►
From YouTube: CESM Climate/Land Ice/Earth System/Polar Climate/Paleoclimate Working Group Meetings Day 1 PM
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Okay,
can
you
see
the
presentation
yes
hold
on
a
second
and
Sarah?
Can
you
introduce
this
special
session
sure
yeah?
So
the
the
motivation
behind
this
session
was
a
couple
different
conversations
with
folks
about
how
there
seems
to
be
a
lot
of
developments
in
the
ocean
model
hierarchy
realm,
and
we
just
wanted
to
highlight.
What's
going
on,
you
know
around
the
country
and
and
then
use
that
information
to
prompt
a
little
discussion
at
the
end.
A
Okay,
go
ahead:
okay,
cool
yeah!
Thank
you
for
the
opportunity
to
speak
today,
I'm
glad
I'm,
a
graduate
student
in
the
MIT
Hui
joint
program
today,
I'm
going
to
present
our
work
on
understanding
the
drivers
of
Atlantic
multitudinal
variability
using
a
stochastic
model
hierarchy.
This
was
recent
I
want
to
acknowledge
my
collaborator
collaborators
younger
Claude
NTN,
and
this
has
been
published
recently
in
the
Journal
of
climate
QR
link
QR
code
is
on
the
lower
left,
so
to
start
off,
I'm
just
quickly,
defining
Atlantic,
multiducato,
variability
or
AMD.
A
Essentially,
this
is
the
quasi-periodic
oscillation
of
sea
surface
temperatures
averaged
over
the
North
Atlantic
and
it
has
the
characteristic
horseshoe-like
pattern
of
forming
with
a
maximum
in
the
subfolder
region.
One
of
the
interesting
things
about
this
phenomenon
is
there's
still
an
ongoing
debate
as
to
the
relative
importance
of
ocean
versus
atmospheric
Dynamics,
for
example,
some
believe
that
AMD
arises
mainly
through
ocean
Dynamics,
such
as
variations
and
import
heat
transport
due
to
the
Atlantic
meridional
overturning
circulation.
While.
A
So
one
of
the
challenges
of
understanding
AMD
is
that,
since
the
observational
record
is
so
short,
we
have
to
often
rely
on
model
simulations
to
make
statistically
robust
statements
about
this
phenomenon
and
to
really
try
to
disentangle
these
two
components.
We
often
rely
on
simplifying
one
of
them.
So
an
example
of
this
approach
is
if
we
simplify
the
ocean
component
and
use
a
slab
ocean
model.
A
So
a
slab
motion
model
is
essentially
an
atmospheric
GCM
coupled
to
a
slab
of
fixed
depth,
and
you
can
see
here
looking
at
the
slab
ocean
temperature
equation,
that
the
temperature
is
primarily
determined
by
the
net
heat
flux
between
the
atmosphere
and
the
ocean.
There's
no
interactive
ocean,
Dynamic
Dynamics,
no
Communications
between
grid
cells,
laterally
or
vertically,
and
additionally
there's
an
additional
heat,
flux.
Correction
term:
that's
applied
to
the
model
to
to
obtain
a
realistic
seasonal
mean
SSC,
climatology
and
generally,
the
mixed
layer.
A
A
So
in
a
kind
of
famous
2015
paper,
Clement
had
all
found
that
slab
motion
models
on
the
right
were
able
to
reproduce
both
the
spatial
and
temporal
characteristics
of
A
and
B
that
are
comparable
to
observations
on
the
left
and
fully
coupled
models
in
the
center.
Raising
the
question
of
is
ocean
circulation
really
necessary
to
generate
a
and
b
more
recently,
there's
been
a
lot
of
work
that
has
found
that
slap
simulations
actually
have
higher
SST
variants
compared
to
fully
coupled
models.
A
This
suggests
that
the
addition
of
ocean
Dynamics
acts
to
damp
SST
variability,
particularly
at
low
frequencies
and
here,
is
part
of
where
the
challenges
of
using
the
slab
approach
arises.
In
that
when
we
remove
ocean
Dynamics
altogether,
it
becomes
a
little
bit
challenging
to
determine
what
exact
processes
lead
to
these
differences
between
slab
and
fully
coupled
simulations.
So
there's
need
for
a
more
transparent
process-based
way
to
understand
why
what
like?
