►
Description
The 27th Annual CESM Workshop will be a virtual event. Specifically, the Workshop will begin with a full-day schedule on 13 June 2022 with presentations on the state of the CESM; by the award recipients; and two presentations from our 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.
To learn more:
https://www.cesm.ucar.edu/events/workshops/2022/
E
Thanks
yeah
just
thank
you,
sir
just
realized
beauty.
You
guys
see
my
screen
and
hear
me
well.
A
Yeah
you
can,
you
can
stop
sharing
for
a
minute,
we'll
just
yeah
just
go
through
the
logistics
and
then
and
then
we'll
start,
but
okay,
okay,
I
guess
we'll
start
now,
so
welcome
to
the
climate,
variability
and
change
working
group
sessions,
I'm
isla
simpson
and
we're
here
with
sarah
larson
and
ishihu,
the
other
co-chairs
of
the
working
group.
A
So
we
have
a
great
lineup
of
talks
and
in
the
middle
kind
of
at
the
end
of
the
first
part,
we
want
to
have
lots
of
discussion
about
the
simulations
that
we're
gonna
propose
for
the
next
computing
allocation.
So
for
the
speakers,
it's
a
12
minute
slot
so
10
minutes
for
talking
and
then
we'll
have
a
couple
of
minutes
for
questions
and
change
over
and
we'll
give
you
a
warning
at
eight
minutes,
so
we'll
just
get
started
and
I'll
hand
it
over
to
aishu
who's.
Gonna
chair
the
first
session.
B
E
Okay,
so
thank
you
very
much
for
having
me
here,
I'm
with
you
from
uc
urban
sites.
So
today
I'm
going
to
talk
about
disentangling
the
impacts
of
stropheric
ozone
depletion
and
tropic
also
increases
on
southern
ocean
in
terror
roaming.
So
this
work
I
collaborate
with
makila
hackling
and
rimmel
czechka,
garcia.
E
E
So
the
figures
I'm
showing
here
is
prescribed
ultimate
change
in
the
two
generations
of
models,
and
basically
is
the
anu
and
zuno
mean
fractional
change.
So
in
odon4c
on
the
right,
I
put
the
formula:
basically
is
ozone
in
95
to
99
minus
55
to
59
normalized
by
the
59
55-59.
E
So
from
this
figure
we
can
see,
we
can
see
the
depletion
of
chopped
ferrous
ozone,
especially
in
the
south
pole
region,
in
the
highlight
tubes
in
high
elevation,
and
that
would
be
the
well-known
stratofabric
ozone
hall
which
had
been
observed
in
the
past
few
vacations.
E
So
now
the
essential
question
is
that
what
will
be
the
only
impact
for
our
climate
system,
because
it
has
been
well
studied
for
the
overall
impacts
on
the
many
things
like
the
homophysical
circulation
and
also
the
radiation
balance
at
the
toa.
E
However,
relatively
less
attention
has
been
put
on
the
ocean,
especially
if
we
think
about
the
ocean
heat
uptake,
especially
in
the
solar
ocean.
So
that
is
a
very
important
issue
and
we
want
to
look
at
so.
There
have
been
studies
talking
about
the
effects
of
strategic
authentication,
for
example,
in
swat
at
all
2018
interjections
paper
and,
however,
one
of
these
importance
would
be
like
what
would
be
the
impact
of
the
tropical
odon
increase,
because
we
can
see
this
choco
increase
is
very
notable
signal
from
the
figure
here.
E
So
what
we
used
would
be
the
simplified
and
76
odom
single,
forcing
experiments
so
for
those
who
are
not
familiar
with
the
experiment,
basically
in
either
semi-finals
make
six
and
we
have
the
single
sourcing.
Basically,
we
only
force
the
model
using
the
historical
changes
of
ozone,
but
for
the
other,
factors
were
set
to
pre-industrial
level.
So
on
the
left,
I'm
showing
the
sigma
five
result
and
on
the
right
would
be
the
semi-six
results.
E
So
the
top
panels
showed
the
annual
and
zoono
mean
to
no
wings
change
during
1955
to
2000
and
the
baltimore
ii
panels
shows
the
annual
zoono
mean
ocean
temperature
during
this
period.
So
let's
first
look
at
the
sigmify
and
we
notice
in
the
signifying
models
and
the
experiments
were
fought
by
total
odon
change
so
which
includes
both
strategic
and
tropical
odon.
However,
it's
different
for
the
semi-six
models
in
which
we
only
have
the
ozone
14
from
the
stratosphere
so
for
semi-six
experiments,
it's
forced
only
by
spherical,
okay.
E
So
let's
first
look
at
atmosphere
and
we
can
see
a
polar
identification
of
the
vector
image.
So
here
the
contour
overlapped
is
the
pre-industrial
climatology
when
filled.
So
we
can
see
this
very
robust
feature
for
the
polar
world
intensification
of
southern
hemisphere
recipient,
which
has
been
reported
in
many
previous
studies
and
also,
if
we
compare
sigma,
five
and
sigma
six
and
we
can
see
yeah,
there
are
not
too
much
differences
difference.
We
can
so
we
can
see
from
these
two
panels.
E
However,
situation
will
be
different
if
we
look
at
the
ocean
so
for
the
ocean
we
can
see
in
the
southern
ocean
simplified
model
simulates
a
very
strong
warming
in
the
entire
region
and,
however,
in
semi-6,
the
warming
is
much
weaker.
So
there
could
be
many
reasons
why
we
can
get
these
different
results
right.
So
first,
the
odon
forcing
we
can
have
some
slight
differences
between
the
semi-5
in
e6.
E
If
we
look
at
the
previous
slide,
the
second
and
we
can
see,
there
would
be
different
models
for
simplify
and
I
have
these
four
models.
E
So
the
figure
shows
multimodal
mean
over
these
four
models,
but
for
cm6
they
are
totally
different
models,
so
these
different
models
could
have
different
impact
because
we
are
not
quite
sure
the
model
will
have
different
response
to
the
university,
but
this
figure
can
give
us
some
implications
like
one
missing
part
or
one
important
factor
would
be
the
spherification,
because
if
we
look
at
simplified
and
we
can
see
it's
forced
by
both
traffic
control
variables,
however,
only
stratford
also
is
input
into
the
semi-sig
simulations.
E
E
This
is
the
cm6
error
model
so
for
this
model,
and
we
have
the
data
available
from
the
cm6
archive
with
two
seals
of
experiments.
So
the
first
suit
would
be
the
total
ozobot
stratospheric
plus
tropospheric
ozone
are
the
results
shown
in
the
left
and
another
suit
of
experiments
would
be
stratford
ozone.
Only
that
is
consistent
with
other
76
models,
so
the
result
is
shown
in
the
middle
and
on
the
right
would
be
the
difference
between
the
first
two
so
which
indicates
the
impacts
of
trouble
for
open
change.
E
Okay,
so
similar
to
the
previous
ones,
and
the
top
panel
showed
the
annual
and
zuno
means
no
wings,
and
so
that's
the
difference
between
the
1955
to
2000
and
pre-industrial.
So
we
can
see
the
controller
indicates
pre-industrial
climatology
and
also
the
cheating
where
we
have
the
dotted
region,
which
means
is
not
statistically
significant.
E
So
if
we
look
at
the
top
panels,
let's
look
at
the
wings
and
we
can
see
a
very
robust
polaroid
identification
for
the
total
odometry.
So
that's
very
similar
to
the
previous
results
and
for
the
strategic
ozone.
We
can
still
see
this
polaroid
intensification
of
southern
hemisphere
recipes
and
the
difference
for
the
tropical
urban
impacts
we
can
see.
There
is
upward
intensification
in
the
upper
levels
and
at
the
lower
levels
we
can
see
a
signal
of
polar
world
incidentification
of
battery
wings.
E
So
it's
worth
noting
that
as
table
to
the
surface
and
the
magnitude
of
wind
change
is
comparable
between
specific
ozone
change
and
tropical
change.
So
that's
for
the
top
panels.
Then,
let's
look
at
the
bottom
panels,
which
shows
the
lunar
and
anu
mean
ocean
temperature
change
during
95
1990
1955
to
2000,
so
we
can
see
very
robust
vomit
occurs
in
both
stropharic
ozone
simulation
and
also
the
difference
which
indicates
tropical
verb
also
impacts.
So
from
this
figure.
To
summarize,
we
can
see
southern
ocean
in
terror.
E
Then
we
also
look
at
the
ocean
heat
content
change,
so
here
I'm
showing
the
time
series
for
all
these
experiments.
So
first,
let's
look
at
the
results
from
cam
esm5.
So
let
me
walk
through
you
from
these
figures.
So
if
we
look
at
this,
so
this
very
sick
blue
curve.
So
that
is
the
result
from
the
this
model
for
the
stratfor
ozone
only
experiment,
and
meanwhile
we
can
see
this
dark
purple
dark
purple
curve
which
indicates
ocean
heat
content
change
in
the
historic
room
for
this
model.
E
So
then,
if
we
compare
these
two
curves,
we
can
see
the
total
oil
change
contributes
about
one
third
of
the
net
historical
ocean
heat
content,
increase
that
is
within
upper.
Two
thousand
meter
between
30
to
60
degree,
notes
a
degree
source
over
1955
to
2000.
So
that's
the
first
result
and
then
we
want
to
participate
the
contribution
from
both
strategic
ozone
and
tropicals,
which
has
been
illustrated
in
the
left
panel.
So
that's
the
blue
curve
I
just
mentioned,
and
then
the
red
curve
is
the
ocean.
His
content
changed
due
to
stratheric
depletion.
E
Well,
the
black
one.
Thank
you
between
the
two.
So
we
can
see
the
50
percent
is
from
tropical
odon
increase,
but
forty
percent
is
from
traffic
over
and
also
the
result
is
consistent
with
simplifying
model
simulation.
So
what
will
be
the
physical
mechanisms?
So
we
decompose
the
total
temperature
change
into
the
spacing
exchange
and
save
change
so
on
the
left,
I'm
showing
the
speed
change
for
the
three
total
stropharic
and
tropical
odome
foxy
and
on
the
right
is
the
hip
part
so
very
clearly,
and
we
can
see
for
the
species
for
the
strawberry.
E
Okay,
so
due
to
the
time
I
just
gone
to
my
conclusion,
and
so
basically
we
can
see
strategic
ozone
and
tropical
ozone
can
both
contribute
to
solar
ocean
warming,
and
the
tropical
ozone
is
very
important
and
it's
more
than
an
air
pollutant,
it's
as
greenhouse
gas.
It
can
be
very
important
for
the
global
warming,
especially
for
the
ocean
heat
up.
E
A
Yeah,
sorry,
just
a
clarification,
I'm
not
an
oceanographer,
I
didn't
what
is
what
does
heave
mean
and
what
what
I
didn't
follow
the
terminology.
E
That's
a
lot
yeah
due
to
the
time
just
cut
off
the
information.
B
E
So
here,
basically,
you
can
understand
like
the
range
driven
parts.
Like
my
pumpkin,
you
know
kind
of
depressed
or
athletes
summer
clients,
so
this
will
increase
the
temperature
change.
On
the
other
hand,
for
this
business
is
something
like
the
subduction
through
the
mood
water
or
intermediate
water.
Basically,
if
we
have
the
surface
heat
flux
or
fresh
water,
forcing
so
the
time
to
transplant
can
propagate
along
the
ip.
B
F
Let
me
share
my
screen
here:
can
you
all
see
the
screen.
B
F
Okay,
great
wonderful,
well,
hello,
everyone,
I'm
carlos
martinez,
I'm
an
asp
postdoc,
I'm
with
cgd
m
cubed,
and
my
presentation
is
on
the
seasonal
hydroclimate
of
the
caribbean
under
various
cesm,
as
well
as
some
siemens
6
models.
So
why
the
caribbean?
Well
I
looked
at
the
caribbean
in
its
observed
hydroclimate,
and
one
of
the
reasons
why
it's
such
a
complex
region,
perhaps
for
models,
is
because
they're,
it's
not
a
homogeneous
region.
You
have
different
hydroclimates
in
different
parts
of
the
caribbean,
for
example
in
the
northwestern
caribbean.
F
You
have
two
peaks,
the
first
being
the
early
rainy
season
from
mid-april
through
mid-june
and
the
late
rainy
season,
which,
for
several
caribbean
regions,
is
the
largest
portion
of
rainfall
across
the
hydroclimate.
But
if
you
look
at
the
eastern
caribbean,
so
over
the
lesser
antilles
and
barbados,
you
don't
see
an
early
rainy
season
and
other
regions
have
different
amplitudes
or
magnitudes
of
each
seasonal
component
of
the
rainfall
cycle.
F
In
addition
to
this,
it's
a
rain-fed
region,
many
stakeholders
rely
on
rainfall
for
their
socioeconomic
needs,
and
the
climate
in
in
the
dynamics
in
this
in
this
region
is
important
for
other
regions
such
as
the
pacific
basin,
as
well
as
the
southeastern
united
states.
So
seeing
the
complexity
of
the
cycle
and
the
mechanisms
which
I'll
explain
later
that
correspond
to
and
influence
the
hydroclimate.
I
wanted
to
look
at
global
circulation
models
like
the
cesm
model,
suite
to
simulate
the
rainfall
cycle
and
see
how
well
they
do
compared
to
my
observational
framework
that
I
did
do.
F
The
models,
for
example,
perform
better
over
ocean
only
spaces
versus
land,
where
there's
less
complexity,
such
as
topography.
Since
we're
dealing
with
islands,
are
there
biases
in
a
specific
sub-region
of
the
caribbean,
or
is
this
general
to
the
entire
basin?
And
is
there
a
relationship
between
model
resolution,
especially
since
we're
dealing
with
islands
versus
between
low
versus
high
resolution
models?
F
And
so
I
looked
at
several
models.
The
top
or
this
box
on
the
right
shows
all
the
models
that
I
used.
There
are
several
cesm
models,
such
as
the
lens
one
and
two,
as
well
as
their
goga
or
aemon
friends
and
the
I
have
high
resolution
cesm
model
as
well.
I
also
looked
at
the
cmip
high
res
mib
models,
which
the
only
difference
for
each
model
group
is
just
the
resolution
looking
at
their
low
versus
high
resolution
versions.
F
To
get
into
that
question
on
the
resolution
component
of
the
study
and
then
looking
at
observations.
I
use
several
gridded
precipitation
data
sets
to
have
an
observational
estimate,
as
well
as
using
era
5,
which
will
also
be
used
as
the
benchmark
observation
for
the
dynamical
variables
that
I'm
using
in
my
analysis,
as
well
as
other
station
data.
F
So
for
the
methodology
I
looked
first
at
precipitation,
scatter
plots
between
ocean
only
and
land
only
precipitation
totals
to
see
if
there
are
differences
there
and
then
I
regressed,
using
all
of
the
models.
Several
dynamical
variables
that
I
know
from
my
previous
work
to
inform
the
hydroclimate
in
the
caribbean
onto
the
precipitation
indices
across
the
different
sub-regions.
F
Then
I
looked
at
scatter
plots
related
to
where
I'm
finding
the
model
regression
and
those
inter-model
differences
to
be
highest
in
the
correlations
looked
at
spatial
averages
of
those
to
produce
dynamical
indices
versus
those
precip
indices
and
then.
Finally,
for
the
sake
of
time,
I'm
just
going
to
look
at
one
seasonal
component
of
the
rainfall
cycle,
the
early
rainy
season,
which
is
important
for
the
onset,
the
start
of
the
rainy
season,
as
well
as
the
growing
season
for
many
stakeholders.
F
So
just
for
clarity.
These
are
the
different
subregions
of
the
caribbean
and
for
the
first
example
I'll
be
looking
at
the
northwestern
corridor,
and
this
is
like
the
spatial
grid
box
that
I'm
looking
at.
But
there
are
several
other
subregions
that
I
have
looked
at
for
the
analysis,
so
the
first
part
is
looking
at
scatter
plots
between
land
versus
ocean
only
precipitation.
F
You
can
see
here
in
the
legend
all
of
the
models
that
I
use,
the
ones
in
black
or
the
observational
data
sets,
and
then
the
low
res
models
are
in
blue
and
the
high-res
models
are
in
red
and
what
you
see,
generally
speaking,
is
that
most
models
underestimate
land
only
and
also
ocean-only
precipitation.
F
We
do
see
some
differences,
especially
when
I'm
looking
from
model
group
to
model
group.
I
do
see
that
the
high
resolution
models
tend
to
show
values
that
are
closer
to
the
observational
estimates
than
their
low
resolution
counterparts.
