the
temperatures
are
correct
in
the
polar
region.
These
are
polar
cap
averages,
the
top
is
wacom
6
or
fd1.
The
middle
is
amera
simulation
and
then
the
difference
and
and
where
the
arrow
is
going
over
to
the
right
you're,
seeing
that
in
the
lower
stratosphere
this
these
these
line
plots
are
85
hectopascals.
S
We
do
a
really
good
job
of
getting
the
temperature
right
and
within
plus
or
minus
one
degree,
and
you
have
to
in
our
current
chemistry
setup.
It's
really
sensitive
chemistry
and
if
you
don't
do
that,
you're
going
to
get
too
much
depletion
or
too
little
and
last
week,
bill
randall
sent
me
this
figure.
He
and
postdoc
john
starr
have
been
working
on
looking
at
temperature
evolution,
anomalous,
temperature
evolution
and
what
they
are
using.
S
Of
course,
comparing
to
the
strasbourg
sounding
unit
and
the
microwave
sounding
unit
bill
published
a
paper
on
this
for
wacom
4
in
his
2017
paper,
and
he's
now
updated
this
for
the
ref
d1
and
frankly,
we
do
a
really
excellent
job
of
getting
the
apparent
trend
and
also
representation
of
volcanic
influences
and
then
included
in
this
is
the
effect
of
ozone
depletion
and
greenhouse
gases
that
have
on
these
trends.
So
I
will
look
forward
to
them
furthering
this
analysis,
but
it
really
looks
good.
S
It's
probably
one
of
the
best
representations
of
the
temperature
trans,
all
right,
so
how
about
total
column
ozone,
because
this
is
a
big
issue
for
the
assessment.
I'm
first
going
to
show
you
on
the
left,
the
polar
on
the
right,
the
equatorial
ozone,
the
sbuv,
mod
or
merged
ozone.
Data
set
is
observations,
and
then
this
is
specified
dynamics
using
maritude
and
nudge,
and
you
can
see
with
with
the
merit2
good
temperatures
decent
circulation.
S
We
get
a
very
good
representation
of
the
chemistry
specifically
for
total
column
ozone
in
the
polar
region
for
october
and
also
in
the
tropics,
which
is
an
annual
average
between
20,
south
and
20
north.
If
you
oops.
S
If
you
now
add
in
the
ref
d1,
these
are
four
realizations
little
dots
are
the
the
realizations
and
then
the
the
solid
lines
the
mean
you
don't
expect
the
refd
one
to
match
up
for
every.
You
know
big
depletion
period.
You
know
it's
not.
You
might
have
a
cold
urinary
one
where
it
wasn't
in
the
observations.
However,
you
see
that
the
means
pretty
much
match
up,
so
there's
no
apparent
bias
between
running
the
ref
d1
and
comparing
to
observations
or
what
was
used
in
sd.
S
But
what
you
do
see
in
the
tropics
is
there's
less
total
column
ozone
than
what
is
the
what
is
in
the
observations
and
in
the
sd
and
I'll
come
back
to
that.
S
Now,
if
you
add
in
the
amit
simulations
now,
the
difference
between
the
ref
d1
and
the
a
map
again
is
one
is
an
interactive
qbo
and
one
isn't,
and
you
know
there
really
is
no
difference
in
the
polar
region
and
there
really
isn't
no
difference
in
the
in
in
the
tropics
either.
I
was
curious
if
that
would
have
any
effect
on
total
column
ozone,
and
it
doesn't,
and
so
I
evidently
that,
where
the
deficiency
in
the
qbo
is
apparent
in
the
lower
stratosphere
in
the
tropics,
it
doesn't
affect
the
total
column
ozone.
S
Now,
if
you
add
in
the
coupled
simulation
interactive
ocean,
interactive
qbo,
again,
there's
no
apparent
bias
in
the
polar
region.
You
know
the
variability
similar
to
what
you
see
in
observations
in
the
sd
and
so
everything's.
Looking
really
good.
This
is
this
is
probably
some
of
the
best
total
column
ozone
we've
ever
ever
used
or
derived
in
wacom.
S
But
then
you
see
on
the
right
in
the
tropics
that
the
coupled
system
has
less
total
combos
and
even
even
more
so
than
the
amip,
and
so
why
is
this-
and
we
talked
about
this
a
little
bit
in
the
in
the
galmanol
paper
and
if
you
look
at
the
vertical
velocities,
as
derived
using
the
tem
diagnostics
or
w
star
what
you
see
in
the
lower
lower
stratosphere
here,
the
blue
is
the
fastest
than
the
then
the
ref
d1
and
the
a
mapper
pretty
similar
and
then
the
sd
is.