What's
the
contribution
of
ocean
Dynamics
to
AMV?
A
A
The
atmospheric
forcing
of
the
system
is
on
much
shorter
time
scale,
so
it
can
be
parameterized
as
white
noise
on
a
certain
amplitude
and
given
some
anomaly
that
anomaly
is
damped
to
the
atmosphere
via
Heat
fluxes
and
the
strength
of
that
damping
can
be
determined,
can
be
parameterized
as
this
sort
of
linear
coefficient
than
the
sub
a
so
essentially
on
this
model
is
pretty
successful
at
capturing
the
mid-latitude
characteristics
of
SST,
but
all
these
parameters
do
vary
on
a
seasonal
basis.
A
Furthermore,
there's
also
the
consideration
of
mixed
layer
entrainment.
So
in
the
winter
time
the
mixer
temperature
anomalies
are
mixed
much
deeper
into
the
ocean.
However,
during
the
summer,
there's
a
formation
of
a
shallow
summer,
thermocline
this
summer,
client
effectively
insulates
temperatures
below
the
mixed
layer
from
further
interaction
with
the
atmosphere.
A
The
Following
fall
as
the
mixed
failure
deepens.
Again,
the
anomalies
are
re-entrained
back
into
the
surface.
This
affects
conditions
the
following
year
and
allows
a
year-to-year
recurrence
or
re-emergence
of
conditions.
This
can
be
described
in
the
stochastic
model
of
the
addition
of
an
entrainment
term
note
that
there
are
both
forcing
and
damping
components.
A
A
So
the
equation
looks
pretty
complicated,
but
there
are
only
really
three
input
parameters:
the
amplitude
of
forcing
the
heat,
flux,
feedback
and
the
mix
layer,
Def
and
I'm
going
to
go
really
quickly
over
how
we
estimate
these
parameters.
But
if
you
have
any
questions,
feel
free
to
ask
me
at
the
end.
A
So
for
the
forcing
we
essentially
do
an
eof
analysis
on
the
stochastic
component
of
the
heat
flux
in
csm1.
We
replace
the
principal
components
with
random
White
Noise.
This
essentially
allows
us
to
have
forcing
that
is
random
in
time,
but
has
spatial
coherence
and
varies
on
a
monthly
basis
for
the
heat
flux
feedback.
We
use
the
statistical
covariance
based
approach
from
the
literature
and
from
the
for
the
Mixed
layer
death.
We
use
the
seasonal
climatological
cycle
from
csm1
full,
so
it's
repeating
on
a
12-month
cycle.
A
We
estimate
all
these
parameters
from
csm1
pre-industrial
control
runs,
and
we
compare
the
resultant
SST
to
these
these
respective
simulations
to
really
try
to
understand.
What's
going
on
in
the
physics
of
this
model
world,
so
we've
developed
a
hierarchy
to
answer
three
key
questions
and
I'm
going
to
walk
you
through
on
how
how
we're
going
to
accomplish
this.
A
So
the
first
questions
we're
trying
to
investigate
again
how
the
seasonal
variation
in
these
model
parameters
impact
variability
to
do
that,
we
start
with
a
fixed,
the
classic
Model
with
fixed
inputs,
and
we
individually
add
in
seasonal
variation
in
damping
and
forcing
together
in
level
three
which
is
analogous
to
the
slab
simulation
and
then
variation
in
mixed
layer
depth
to
understand
how
entrainment
impacts,
SSC
variability
We
compare
what
happens
with
and
without
the
entrainment
term.
A
Finally,
we
focus
on
these
two
comparisons
to
see
if
we
can
recover
the
spatial
characteristics
of
a
and
b
using
stochastic,
forcing
so
to
kind
of
get
started
and
to
really
understand.
What's
going
on,
we've
chosen
a
single
point
in
the
subpolar
gyre
for
three
reasons:
it's
located
where
the
A
and
B's
maximum
in
the
csn1
full
simulation.
So
the
signal
is
the
largest
there.
A
It's
also
away
from
effective
regions
like
the
western
boundary
current,
allowing
us
to
focus
on
the
interplay
of
local
processes
and,
finally,
there's
a
very
typical
seasonal
variation
in
parameters
here
so
notice
that
all
the
parameters
are
high
in
the
winter
and
low
in
the
summer.