So
this
is
just
an
example
here,
the
high
res,
as
well
as
the
aim
of
brent
from
high
res
cesm,
and
then
just
to
highlight
this.
We
see
this
pattern
across
most
of
the
models.
F
In
addition
to
that,
I
do
have
the
large
ensembles
and
their
spread
within
the
individual
members
and
comparing,
for
example,
cesm2
with
the
csm2
gogo
runs.
You
do
see
that
the
aim
of
brands
show
more
precipitation,
but
under
a
same
land,
ocean
ratio
and
that's
what
the
dotted
lines
show
where
we
also
see
the
model
show
a
less
ratio
than
the
ratios
that
we're
seeing
between
land
versus
ocean
in
the
observational
spread.
F
So
if
I
look
at
another
region,
for
example,
we
see
a
similar
case
and
the
spread
between
the
blues
and
the
reds.
The
only
difference
is
the
aim
of
friends
showing
more
precipitation
than
the
fully
coupled
runs
from
the
cesm
models.
So
it's
clear
here
that
there
seems
to
be
an
underestimation
generally
speaking.
Why
is
that
and
it's
clear
that
there
seems
to
be
a
separation
between
the
high
res
and
the
low
res
models?
Could
dynamics
be
a
role
in
that
bias
in
that
dry
bias
and
so
going
into
the
model?
Spatial
regressions?
F
We
see
some
patterns
here
that
might
suggest
this.
So
what
I'm
presenting
here
is
an
example.
Looking
at
may,
precipitation
regressed
onto
central
caribbean
may
precipitation
index
on
the
left
is
the
regression
beta
coefficient
and
on
the
right
is
the
correlation
coefficient
from
that
model
based
regression,
and
what
you
see
is
that
there
are
patterns
here
that
go
well
beyond
the
caribbean
basin.
F
You
have,
for
example,
this
couplet
over
the
atlantic
and
eastern
pacific
pacific
basins
and
this
widespread
of
precipitation
or
positive
values
in
the
caribbean
basin
off
to
the
tropical
north
atlantic
and
when
we,
when
we
look
at
the
correlation
of
that
plot,
we
see
it
well
defined,
and
so
this
is
where
we
will
see
most
likely
those
intermodal
differences.
F
But
what
are
these
mechanisms
right?
So,
in
my
observed
framework,
there
are
three
main
facilitators
of
moisture
convergence
that
I
found
for
the
hydroclimate
in
the
caribbean.
One
is
over
the
western
flank
of
the
north
atlantic
subtropical
high
and
then
the
eastern
and
atlantic
branches
of
the
intertropical
convergence
zones
during
the
season
prior
to
the
early
rainy
season.
F
So,
in
this
case,
for
the
hydroclimate
in
february,
looking
at
total
moisture
flux
and
convergence,
we
see
the
high
is
very
strong
and
with
a
divergence
and
easterly
winds
across
the
caribbean
corridor,
we
see
some
patches
of
green
right
over
the
islands
that
denotes
perhaps
the
small-scale
convergence
over
the
islands.
But,
generally
speaking,
we
see
this
widespread
divergence
when
we
move
into
the
month
of
may
or
the
beginning
of
the
early
rainy
season.
We
see
a
lot
of
changes
here.
We
see
the
divergence
band
weaken
across
the
caribbean
basin.
F
The
easterly
winds
become
southeasterly,
so
we're
having
an
influx
of
moisture
from
the
tropics
onto
the
caribbean
basin
and
the
high
produces
on
its
western
side.
This
convergence
band
that
moves
northwest
into
the
northwestern
caribbean
alongside
the
itcz
bands,
migrating
northward,
and
so
this
is
why
we
see
for
the
early,
the
early
rainy
season
and
the
transition
from
the
winter
dry
season.
F
So
when
I
look
at
the
regression
of
other
variables
that
I
know
are
to
inform
the
hydroclimate
in
the
caribbean,
we
see
patterns
that
resemble
these
large
scale
features
so
slp
with
the
zonal
mean
removed,
showing
more
precipitation
with
lower
slp
values
for
ssts.
We
see
warmer,
caribbean
sea
and
the
southern
portion
of
the
north
atlantic
and
under
the
zonal,
winds
low-level
winds.
F
We
see
more
westerlies
or
weaker
or
more
easterlies,
which
denote
a
weakening
of
the
circulation
itself
and
consequently,
thus
a
weaker
high
for
more
precipitation
across
the
caribbean,
and
this
resembles
my
paper
earlier
from
the
intra-annual
variability
of
the
early
and
late
rainy
seasons,
where
these
anomalous
patterns
resemble
the
winter
induced
north
atlantic
oscillation,
where
under
a
wet
phase
a
negative
neo
induces
these
anomalous
signals.
So
there's
something
here
at
play
that
the
models
are
going
to
have
intermodal
differences
with.
F
So,
just
to
then
conclude
here
I
took
spatial
averages
of
these
various
dynamical
variables
between,
for
example,
sea
level,
pressure
anomalies
with
the
zonal
mean
removed
versus
the
land
precipitation
totals
over
the
central
caribbean
for
the
two
months,
and
we
see
very
clearly
this
relationship
where
you
have
lower
sop
and
more
precipitation
compared
to
the
observations,
all
of
which
have
the
slp
value
from
era
5..
F
More
influx
of
divergence
easterlies
across
the
caribbean,
which
might
be
a
reason
for
the
dry
bias
that
we
see
during
the
early
rainy
season.
And
again
we
see
this
simultaneous
across
all
of
the
sub-regions
as
well,
where
all
the
models
show.
The
high
to
be
further
westward
than
what
we
see
in
the
observations,
and
there
is
some
suggestion
of
differences
between
the
high
versus
the
low
resolution
in
in
that
case
as
well.
F
So,
just
to
summarize
and
conclude,
I
performed
several
simulations
of
the
cycle
and
the
mechanisms
under
a
suite
of
these
models.
I
do
see
that
most
models
underestimate
the
land
and
ocean
precipitation,
where
you
would
find
you
know
less
complexity,
for
example,
because
there's
no
land
right.
So,
interestingly,
looking
at
the
model
regressions,
they
suggest
that
perhaps
the
dry
bias
is
in
part
due
to
a
stronger
and
or
westward
shift
of
the
high,
which
causes
the
enhanced
easterlies
and
divergence
and
potentially
also
a
southern
displacement
of
the
itcz.
F
But,
to
conclude
here,
I
will
be
looking
at
some
of
these
other
variables
like
mariano
wind
component
and
look
at
these
other
regressions
indices
and
the
other
seasons.
The
mid-summer,
drought
and
the
late
rainy
season
to
see
if
there
is
consistency
there,
and
so
with
that.
I
I
conclude
my
presentation.
Thank
you
so
much.
F
Looking
at
the
multi-decadal
variability
is
that
correct
yeah
I
haven't,
I
haven't,
looked
at,
you
know
decadal
or
multi-decadal
scales
to
see
if
there
are
differences
here.
What
I
did
find
with
the
anomalous
patterns
that
I
or
in
the
model
regressions
was
there.
F
There
appear
to
be
similarities
between
my
anomalous
plots
that
I
showed
from
a
previous
study
that
looked
at
the
intra-annual
variability
in
the
nao
inducing
those
signals,
as
the
early
rainy
season
is
primarily
like
changes
between
its
dry
and
wet
state,
is,
is
due
to
how
the
nao
how
it
works
during
the
winter
time
preceding
the
early
rainy
season,
and
then
you
see
a
persistent
slp
and
sst
signal
from
ocean
memory
that
continues
on,
but
I'd
be
interested
in
looking
at
those
other
time
scales,
as
those
likely
do
inform
the
hydro
climate
in
the
region
as
well.
B
Yes,
you
can
go
and
share
your
stream
by
the
way,
I
would
like
to
remind
everybody
and
all
the
messages
will
be
recorded.
So
if
you
have
any
questions,
you
can
just
ask
the
previous
speaker.
Okay,
I'm
recording.
G
G
As
you
know,
the
american
mediterranean
sea
basin
is
important
from
the
perspective
of
climate
and
population,
so
we
focused
on
in
this
region.
In
climate
studies,
one
of
the
widely
used
methods
is
utilizing.
Esm's
ssd
is
one
of
the
problematic
parameters
for
esm's
biases
of
ssd
effect
arc
interaction
in
the
model.
We
are
motivated
to
utilize
csm
and
use
the
magic
method
to
make
a
better
representation
of
the
mediterranean
climate.
So
our
purposes
are
applying
sst
imaging
for
only
mediterranean
sea
and
creating
ensemble
using
this
mediterranean
pacemaker
experiment.
G
Then
we
aim
to
investigate
the
role
of
mediterranean
sea
on
the
basin
climate
in
detail.
First,
we
obtain
long-term
simulations
with
pre-industrial
conditions
and
unsurprising
anthropogenic
conditions
separately.
We
prepared
observation
data
and
customized
the
nameless
file
for
the
nudging
method
to
obtain
in
some
simulation.
We
applied
micro,
perturbation
in
the
atmospheric
component
and
we
used
magic
method
on
the
ocean
component
with
the
mediterranean
and
mask
file
and
observation
data.
In
the
figure
located
left
hand,
side
gray
line
shows
the
daily
mean
of
mediterranean
sea
ssd
of
control
simulation.
G
G
We
just
compared
temperature
and
precipitation
with
observation
and
three
upper
layer
parameters
with
three
analysis
data.
We
used
a
simple
index
to
evaluate
the
mediterranean
experiment.
The
index
basically
uses
differences
between
the
control
simulation
and
the
mediterranean
experiment
with
observation.
G
If
the
index
is
higher
than
one,
then
it
means
the
mediterranean
experiment
will
produce
parameter
better
over
that
region.
First,
we
run
the
model
for
200
years
under
pre-industrial
conditions.
Then
we
continue
with
entrepreneurship
conditions
until
2014.
also,
we
used
er,
sst
and
oict
observation
data
to
generate
nudging.
G
Similarly,
we
produced
new
simulations
with
microperturbation.
Basically,
we
have
two
main
experiments,
control
and
mediterranean
simulation.
Each
ensemble
has
nine
members.
This
plus
show
global
means
of
radiation,
temperature,
precipitation
and
sst.
The
left
column
belongs
to
pre-industrial
conditions.
Right
column
belongs
to
control
and
peacemaker
simulation
values,
range
and
trends
in
time
series
low
consistent.
So
we
can
say
that
we
have
a
stability
and
balance
in
the
model.
G
This
map
shows
temperature
mean
of
the
control
simulation
for
a
30
year
period,
figure
b
shows
differences
between
control
and
observation,
control,
reproduce,
run
buyers,
except
for
amazon
based
in
sahara,
region
and
green
land,
also
control
reproduce,
cold
bias
for
north
side
of
the
europe
and
bias
around
the
mediterranean
sea.
This
map
shows
the
differences
between
mediterranean
and
control
simulation
in
general.
Higher
values
are
reproduced
by
midfielders
stimulation
over
the
northern
hemisphere.
G
This
is
the
index
map.
Green
and
blue
colors
show
a
better
representation
of
the
temperature
reproduced
by
mediterranean
simulation.
Accordingly,
we
have
some
improvements
around
the
mediterranean
sea
and
over
northern
europe.
The
meaning
of
these
maps
is
similar
to
previous
one.
In
figure
b,
the
bias
of
precipitation
amount
is
higher
over
equatorial
region.
G
Control
stimulation
overestimates
the
precipitation
of
europe
in
general,
dotted
region
on
the
index
map
show
regions
where
correlation
is
in
significance.
Mediterranean
simulation
is
better
representation
over
the
entering
souls:
digital
water,
street,
turkey,
balkans
and
northeast
part
of
europe.
These
maps
were
plotted
for
850
millibar
temperature
infigurability
counter
simulation
has
cold
bias
in
north
africa
and
europe.
When
we
look
at
index
map,
the
mediterranean
simulation
has
a
better
representation
of
temperature
over
the
north
part
of
europe
along
the
upper
latitudes.
G
This
is
the
wind
wind
is
one
of
the
top
parameters
to
simulate.
When
we
look
at
the
biases
of
figure
b,
we
can
see
latitudinal
patterns.
The
index
map
is
not
that
clear,
but
asia
is
part
of
america.
Africa
and
the
indian
ocean
have
some
improvements
for
now.
As
I
said,
we
have
only
preliminary
results.
We
are
trying
to
obtain
better
stimulation
for
mediterranean
basin.
Accordingly,
the
mediterranean
experiment
has
some
improvements,
but
also
poor
representation
of
temperature
and
precipitation
over
some
regions
to
understand.
Well,
we
will
continue
to
analyze
and
assemble
data
in
detail.
G
We
will
compare
various
parameters
with
other
observations
and
the
analysis
data.
As
I
said,
we
use
monthly
assist
observation
for
this
study.
We
might
write
daily
data
for
logic
method
as
further
study
and
we
might
think
of
member
size
of
ensembl,
and
we
can
try
other
version
of
origin
component
of
csm.
So
thank
you.
Thank
you
for
listening.
A
G
A
G
Not
climatically,
we
use
step
by
step
in
in
each
calculation
in
in
the
ocean.
Part
of
of
the
model.
G
We
haven't
checked
for
the
for,
for
use,
did
you
mean,
might
use
yearly
ssd
data
or
just
analyze
data
year
by
day.
G
For
now,
we
use
it
step
by
step
when
we
nudge
the
model
and
in
the
results
we
use
30
years
time
period.
But
when
we
look
at
the
seasonal
and
day
or
year
by
year,
probably
it
will
change
the
effect
yeah
yeah.
I
Yeah
thanks:
can
you
see
my
screen.
I
Thanks
great
hi
everyone,
my
name
is
teacher
sharon
gu,
I'm
a
phd
student
from
penn
state
university.
Today,
I'd
like
to
talk
about
the
study
done
by
me
and
my
advisor
dr
moise
gervais.
I'm
diagnosing
two-way
coupling
indicator
on
nostalgic
c-surface
temperature
variability
using
time-evolving
self-organizing
maps,
and
the
study
has
just
been
published
in
gl.
I
So,
as
I
have
mentioned
in
the
title
page,
we
are
using
a
method
called
self-organizing
map.
So
what
is
self-organizing
map,
or
sum
so
sum
is
an
unsupervised
machine
method
that
can
be
used
to
classify
high-dimensional
data
sets
compared
to
some
conventional
methods
such
as
eof.
It
is
not
restricted
by
optimality
or
stationarity
and
therefore
it
can
identify
physical
relevant
patterns
so
for
conceptualized
understanding.
Some
is
changed
by
constantly
seeing
the
data
which
is
shown
in
this
purple
data
cloud
here.
I
So
sums
are
often
changed
by
input
data
vectors
that
contain
a
spatial
pattern
for
a
single
time
step,
and
we
call
this
as
an
honors
sum.
So
you
can
understand
the
ordinary
sum
as
classifying
pictures.
However,
the
other
sum
is
limited
in
terms
of
its
ability
to
capture
the
spatial
temporal
evolutions.
So,
for
example,
here
is
a
picture
of
the
kitty
staying
aloft,
but
by
only
looking
at
this
picture,
you
are
not
able
to
understand
or
know
what
happens
before
and
after
this
jump.
I
Therefore,
we
divide
this
new
evolution,
sound
method
or
time,
evolving,
synchronizing
method
based
on
the
ordinary
sum,
so
for
an
evolution
sum
an
input.
Data
vector
is
a
number
of
consecutive
spatial
patterns,
so
basically
we're
classifying
videos
so
by
looking
at
the
same
kitty
example
here,
if
you,
if
that
is
a
video
here,
you
are
able
to
understand.
Why
does
this
kitty
drum
and
what
happens
afterwards
right?
I
So
this
is
an
advantage
of
the
evolution
sound
that
is
able
to
characterize
variability
in
space
and
time
simultaneously.
So,
in
our
study,
we
applied
this
method
to
classify
the
spatial
temporal
evolutions
of
winter,
north
atlantic
essentia
anomalies
over
the
course
of
10
consecutive
years.
In
this
way
we
are
able
to
identify
evolutions
without
time
fieldering.
Therefore,
we
can
capture
both
inter-annual
and
decay
variability,
and
this
would
be
more
similar
to
what
would
be
produced
in
a
decadal
prediction.
I
Then,
let's
look
at
some
results,
so
these
are
the
24
generalized
10-year
evolutions
we
classified
from
the
non-sedantic
sst
anomalies
in
the
csm1
pi
control
simulation.
So
these
results
highlight
the
diversity
of
this
severability
in
the
north
atlantic
and
in
our
forum
study
we
are
exploring
these
different
modes
of
variability.