S
Is
the
slowest-
and
the
idea
here
is
that
if
you
have
faster,
faster
circulation,
lower
stratosphere
where
the
total
column
is
going
to
be
most
sensitive,
you're
going
to
be
bringing
up
tropospheric
air
which
has
lower
ozone,
abundance,
etc?
And
so
this
is
this
all
kind
of
makes
sense.
S
We
haven't
really
looked
into
what
what
the
the
wave
properties
are
changing
between
the
couple
simulation
and
these
other
ones
yet,
and
that's
something
that
we
still
need
to
do
all
right
and
just
to
finish
up,
I
wanted
to
show
the
forecast
simulations
we're
currently
running
these.
This
is
just
the
polar
region.
The
ref
d2
is
the
red
lines
there
and
I
have
three
realizations
but
they're
only
through
2056.
S
As
far
as
the
mean
is
is
concerned,
and
the
green
is
what
you
would
get
with
the
ssp
245
from
cmx6
or
four
realizations
there
and
the
the
ref
d2
started
off
of
the
blue
or
the
coupled
simulations,
and
at
this
point
it's
really
hard
to
see
that
there's
going
to
be
any
difference
in
the
return
date
to
say
1980
conditions.
S
S
You
really
have
to
run
a
lot
ensemble
members,
and
this
will
come
out
if
you
have
say
15,
ccm's,
running
three
realizations
and
when
you
finally,
you
know
create
this
multi-modal
mean
you
will
see
a
return
date,
that's
different,
and
it
should.
It
should
follow
the
the
forcing
of
the
equipment,
effective,
stratospheric
chlorine.
So
that
being
said,
I
think
everything's
good
oops.
S
I
keep
hitting
the
wrong
button,
so
so
this
was
just
basically
a
first
look
at
the
ref
d1
and
the
refte2
simulations
and
it
the
results,
look
very
promising,
and
I
just
want
to
let
the
community
know
if
you're
interested
in
these
simulations.
Please
let
me
know
because
the
more
eyes
on
this
stuff,
the
better
and
if
you
want
to
use
them
for
your
research,
that
would
be
great
so
I'll
in
there,
and
I
think
that's
it.
Thank
you.
Q
Thanks
doug,
so
we
do
have
time
for
any
questions
or
discussion
on
the
last
few
presentations.
There
were
several
questions
in
the
chat,
but
I
think
all
of
those
were
answered.
If
not
and
anyone
wants
to
clarify
anything
they
asked
or
their
question
was
not
answered.
Please
speak
up.
Q
All
right,
I
guess
everyone's
questions
are
answered.
Go
ahead
down
you,
you
raise
your
hand.
G
Yeah,
just
interested
in
in
this
idea
about
the
implementing
gravity
wave
effects
on
chemistry.
I
think
this
is
an
area,
a
really
interesting
area
that
we
could
coordinate
some
activities
at
the
university
of
leeds.
G
We
have
a
nerc
project
looking
at
the
subscale
transport
in
the
mlt
bringing
knox,
you
know,
impacts
on
nox
and
metal
layers
in
the
mlt
and
how
we
are
doing
something
quite
similar,
we're
taking
the
gravity
wave
spectra
that
are
not
have
yet
to
dissipate
or
be
dissipated
in
the
code,
so
they
haven't
reached
the
breaking
level
and
we're
using
that
information
to
try
and
derive
some
kind
of
additional
vertical
transport
so
and
that
impacts
things
like
temperatures
as
well.
G
So
I
think
I
think,
maybe
at
the
next
gravity
wave
me
I'm
sorry,
a
future
wacom
meeting.
We
can
kind
of
give
an
update
to
the
group
and
maybe
see
if
there's
some
overlap.
G
G
Okay,
yeah
we'll
try
and
let's
schedule
a
meeting
and
try
and
get
people
talking,
it'll
be
great,
it's
also
with
chet
gardner
and
at
the
university
of
illinois.
So
it's
a
nsf
nurik-funded
crime.
Q
K
Sure
I
mean
you
know
we're
doing
we're
doing
these
regionally
refined
simulations
here
with
wacom,
and
so
like
I
mean,
if
you're,
if
you're
also
just
interested
in
general
and
output
or
anything
there's,
no
there's
no
there's
no
pressure,
but
it's
kind
of
like
I
can
we're.
I
think
we're
happy
to
point
people
to
where
some
of
this
stuff
is
because
there's
probably
an
infinite
array
right
now
of
things
people
can
look
at
in
these
runs
and
we're
really
just
trying
to
get.