We
integrate
each
level
of
the
hierarchy
for
about
10
000
years.
At
this
point,
and
first
we're
going
to
talk
about
the
Persistence
of
SST,
so
here
I'm
showing
the
February
SST
light
Auto
correlation
we
choose
February
because
it's
the
month
the
deepest
mixed
layer
death
at
this
point.
A
So
it
allows
us
to
understand
this
extreme
case
where
we
expect
cases
temperatures
to
be
the
most
persistent.
So
the
canonical
Solutions,
the
stochastic
bottle,
is
an
exponential
decay
of
time.
So
this
comes
as
no
surprise,
but
when
we
look
at
what's
happening
in
csm's
lab
at
the
same
point,
you'll
notice
that
there's
a
sudden
slowdown
in
decoration
that
happens
during
the
summer
months.
A
So
the
reasons
for
the
C
correlation
becomes
clearer
as
we
advance
up
the
hierarchy.
Specifically,
when
we
look
at
it's
the
addition
of
seasonal
variation
enforcing
in
cyan
and
red
that
allows
for
the
reproduction
of
this
sort
of
slowdown
and
decoration,
and
so
what
exactly
is
happening
here.
So
if
you
remember
from
the
seasonal
variation
of
parameters,
the
stochastic
forcing
is
much
weaker
in
the
summer.
However,
in
these
slide-like
configurations
that
makes
their
death
is
constant.
A
As
a
result,
you
have
this
weaker
stochastic,
forcing
and
more
persistent
conditions
that
happen
during
the
summer,
suggesting
that
slap
like
configurations
have
this
sort
of
unrealistic,
SSC
persistent
and
to
bring
that
home.
What
happens
when
we
add
in
seasonal
variation
and
mix
layer
depth.
So
here
in
the
magenta
line,
You'll
see
that
when
we
add
that
in
this
persistent
shoulder-like
feature
it
disappears
and
to
kind
of
explain
that
again
in
the
summertime,
we
have
much
more
shallow
mixer
depths.
A
So
the
memory
of
that
is
no
longer
present
and
we
no
longer
have
persistent
conditions
through
the
summer
months,
so
going
on,
extending
that
a
little
bit
further
and
comparing
now
to
the
full
physics
ocean
model,
we
see
that
the
simple
addition
of
seasonal
variation
in
mixed
layers
isn't
enough
to
capture
the
winter
to
Winter
recurrence
of
conditions,
and
it's
only
with
the
addition
of
the
entrainment
term
that
we
actually
begin
to
capture
this
behavior
and
it
does
pretty
well
considering
that's
a
simple
model
of
just
three
terms.
A
When
we
add
in
treatment,
we
see
that
there's
an
excite
Improvement,
but
it
still
doesn't
quite
capture
everything.
The
main
effect
of
adding
in
entrainment
is
a
damping
at
intranial
time
scales
and
there's
still
an
underestimate.
That's
going
on
at
low
frequencies,
suggesting
that
the
Dynamics
Beyond
entrainment
really
come
into
play
at
these
lower
frequencies.
A
A
So
in
the
broader
subpolar
dryer
region,
we
see
a
similar
story
where
there's
stamping
at
inter-annual
time
scales
and
enhancement
and
authenticated
time
scales.
However,
in
the
Western
subtropics
and
the
Eastern
subtropics,
we
see
that
while
there
is
a
similar
thing
happening,
this
actually
widens
the
disagreement
with
what's
happening
in
the
full
physics
ocean
model
in
white.
So
this
is
no
surprise,
as
in
this
region,
there
are
other
ocean
dynamics
that
are
not
included
in
the
stochastic
model,
such
as
infection
and
subduction.
A
So
the
next
step
would
be
trying
to
understand
how
different
contributions
of
other
processes
in
these
in
the
subtropical
regions
could
lead
to
better
agreement
with
the
full
physics
ocean
model
and
we've
been
talking
a
lot
about
the
temporal
characteristics.
So
what
about
the
AMD
spatial
pattern,
so
here
I'm
comparing
to
the
CSM
slab
and
its
analogous
stochastic
model
hierarchy.
So
you
can
see
here
that
the
stochastic
model
does
a
pretty
good
job
of
capturing
the
major
features
of
the
canonical
horseshoe,
but
underestimates
its
amplitude
We,
compare
CSM,
slab
and
CSM
full.