I
I
This
is
the
pattern
to
a
positive
amb
like
or
authentic,
multi-decadal
variability,
like
ssd
pattern
to
a
weak,
negative
online
tc
pattern
over
10
years,
as
you
can
see
in
the
years,
2
4,
6
and
8
and
10
here
so
focusing
on
this
specific
node
allows
us
to
demonstrate
how
the
evolution
sound
method
can
provide
new
insights
into
this
into
the
mechanisms
involved
in
this
well-known
aomb
transition
and
with
this
evolution
sound
method,
we
identify
spatial
patterns
that
cannot
be
well
captured
by
the
av
index,
as
shown
in
the
top
here,
and
we
are
able
to
capture
the
io
through
amd
transition
without
using
any
time
filtering,
and
we
found
this
evolution
occurs
within
six
years.
I
So
in
the
paper
we
talked
about
three
dominant
mechanisms,
including
the
buoyancy
journal,
ocean
circulation,
wind
german
circulation,
as
well
as
the
chain
84
scene.
So
in
this
presentation,
due
to
time
limits,
I
will
just
focus
on
the
trans-eddy
forcing
mechanism,
because
this
localized
chain,
edit,
forcing
might
be
hard
to
capture
in
some
other
frameworks.
I
We
use
this
diagnostic
method
proposed
by
hosting
adult
called
e-vector,
so
the
e-vector
can
be
used
to
describe
vertical
adhesivity
propagation
and
its
feedback
onto
the
mean
flow,
and
there
are
two
components
associated
with
the
e-vector.
The
virtual
component
is
the
mean
v,
prime
t,
prime
at
850
hecta
casco,
and
it
is
the
lower
troposphere,
eddy
heat
transport
and
it
can.
I
It
is
a
measure
of
the
vertical
eddy
activity,
propagation
and
the
horizontal
component
of
the
vector
is
showing
this
math
expression
at
200
hecasco,
and
it
is
an
estimate
of
the
eddy
momentum
forcing
of
the
zomotomy
flow.
So
in
order
to
isolate
the
bioclinic
adhesivity,
we
use
our
high
pass
filtering
to
the
wing
and
temperature
fields.
I
Then
let's
look
at
the
composite
some
composite
analysis.
So
this
is
the
sea
level
pressure
and
surface
wind
composite
onto
the
year
2
of
this
node
1
6..
So,
as
you
can
see,
the
dipole
in
the
cll
pressure
field,
which
is
the
shading
here,
indicates
a
strong
positive
and
a
light
signal
and
consistent
with
this
co
pressure
field.
I
Tropical
nozzle,
antique
and
on
field
counters
here
shows
the
mean
v,
prime
t,
prime,
at
850
hectacosco
and,
as
you
can
see,
due
to
the
surface
diagonality
associated
with
this
large
sst
gradient,
there
is
enhanced
and
upward
gated
eddy
activity,
so
this
low-level
eddy
activity
will
lead
to
some
change
in
the
upper
troposphere,
as
shown
in
this
figure.
So
this
figure
shows
the
composites
of
the
horizontal
e
vector
anomalies
at
200,
heck
costco
in
vectors
and
the
e-vector
divergence
anomalies
in
shading.
I
So,
as
you
can
see
associated
with
the
lower
level
at
the
activity,
there
is
anomalous
divergence
of
the
horizontal
vector,
anomalies,
components
in
the
upper
top
sphere,
as
shown
in
the
warm
shadings
in
this
region
marked
by
this
green
box.
So
this
implies
that
the
enhanced
adiactivity
is
providing
energy
to
the
mean
flow
in
the
upper
troposphere.
I
So
this
energy
provided
by
the
80
will
lead
to
an
acceleration
of
the
jet.
So,
as
you
can
see
in
the
third
figure
where
the
shading
is
the
composite
of
the
wing
speed
anomalies
at
200
hex
glasgow,
there
is
a
port
shifted
and
eastward
extended
jet,
as
shown
in
the
red
shading,
compared
to
the
climatology,
as
shown
in
the
purple
counters
here,
and
the
location
of
the
positive
anomalies
of
the
jet
corresponds
well
with
the
e-vector
divergence
region
here.
I
So
I'm
sorry
that
I
didn't
have
time
to
touch
all
three
mechanisms
in
details
here
because
of
the
time
limit,
but
please
feel
free
to
check
this
paper
or
reach
out
to
me
if
you
would
like
to
learn
more
and
I'm
currently
visiting
in
car.
So
if
you
are
in
boulder,
I'm
also
very
happy
to
talk
in
person
with
that.
I'm
happy
to
take
any
questions
thanks.
I
They
are
like
in
terms
of
the
like
pattern
itself:
it's
not
really
independent,
because
self-managing
map
captures
a
continuous
change
of
the
patterns,
but
in
terms
of
the
years
associated
with
each
pattern
like
each
pattern,
they're
independent,
so
one
year
can
only
be
a
one.
Ten
year
period
can
only
be
belong
to
one
of
the
sounds
self
self,
organizing
or
self
going
maps
nodes.
Does
that
make
sense.
D
Yeah
thanks
for
that
talk,
I'm
wondering
if
the
som
nodes
are
multivariate
nodes
or
are
they
just
based
on
sst?
They.
I
Oh
sure,
so
let
me
show
you
a
schematic
here.
I
Oh
wait
a
minute,
so
basically,
so
each
of
the
input
data
vector
will
belong
to
one
of
the
sum
node.
So
here,
for
example,
I
have
a
by
six
sum
over
here
and
the
each
of
the
input
data
vector
will
belong
to
one
of
the
24
nodes
here.
So,
for
example,
year
2
1
will
belong
to
this
node
and
to
do
a
composite
analysis.
We
can
just
select
these
year
numbers
that
belongs
to
one
node
and
do
the
composite
here.
Does
that
make
sense.
I
Not
one
year
so
each
node
has
10
year
of
the
has
a
10
year
and
the
data
the
input
data
vector
actually
all
contains
10
years.
So,
for
example,
the
first
input
data
vector
is
from
year
1
to
year,
10
and
second
is
from
year
2
to
year
11..
So
each
imported
vector
contains
10
so
yeah.
So
when
you
do
the
composite,
you
are
just
like
averaging
over
all
the
input
data
vectors
that
belongs
to
this
node,
which
all
contain
10
years.
Does
that
make
sense.
I
I
Oh,
that's
a
good
question,
so
actually,
in
our
full-on
study,
we
are
actually
looking
at
all
the
nodes
and
they
are
all
have
different
behaviors.
But
in
this
particular
study
we
looked
at
this
node
because
this
node
shows
an
aom
transition.
That
is
very
well
known
and
we
would
like
to
like
kind
of
prove
that
our
method
is
used
for,
and
we
can
also
provide
some
new
insights
in
this
and
focusing
on
onenote
allows
us
to
use
a
large
number
of
diagnostic
variables.
J
K
K
So
I
believe
everyone
knows
about
enso.
It's
the
primary
source
for
inter
annual
variability
over
the
tropical
pacific
ocean,
and
it
has
it
has
wide-ranging
global
impacts,
and
one
aspect
is
that
enzo
can
effectively
modulate
global
tropical
cyclone
activity
by
altering
the
large-scale
conditions.
For
example,
during
el
nino
years
we
typically
have
more
tcs
tropical
cyclones
in
the
western
pacific
and
less
tissues
in
the
north
atlantic,
but
today's
talk
focuses
on
the
reverse
relationship
or
tc's
impact
on
them.
K
So
you
may
be
wondering
why
do
we
ask
this
question
in
the
first
place,
it
helps
if
we
take
a
look
at
the
enzo
dynamics.
So
essentially,
enso
is
a
neutrally
stable,
natural
mode
of
oscillation
resulting
from
errors
interactions.
It
is
sustained
by
stochastic
atmospheric.
Forcing
and
enzo
characteristics
depend
on
three
important
factors:
the
ocean
state,
the
wesley
wind
bursts,
as
well
as
the
ocean
hemisphere
processes,
including
the
alpha
kelvin
wave
propagation
and
a
warm
pool.
Expansion
and,
interestingly,
tropical
cyclones,
can
influence
all
of
these
three
critical
factors.
K
Let
me
elaborate
on
each
of
them
so
for
the
initial
ocean
state.
Recent
studies
have
found
that
tropical
cyclones
can
actually
increase
the
pacific.
Tropical
pacific
ocean
heat
content
through
their
interactions
with
the
upper
ocean.
Specifically
tcs,
can
cause
intense
vertical
mixing
that
eventually
pump
heat
into
the
ocean
thermocline,
for
example,
in
this
experiment.
K
We
add
transient
tropical
cyclone
winds
in
the
global
ocean
model
and
found
that
the
added
tcs
can
cause
a
anomalous
warming
in
the
equatorial
thermocline,
especially
in
the
eastern
pacific,
and
this
kind
of
pattern
is
conducive
to
el
nino
development,
and
the
second
factor
is
the
wesley
wind
breasts,
and
recent
studies
found
have
found
that
the
majority
of
the
near
equator
tcs
are
associated
with
the
wesley
weinberg
events.
So
this
provides
another
ingredient,
important
ingredient
for
el
nino
characteristics
and
thirdly,
the
ocean
equatorial
cowboy
propagation
that
can
help
deepen
the
thermocline
depth.
K
So
in
this
animation,
I'm
showing
the
ocean
temperature
anomalies
in
response
to
tropical
cyclone
forcing
these
moving
black
dots
are
the
tropical
cyclone
locations
and
blue
and
yellow
at
temperature
anomalies.
So
we
can
see
some
near
equator.
Tcs
in
this
model
can
generate
a
yeast
propagating
eastward
propagating
kelvin
waves
and
carrying
with
them
are
the
tropical
cyclone-induced
thermocline
heating.
So
so
far
we
see
that
tropical
cyclones
can
influence
all
of
these
three
important
factors
for
enzo
characteristics.
K
So
naturally
our
question
is
to
to
explore.
Our
goal
is
to
explore
the
impact
of
tropical
cyclone
winds
on
enso
characteristics
and
in
the
context
of
a
fully
coupled
earth
system
model.
K
K
This
simulation
is
mainly
used
to
provide
tc
when
forcing
so
at
this
resolution,
the
model
can
generate
their
own
tropical
cyclones
with
climatology
generally
consistent
with
the
observations.
So
our
first
step
is
to
find
and
detect
all
these
tcs
in
the
model
and
for
each
tc
pc
day.
We
would
take
a
snapshot
of
the
tc
surface
wind
within
a
20
degree
domain
box
and
when
they
are,
and
then
we
archive
the
surface
winds
into
a
tcu
enforcing
inventory,
oops
and
and
then
we
start
running
this
low
resolution
fully
compound
model.
K
This
model
has
a
one
degree
atmosphere
which
does
not
resolve
tc
strength,
ones,
at
least
very
little,
so
we're
gonna
add
these
tc
snapshots
into
the
low-res
model.
As
the
model
integrates
so
that
we
can
come
up
with
this
low-res
tc
simulation,
where
the
tcs
impact
is
embedded
in
this
void
cover
model,
so
the
difference
between
the
low-res
control
and
the
lower
stc
is
just
going
to
be
the
added
transient,
surface
tc
winds,
and
we
can
diagnose
the
impact
of
tc
when
forcing
within
this
couple
climate
system
by
comparing
these
two
simulations.
K
So
it
turns
out
the
tcs
have
a
wide
ranging
impact
on
the
global
climate
in
the
simulation,
and
one
was
one
of
the
most
interesting
aspects
is
its
impact
on,
and
so
so
here
I'm
looking
we're
looking
at
the
time
series
of
the
nino
index
of
this
100
simulation
left
is
in
the
control
and
the
right
is
in
the
tc
run.
So
just
by
eyeballing,
we
can
see
that
the
tc
run
has
more
strong
to
extreme
el
nino
events,
and
then
we
did.
We
look
at
the
monthly
standard
deviation
of
the
nino
index.
K
This
black
curve
shows
what
it
looks
like
in
the
observations
and
the
blue
is
from
the
control
and
the
red
is
from
the
tc
run.
So
we
see
that
after
you're
adding
these
transient
winds,
your
lineal
standard
deviation
has
increased,
but
your
the
shape
of
your
annual
cycle
in
the
tc
run
is
actually
more
consistent
with
the
observations
compared
to
the
control
and
then
on
the
bottom
right.
We're
looking
at
the
power
spectra
of
the
nino
index.
Again,
the
black
shows
what
it
looks
like
in
the
observations.
K
K
We
can
see
that
the
tc
simulation
has
a
lot
more
strong
el
nino
events
with
warmer
sst
in
the
central
to
eastern
pacific
and
in
fact
there
are
18
el
nino
events
in
the
control
simulation
and
two
of
them
are
el
nino
extremes
and
there
are
19
el
nino
events
in
the
tc
run,
and
six
of
them
are
el
nino
extremes,
it's
just
quite
a
increase
and
on
the
right
figure
we're
looking
at
a
linear
regression
between
the
nino
3.4
index
of
the
strong
el
nino
versus
the
near
equator,
integrated
tc
power,
a
tc
wind
power.
K
So
we
can
see
we
find
this
very
strong,
positive
relationship
between
tc
wins
and
el
nino
strength,
and
especially
there
is
a
very
strong
event.
It's
this
guy
right
here
and
it's
also
associated
with
the
largest
tc
when
tc
went
forcing
and
then
to
better
understand.
K
What's
going
on,
we
have
been
looking
we're
going
to
look
at
the
composite
analysis
of
the
el
nino.
So
here
in
the
top
figures
are
the
composite
el
nino
sst
the
left
is
control,
the
middle
is
the
tc
and
the
difference
are
shown
in
the
rightmost
column.
So
we
can
see
that
the
tc
run
has
a
warmer
el
nino
sst,
especially
during
the
peak
as
well
as
when
the
el
nino
is
coming
to
an
end,
which
is
quite
interesting,
and
the
bottom
shows
the
composite
thermocline
equatorial
pacific
thermocline
depth
anomalies.
K
We
can
see
that
there
is
a
deeper
thermocline
in
the
early
season
in
the
western
pacific
and
it
kind
of
propagates
eastward
as
the
el
nino
develops,
and
there
is
a
strong,
deep,
deeper,
deeper
thermocline
in
the
at
the
time
when
el
nino
is
developing
and
also
at
the
time
when
el
nino
is
going
to
an
end.
I
need
to
mention
that
this
black
contours
on
top
of
the
colors
are
the
composite
tropico
cyclone
wind
forcing
so
we
see
that
this
deeper
thermocline
largely
overlapped
with
the
enhanced
tc
winds.
K
Additionally,
we
also
looked
at
the
zonal
equatorial
current
anomalies
and
we
see
that
there
is
enhanced
zonal
current
in
the
tc
simulation,
which
is
also
conducive
to
el
nino
development.
So,
put
all
of
this
into
perspective.
We're
going
to
take
a
look
at
the
el
nino
dynamics,
specifically
we're
going
to
look
at
the
heat
budget
of
the
ocean
mixer
in
equatorial
pacific.
K
So
the
heat
budget,
the
two
dominant
terms
of
the
heat
budget,
is
the
thermocline
feedback
term,
as
well
as
the
vaccine
feedback
term.
So
we're
going
to
look
at
the
composite
of
these
two
runs
in
terms
of
these
two
terms,
so
the
top
shows
the
thermocline
feedback
term.
So
we
can
see
that
the
tc
wind,
the
tc
simulation,
has
a
stronger
thermocline
feedback,
especially
in
the
eastern
pacific.
Actually,
this
pattern
is
quite
consistent
with
the
thermocline
type
phenomena
we
saw
in
the
last
slide
and
during
the
peak
season.
K
The
increase
of
the
thermocline
feedback
is
about
20
percent
compared
to
the
control
20
more
and
then
the
bottom
shows
the
reaction,
feed
feedback,
so
a
lot
of
action
feedback.
We
see
these
patterns
actually
very
consistent
with
the
increase
of
zonal
current.
We
saw
in
the
last
slide
so
and
the
magnitude
is
comparable
with
the
thermocline
feedback.
So
from
this
dynamic
analysis,
we
found
that
a
tcs
can
impact
el
nino
through
enhanced
the
thermal
client
feedback
and
zone
of
action
feedback
in
the
heat
budget
terms,
so
yeah.
K
This
is
the
summary
of
what
what
I
just
said.
So
we
basically
we
are
interested
in
the
impact
of
pc
wins
on
enso.
So
we
designed
this
experiment
where
we
can
prescribe
high
resolution.