K
The
data
volume
of
these
runs
is
difficult
to
work
with,
but
like
it's
basically
just
sitting
around
waiting
for
people
to
to
use
right
now,
so
we've
got
like
10
kilometer,
resolve
output,
chemistry
and
dynamics
like
four
minute
time
resolution,
so
you
can
really
exactly
see
what
the
waves
are
doing.
It's
just.
Obviously
that
means
for,
like
a
year-long
simulation,
that's
just
too
much
for
a
few
people
to
look
through.
So
if
that's
of
interest
to
you
or
anybody
else
here,
you
know
let
us.
G
Know
and
if
it's
going
to
be
huge,
I
think
we're
gonna,
I
mean
it
may
not
be
practical,
for
you
know
most
of
most
simulations,
but
we're
gonna
learn
a
lot.
I
think
of
the
small
scale
interactions
from
those
those
simulations.
K
This
is
definitely
a
case
where
it's
like
it
takes
a
while
to
get
yourself
set
up
with
like
processing
output,
because
you
definitely
need
to
make
use
of
like
you
know,
some
nodes
to
to
work
through
all
that
data.
G
G
You
know
park
it
somewhere,
like
I,
don't
know
mid-latitudes
over
an
ocean
even
or
something
and
just
just
get
an
idea
about.
You
know
where
you
know.
If
you
did
turn
off
the
gravity
waves,
it
wouldn't
really
impact
the
global
circulation,
but
you'd
really
at
least
get
to
see.
What's
going
on
in
the
vertical
quite
easily,
with
the
data
volumes
being
a
bit
less.
K
G
Yeah
I
I
know
there
was
some
discussion
about
you
know
because
I
wanted
to
put
higher
res
over
the
polls
and
that
the
grid
gridding
software
was
kind
of
a
little
tricky
to
when
you
get
near
the
polls
and
that
that
there
was
some
talk
about
saying.
Well,
look,
you
just
have
a
square,
and
I'm
just
going
to
move
that
pole
or
move
that
to
set
to
be
centered
somewhere,
so
that
once
you
set
up
that
one
grid,
you're
just
free
to
then
just
go
easily
move
it
somewhere
else.
G
Do
another
simulation,
move
it
somewhere
else
and
kind
of,
without
necessarily
having
to
go
through
this
a
lot
of
this
process
of
setting
up
a
new
grid
and
all
that
kind
of
stuff.
So
I
think
something
a
bit
more
general
and
smaller
scale
like
we
might
be
easier,
might
be
easier
to
kind
of
probe.
These
questions.
Q
Q
G
Yeah,
I
think
yaga
knows
kind
of
like
the
the
angle
or
you
know,
as
pictures
of
the
angles
of
what
these
things
are
coming
up.
So
as
long
as
it
was,
you
know
not
obviously
not
100
kilometers
square,
but
maybe
you
know
something
you
know
500
kilometers.
Maybe
I
don't
know
what
what
that
sort
of
angle
is
at
which
you
would
expect
a
wave.
I
mean
I've
seen
some
of
when
holly
did
his
early
simulations
at
quarter
degree.
G
L
Yeah
well
in
the
model,
it's
just
nothing
propagates
away,
but
in
I
think
in
real
life
they
could
go
quite
quite
a
few
grid
boxes.
Now.
G
K
G
K
But
just
just
remember
too,
though,
like
no
matter
how
small
you
make
the
refinement,
you
still
have
to
run
globally
with
the
high
time
step.
So,
like
I
know,
in
terms
of
storage,
you
save
a
lot,
but
the
model
cost
is
still
going
to
basically
be
the
same
as
soon
whatever.
Whatever
you
choose
for
the
refinement
factor
is
really
going
to
set
a
lot
of
the
cost
just
off
the
bat,
no
matter
how
small
it
is.
B
Nick,
I
think
it
would
be
awesome
to
compare
the
gravity
wave
amplitudes
that
you're
getting
in
this
version.
With
with
what
we
get
out
of
the
parameterization,
I
mean
I
would
think
that
would
be
of
interest
from
a
dynamical
point
of
view
too,
but
we'd
be
happy
to
to
look
at
those
t,
primes
and
think
about
what
that
means
with
you.
K
Yeah
sure
we
can
include
you
and
whenever
we
have
some
of
these
targeted
discussions,
I
think
probably
we're
gonna
do
some
longer
runs
soon.
We're
just
sort
of
feeling
out
with
some
short
runs
like
what
we
can
and
cannot
do
in
terms
of
the
analysis-
and
you
know
turning
parameterizations
on
and
off,
but
yeah
happy
to
include
you
in.