A
A
The
answer
is
kind
of
adding
in
and
treatment
dense
things
way
too
much
and
also
causes
the
maximum
the
supply
which
either
shift
to
the
Northeast,
where
the
mixer
depth
is
maximum.
So
this
kind
of
suggests
that
the
missing
Dynamics
here
really
come
into
play
when
considering
the
small
scale.
Details
of
the
AV
pattern,
as
well
as
getting
the
correct
amplitude
of
variance
so
sort
of
the
main
takeaways
I
think
I'm
kind
of
running
a
bit
short
on
time.
A
So
I
might
just
leave
this
here
and
open
the
floor
to
any
questions
you
might
have.
Thank
you
for
listening
and
I
have
the
link
to
our
paper
on
the
lower
left
and
happy
to
take
any
questions
now.
A
This
is
a
clarification.
So
would
you
argue,
then
that?
Because
when
you
include
the
entrainment,
you
get
a
much
weaker
anomalies
in
the
North
Atlantic
that
they
kind
of
climate.
That's
all
studied
and
find
similar
amplitude
anomalies
to
the
couple
models,
kind
of
getting
things
right
for
the
wrong
reasons.
A
Right
yeah,
there's
the
because
of,
for
example,
in
the
in
the
slab
models,
because
of
that
sort
of
fixed
mixed
layer,
def,
there's
sort
of
more
I
guess,
like
kind
of
like
more
ocean
memory
in
that
region.
So
you
could
see
it
as
getting
sort
of
the
right
answer
for
the
wrong
reason.
I
guess
and
it'll
be
interesting
to
kind
of
interrogate
a
bit
more
and
I
guess
quantify
how
much
that
impacts.
The
yeah,
I
guess
kind
of
modify
that
a
bit
more
rigorously,
but.
A
Yeah
I
actually
had
a
question
yeah
thanks
for
the
the
nice
talk.
Do
you
have
any
idea
why
entrainment
is
damping
too
much
in
the
stochastic
model?
Does
it
have
anything
to
do
with
just
the
lack
of
motion
dynamics
that
ramps
up
the
amplitude
of
the
anomalies,
or
is
that
kind
of
an
unrelated
problem.
A
Right
so
the
idea
of
entrainment
damping
too
much
in
a
stochastic
model
I
need
to
think
a
little
bit
more
about
that,
and
we
haven't
investigate
that
in
much
more
detail.
There
has
been
other
works
in
the
literature
that
I
found
that
the
primary
contribution
of
entrainment
and
sort
of
these
heat
budget
calculations
is
to
Dem
anomalies,
particularly
in
the
subpolio
agile
region,
but
yeah
I
I
haven't
kind
of
specifically
investigated
within
the
stochastic
model.
A
A
comment:
if
you
haven't
first
or
Auto
regressive
process-
and
you
are
perturbing
the
auto
regressive
parameter,
it
will
act
as
a
damping,
so
so
that
that
would
then
give
you
the
variation.
You
could
even
use
this
to
deduct.
How
much
perturbation
you
need
to
get
the
correct
damping.
A
We
use
the
fluctuation
dissipation
Theory,
so
just
as
a
comment,
no
thank
you
yeah
I'd
love
to
talk
more
about
that
if
we
could
yeah
potentially
talk
after
or
thank
you
thanks
for
the
comment,
though,.
A
A
A
A
A
So
to
start
off
this
talk,
I
will
show
you
two
different
response,
and
this
is
the
response
of
the
ocean
model.
Oh
sorry
of
atmosphere
model
the
annual
mean
precipitation
rate
if
the
atmosphere
is
coupled
to
two
different
Ocean
Models,
which
is
the
establishing
model,
and
the
ogcn
in
this
case
is
the
pop2,
and
you
can
see
that
in
this
lab
motion
model,
the
itcs
actually
shifted
northward
in
the
north
hemisphere,
whereas
in
the
ogcm1
it
doesn't
shift
at
all.
A
Actually,
a
contract
intensified
at
the
equator
and
I
think
Glenn
did
a
really
good
job.
Introducing
what
celebration
model
is
so
basically
slow
motion
model
is
a
land
model.
It's
an
ocean
that
doesn't
flow
so
it
doesn't
respond.