Tc
wins
into
a
low
resolution
fully
coupled
csm,
and
then
we
find
that
added
tc
wins
can
increase
the
enzo
power
and
shift
it
shift
to
the
end.
K
So
frequency
more
approaching
the
observations
and
we
found
that
there's
more
strong
to
extreme
el
nino
events
when
we
add
in
the
tcs
and
there's
a
positive
correlation
between
strong
el
nino
and
tc
went
to
power,
and
we
also
found
that
the
the
reason
why
diseases
can
impact
and
so
is
through
the
enhanced
zone
of
action,
feedback
and
thermal
cloud
feedback.
K
So
our
results
indicate
that
these
tcs
can
actively
influence
and
so
in
a
fully
coupled
comprehensive
earth
system
model
pointing
to
an
important
two
interactions
between
transient,
tc
events
and-
and
so
that's
all.
I
have
thank
you.
B
C
Hi
thanks
for
your
presentation,
I
was
curious
if
you've
looked
at
how
the
spectra
of
the
zonal
winds
changed
over
the
equatorial
pacific,
when
you
include
the
tc
ones,
particularly
because
I
think
it'd
be
interesting,
because
a
lot
of
models
don't
get.
The
statistics
of
wind
bursts
correctly
and
I'm
wondering
if
by
adding
in
the
tc
wins
you
get
a
little
bit
closer
to
observations.
K
B
D
K
B
D
K
I
K
Sorry
about
that,
my
computer
is
frozen.
A
A
I'll
just
give
a
summary
of
kind
of
basically
the
new
simulations
that
are
about
to
be
available
or
have
just
become
available
and
the
new
simulations
that
we'll
be
doing
in
the
next
few
months.
So
I
I
only
want
to
spend
a
few
minutes
here,
because
I
think
the
important
thing
is
to
have
the
discussion
so
one
of
the
new
simulations
that
is
available,
the
website's
not
quite
up
yet,
but
it
will
be
soon-
are
these
pre-industrial
controls
with
the
new
vertical
grid
of
cam.
A
So
on
the
left
and
the
colors
there,
you
see
the
new
vertical
grids
that
cam
will
be
using
in
the
next
generation,
and
you
can
compare
that
with
the
gray
and
the
black,
which
is
the
old
vertical
grids.
A
And
so
the
grid
spacing
is
on
the
x-axis
and
the
height
in
the
model
is
on
the
y-axis
and
so
basically
the
red
one
here
is
kind
of
the
workhorse
one
and
it
has
a
higher
top
than
cam
and
it
has
higher
vertical
resolution
a
grid
spacing
of
about
500
meters
in
the
free
troposphere
and
lower
stratosphere,
and
so
we've
run
two
pre-industrial
controls
that
are
kind
of
with
the
next
grid,
except
the
boundary
layer
additions
that
are
circled
here.
A
A
So
there
are
two
pre-industrial
controls
of
100
years
length
with
this.
This
new
vertical
grid-
and
this
has
a
representation
of
the
qbo.
It
has
a
higher
model
top
so
hopefully
an
improved
representation
of
stratospheric
variability.
And
so
these
are
unsupported
and
undocumented
at
the
moment,
so
use
these
with
care.
They
are
available
on
cheyenne
and
we
hope
to
document
them
at
some
point
in
the
near
future.
A
We're
also
probably
going
to
run
some
aim
up
simulations
and
a
historical
simulation
with
this
grid
too.
In
the
near
future,
single
forcing
large
ensemble
is
coming
extremely
soon.
I
had
hoped
to
get
the
description
paper
done
before
this
meeting,
but
I
have
failed
to
do
that,
so
I
apologize
it
will
hopefully
be
released
in
july.
A
So
this
is
complementary
to
the
second
50
members
of
the
csm
two
large
ensemble,
with
forcings
held
fixed
at
1850s,
except
those
of
interest,
and
there's
going
to
be
four
sub-ensembles
with
greenhouse
gas,
forcing
in
one
anthropogenic
aerosols
and
one
biomass
burning
aerosols
in
one
and
then
everything
else
other
than
those
three
forcings
is
are
in
the
fourth
ensemble,
and
I
won't
go
through
this.
A
But
this
is
a
kind
of
global
mean
temperature
response
in
the
single
forcing
large
ensemble,
and
I
should
mention
that
that
non-rose
and
bloom
did
the
the
work
towards
producing
this
okay.
So
the
new
new
simulations
that
we're
going
to
do
in
the
remainder
of
this
allocation
we're
going
to
do
a
regionally
refined,
1
8
degree
north
atlantic
amip
simulation
using
this
grid
that
was
developed
by
rob
england,
wills
and
adam
harrington.
A
When
we
go
to
this
kind
of
somewhat
unprecedented
high
resolution
for
for
this
length
of
simulation,
there's
going
to
be
a
100
1
000
year
long
pre-industrial
control
with
this
pencil
model,
which
has
got
kind
of
single
column,
ocean
models
at
each
grid
point
so
somewhere
between
the
slab
ocean
and
a
full
3d
ocean,
and
and
has
the
advances
over
the
slab
ocean
that
will
be
like
prognostic
mixed
layer
depths
and
things
and
sarah's
also
been
working
with
her
group
on
this
mechanically
decoupled
csm,
which
she'll
give
an
update
on
probably
at
the
winter
meeting.
A
And
then
the
final
big
project.
With
the
remainder
of
the
allocation
is
led
by
judith
burner,
and
this
is
going
to
be
a
1
000
year,
long
pre-industrial
control
with
stochastic
physics,
and
so
I
will
not
go
into
any
more
details
there.
A
I
just
wanted
to
give
a
summary
in
case
people
missed
it
on
monday
of
the
things
that
are
happening,
but
what
we
really
want
to
do
now
is
have
the
discussion,
so
we
need
to
come
up
with
our
proposal
for
the
simulations
for
the
next
two
years
of
computing
allocation
and
we're
gonna
go
into
breakout
rooms.
To
do
this,
I
guess,
unless
anyone
has
any
questions
on
what
I
just
presented
kind
of
rapidly
there,
so
we
don't
have
long
we're
supposed
to
have
a
10-minute
break
in
10
minutes
time.
A
A
Okay,
that
was
a
fast,
fast
10
minutes,
but
we'll
put
the
the
google
doc
in
the
chat.
So
if
you
have
any
other
ideas,
we'll
probably
email
around
to
the
working
group
as
well
as
we're
getting
the
proposal
together
over
the
next
month
or
so
so,
yeah
keep
thinking
about
experiments
that
you'd
like
to
see.
A
C
You
are
correct,
hassan
is
filling
in
for
that.
Oh
okay,
so
we've
got
a
last
minute
edition,
which
is
great.
A
C
All
right,
it
looks
like
we're
at
4
30..
So
just
a
quick
reminder
for
the
speakers.
It's
a
12
minute
slap
but
you'll
have
10
minutes
to
give
your
presentation
if
you're
starting
to
go
over
I'll.
C
Give
you
a
a
reminder
around
eight
minutes
that
you've
got
about
two
minutes
to
finish
up
and
if
I
mispronounce
any
of
your
names,
please
when,
when
you
begin
your
presentation,
reintroduce
yourself
with
the
correct
pronunciation,
so
we
can
all
learn
how
we
should
be
saying
it,
because
we
want
to
make
sure
we
say
your
names
correctly.
M
All
right,
so
thank
you.
I'm
adi
hassan,
a
psd
student
at
nc
state
university.
I
work
with
I
work
in
dr
salaris
and
group,
so
today,
I'll
be
talking
on,
as
sarah
said,
projected
future
changes
in
the
role
of
pikmin
heat
flux
on
ssd
variability
I'll
quickly
start
with
a
very
brief
description
of
a
particular
type
of
atmospheric
circulation,
which
is
a
headless
circulation.
M
As
you
can
see
in
the
figure
right
headless
circulation,
it
is
basically
a
thermally
driven
medium
circulation
and
it
is
it
transfer
its
transport,
energy
and
momentum
towards
the
pole
and,
as
you
can
see
in
the
figure,
the
ages
of
heavenly
cell
is
where
the
surface
is
steadily
changed
to
the
westerlies,
and
there
has
been
a
prior
if
research
was
showing
that
the
climate
change
impact,
the
hadley
circulation
in
particular,
there
is
a
polar
expansion
of
the
headley
boundary
and,
as
you
can
see,
from
the
figure
on
the
right
from
president
davis
paper,
where
the
blue
line
is
showing
the
headley
boundary
computed
from
the
reanalysis
products.
M
This
is
from
southern
hemisphere
over
the
historical
period.
Consequently,
there
is
a
polar
displacement
of
the
subtropical
wind
build
as
a
result.
This
will
impact
the
winter
even
mean
ocean
circulation,
and
there
is
a
lot
of
literature
that
shows
how
the
climate
change,
and
especially
the
headley
expansion,
can
impact
the
mean
ocean
circulation.
M
For
the
sake
of
time,
I
cannot
get
into
details
of
all
of
these,
but
I
show
some
snapshot
of
few
papers
were
people
looking
at
the
mean
ocean
circulation
chains
due
to
the
climate
change
or
hadley
expansion,
either
in
a
regional
or
or
global
perspective.
M
However,
it
is
less
clear
how
the
hadley
expansion
changes
the
ocean
variability,
and
this
sets
up
the
motivation
of
my
research,
and
that
is
any
change
in
the
prevailing
surface.
Wind
due
to
the
headley
expansion
may
have
potential
influence
on
the
sst
variability,
because
two
of
the
important
drivers
of
the
existing
variability
that
are
turbulent
erc
heat
flux
and
heat
flux
due
to
the
ecma
ecman
heat
advection,
are
related
with
the
prevailing
surface
swing.
M
Therefore,
on
a
large
scale,
both
of
these
fluxes
are
related
through
their
mutual
dependence
on
the
on
the
wing,
and
we
can
see
that
relationship
from
this
figure
and-
and
this
is
showing
the
special
pattern
of
the
anomalous
turbulent
hip
flux
and
ecman
hip
flux.
Where
you
see
the
blue.
This
is
where
the
anomaly
fluxes
are
opposite
sign,
and
so
they
are
negatively
correlated
where
you
see
the
red,
they
have
the
similar
sign
of
the
anomalies
and
they
are
positively
correlated.
M
So
if
we
look
at
the
subtropic,
where
there
is
a
mean
easterly
wind,
this
is
where
the
anomalous
eggman
opposes
the
anomalous
turbulent
heat
effective
at
sst
variability,
whereas
in
in
mid-latitude
they
reinforce
anomalous,
ecman
green
forces
dynamics,
turbulent
refractive
and
ssd
variability.
M
M
So,
based
on
this,
we
make
a
hypothesis
saying
that,
in
a
future
climate,
the
polar
expansion
of
the
headless
cell
and
the
associated
changes
in
the
prevailing
wind
will
result
in
a
polar
shift
of
the
eightmen
transition
boundary
and
it
will
have
an
important
implication
towards
the
subtropical
ssd
variability
that
we've
just
seen
in
the
previous
figure
and
the
objective
of
this
research
is
to
quantify
and
compare
the
anthropogenic
trends
in
both
the
hadley
and
action
transition
boundary
based
on
the
re-analysis
product
and
siemens
couple
models
over
the
historical
and
future
climate
scenarios.
M
I
use
three
different
free
analysis:
data
set
that
are
era
five
and
sub
n
current
jr55,
along
with
eight
sim
six
models
that
are
listed
here,
with
their
corresponding
ensemble
member
for
historical
simulation
and
also
the
future
climate
simulation
for
future
climate.
I
use
a
scenario
where
there
is
a
highest
greenhouse
gas
emission
rate
and
the
period
is
1979
to
2014
for
historical
period
and
2015
to
2100
for
future
run
simulation.
M
So
how
did
you
calculate
the
headley
boundary?
So
we
we
noticed
that
from
subtropic
to
mid
latitude,
the
prevailing
surface
surface,
wind
change
from
easterly
towards
to
the
westerly
wind,
and
if
we
take
a
look
at
the
simple
scam
schematic
showing
on
the
right,
if
we,
if
we
consider
I'm
sorry
if
we
consider
the
zonal,
mean
zonal
wind
stress,
it
changes
sign
from
subtropic
to
mid
latitude
and
we
grab
the
zero
crossing
where
it
changes
the
sign,
and
this
is
basically
the
headley
boundary
and
similarly
for
the
ekman
transition
boundary.
M
M
This
is
the
plot
showing
the
correlation
between
anomalous,
turbulent
heat,
truss
and
anomalous.
Ecman
heat
flux,
similar
to
what
we
we've
just
seen.
But
this
is
a
time
versus
latitude
plot
where
the
special
pattern.
This
is
a
zonally
average.
So
in
the
x-axis
we
see
the
time
and
y-axis
we
see
the
latitude.
So
in
the
subtropic
we
see
that
the
fluxes
are
negatively
correlated
and
mid-latitude
they're
positively
created,
and
we
see
a
consistency
of
this
relationship
throughout
the
historical
period
and
the
black
solid
line.
M
Here
is
the
ecman
transition
boundary,
which
is
is
showing
the
where
the
ekman
changes
its
role
and
the
dotted
line
is
showing
the
headley
boundary.
So
over
the
time
they
they
move
in
tandem,
and
this
result
is
based
from
the
era
fibrino
and
the
analysis
data
set.
Now
if
we
combine
result
from
re-analysis
and
all
the
semi-six
model
that
we
consider.
M
So
a
positive
value
in
northern
hemisphere
means
a
polaroid
expansion.
This
is
the
quantile
which
shows
the
polar
expansion
of
both
boundaries
and
the
the
small
circle
are
the
result
from
each
individual,
ensemble
members
and
the
large
closed
circle.
Are
the
ensemble
mean
results?
The
black
triangle
is
the
multimodal
ensemble
and
the
black
triangle
is
the
reanalysis
mean.
So
we
see
that
the
majority
of
the
ensemble
mean
results
are
falling
in
in
this
quantile,
which
is
a
polar
expansion
of
of
the
both
the
headline.
M
Etman
transition
boundary
and
the
the
shading
you
see
like
the
oval
shape
is
the
confidence
elite
calculated
from
the
ensemble
mean
we
see
that
the
reanalysis
mean
falling
out
of
the
confidence
elite,
but
but
it
is
within
the
spread
of
the
ensemble
results.
So
that
means-
and
we
use
three
re-analyzes
over
a
relatively
shorter
period
of
time,
so
natural
variability
can
dominate
from
the
re-analysis
result.
Alternatively,
the
semi-six
result
shows
the
force
response
and
it
it's
clustered
around
zero.
M
So
there
is,
we
do
not
see
an
obvious
force
response
and
similarly,
if
we
look
at
the
southern
hemisphere,
we
see
that
the
the
change
in
the
boundary
is
larger
than
the
northern
part
and
we
sort
of
see
a
linear
relationship
between
the
hadley
boundary
and
ecma
transition
boundary
for
southern
hemisphere.
This
is
the
quantile
where
both
the
boundary
show
a
polar
expansion.
M
Now,
if
we
look
at
the
future
simulation,
thank
you.
We
quickly
noticed
that
the
trend
does
emerge
in
northern
hemisphere
and
both
in
the
southern
hemisphere.
The
the
expansion
is
more
pronounced
in
the
southern
hemisphere,
where
the
multimodal
ensemble
mean
for
both
the
hadley
boundary
and
equine
transition.
Boundary
has
a
relatively
1.5
close
to
1.5
degree
expansion,
and
this.
This
dash
line
is
showing
us
the
perfectly
linear
relationship,
so
that
we
can
have
some
sort
of
idea
how
the
headley
boundary
modulates
the
equipment
transition
boundary.
M
As
conclusion,
I
make
a
overall
general
hypothesis
showing
that
the
equine
transition
boundary
migrates
with
the
header
boundary
and
based
on
the
real
analysis
and
the
cmipsip
cmx6
simulation
results.
We
see
that
the
equine
transition
boundary
is
generally
modulated
by
the
hadley
boundary
both
over
the
historical
and
future
climate
simulation
and
anthropogenic
trend
is
a
small
over
the
historical
period.
M
However,
in
future
climate
projection,
the
trend
does
emerge,
and
these
are
the
exact
number
of
the
expansion
coming
from
the
future
climate
projection.
M
And
so
it
is
critical
for
climate
models
to
predict
the
migration
of
the
headley
boundary,
with
a
reasonable
accuracy
to
simulate
potential
changes
in
the
ocean
variability
in
future
climate,
and
also
the
next
thing
that
I
would
like
to
look
up
is
to
to
analyze
the
temperature
budget
at
this
expanded
latitude
to
see
or
to
quantify
the
overall
sst
variability.