K
A
So
maybe
we
should
move
on
to
some
more
general
discussion.
We
didn't
have
a
lot
of
formal
items
that
we
wanted
to
discuss,
but
we're
hoping
that
this
is
a
time
for
getting
feedback
on
things
that
you
think
are
missing
in
the
model.
What
you
would
like
to
see,
such
as
I'll
just
put
our
little
bullets
in
the
chat
there.
So
you
know,
as
always,
the
model
is
evolving
and
there
probably
will
be
a
release
of
cesm
2.3
in
the
fall.
A
It's
a
chance
for
you
to
request
comp
sets
other
features
and
we'd
also
like
to
hear
about
your
research
interests
and
what
are
maybe
some
of
the
model
developments
needs
for
addressing
those
topics
and
we'd
really
like
to
hear
what
how
people
are
using
the
model
or
think
what
is
missing
or
are
doing
any
developments
that
we
haven't
heard
about.
Yet
today,.
B
Yeah,
I
think
I
got
it
up
a
little
too
late.
I
just
wanted
to
make
one
more
comment
on
the
gravity
wages.
That's
all
right!
Oh.
B
Yeah,
I
just
wanted
to
point
out
to
people
that
there's
a
an
effort
being
led
by
joan
alexander
and
chris
cruz
who's.
I
think
on
this
telecon
is
working
on
it,
a
lot
it's
to
look
at
high
resolution
simulations
using
wharf
and
other
kind
of
high
high-resolution
non-hydrostatic
models
around
the
and
the
southern
tip
of
the
andes
and
they're.
Comparing
with
air
satellite
observations,
I
think
that
there
was
a
there
was
an
item
in
the
chat
on
that.
G
Oh,
I
I
was
just
saying
that
for
those
that
might
have
missed
the
amwg
meeting
yesterday,
that
there
is
a
document
on
the
model,
development
priorities
and
I
think
it's
open
to
comment
and
julius
set
it
up,
so
that
people
can
write
on
the
second
page
ideas
or
things
that
they
think
are
important.
And
so
there
is
there's
a
geospace
and
chemistry
section
here.
But
if
there
are
other
critical
things,
this
would
be
a
good
place
to
put
things
in
right.
Julio.
B
Oh
yeah
document
the
document
sure.
Yes,
sorry,
I
I
was
checking
an
email.
A
No,
that's
great
all
right,
yeah
I'll,
take
notes.
If
anyone
has
any
thoughts,
they'd
like
to
bring
up
here.
G
Yeah,
we've
done
lots
of
evaluation
on
the
on
the
chemistry
side
of
it,
comparing
it
to
satellites
and
trying
to
understand
how
it
improves
the
representation
of
the
response
to
energetic
particle
events
for
for
knox
production
at
the
high
latitudes
and
it's
impact
on
ozone.
So
so
yeah.
I
think,
and
we
there
have
been
some
extended
simulations
trying
to
understand
whether
or
not
that
those
effects
in
the
upper
stratosphere
from
energetic
particles
might
propagate
down
to
the
surface.
G
So
if
you
want
to
know
some
specific
papers,
I'm
happy
to
share
them.
They're,
they're,
typically
coming
out
of
the
group
in
fmi.
B
Cool
thank
you
very
much
daniel.
I
will
maybe
drop
an
email
to
ask.
A
A
G
Good,
well,
I
must
say
that
I
mean
it
hasn't
been
brought
up
in
this
meeting,
but
there
there
does
seem
to
be
an
issue
with
a
difference
between
re-analysis
and
wacom
at
the
stratopause
that
we're
trying
to
track
down.
We
don't
quite
know
yet
whether
or
not
it's
got
to
do
with
differences
in
the
and
radiative
transfer
between
what
mara
uses
and
what
we
have.
G
But
there
does
seem
to
be
10
to
15k
difference
when
we
that
we
notice,
when
we're
doing
our
initialized
forecasts,
where
we
initialize
it
to
mira
and
then
and
then
let
it
free
run,
and
you
see
a
big
jump
so
we're
trying
to
track
that
down
and
wondering-
and
in
fact
it
looks
like
from
preliminary
comparisons
to
observations
that
maybe
mera
is
actually
closer
to
ops
in
the
lower
mesosphere.
G
B
And
did
you,
I
presume
that
you
got
sorted
out
that
issue,
that
originally
was
there
with
the
diurnal
averaging
of
the
ozone
that
was
fed
to
the
radiative
transfer.
G
Now
this
is
all
fully
interactive
chemistry,
we're
doing
these
forecasts
with,
so
that
that's
that
was
only
with
wacom,
with
prescribed
chemistry
prescribed
ozone
that
there
was
a
bit
of
a
heating
bias
there.