The
ocean
float
doesn't
respond
to
the
atmosphere,
so
there's
no
horizontal
transport
response
at
all,
whereas
ogcn
contains
everything,
so
it's
very
complicated
and
so
there's
a
really
big
gap
between
them.
So
apparently
the
ocean
process
modulates
the
itcc
shift.
A
So
if
the
atmosphere
is
coupled
to
the
slab
ocean
model,
then
all
the
actions
have
to
happen
in
atmosphere.
So
there's
a
really
strong
itcc
shift
in
general.
If
ocean
is
not
reacting,
but
if
the
ocean
can
react
such
as
ocean
heat
uptake
either
I've
I've
taken
by
an
amok
or
the
mixed
layer,
then
you
can
directly
inhibit
the
reaction
by
doing
this,
but
a
really
popular
Theory
upon
this
stamping
effect
is
actually
the
acne
flow.
A
So
here
I
show
the
response
of
the
zono
wing
anomaly
in
the
gray
arrows
and
the
cross-equatorial
wind
is
shown
in
the
red
of
flat
arrows
and
you
can
see
the
corresponding
driven
eggplant
flow
shown
in
the
blue
is
transporting
the
heat
Southward
across
the
Equator,
so
that
because
the
ocean
is
helping
sending
the
heat
to
the
South
hemisphere,
so
the
atmosphere
doesn't
have
to
do
that
much.
So
that
means
that
Eggman
flow
is
damping,
the
itcz
shift,
but
in
the
actual
Earth
all
these
processes
will
happen
all
together.
A
So
it's
a
little
bit
hard.
You
can
theoretically
kind
of
verify
this
idea,
but
if
you
want
to
kind
of
see
these
things
in
the
actual
Earth
configuration
it's
kind
of
hard
because
they
all
happen
together
and
why
might
Shield
another
so
a
way
to
resolve
this
is
to
use
the
hierarchy
model
hierarchy,
and
here
is
the
model
hierarchy
we
developed.
It's
called
the
email
model
hierarchy,
Iman
stands
for
equine
mix,
layer,
ocean
model
and
basically
it's
a
model
that
resolved
two
different
processes.
A
One
is
the
very
mixed
layer
depth
in
time
and
the
other
one
is
the
Eggman
flow
that
respond
to
the
wind
stress
so
because,
in
this
particular
design,
I
can
turn
on
or
off
each
processes
independently.
So
it
actually
effectively
formed
four
different
models.
The
only
three
of
them
will
form
a
strict
hierarchy
in
terms
of
the
complexity.
So
if
you
turn
off
both,
that
means
you
have
a
constant
mixer
depth
and
non-responsive
ocean
flow.
A
So
that
gives
you
this
level
shape
model
and
if
you
turn
on
the
mix
layer
model
on
sorry
next
layer,
the
temporal
variation
of
mixture
death
on.
So
that
gives
you
the
mixer
model,
and
this
is
very
similar
to
a
Glenn's
mixer
model.
But
we
have
a
little
bit
diffusion
over
here
so
that
they
can
be
comparable
and
on
top
of
it,
if
you
even
turn
on
the
economics
layer.
A
A
So
here
I
need
to
introduce
the
admin
flow
parametrization,
which
is
done
introduced
in
kodron
2012,
so
basically,
equivalent
flow
here
is
the
result
of
the
force
balance
between
coriolis,
Force,
friction
and
the
wind
stress,
and
in
this
particular
formation,
because
you
introduce
the
friction,
you
can
resolve
the
The
Vanishing
of
the
eggplant
flow
at
the
Equator.
So
there's
no
singular
Point
in
this
parametrization.
A
So
the
result,
the
flow
is
the
superposition
of
the
traditional
equivalent
flow
and
the
frictional
flow,
and
here
you
have
to
notice
that
the
frictional
flow
is
particularly
important
at
the
equator
and-
and
we
pick
this
number
so
that
the
frictional
flow
will
dominate
between
five
to
five
and
it
will
gives
you
a
proper
upwelling
at
the
equator.
That's
how
this
Epsilon
number
is
chosen.
A
So
here
let's
come,
we
come
to
the
experiment,
preservation,
experiment,
which
is
Arctic
size
loss,
and
here
we
force
the
simulation
with
the
RCP
8.5
CI
slots.