C
N
N
Perfect,
okay,
so
in
a
moment
I
will
be
talking
about
how
consideration
of
the
variability
in
biomass
burning
emissions
can
act
as
an
important
contribution
to
the
total
aerosol
forcing
but
first
I'd
like
to
acknowledge
my
graduate
advisor
hunty
singh,
as
well
as
our
collaborators
in
particular
phil
rash,
whose
ideas
really
form
the
basis
for
this
work,
but
also
help
from
patricia
dirk
and
kidney
as
well
as
haruki.
Hirosawa.
N
This
was
in
contrast
to
the
csm1
csm1
contributions
to
csm5,
where
the
arctic
sea
ice
loss
was
much
smoother
and
what
they
were
able
to
do
is
actually
attribute
this
to
the
way
that
the
biomass
burning
emissions
were
prescribed
within
siemens
six.
N
So
if
we
look
at
the
emissions
over
time,
what
we
can
see
is
that
in
the
at
the
end
of
the
historical
period,
there
is
this
large
step
change
in
the
prescribed
variability,
and
this
is
due
to
the
use
of
satellite
observations
during
this
period.
N
And
what
patricia
derektini
and
colleagues
were
able
to
show
was
that
this
rapid
acceleration
in
arctic
sea
ice
loss
was
actually
directly
attributable
to
this
step.
Change
towards
a
larger
variability
in
the
prescribed
biomass
burning
emissions
and
then,
in
conjunction
with
this
work,
a
study
led
by
john
pasulodol
this
year
was
able
to
link
this
increase
in
biomass
burning
emissions,
variability
with
a
cloud-thinning
effect,
which
was
particularly
strong
in
the
northern
hemisphere,
mid
to
high
latitudes
and
then
alongside.
N
Then,
following
this,
we
were
able
to
use
the
csm2
large
ensemble
to
show
that
this
increase
in
biomass
burning
emissions
variability
also
amplified
the
hydrologic
cycle.
So
in
particular,
we
saw
increases
in
evaporation,
atmospheric
water
vapor,
as
well
as
precipitation,
again
all
concentrated
in
that
northern
hemisphere
mid
to
high
latitude
region
and
then
to
put
some
of
these
numbers
into
context.
N
N
So
from
these
past
three
studies,
we
can
see
that
consideration
of
this
variability
of
biomass
burning
emissions
is
quite
important
when
it
comes
to
properly
representing
our
simulated
climates.
But
a
big
question
that
still
remains
is
how
important
is
this
to
trying
to
quantify
the
aerosol
forcing
so
in
order
to
tackle
this,
we
designed
and
ran
a
set
of
idealized
simulations.
So
we
did
these
within
csm2
using
2000's
climatology.
N
N
So
what
I'm
showing
here
is
two
of
these
simulations,
where
first
we
ran
a
fixed
emissions
baseline,
where
we
held
annual,
mean
biomass,
burning
emissions,
constant
and
then
we
also
ran
a
second
simulation
where
we
exposed
the
model
to
this
high
variability
period
from
the
cmip6
prescribed
biomass
burning
emissions.
N
So,
by
looking
at
the
difference
between
these
two
models,
this
allowed
us
to
see
the
difference
that
this
variability
in
the
biomass
burning
emissions
has
on
the
overall
radiative
forcing
so
when
looking
at
the
global
mean,
we
can
see
that
this
tends
to
increase
the
radiative
forcing
by
about
order
of
0.1
watt
per
meter
squared
in
the
annual
mean,
but
we
can
see
that
this
is
largely
driven
by
increases
in
this
high
impact
region
in
the
northern
hemisphere
mid
to
high
latitudes,
and
when
we
look
at
the
annual
mean
increase
due
again
solely
to
that
increase
in
biomass
burning
variability.
N
So
we
can
see
from
these
experiments
that
these
are
consideration
of
the
variability
in
these
aerosol
emissions
are
important
when
I'm
quantifying
the
aerosol
forcing.
However,
a
question
that
still
remains
is
why
exactly
do
we
see
these
changes?
Even
though
we
have
the
same
amount
of
emission
integrated
over
time?
Why
do
we
see
differences
with
the
radiator
forcing
so
to
tackle
this?
We
ran
another
set
of
experiments
where
we
idealized
this
biomass
burning
emissions,
variability
and
essentially
isolating
a
temporal
variability
component.
N
So
using
these
idealized
variability
scenarios,
we're
actually
able
to
reproduce
a
lot
of
the
same
quantitative
or
sorry,
qualitative
features
that
we
saw
in
the
fully
coupled
runs
and
within
the
cmip6
forced
emissions
variability.
So,
firstly,
we
see
that
cloud
thinning
effect
shown
here
in
reductions
in
the
cloud.
Droplet
concentration
concentrations,
as
well
as
a
reduction
in
clouds,
indicated
here
through
the
liquid
water
path,
and
then
these
act
in
conjunction
together
to
increase
the
cloud
radiative
effect.
N
N
We
can
then
zoom
in
on
these
high
impact
regions
and
look
at
how
they
evolve
over
time,
and
when
we
do
that,
we
can
see
that,
within
these
idealized,
very
ability
simulations
that
we
essentially
have
two
different
regimes,
one
that
is
associated
with
the
high
emission
years
and
another
that's
associated
with
the
low
emission
years.
N
What
clearly
pops
out
from
this
is
that
we
tend
to
have
a
bias
towards
these
low
emission
years,
where
the
effect
of
lower
emissions
really
has
a
higher
impact
to
the
time
integrated
effect
in
the
high
emission
years,
which
really
starts
to
indicate
to
us
possible
non-linearities
in
these
fields.
N
So
to
better
tackle
this,
what
we
did
was
re-ran
our
fixed
baseline
emissions
at
varying
emissions
levels.
This
allowed
us
to
see
the
smooth
response
of
these
cloud
fields
to
varying
aerosol
concentrations
in
the
atmosphere.
N
What
we
find
is
a
strong
non-linear
response
in
both
the
cloud
droplet
number
concentrations,
as
well
as
the
liquid
water
path,
and
this
is
in
agreement
with
a
a
previous
understanding
of
how
these
cloud
fields
react,
to
changing
aerosol
concentrations
and
most
per
famously
noted
by
karsala
at
all
in
2013..
N
So
what
these
cloud
these
non-linearities
in
this
cloud
fields
results
in
is
actually
a
non-linearity
in
the
total
forcing
with
varying
aerosol
concentrations.
N
We
can
then
look
at
those
cmip6
variability,
in
particular
the
individual
years
from
those
fixed
sst,
runs
and
see
that
we
see
that
that
non-linearity
pop
out
within
those
runs
as
well,
and
we
can
see
here
that
it's
those
high
emission
years
that
really
have
less
unit
impact,
both
in
the
cloud
fields
and
radiatively
than
the
low
emission
years.
N
Thank
you,
okay,
so
we'll
finish
off
with
a
few
takeaway
points,
so
first
consideration
of
this
variability
in
aerosol
emissions
can
have
a
large
impact
to
the
overall
aerosol,
forcing
increases
in
this
variability.
N
We
find
tends
to
reduce
the
total
negative
forcing
and
that
these
differences
in
the
forcing
are
due
to
non-linear
cloud
aerosol
responses.
N
So
thank
you
very
much
for
your
attention
and
I'm
happy
to
take
any
questions.
If
you
have
any
remaining
time.
A
It
was
really
interesting
and
just
to
make
sure
I
understood
your
conclusions,
your
your
aim
at
bronzer
look
very
similar
to
the
coupled
ones.
So
are
you
suggesting
that
you,
you
don't
need
any
kind
of
like
rectification,
effect
of
the
ocean
to
to
kind
of
produce
the
overall
temperature
trends?
It's
really
just
kind
of
the
direct
influence
of
the
forcing
from
the
atmosphere.
N
D
Hey
kyle,
I
guess
my
question
is
whether
you
found
any
particularly
sensitive
cloud
regimes
or
if
it's
really
just
whatever
kind
of
clouds
are
there
happen
to
be
sensitive
to
this
variability
in
inhibitions.
N
You
know
that's
an
excellent
question
and
something
that
we
haven't
dug
in
too
deeply,
yet
we
do
find
fairly
strong
sensitivity
in
more
pristine
atmospheres.
It
doesn't
really
answer
your
or
your
cloud
version
question,
but
we
do
see
a
higher
sensitivity
in
more
pristine
atmospheres,
as
you
know,
that
cloud
sensitivity
is
stronger
at
lower
concentrations,
but
that
that
is
an
excellent
question
and
then
something
that
we
should
look
into.
Thank
you.
O
Okay,
oh
so
good
afternoon.
My
name
is
it's
kinda
long
so,
but
you
can
call
me
dia
and
I'm
from
oh
I'll,
just
change
this
point,
I'm
from
I'm
a
phd
student
from
the
hong
kong
university
of
science
and
technology
and
I'm
currently
visiting
university
of
wisconsin
medicine
working
with
professor
elisabeth
maroon
and
I'm
here
to
share
my
thought.
My
research
on
the
changes
in
the
north
american
temperature
extremes
associated
with
the
chefs
in
the
northern
angular
mode
or
nan.
O
So
primary
influences
on
middle
relative
weather
change
are
well
studied
that
it
state
that
it's
influenced
from
the
changes
in
the
storm
track
and
jet
streams
and
if
we
trace
back
some
precursor
also
coming
from
the
global
climate
changes
and
internal
climate
modes,
which
northern
annual
mode
is
the
primary
mode
over
the
northern
hemisphere.
O
And
if
you
see
this
work
from
thomson,
evol
2001
you'll
clearly
see
that
how
nam
is
related
to
the
wind
colder
surface
temperature
over
the
north
and
american
continent
and
the
high
pressure
level
pressure
over
the
pacific
and
atlantic
white.
It's
lower
over
low
syllable
pressure
over
the
arctic
region.
O
So
from
this
we
came
up
with
our
research
question
on
how
much
is
the
northern
england
motor
nem
influences
the
winter
northern
american
extreme
temperature
in
a
warming
climate?
So
we
use
40,
ensemble
members
of
cs
csm1,
large
ensemble
to
analyze
the
mean
pattern,
considering
the
internal
climate
for
everybody
over
the
northern
hemisphere.
O
So
coming
to
the
first
result
of
this
work
here
we
want
to
see
how
csm
the
ensemble
mean
of
csm
one
large
ensemble,
showing
the
main
dipole
pattern
of
the
nam.
So
this
is
the
positive
nand
pattern
that
showing
a
high
syllable
pressure
anomaly
of
the
atlantic
and
pacific
while
it's
low
over
the
arctic
region,
and
this
is
calculated
from
the
monthly
data.
O
O
So
from
the
future
from
the
future
rcp
8.5
scenario,
it's
clear
that
there
is
a
change
in
pacific
and
atlantic
center,
but
in
the
pacific,
it's
stronger
in
the
future
and
there's
a
shift
also
while
it's
weakened
over
the
atlantic-
and
you
can
see
the
more
clear
here
in
this
panel-
see
that
the
shading
is
difference
between
future
and
historical.
So
there
is
a
polar
chef,
also
in
the
pacific
center
of
nam.
O
If
you
refer
to
the
contour
line
as
the
historical
pattern-
and
this
result
is
supported
by
15,
symmetry6
model
mean,
which
also
indicates
stronger
and
the
polar
shift
of
the
non-pacific
center.
O
So
we
also
try
to
look
at
the
changes
in
them
to
the
connection
pattern.
In
a
climb
in
a
warming
climate
using
a
daily
extreme
temperature,
so
from
the
minimum
temperature
in
the
gfm
winter
season,
so
to
see
non-positive
in
historical
main
characteristic
and
the
minimum
temperature.
O
O
It's
it
supports
the
stronger
westerly
and
from
figure
b
here,
the
non-negative
composite,
showing
that
in
historical
run,
there's
an
animal
school
that
we
all
focus
on
of,
like
nem
teleconnection
pattern,
how
it
changed
the
temperature
over
northern
american
continent,
but
there's
a
small
change
over
the
western
coast
western
coast,
so
to
see
how
a
better
represent
the
composite.
We
plot
another
plot
here
to
find
the
difference
between
minimum
temperature
and
then
positive
and
I'm
negative.
O
So
it's
well
represent
the
colder
than
positive
in
the
western
coast
and
colder
central
north
american
minimum
temperature
and
in
the
future,
the
special
pattern
of
this
minimum
temperature
in
different
name
phase.
Num
phase
is
similar
with
a
different
magnitude.
So
to
to
compare
more
clearly,
you
can
take
a
look
at
these
two
figures.
This
is
the
historical
and
future
of
the
composite
difference,
so
the
colder
extreme
temperature
in
this
western
coast
of
northern
american
continent
is
intensified
in
the
future.
While
this
colder
nem
negative
is
weakened
in
the
future.
O
So
you
can
see
here
that
there
are
changes.
This
is
the
difference
between
future
and
historical,
so
this
blue
color
here
means
there
are
changes
in
both
region,
while
it's
weakened
in
this
weekend
in
the
central
northern
american,
but
it's
intensified
here
in
the
western
coast-
and
this
is
also
similar
in
1576
model-
mean
analysis.
O
So
we
can
suggest
that
there
is
a
temperature
of
action.
That's
play
a
role
in
this
cold
air
western
coast
due
to
the
polish
chef
of
the
pacific
center
and
to
see
to
support
our
result.
We
plot
another
standard
deviation
of
minimum
temperature
from
daily
data,
ensemble
mean
of
csm
csm1,
large
ensemble.
So
in
the
historical
it's
clear
that
the
minimum
temperature
is
very
small
over
the
mid
latitude
in
the
northern
hemisphere,
but
it's
weakened
in
the
future.
O
It's
even
more
clear
here
in
the
panel
see
that
difference
between
this
two
period
is
showing
weakening
of
this
minimum
temperature
in
the
winter.
So
it's
possible
that
nam
is
possibly
contributed
to
such
weakening
and
respect
to
another
precursor
of
such
change.
In
the
nam
teleconnection
pattern,
we
take
a
look
at
the
storm
track,
changes
in
the
historical
and
the
future
run.
O
So
this
plot
is
an
eddie
kinetic
energy
composite
during
non-positive
and
I'm
negative.
So
the
warm
color
here
showing
that
eddie
kinetic
energy
is
stronger
in
the
positive
in
the
pacific
center.
So
it's
a
csm1
ensemble
member,
showing
this
two
characteristic
of
the
storm
in
the
northern
hemisphere,
but
in
the
future
from
this
panel
b.
So
this
is
high-pass
10-day,
high-pass
filter
and
it
is
intensified
in
the
future
and
if
we
find
a
difference,
it's
clear
that
there
is
a
polar
shift
of
this
pacific
storm
track.
O
Well,
it
is
match
the
location
of
the
change
of
the
polar
shape
of
the
name
showing
by
the
contour
line
in
this
figure.
So
to
summarize
my
talk,
the
40
ensemble
members
mean
of
csm.
One
large
ensemble
suggests
that,
in
terms
of
nand
change,
it
is
intensive,
it
will
be
intensified
and
there
will
be
a
polaroid
or
not.
It
will
north
eastward
shift
in
the
name
pacific
center
and
in
terms
of
nematode
connection
pattern.
O
There
will
be
a
lesser
stream
temperature
variability
in
the
future,
and
this
is
coincides
with
the
weakening
in
winter,
an
environmentalist
cold
over
the
central
north
america
during
non-negative
phase
and
another
one
here
related
to
non-declination
pattern
is
the
cold
advection
associated
with
the
polar
shift
of
the
nam
pacific
center,
its
result
in
a
colder
winter
over
the
west
coast
or
western
north
american
during
them
positive.
It's
also
consistent
with
the
change
in
storm
track
that
is
intensified
and
there
is
also
polar
shift
in
the
specific
strum
stone
track
doing
the
none
positive.
C
Hey
thanks
yeah.
I
have
a
quick
question.
You
showed
the
can
you
hear
me
yeah,
yeah,
okay,
great
yeah.
You
showed
that
the
namm
intensification
changed
some
of
the
centers
of
action.
Were
you
able
to
connect
that
to
any
changes
in
sea
ice
over
the
same
time
period.
O
O
C
C
C
O
C
Okay,
great
all
right,
so
our
next
speaker,
we
have
xiaowei
li
and
the
title
of
the
talk
is
impacts
of
arctic
sea
ice
loss
on
global
ocean
circulations
and
inner
basin
ocean
heat
exchanges.