But
now
this
is
all
just
you
know
everything
ozone
does
have
its
full
cycle
up
there.
G
So
it's
not
it's
not
related
to
that.
I
have
a
feeling,
though
it
could
be
related
to
rtmg
being
a
bit
tuned
for
the
lower
atmosphere
and
not
really
caring
too
much
about
the
stratopause.
So
I'm
hoping
to
get
a
different,
compare
radio
transfer
code
in
for
a
number
of
different
ideas
like,
for
instance,
being
able
to
run
with
more
extreme
atmospheres
by
so
switching
out
to
the
socrates,
radiator
transfer
code
that
met
office
uses
and
seeing
if
that
is,
is
better,
better
or
at
least
gives
different
results.
G
Q
Yeah
I
I
would
certainly
agree
it's
it's
an
opportunity.
I
don't
have
anything
just
like
right,
ready
to
show
in
terms
of
what
you're
saying,
but
it
does
certainly
have
an
impact.
You
know
if,
if
the
temperatures
change
10k
within
two
days
or
something
that's
certainly
an
issue
in
terms
of
analysis
of
any
sort
of
forecasting
in
the
evaluation.
S
S
S
What
yoga
has
done
with
the
one
degree,
and
we
got
a
really
nice
looking
qbo
except
then
we
ran
into
an
issue
with
dehydration
of
the
stratosphere,
which
is
is
somehow
coupled
back
to
the
convection
routine,
and
so
that
I
think
that's
something
would
be
really
nice
to
get
handled,
and
I
know
cgd
is
working
on
that.
S
But
I
just
thought
I
would
point
that
out
that
that
that's
something
could
be
very
valuable
for
this
community,
because
you
don't
always
need
to
have
a
one
degree
model
to
look
at
middle
atmosphere,
studies
and
you
don't
even
necessarily
have
to
have
tsmlt
type
chemistry.
So
I
think
it'd
be
nice
to
have
a
faster
model
where
universities
could.
You
know,
do
some
science
with
so.
S
A
And
sort
of
related,
I
know,
there's
interest
in
having
a
faster
model
for
tropospheric
chemistry,
and
some
groups
are
still
using
the
trope
mozart
mechanism,
which
is
the
old
mozart
for
chemistry,
and
I
know
I've
said
several
times,
but
I
will
try
to
get
a
new
trope
mozart
scheme
running
that,
if
you
for
looking
at
surface
air
quality-
and
you
don't
need
full
stratospheric
chemistry,
you
know
particularly
with
specified
dynamics
and
that
sort
of
thing,
I
think,
there's
still
some
interest
in
that.
J
B
Hang
along,
if
I
just
freeze
so
so
the
question
was
actually
since
I
originally
came
from
astroparticle
physics,
I
would
really
love
to
see
a
cosmic
ray
cascade
called
actually
working
at
esm,
because
then
you
would
really
see
all
the
small
scale
effects
in
the
atmosphere
like
gravity
waves
and
so
on.
If
this
will
have
would
have
an
influence
in
the
actual
propagation
of
these
particles.
G
There's
a
group
at
cu
working
with
jeff,
fair
that
did
the
at
least
the
global
electric
circuit
and
then
was
trying
to
tie
that
to
lightning
and
and
those
sort
of
effects
as
well.
I
think
there.
E
I
G
Three
or
four
years
ago
which
they
showed
their
their
parametrization,
but
I
don't
think
it
ever
got
into
the
release
code.
G
But-
and
I
know
there
are
other
groups
that
work
on
ion-mediated
nucleation
for
clouds,
but
no
one's
tied-
that
to
any
enhancement
you
might
get
from
ionization
that
occurs
from
galactic
cosmic
rays,
so
the
pieces
are
around.
I
think,
to
do
some
of
these
things.
They've
just
not
been
linked
together.
B
B
A
So
people
don't
have
anything
else
to
say
to
the
big
group
we
can
try
interacting
in
breakout
rooms.
We
thought
there
might
be.
You
know,
sort
of
nominal
topics,
but
very
informal
and
just
a
chance
to
chat
with
people.
A
B
Says
I
am,
I
can
go
ahead
and
open
the
rooms.
Okay,
if
you
want
to
suggest
topics,
I
can
name
the
room
rooms
with
certain.
A
Topics
or
right
now,
they're
just
numbered
one
through
seven
and
you
can
go
ahead,
and
so
we
had
thought
there
might
be
four
topics.
People
come
here,
put
them
in
the
chat.