So
blue
here
is
the
annual
main
sea
ice
thickness
in
the
pre-industrial
control,
and
the
red
here
is
the
sea
ice
in
the
last
20
years
of
the
20th
century
in
the
RCP
8.5.
A
So
the
difference
between
them
is
how
much
CIS
loss
we
force
using
the
notching
method
and
on
the
right
I
just
want
to
show
you
all
the
system
quickly,
equilibs
starting
to
equilibrate
after
80
years.
So
we
pick
the
last
100
years
as
the
statistics
on
the
climatological
mean
in
the
in
the
Cs
laws
and
subtract,
the
the
pre-industrial
mean
out
as
the
response.
A
Okay,
so
I
think
it's
enough
to
show
results
with
SST
and
the
precipitation
to
demonstrate
how
different
they
are.
So,
first
of
all,
the
SSD
in
general
is
similar
because
there
is
a
much
a
very
strong
woman
in
the
north
hemisphere
because
that's
exactly
how
we
force
the
sea
ice
loss,
but
their
precipitation
response
is
very
different.
A
So
the
top
and
the
bottom
are
the
precipitation
response
in
the
ogcn
and
this
love
ocean
model,
in
which
we
have
we
kind
of
understand
what
happened
because
their
celebration
model
doesn't
react,
doesn't
cause
any
ocean
heat
transport,
so
it
has
really
pronounced
pronounce
ipcc
shift.
A
But
in
the
mix
layer
model
an
idct
shift
is
damped,
because
the
mixed
layer
depth
is
exact
effectively
deeper
in
the
mixer
model.
Because
in
the
winter
it's
much
deeper
than
the
annual
mean
so
introduce
more
effective
ocean
thickness.
So
that's
why
it's
damped
in
the
mixer
model,
but
in
the
Imam
is
the
what
is
very
interesting
because
in
the
email
we
really
don't
see
the
damping
of
itcc
shift.
Instead
it
actually
amplifies
it.
A
So
if
you
see
the
zonal
mean
of
person
on
the
right,
the
red
is
the
shift
in
the
sorry,
the
precept
response
in
the
slow
motion
model
and
the
blue
is
depressive
response
in
the
Imam,
which
is
almost
double
the
strength
I
mean
in
terms
of
the
precip
compared
to
slab
ocean
model.
So
this
is
very
different
from
what
we
initially
expected.
A
So
if
you
dig
into
the
ocean
heat
transport
and
it's
even
more
comparable
in
this
way
that
this
level
should
model
here.
A
Sorry
here
are
the
three
lines:
the
black
line
is
the
the
total
heat
transport
and
the
breakdown
is
the
red
atmosphere,
heat
transport
and
the
blue
is
the
ocean
heat
transport,
so
you
can
see
in
the
slow
motion
model,
the
all
the
heat
transport
at
the
equator
is
done
by
the
atmosphere,
whereas
in
the
ogcm
the
heat
transport
at
the
equator
is
mostly
done
by
the
ocean
and
in
particular
it's
in
Atlantic
Basin,
so
that
that's
a
really
strong
signal
of
a
mark
action.
A
So
we
know
that
in
the
ogcn
there's
less
effect
of
ecma,
because
the
the
a
mark
is
damping,
everything
is
transporting
the
heat
totally.
So
there's
no
role
for
the
atmosphere
to
act,
but
for
the
email
you
can
see.
The
ocean
heat
transport
is
not
really
helping
atmosphere,
it
actually
transport
the
heat
to
the
north
hemisphere,
so
that
means
atmosphere
had
to
react
even
more
to
trans
to
rebalance
the
inter-emistic
energy
imbalance.
So
that's
why
there's
more
its
easy
shift
in
the
Imam,
so
I
will
do
really
quickly.
A
So
this
is
the
final
slide
that
the
after
we're
digging
into
the
reason.
We
think
the
reason
is
because
there's
a
strengthened
Tradewind
across
right:
it's
not
really
Trend.
This
is
transcend
I'm
already
down
the
wind
across
the
equator
and
especially
it's
blowing
the
ocean
through
the
north,
causing
the
frictional
flow
and
Associated
Ultra
heat
transport.
A
So
if
we
add
that
understanding
into
the
image
on
the
right,
that
means
we
are
causing
and
we're
going
to
win
anomaly
and
the
frictional
Oceanic
heat
transport
to
the
north,
which
causing
more
heat
transport
into
the
north
hemisphere.