D
H
H
H
H
In
particular,
to
replicate
the
cs
laws
in
the
prosecuted
case.
Here
we
reduce
the
orbital
cs.
We
have
performed
november
members
and
other
alumni
members
lost
for
200
years.
We
first
compared
the
simulated
arctic
sas
in
our
experiment.
With
the
observations,
the
top
three
panels
are
showing
the
observed
active
cs
concentration
in
19,
18
to
1989,
200,
cs
and
2015,
while
dependency
here
actually
is
showing
the
difference
between
the
two
periods
here
on
the
bottom.
Three
panels
are
showing
the
simulated
arctic
csgr
experiment.
H
The
bottom
part
of
the
green
line
here
is
showing
the
changes
in
rtx
cs
area,
so
we
can
bend
that
it
provides
a
force
response
within
the
first
10
years,
which
is
a
very
strong
retention
in
this
period.
However,
the
ocean
circulations,
for
example
the
atlantic
meridian
overturning
circulation,
the
ammo.
It
provides
a
slow
response
based
on
a
multi-killer
time
skill.
It
suggests
bracelet
changes
within
the
first
two
years.
H
However,
after
that
the
emo
suggests
a
very
strong
weakling
trend,
the
aim
of
quickly
lasts
for
about
100
years
and
by
the
end
of
this
experiment,
the
emo
has
been
weakened
by
about
six
super
jobs
and
she
represents
the
ocean
circulation
in
the
atlantic
basin.
We
find
that
the
indonesia
through
flow
the
atf
here
which
connects
in
the
ocean
and
the
pacific
ocean
at
lower
latitudes.
H
H
The
panel
and
the
penalty
here
is
showing
the
force
response
within
the
first
two
years,
so
the
final
issue
we
can
find
is
that
the
surface
heat
force
suggests
positive
downward
anomaly
in
the
north
atlantic,
so
which
means
the
head
is
entering
the
ocean
in
this
region,
and
the
panel
b
here
is
showing
the
ocean
vertical
integrated
ocean
heated
content
dependence.
So
we.
H
Positive
anomaly
here
so,
which
means
most
of
the
ticking
head,
is
locally
stored
in
this
region,
contributed
to
the
ocean
height
contact
increasing
in
the
force
response.
However,
for
the
slow
response
for
dependency
and
penalty
here
we
find
that
even
the
positive
anomaly
is
still
there
for
the
surface.
It
forces
here,
however,
for
the
penalty,
we
find
that
changes
in
the
ocean
containers.
H
H
H
Mechanisms
may
add
a
plea
for
different
for
the
force
response
and
the
snow
response,
and
here
we
first
enter
exchanges
in
the
first
10
years
to
understand
the
force
response
on
the
left
panel.
Here
is
the
other
changes
for
the
atmospheric
temperature,
so
we
find
that
the
active
status
loss
leads
to
strong.
L
M
H
H
The
ocean
head
storage
in
each
ocean
basin
equals
to
the
ocean
heat
uptake
in
this
region
minus
the
sum
of
the
outgoing
interfacing
ocean
head
exchange
at
each
ocean
based
on
boundaries
here.
So
this
is
the
description,
and
here
on
the
left
is
the
lrs
for
the
head
project
in
the
atlantic
basin.
There
are
four
layers
here:
showing
the
ocean
head
storage,
the
black
curve,
the
ocean
head
object
is
a
red
curve
and
the
oceanic
transport
under
the
ocean
boundary
is
here.
H
H
The
southern
ocean,
so
we
find
that
for
the
first
response
within
the
first
10
years,
the
atlantic
suggests
a
very
strong
positive
anomaly
for
the
ocean
heat
object
in
this
region.
It
was
period,
however,
in
for
the
fourth
response,
since
the
ocean
circulation
suggests
slight
changes
in
this
period,
so
we
find.
Q
H
Most
of
the
ticking
hits
is
locally
stored
in
the
atlantic
in
the
atlantic
basin,
contributing
to
the
increasing
chain
of
ocean
field
storage
corresponding
to
the
increasing
chain
for
the
ocean.
Heat
storage
on
the
red
panel
is
the
only
average
of
the
temperature
trend.
We
can
find
the
very
strong
woman
chain
between
40
degrees,
north
to
60
degree
north
from
the
surface
down
to
about
3000
meters
level.
H
R
H
Surface
temperature,
so
we
can
find
that
the
temperature
contrast
is
amplified
in
the
slow
response
it
can
contribute
to
the
downward
non-interpreted
heat
frost
in
the
through
response
on
the
bottom
panel.
Here
is
the
difference
between
the
first
response
and
the
through
response,
so
we
can
find
that
compare
with
the
first
response,
more
height
can
be
absorbed
by
the
ocean
in
the
snow
response.
H
H
The
gold
curve
here
is
due
to
the
velocity
of
knowledge,
so
we
can
find
that
it
dominates
the
alarming
ocean
transport
at
the
atlantic
and
the
motion
boundary
as
well
as
the
total
changes
there.
Well.
The
negative
number
here
is
showing
that
the
atlantic
is
exporting
heat
to
the
southern
ocean
so
which
means
further
slow
response.
Even
more
heads
can
be
absorbed
by
the
atlantic
ocean.
However,
most
of
the
taking
hit
is
transported
away
to
other
ocean
basins,
a
further
story
smalls
in
the
pacific.
We
found.
L
D
D
H
D
H
D
D
H
H
Here
we
find
that
the
thermal
ocean
is
important,
hit
from
other
ocean
basins
and
here
combine
the
ocean
transport
between
the
southern
ocean
and
intellectic
base
the
orange
curve
between
the
the
ocean
and
the
individual,
the
green
and
the
blue
curve.
Here
we
find
that
about
two
sources
of
the
head
redistributed
from
the
atlantic
basin
is
imported
by
the
southern
ocean.
Even
part
of
the
heat
is
further
further
reduced
back
to
the
atmosphere
in
the
form
of
negative
ocean
heat
object.
H
The
southern
ocean
is
still
suggesting
an
increasing
trend
of
ocean
high
storage,
and
here
are
the
summer
conclusions.
We
find
that
the
active
cs
laws
promote
ocean
hydrogen.
In
the
north
atlantic
or
wonderful
region,
however,
the
changes
for
the
ocean
health
storage
suggests
different
patterns
based
on
different
time
skills
for
false
response.
Within
the
first
10
years.
The
ocean
circulation
suggests
flat
changes.
Therefore,
most
of
the
ticking
hits
is
locally
stored
in
the
atlantic
basin.
However,
for
for
slow
response
based
on
multiplication,
temp
skill
will
strengthen
more
quickly.
H
C
All
right:
well,
if
you
come
up
with
a
question
as
we
move
into
the
next
presentation,
feel
free
to
send
it
in
the
chat
to
xiaowei,
if
you
can
address
it.
S
S
In
my
presentation,
the
baron
chunky
sea
is
called
pacific
sector.
I
want
to
start
with
the
brief
overview
of
the
sea
ice.
Previous
studies
have
suggested
a
possible
linkage
between
arctic
sea
ice
loss
in
autumn
and
subsequent
weakening
of
the
polar
vortex.
Some
studies
have
argued
that
cs
loss
has
a
strong
impact
for
mid-latitude
weather
and
climate.
S
S
The
finger
on
the
bottom
is
a
time
series
of
cs610.
The
observed
arctic
sea
ice
over
pacific
sector
in
september
shows
a
strong
decline
in
recent
decades.
However,
it
has
received
much
less
attention
compared
with
cs
laws
over
the
barren
krc.
The
the
stratospheric
response
to
cs
loss
is
sensitive
to
cs
laws
location.
S
Therefore,
the
impact
of
cs
laws
or
the
pacific
sector
requires
further
investigation.
Besides
the
cis
laws
location,
this
linkage
could
also
be
modulated
by
qbo
phase.
Therefore,
in
our
study
we
focus
on
easter
egg
qbo
and
try
to
access
the
impact
of
cs
laws
in
autumn
over
the
pacific
sector
on
the
stratosphere,
particularly
for
ssw
event,
the
new
the
numerical
experiments
were
performed
using
ces
and
welcome.
6.
welcome
6
has
improved
stratospheric
variability,
including
internally
generated
qbo
and
the
realistic
assessment
climatology
to
understand
the
impact
of
cs
loss.
S
We
conducted
a
pair
of
one-year
experiments
named
the
control
and
the
law.
This
experiment
starts
from
1858
june
1st
and
the
each
contains
75
ensemble
members.
The
finger
on
the
left
shows
ci's
concentration,
difference
between
control
and
the
law.
The
regional
cis
loss
is
greatest
in
autumn,
since
the
qbo
might
modulate
the
stratospheric
response
to
ci
slots.
During
our
one
year
simulation
we
have
to
ensure
the
qbo
does
not
switch
phase.
S
Here
we
show
the
time
series
of
zonal
means
on
a
wind
average
over
qbo
region.
It
starts
from
1858
june
1st
each
line
represents
one
ensemble.
Member
control
is
showing
grey
line.
Low
is
showing
right
align
overall,
the
easily
qbo
is
consistent
and
similar
between
control
and
the
load
during
the
winter.
S
S
S
The
x-axis
is
100
high,
pascal
vertical
wave
activities
average
in
five
days
before
the
onset,
the
y-axis
is
ssw
intensity,
which
is
defined
by
10
hepatica
window
reversal.
During
the
event,
the
bubble
size
indicates:
ssw
duration,
due
to
upward,
wave
propagation
perturbs
the
polar
vertex.
A
linear
relationship
exists
between
x
and
y
by
comparing
the
green
bubbles
and
the
right
bubbles.
S
We
notice
that
the
x
value
y
value
and
bubble
size
tend
to
be
larger
and
right,
meaning
that
the
cs
laws
over
pacific
sector
triggers
stronger
upward
wave
activities
and
leads
to
more
persistent
and
intense
ssw
events.
The
changes
of
these
characteristics
are
significant
to
understand
how
ci
slots
impact
the
upward
wave
activities.
S
We
focus
on
10
days
before
the
onset
we
start
from
the
temperature
field.
The
field
contours
are
the
temperature
difference
between
control
and
low
on
the
left.
It's
average
at
high
latitude
and
showing
longitude
versus
altitude
cross-section
on
the
right
is
average
over
pacific
sector
and
showing
latitude
versus
altitude
cross-section.
S
S
S
A
positive
marijuana
temperature
gradient
is
associated
with
zonal
wind,
decreasing
with
altitude
on
the
left,
six
synonymous
69th
degree
meridian
wind
is
shown
as
purple
line
contours
on
the
right.
The
zonal
wind
average
over
pacific
sector
is
showing
green
light
contours
associated
with
horizontal
temperature
gradient.
There
is
a
stronger
southward
motion
over
now
season
eurasia
and
a
stronger
non-sword
motion
over
the
time,
as
well
as
a
stronger
westward
motion
at
high
latitude.
S
Next,
we
show
the
850
hypasca
geopotential
height
before
the
onset
in
control
low
and
their
difference
in
control.
We
notice
that
the
anti-cyclonic
anomalies
are
over
europe
and
significant
cyclonic
anomalies
are
over
now
specific,
as
well
as
canada.
These
patterns
of
height
anomalies
are
precursors
of
assassination.
S
Once
the
ice
loss
occurred,
the
cyclonic
anomalies
are
not
specific,
are
significantly
amplified.
It
can
enhance
the
upward
planetary
wheel,
propagation,
the
cyclonic
circulation
along
the
cyclonic
circulation
response
over
now
specific
also
suggests
a
stronger
southward
motion
or
nonseizing
erasure
and
a
stronger
northward
motion
or
alaska,
which
is
consistent
with
the
marijuana
wind
pattern.
We
mentioned
in
the
previous
slide
to
further
understand
how
anomalous,
wind
and
the
temperature
enhance
the
upward
wave
activities.
S
We
decompose
the
power
added
heat
flux
into
its
linear
and
nonlinear
component.
These
fingers
are
shown
at
850
hypascape
on
the
left.
It
is
the
total
added
heat
flux.
Control
is
showing
line
contours
the
difference
between
low
end
control
are
showing
field
contours.
In
control
before
the
onset
upward
wheel,
activities
are
mainly
dominant
oversea
of
japan.
S
There
are
some
small
patches
over
now
specific,
as
well
as
not
america
once
their
slots
occurred.
The
upward
wave
activities
over
sea
of
japan
are
significantly
enhanced
by
decomposition.
We
notice
that
the
enhancement
of
upward
wave
activities
over
sea
of
japan
are
mainly
contributed
by
linear
term
whereatc.
S
S
All
the
variables
are
the
difference
between
control
and
low
average
in
10
days
before
the
onset.
The
line
contours
is
the
wind
field.
Field
contours
is
the
divergence.
The
vectors
are
the
ep
flux.
The
all
wave
numbers
are
shown
on
the
left
wing
number
one
is
in
the
middle.
With
number
two
is
on
the
right
from
the
vectors
cs:
reduction
enhances
upward
planetary,
propagation,
the
stronger
will
activities,
convergence
in
the
stratosphere.
S
S
S
The
zonal
temperature
gradient
enhance
meridian
wind
throughout
the
troposphere,
the
coupling
of
wind
anomalies
and
the
temperature
increase.
The
upward
wave
activities
the
wheel
drag
induced
could
increase
asset
up
ssublu
duration
and
intensity
with
impact
on
cold
air
cold
air
outbreaks.
At
the
surface.
S
A
Yeah,
I
was
just
wondering
kind
of
based
on
your
understanding
of
how
this
affects
the
waves.
If
you
have
the
pacific
loss
accompanied
by
the
loss
in
all
the
other
places
as
like
in
the
real
world,
do
you
think
you
would
see
the
same
thing
like?
Would
you
expect
to
see
this
in
coupled
simulations
forced
by
greenhouse
gases,
for
example,.
S
Yeah,
I
think
yes,
so
I
think
in
our
simulation
we
only
focus
on
the
ci
slots
and
we
try
to
exclude
other
factors
but
yeah.
I
think
other
factor
will
have
an
impact
on
the
result
may
be
required
for
the
investigation.
D
J
Great
well,
thank
you,
everyone
for
coming
out
today.
My
name
is
matt
jenkins
and
I'm
a
third
year
phd
student
at
the
university
at
albany
in
new
york
and
today
I'm
going
to
discuss
the
impact
of
sea
ice
loss
on
arctic
climate
feedbacks
using
model
simulations
and
the
rfi
re-analysis
data
set.
So
just
a
brief
overview
for.
J
Realized
in
very
early
modeling
studies
with
increased
co2,
so
we
can
see
in
this
figure
from
hansen
that
all
1984
that
in
that,
in
the
cold
season,
the
largest
arctic
warming
occurs
and
there's
about
eight
to
twelve
to
eight
to
five
degrees
celsius
of
warming
in
the
cold
season,
but
but
in
the
summertime
only
about
one
to
three
degrees
celsius
of
warming.
J
So,
since
this
is
a
talk
about
climate
feedbacks,
I
just
want
to
give
a
brief
overview
of
how
our
how
client
feedbacks
impact
the
arctic
region,
so
in
this
study
from
pit
hand
and
morrison,
they
used
the
cm5
on
quadruple
co2
runs
and
they
computed
the
warming
contributions
of
different
feedbacks
in
the
arctic.
So
if
a
process
induces
greater
warming
in
the
arctic
region
than
in
the
tropics,
then
that
process
is
said
to
contribute
to
arctic
amplification.
So
that's
the
region
right
above
this
dash
gray
line.
J
So,
for
example,
the
surface
albedo
feedback
is
contributes
to
greater
warming
in
the
arctic
in
the
tropics.
So
that's
a
process
that
contributes
to
the
amplification,
however,
that
process
that
induces
greater
warming
in
the
lower
latitudes
is
said
to
oppose
arctic
amplification.
So
another
example
of
this
is
the
water
vapor
feedback,
since
its
effect
is
stronger
in
the
tropical
regions
in
the
arctic
and
even
though
it
does
cause
a
little
bit
of
warming
in
the
arctic,
since
its
effect
is
stronger
in
the
lower
latitudes.
As
I
said
to
to
oppose
the
arctic
amplification.
J
However,
many
studies
have
looked
at
have
cited
the
lab
straight
feedback
as
a
key
cause
of
the
arctic
amplification,
since
it
induces
warming
in
the
arctic
region
and
cooling
in
the
tropics,
so
the
lantern
feedback
was
attributed
to
the
high
stability
of
the
arctic
atmosphere,
meaning
that,
on
this
adult
I
mean
that
an
atmosphere
with
high
stability
can
find
any
warming
signals
near
the
surface
and
and
that
that
prevents
any
vertical
mixing
within
the
upper
troposphere.
So
any
warming
is
effectively
confined
to
the
surface
and
under
a
bottom
heavy
warring
profile.
J
So
to
answer
this
question,
we
used
two
ces
and
one
model
experiments
which
were
used
previously
by
the
study,
diadel
2019
and
our
two
roms.
The
first
experiment
is
a
standard.
One
percent
per
year
increase
in
co2
run
with
fully
coupled
dynamic
sea
ice
and
there's
also,
we
also
run
a
fixed
sea
ice
run,
which
is
the
same,
which
is
the
same
as
the
1
co2
run,
so
the
co2
still
increases
by
1
each
year,
except
the
effects
of
the
arctic
sea
ice
loss
are
excluded
and
and
in
the
simulation.
J
J
With
the
oceanic
heat
uptake,
which
is
to
the
very
far
right
of
the
graph,
we
can
see
that
there's
large
oceanic
heat
release
of
energy
to
the
atmosphere
in
the
winter
which
which
causes
much
of
the
bottom
heavy
warming,
and
we
don't
see
this
in
fixed
ice
case
so
now,
moving
on
to
a
couple
other
results.
Looking
at
the
plank
and
lateral
feedbacks
in
the
winter
time,
we
can
see
in
the
gray
bar
in
the
standard
run
with
the
dynamic
sea
ice
that
these
the
plank
feedback
is
largely
negative
and
latitude
back.
J
And
lastly,
we
look
at
the
atmospheric
heat
transport,
as
this
has
also
been
cited
as
a
cause
of
arctic
amplification.
We
can
see
that
with
fully
coupled
dynamic
sea
ice
and
the
large
bottom
heavy
warming
that
the
atmospheric
heat
transport
is
actually
equator
work.
So
this
on
opposes
the
arctic
amplification,
but
in
the
fixed
ice
case,
so
the
heat
transport
is
anomalously
on
forward.
J
So
now,
looking
at
some
spatial
patterns
for
some
of
the
feedbacks
here
are
the
spatial
patterns
of
the
latching
feedback,
which
is
in
the
shading
in
the
contours
of
the
sea
ice
loss,
and
in
this
panel
we
show
the
results
from
the
one
percent
co2
run
during
the
winter
time.
J
So
now
we
wanted
to
compare
our
model
results
from
the
aero
5
to
the
to
the
arrow
five
data
set,
and
to
do
this,
we
look
at
the
years
1990
to
2019
relative
to
years,
1950,
1979
and
once
again
using
the
radiative
kernel
method,
and
here
we
want
to
emphasize
what
the
spatial
patterns
of
climate
feedbacks
are
in
the
arctic
region
and
just
for
the
sake
of
time,
I
just
focus
on
the
electric
feedback,
but
our
studies
do
look
at
the
other.
J
So
we
estimate
the
temperature
inversion
as
the
temperature
difference
between
at
the
temperature
850
had
to
have
scales
and
a
thousand
hectopascals,
and
that
that's
represented
by
the
shading,
where
the
red
is
more
stable.
Conditions
in
the
blue
are
more
unstable
conditions
and
the
contours
or
the
strength
of
the
light
are
the
latter
feedback.
J
The
spatial
patterns
of
both
of
both
on
processes
do
not
line
up
very
well,
but
looking
at
a
map
of
electric
feedback
and
the
sea
ice
sloths,
where
the
shading
in
this
figure
is
the
sea
ice
loss,
the
green
contours
or
the
surface
warming,
the
red
contours,
electric
feedback.
We
can
see
that,
where
there's
large
sea
ice
loss
along
the
atlantic
sector
in
the
arctic,
that's
where
the
larger
surface,
warming
and
latitude
feedback
response
occurs,
and
we
also
get
a
similar
relationship
in
autumn.
J
J
So
this
suggests
that
large
a
causes,
an
equator
word
on
atmospheric
energy
transport.
So
here
are
the
studies
that
we
have.
What
one
of
them
is
already
published,
the
ces
and
one
simulations
in
grl,
the
other
one.
We
just
got
back
for
our
revisions
and
we're
currently
revising
that
paper
and
in
both
studies
we
discuss
each
feedback.
Besides
just
electric
feedback
in
greater
detail,
so
I'll
stop
right
there
and
I'll
be
willing
to
take
any
questions
that
you
may
have.
C
J
C
P
C
Yes,
your
your
sound
is
a
little
echoey
right.
Oh.
C
Yeah,
if
you're
also
on
the
youtube
channel
exit
out
of
that
that
sometimes
helps.
P
Okay,
okay,
sure,
okay,
so,
okay,
so
hi.
Everyone
today
we're
talking
about
the
seasonal
illusion
of
artification,
and
we
also
analyze
is
decomposed
uncertainty
in
projection.
I'm
jotin
wu
from
national
taiwan
university
and
my
advanced
advisor
is
also
things
for
all
these
people
to
improve
these
studies.
Okay,
so
let's
get
started
so
artisan
factor
which
is
which
means
on
the
arctic
once
more.
P
P
Hello
find
out
is
this
known
as
metric
of
arty
wormy,
which
arctic
burns
more
in
winter
than
in
summer,
and
also
indicate
that
this
is
driven
by
the
ci's,
effective
heat
capacity,
and
the
right
figure
shows
that
leonardo
use
larger
ensemble
and
find
out
a
seasonal
shifting
of
maximal
arctic
warming
from
the
autumn
to
the
winter,
and
so
as
they
are
dedication
factor.
However,
this
results
also
shows
a
large
modern
model
difference.
P
Of
artification
is
in
seasonality
is
still
unclear.
Okay,
so,
following
these
questions,
here
is
my
scientific
questions.
First,
how
do
authentication
in
seasonality
and
what
mechanism
drives
the
seasonal
evolution
of
authentication
and
the
third
one
is.
How
do
uncertainty
looks
like
in
these
signals.
P
So
now
from
the
question
one
and
two,
I
use
the
sixth
single
modulation
foundation,
symbol,
which
is
under
rcp
8.5
scenarios,
we
analyze
the
arctic's
surface,
air
to
measure,
precipitation
sea
ice
area,
temperature
flux
and
also
the
arctic
application
factor.
P
Also,
we
use
these
simplified
simulations
data
with
different
rcp
scenarios.
Okay,
and
about
the
questions
on
the
third
question.
We
use
the
uncertainty.
Conversation
analysis
following
learning
out
hello
and
bonanno,
which
we
decompose.
The
total
uncertainty
into
three
main
sources,
which
is
muscle
structurality,
mod
scenario,
uncertainty
and
internal
variability
uncertainty
and
muscle
structure
means
mainly
varies
across
the
ensemble
means
of
six
months
and
the
scenario.
Uncertainty
means
the
variance
in
the
multiple
mean
under
three
different
rcp
scenarios
and
the
general
variability
means
the
mean
of
the
variance
across
the
ensemble
means
english
smiles.
P
Okay.
So
in
this
analysis,
we're
focusing
on
the
articulation
factor
on
architecture,
precipitation
and
also
cs
area,
okay,
so
so
first
we
will
look
at
the
seasonal
evolution
in
arctic
climate
in
large
ensemble
and
find
out
its
possible
mechanism.
P
P
Here
shows
the
four
variables
in
seasonality
and
in
each
column
shows
different
models
and
in
each
panel
the
x-axis
stands
for
a
year
while
the
y-axis
stands
for
months
and
the
thought
in
the
panel
stands
for
the
maximum
within
12
years
in
each
year.
So
first
we
compare
presentation
and
temperature.
P
And
next
we
move
on
to
the
cs
area,
because
we
know
that
in
under
global
warming,
the
more
sea
ice
loss
and
the
more
of
the
water
heat
release
and
the
more
heat
from
the
ocean
to
atmosphere
heat
release.
So
as
it
already
flux.
So,
as
you
can
see,
the
cs
area
also
shows
the
seasonal
shifting,
which
is
from
about
september,
to
october,
to
november
or
december,
but
compared
with
terminology
locks.
P
The
shifting
is
about
one
month
earlier
than
terminating
flux,
okay,
so
the
permian
mechanism
is
that
the
seasonal
evolution
of
ci
loss,
what
drives
the
seasonal
shifting
of
turbulent
influx,
which
is
in
advance,
drives
the
arctic
temperature
increase
and
precipitation
increase,
and
another
interesting
thing
is
that
compare
models
model
that
also
shows
the
large
amount
of
spread
in
of
these
signals.
Okay
and
next
we
move
on
to
the
artification
factor,
which
is
defined
as
the
arctic
warming
divided
by
the
global
warming,
and
in
this
panel
it
also
shows.
P
While
you
can
see
that
the
arctic
education
factor
will
reach
its
maximum
about
about
2050
and
then
decrease
okay,
so
here's
the
difference
and,
however,
we
compare
on
each
panel,
which
also
shows
the
large
model
spread
in
the
af
and
in
addition,
we
also
test
different
scenarios,
and
so
in
this
figure
shows
the
rtmp
arctic
education
factor
at
the
end
of
the
21st
century
and
in
each
panel
the
y-axis
shows
the
18
models
in
semi-5,
and
the
number
in
brackets
shows
the
maximum
the
shifting
months
of
the
maximum
af
from
rcb
2.6
to
8.5.
P
So,
as
you
can
see,
the
maxima,
af
and
rcp
2.6
are
located
in
november,
while
the
when
we're
looking
at
the
rcp
0.5,
the
maximum
aff
will
locate
in
about
november
to
december,
which
means
that
they
also
exist.
The
scenario
differences
in
the
af
and
so
following
this
question.
Well,
next,
move
on
to
do
the
next
analysis,
which
we
do.
P
The
answering
decomposition
analysis
in
larger
example:
okay,
so
this
figure
shows
the
four
variables
we're
focusing
on
valve
and
we
do
the
uncertainty,
composition,
so
the
first
column
shows
the
total
variance,
which
is
the
total
answering
values
in
seasonality,
and
the
other
three
columns
is
the
corresponding
uncertainty
values.
Okay,
so
first
we're
looking
at
the
first
column,
it
shows
that
these
four
variables
shows
the
clear
seasonal
evolution,
which
is
also
shifts
from
the
from
the
ocean
to
the
winter
and
compared
with
the
model
variance.
P
It
also
shows
the
clear
system
evolution
and
this
signal.
This
signal
is
quite
similar
and
we
next
compare
with
scenario
and
internal
variability
uncertainty.
This
scenario
and
the
internal
variability
do
not
show
clearance
is
an
illusion
and
its
value
is
is
quite
smaller
compared
with
model
variance
okay.
So
we
can
know
that
the
model
uncertainty
will
dominate
the
total
uncertainty,
and
this
is
an
evolution.
P
P
In
the
af,
which
also
means
that
we
have
to
be
more
cautious
about
the
model
spread
in
af
when
calculating
in
seasonality.
Okay,
so
to
conclude
our
study
first,
we
analyzed
the
seasonality
of
artification
and
presentation
and
we
found
out
that
the
annual
maxima
of
aaa
and
its
presentation
will
shift
from
autumn
to
winter
in
the
21st
century,
and
this
is
due
to
the
sea
ice
loss
and
its
corresponding
turbulent
reflux.
And
next
we
also
do
finger
analysis
which
is
not
shown
in
these
slides.
P
Okay,
and
the
second
thing
is
that-
and
it's
certainly
is
that
we
do
uncertainty,
decomposition
analysis
and
we
found
that
the
model
uncertainty
will
dominate
the
total
uncertainty
with
respects
to
how
months
is
shift
and
in
what
here
it
shifts,
and
the
next
things
is
that
we
found
out
the
smaller
scenario,
uncertainty
of
af
in
seasonal
illusion
and
which
is
not
not
not
like
the
arctic
warming
and
presentation.
So
they'll
have
to
be
more
taken
about
calculating
af
in
seasonality.
Okay,
so
that's
it
and
thanks
for
listening.
C
Hey
thanks,
you
ting
looks
like
we
have
a
question
from
laura
landrum.
R
I
I
just
want
to
make
a
comment
with
this
talk
in
the
previous
talk
that
you
once
again,
you're
looking
at
sea
ice
concentration
losses,
and
that
I
think
I
think
we
should
be
careful
when
we're
referring,
particularly
to
heat
fluxes
in
the
winter,
whether
the
sea
ice
loss,
sea
ice
loss
implies
sea
ice
loss,
which
can
also
be
a
volumetric
change
and
a
thickness
change
versus
a
concentration
thing
change
and
in
the
models
that
there
will
be
a
period
of
time
in
each
of
these
models,
when
the
concentration
losses
are
not
great,
but
the
thickness
losses
might
be
significant
and
they
can,
they
can
add
a
significant
source
of
heat
in
the
winter.
R
In
the
winter,
so,
for
example,
the
ces
in
one
large
ensemble
you'll
see
about
a
10
watt
average
or
so
a
10
watt
per
meter
10
to
12
watt
per
meter,
squared
increase
in
surface
in
terp
in
the
conductive
heat
flux
over
sea
ice
covered
areas
just
from
the
thinning
of
the
sea
ice
from
the
beginning
of
the
20th
century,
to
about
2070
and
and
so
maybe
something
either
to
consider,
or
at
least
when
we
communicate
make
sure
that
it's
sea
ice
concentration
versus
sea
ice
thickness.
R
P
Okay,
so
I
see
so.
D
R
P
P
R
P
A
D
C
Q
I
forgot
okay,
great
yeah,
so
thank
you
yeah.
My
name
is
haruki
and
today
I'll
be
discussing
some
work
that
I've
been
doing
with
my
supervisor
at
the
university
of
toronto,
paul
kushner
and
what
we've
been
doing
is
using
csm1
large
on
the
large
ensemble,
as
well
as
slab
ocean
model
simulations
in
order
to
try
to
estimate
the
effect
of
ocean
circulation
changes
in
the
late
20th
century
and
so
yeah.
Q
It
turned
out
to
be
more
challenging
than
at
least
I
expected
so
hopefully
you'll
see,
okay,
okay,
so
the
work
I'm
doing
here
is
building
off
of
the
csm1
large
ensemble
single,
forcing
simulations.
Q
So
here
I'm
showing
the
land,
precipitation
and
sea
surface
temperature
anomalies
from
the
the
large
ensemble
for
the
off,
forcing
on
the
left,
the
alba
aerosol
in
the
middle
and
then
the
difference,
which
is
the
aerosol
effect
on
the
right
for
two
periods
in
the
late
20th
century.
So
on
the
top
row
I
have
the
19
1970s
minus
1950s
and
then
and
then
on
the
bottom.
I
have
the
2
000
minus
1970s.
Q
So
these
are
the
two
periods
I'll
be
using
for
the
rest
of
the
talk
and
they're
sort
of
chosen,
so
that
one
captures
a
period
where
aerosols
emissions
are
generally
increasing,
particularly
over
europe
and
north
america,
and
then
the
second
period
is
sort
of
capturing
the
period
where
we're
getting
an
west
to
east
shift
of
emissions.
Q
And
so,
if
you
look
at
the
precipitation
patterns
here,
you
see
that
you
know
not,
surprisingly,
greenhouse
gas
forcing
is
increasing
precipitation
in
the
asian
and
african
monsoon
regions,
but
actually,
in
the
historical
period
at
least
most
of
the
historical
forced
rainfall
variability
is
being
caused
by
aerosol,
forcing
where
you
know
it's
drying
in
much
of
asia
and
africa
during
the
aerosol
increase
period.
And
then
during
the
aerosol
shift
period
you
get
continued
drying
in
asia
and
increases
in
precipitation
in
africa
and
middle
east.
Q
So
what
we
want
to
do
is
try
to
understand
the
mechanisms
that
drive
that
cause
these
precipitation
changes
and
so
well.
Our
approach
is,
to
you
know,
take
try
to
decompose
the
total
coupled
response
into
different
time
scales,
and
so
this
is
has
been
done.
Q
You
know
by
many
people
with
using
the
ocean
model
hierarchy
in
order
to
separate
out
these
different
time
scales,
and
so
in
particular,
you
know
people
have
used
fixed
sst
simulations
wherein
yeah,
if
you
just
apply
a
forcing
with
fixed
ssds
or
without
changing
ssds.
You
get
this.
Q
What
I,
what
I'm
calling
direct
atmospheric,
but
what's
often
called
the
fast
effect
where
it's
just
atmospheric
radiation
and
circulation,
adjustments
to
the
forcing
plus
the
effect
of
any
land,
surface
temperature
changes,
and
so
these
typically
occur
on
the
day
to
month,
time
scales
and
then
those
are
sort
of
separated
from
the
slower
ocean
mediated
processes
that
are
the
result
of
ocean
surface
temperature
anomalies
that
then
feed
back
onto
the
atmosphere.
Q
So
this
is
often
called
the
slow
effect,
and
so
I
I
think
fast
and
slow
were
used
in
a
previous
talk
to
refer
to
very
different
time
scales,
so
here
so,
which
is
why
I'm
trying
to
use
these
more
precise
terminology,
and
so
in
this
paper.
I'm
sorry
in
this
talk
I'm
instead
using
slab
ocean
model
simulations,
and
so
the
idea
is
that,
instead
of
separating
out
the
direct
atmospheric
effect
with
these
slab
ocean
models,
experiments,
we
can
instead
separate
out
the
ocean
circulation
effect.
Q
Okay,
so
yeah
slab
ocean
models.
So
this
is
a
type
of
model
where
the
ocean
is
represented
by
a
fixed
single
mixed
layer
which
has
diagnostic
temperatures,
but
all
other
ocean
properties
are
fixed
and
so
processes
like
ocean
circulation
and
heat
exchanges,
both
laterally
and
vertically,
are
represented
by,
what's
called
a
q
flux,
which
we
can
calculate
using
pop
two
data
for
the
csms
sum,
and
so
the
experiments
we've
conducted
here
are
essent.
Q
What
we
do
is
we're
trying
to
target
the
50s,
70s
and
2000s
for
the
all
and
all
the
aerosol,
forcing
experiments,
and
so
we
use
the
large
ensemble.
Q
You
know,
coupled
pop
two
outputs
to
calculate
the
q
flux
fields,
which
we
apply
as
boundary
conditions
to
our
simulations,
and
so
here
I'm
showing
sort
of
a
schematic
of
the
different
experiments
we've
conducted
where,
on
the
x-axis
we
have
the
the
q
flux
and
on
the
y
axis,
we
have
this
global
mean
surface
temperature
and
so
just
to
take
a
look
at
the
blue
here,
which
is
the
alba
aerosol
forcing
which
is
a
bit
simpler.
Q
You
know
we,
we
can
first
change
the
q
fluxes
that
we
diagnosed
from
the
large
ensemble
and
so
a
positive
q
flux.
These
are
positive
down,
so
a
positive
q
flux
is
causing
a
cooling
corresponding
to
ocean
heat
uptake,
and
then
we
we
can
then,
while
keeping
key
fluxes
fixed,
impose.
Q
You
know
a
forcing
change,
increasing
co2
and
this
then
warms
up,
and
so
the
idea
is
that
in
principle,
if
we
set
everything
up
correctly,
if
you
add
up
the
effect
of
the
q
flex
and
the
effect
of
the
forcing
this-
should
approximately
equal,
the
full
ocean
model
response,
and
so
with
these
simulations,
we
essentially
get
two
methods
for
estimating
the
ocean
circulation
effect.
Q
The
first
is
to
take
the
difference
between
the
full
ocean
model,
the
large
ensemble
and
the
slab
ocean
model,
where
we
change
forcings
and
the
other
is
to
you
know,
take
a
look
at
the
effect
of
changing
the
q,
fluxes
and
and
we're
gonna
focus
here
on
the
residual
method.
For
reasons
I
will
show
you
right
now,
which
is
that,
unfortunately,
we
couldn't
get
the
q
flux
perturbations
to
give
us
reasonable
results.
Q
So
if
we
compare
the
full
ocean
model,
large
ensemble,
surface
temperature
anomalies
to
the
slab
ocean
model,
surface
temperature
anomalies,
we
see
that
you
know
this
reasonable
agreement
in
some
cases,
but
in
particular
this
alba
aerosol
2000's,
minus
70's
effect,
for
whatever
reason
the
q
flux,
cooling
effect
is
much
too
weak.
So
we
get
way
more
warming
in
the
slop
ocean
model
than
in
the
full
ocean
model.
Q
So
for
the
rest
of
the
talk,
I'll
be
focusing
on
the
residual
method
and
if
anyone
has
any
questions
about
or
sorry
any
suggestions
on
why
this
might
be
occurring.
I
definitely
appreciate
that
right,
so
here
I'm
showing
the
results
from
the
slop
ocean
model
simulations.
So
for
these
figures,
I'm
showing
the
july
august
september,
surface
temperature
anomalies
in
the
large
ensemble
on
top
the
slab
ocean
model
where
we
keep
q
fluxes
fixed
and
then
increase
the
forcings
in
the
middle
and
then
on
the
bottom.
Q
The
difference
between
the
two
which
were
inferring
to
be
the
ocean
circulation
effect,
and
so
you
know
we
see
that
in
general,
the
ocean
circulation
is
damping.
Q
The
transient
warming
signal
corresponding
to
ocean
heat
uptake
and
it's
also
redistributing
heat
away
from
the
arctic
and
north
atlantic
sort
of
you
know
also
damping
a
lot
of
this
north-south
inter-hemispheric
gradient
in
the
ssds,
and
so
this
obviously
has
a
impact
on
the
precipitation
in
the
monsoon
regions
that
we've
that
I
sort
of
pointed
out
before
and
so
again,
it's
the
full
ocean
model
on
the
left,
the
slab
ocean
model
in
the
middle
and
then
the
infrared
ocean
circulation
effect
on
the
right,
and
so
we
can
see
that
the
ocean
circulation
effect
is
sort
of
counteracting.
Q
You
know,
gradients
in
the
atlantic
sst,
and
so
this
is,
you
know,
it's
damping,
this
first
southward,
then
northward
shift
of
the
atlantic.
It
said
in
the
two
periods.
Q
We
also
see
that
it's
counteracting
a
lot
of
the
drying
we
see
in
africa
and
east
asia,
and
it's
also
reducing
precipitation
in
south
asia
and
the
maritime
continent,
and
so
if
we
then
remove
the
alba
aerosol
effect,
we
can
take
a
look
at
the
aerosol
forcing
effect.
Q
And
so
here
the
effect
of
ocean
circulation,
of
reducing
precipitation
in
africa
and
asia
is
much
clearer
and,
interestingly,
actually,
even
though
we
get
a
reduction
in
aerosol
forcing
over
africa
and
the
north
atlantic
in
the
later
2000s
minus
70s
period,
we're
still
getting
a
positive
rainfall
influence
of
ocean
circulation.
Q
So
it
seems
like
perhaps
this
you
know,
maybe
yeah.
So
I
guess
the
ocean
circulation
effect
might
be
reacting
or
like
responding
to
a
you
know:
decades-old
aerosol
cooling,
forcing
effect
yeah,
okay.
So
just
to
conclude
so
in
in
we,
we've
tried
using
these
slop
ocean
models,
and
you
know
diagnosing
the
key
fluxes
from
the
slab
from
the
large
ensemble
to
you
know,
set
up
these
simulations
such
that
we
can
reproduce
the
the
large
ensemble
anomalies
with
the
slab
oceans.
Q
But
unfortunately
this
didn't
work
for
us
so
we're
instead
relying
on
the
more
simple
residual
method.
But
you
know
we
still
see
some
interesting
things
where
the
ocean
circulation
sort
of
reduces
hemispheric
asymmetries
and
you
know
reduces
transient
sst
signals
and
and
in
in
regions
like,
for
example,
does
the
hell?
The
ocean
circulation
is
as
damping
the
effect
of
aerosol
force
drying,
and
you
know
reducing
that
drought
signal
yeah.
So
thank
you.
C
So
do
you
want
to
kick
off
the
the
discussion.
A
A
I
guess
we
can
just
kind
of
go
over
what
was
suggested
in
our
breakout
rooms,
so
I
can
start
with
our
one,
and
so
I
think
there
were
three
main
suggestions
or
or
topics
so
with
this
allocation,
brian
madiros
has
been
working
on
this
regionally
refined
tropics,
so
there
might
be
more
he's
just
been
kind
of
developing
the
configuration
up
to
now,
so
there
might
be
more
things
to
do
with
that.
A
It
might
be
interesting
to
look
at
that
in
the
context
of
the
qbo
mgo
connection.
Now
that
we
will
have
a
qbo
in
the
model.
The
other
thing
was
building
off
of
the
regionally
refined
north
atlantic
simulations.
A
So
our
simulation
that
we're
about
to
begin
kind
of,
depending
on
on
how
that
goes,
we
might
want
to
have
some
control
experiments
and
also
some
kind
of
perturbation
experiments
to
explore
any
influences.
We
find
like
maybe
smoothing
the
sst
anomalies
or
sst
itself,
to
kind
of
see
what
is
the
role
of
having
that
high
resolution
sst
versus
having
a
high
resolution
atmosphere.
So
there's
probably
things
to
think
about
there
with
building
off
of
those
experiments.
Unfortunately,
we
we
don't
have
any
results
yet
from
that
it
hasn't
started.
A
So
we
don't
know
like
how
fruitful
it's
gonna
be,
but
there
could
be
things
to
do
there
and
then
another
suggestion
from
hansee
was.
Maybe
we
keep
having
this
issue
with
new
model
configurations
with
the
labrador
c
freezing
and
maybe
some
perturbed
parameter
experiments
to
explore
kind
of
what
what
produce
it?
What
is
what
is
that
sensitive
to
I'm
not-
and
I
don't
have
enough
knowledge
myself
to
know
exactly
how
this
would
go
and
what
we
would
perturb,
but
something
to
think
about.
A
So
I
think
that's
pretty
much
what
we
had
in
r1.
If
I've
forgotten
anything
other
people
can
jump
in
so
yeah.
Sarah,
I
don't
know
if
you
want
to
go
through
what
just
said
in
your
group.
C
Sure
yeah
we
had,
I
guess
so.
The
categories
that
kind
of
is
two
different
main
topics.
One
was
about
data
availability
being
able
to
access
existing
or
soon
to
be
publicly
available.
Data
sets
and
the
other
one
was
potential
experiments.
So
I'll
start
with
the
the
former.
So
there
was
a
little
bit
of
concern
about
access
to
the
cesm2.
A
Yeah
it
looks
like
he's
not
but
okay,
we
can
look
into
that.
I
was
not
aware
that
there
was
an
issue
there.
Yeah.
C
And
perhaps
there
are
multiple
locations
to
find
these
simulations,
and
maybe
we
could
clear
that
up
and
send
out
a
clarification.
M
So
I
basically
looked
at
the.
I
was
looking
at
the
ssp5
data
set
from
csm2
from
our
system
grid
foundation
where
we
get
the
symmetric
data
set.
So
I
realized
that
there
are
only
three
ensemble
members
in
that
scenario,
so
I
was
hoping
that
there
are
more.
If
there
are,
how
can
I
get
that.
A
I
I
believe
there
might
only
be
three,
I
don't
know
man
do.
You
know,
I'm
not.
L
I'm
sorry,
I
wasn't
following
the
conversation
up
this
one.
Can
you
repeat
the
question.
A
Yeah,
so
the
question
is:
are
there
more
than
three
cement
six
rc
ssp
585
members.
L
There's
only
three:
let
me
double
check,
I'm
right
there.
Okay.
D
A
There's
16
ssp
two
four
fives,
but
I
think
we
haven't
run
any
extra
ssp
585
beyond
what
was
done
for
cm6
and
I
believe
there
was
only
three
but
yeah.
Maybe.
L
Yeah
there
were
only
three
of
the
five
eight
five
and
those
were
re-run
to
fix
an
issue,
but
there
still
remained
only
three
corrected
simulations
for
585.
A
C
Great
I'll
add
that
to
our
groups,
suggestions
and
great
we've
solved
some
problems
too,
the
other.
Oh,
the
other
thing
about
data
availability
were
for
some
simulations
that
aren't
available
yet,
but
will
be
the
the
mechanically
decoupled
and
the
refined
grid
north
atlantic
simulations
and
just
a
general
thought
that
it'd
be
great
if
those
were
made
publicly
available-
and
I
believe
both
are
expected
to
become
so.
A
Yes,
definitely:
the
regionally
refined
north
atlantic
we've
we've
like
budgeted
the
allocation
for
that
in
the
last
proposal,
and
that
will
be
made
available.
I
guess
the
mechanically
decoupled
is
a
little
more
tricky
for
us
just
because
it
was
not
working
group
resources.
But
if
you
have
resources
on
your
side
to
kind
of
make
that
available,
then
at
least
there's
a
portion
of
it
could
be
made
available
as
well.
C
Yeah
yeah
we've
been
working
with
gary
to
reduce
as
much
as
possible,
even
just
the
main.
You
know
the
big
variables
and
and
ignore
a
lot
of
the
others
and
shrink
it
down
to
maybe
just
a
couple.
Terabytes
is
very,
very
much
within
reasonable
expectations
so
and
so
the
other
suggestions
for
potential
future
simulations
one
was,
I
think,
sort
of
motivated
by
one
of
the
talks
this
morning,
which
was
the
ozone
single,
forcing
experiments
with
just
stratospheric
ozone,
forcing
just
tropospheric
ozone
and
then
a
combination
of
both
would
be
interesting
to
be
run.
C
Another
suggestion
was
an
ensemble
with
different
aerosol
emissions
or
different
aerosol
cloud
interaction
parameters,
there's
something
to
investigate
or
the
model
sensitivity
to
aerosols
and
csm2
and
then
finally,
it
was
related
to
the
high
resolution,
pre-industrial
control,
run
and
and
a
question
if
that
could
be
extended
to
be
a
longer
run
in
the
future
as
well.
C
I
think
so
we
kind
of
switched
out
breakout
groups
before
that
could
be
fully
clarified.
Oh
there's
a
message
in
the
chat.
L
Thanks,
I
can
try
to
answer
that,
so
the
eye
has
stimulation
may
actually
have
extended
somewhat
beyond
500.
I
think
they
might
have
actually
gone
out
to
650,
but
we
that
was
run
on
on
the
machines
in
china,
which
is
the
sunway
system.
It
wasn't
run
on
any
system
in
the
us
and
I
don't
believe
that
we
have
the
restarts
for
that
or
the
ability
to
run
that
in
the
same
you
know
it
would
be
a
major
machine
change
to
to
continue
that
run.
L
That
would
be
a
science
question,
whether
that
would
be
a
valid
extension
of
that
simulation.
D
A
So
is
that
was
that
the
summary
from
there
was
was
there
anything
else
to
add
sarah,
oh
nope,
okay,
okay,
cool
yeah,
I
think
also.
We
should
think
more
about
single,
forcing
runs
too
there's
going
to
be
this
single,
forcing
large
ensemble
mip,
which
is
asking
for
quite
a
lot
of
different
forcings.
We
have
some
of
them
already,
but
we
should
think
about
whether
we
want
to
contribute
other
ones.
A
I
don't
have
the
list
with
me
right
now,
but
I'm
not
sure
how
how
interesting
all
of
them
will
be,
but
we
could
think
about
whether
we
want
to
add
any
more
of
those
and
that
the
ozone
forcing
would
come
into
that
as
well
as
like
volcanoes.
Only
and
yeah.
I
don't
remember
the
rest.
A
So
we
could
just
have
an
open
discussion
now.
If
anyone
has
any
thoughts,
I
guess
brian,
you
posted
something
about
a
ppe
in
the
in
the
chat
for
that
was
for
cam,
or
is
that
for
coupled
simulation.
D
Yeah,
that's
it's
cam,
only
and
they're
short
runs
and
they're
focused
on
perturbing
parameters
that
are
in
the
well
convection
cloud
and
microphysics
schemes,
and
I
it
sounded
like
what
one
of
the
one
of
the
questions
that
from
sarah's
group
was
about
that
topic.
So
I
thought
I
would
share
that
there
is
this
ppe
available
and
I
think
it
will
be
made
public
imminently.
I
I
mean
very
soon,
if
not,
if
it's
not
already
public.
A
Okay,
excellent
yeah.
I
think
it
was
hansi's
suggestion
for
the
ppe
for
coupled
particularly
focused
toward
the
labrador
sea
freezes.
A
A
A
We'll
send
out
this
google
doc
to
the
mailing
list,
so
you
can
think
about
it,
some
more
and
put
in
any
ideas
that
you
like,
and
obviously,
if
you
have
a
suggestion-
and
you
want
to
take
the
lead
on
it-
that
you
you're
very
welcome
to
do
that.
So
it's
not
it's
not
like
you're
suggesting
experiments
that
other
people
will
will
be
doing.
You
can
take
the
lead
on
things
if
you,
if
you
suggest
them.
A
Great
well,
we
hope
we'll
be
seeing
you
in
person
next
time.
Thank
you
all
for
joining,
and
thanks
for
the
talks.