A
So
that's
kind
of
the
thing
that
the
traditional
argument-
it's
kind
of
omitting
this
frictional
effect.
A
So
if
you
look
into
the
the
horizontal
map,
you
can
see
the
The
Black
Arrow
is
the
surface
wind
and
the
Green
Arrow
is
the
associated
ocean
flow.
So
it's
actually
at
the
vecting
SST
into
the
north
hemisphere,
causing
the
heat
convergence
in
the
north
hemisphere,
so
that
re-strengthened
ipcc
shift
in
this
case.
A
So
the
take-home
message
is
that
we
control
hierarchy
couple
to
CSN,
so
we
applied
that
to
the
Arctic
says:
laws
to
study
the
modulation,
the
ocean
and
in
the
ogcm.
The
a
mark
is
take
up
the
heat
strongly.
So
itcc
shift
is
inhibited
and
in
the
admin
allowed
model
the
frictional
flow
will
amplify
the
itcz
shift
and,
however,
this
outcome
is
sensitive
to
the
form
of
actual
parametrization,
specifically,
the
choice
of
Epsilon
or
the
frictional
term
will
change
your
answer.
Then
the
extent
of
the
ipcc
shift.
A
A
A
One
of
these
idealized
configurations
are
up
to
the
purposes
of
looking
at
the
response
to
Sea
ice
laws.
Yes,
yeah
I
think
I
think
that's
a
good
point,
because
I
think
that's
also
connected
to
how
we
force
the
Cs
laws,
because
in
this
configuration
the
Ci's
loss
is
completely
caused
by
the
pseudoheating
directly
to
the
sea
ice
model.
A
So
there's
a
really
strong,
fresh
water
flux
into
the
ocean,
so
the
amount
respond
really
strongly.
But
in
the
reality
the
sea
ice
loss
is
caused
by
the
the
CO2
forcing
so
the
response
was
supposed
to
to
be
different,
and-
and
the
second
thing
is
that
a
lot
in
a
lot
of
simulation.
A
If
you
force
the
CIS
loss
with
the
quadruple
CO2,
let's
say
then,
usually
after
a
few
hundred
years,
the
AMAC
will
restore
itself
to
a
pretty
good
extent
and
by
so,
if
you
really,
if
you're,
actually
looking
at
the
equilibrium
kind
of
the
solution,
sometimes
a
not
actually
doesn't
in
terms
of
the
heat
transport.
It
doesn't
change
that
much
especially
at
the
Equator.
So
that
will
be
a
window
of
her
of
opportunity
to
think
of
the
ocean
response.
A
As
a
really
surface
surface
reaction,
rather
than
a
deep
ocean,
so
I
think
one
of
the
message
here
I
didn't
write
it
out
is
that
amok
is
important,
but
mostly
in
the
transient
response.
It
may
not
be
as
important
as
it
is
in
the
equilibrium
ones.
A
A
Yes,
that's
a
good
question.
Sarah
yeah,
it
varies
seasonally,
it's
prescribed
seasonally,
so
the
mix
layer
death
doesn't
respond
to
the
change
of
the
wind
stress
in
the
atmosphere,
and
so
this
is
not
the
same.
This
is
more
similar
to
glenview's
a
mixer
model,
but
not
the
same
as
the
the
pencil
model,
which
you
know
Dr
yohokung
will
introduce
later.
A
No,
could
you
hear
me
and
see
my
screen
all
right,
so
yeah
hi
everyone
thanks
for
having
me
here:
I'm
John,
Faye
So,
currently
I'm
a
poster
working
with
Amy
and
Mark
in
the
University
of
Miami.
So
today,
I'm
going
to
talk
about
a
recent
public
study
on
Gil
that
issues
that
there
is
a
warming
hole
could
arise
in
your
GCM
without
ocean
escalation.
Basically,
the
slab
ocean
model,
I
think
Chen,
yo
and
Clinton
has
introduced
it
quite
a
bit.
So
the
background
here
is
very
interesting.
A
So
on
a
global
warming,
there
is
a
cooling
center
in
the
south
Portland
Atlantic,
and
this
has
been
hypothesis
to
be
caused
by
the
Slowdown
of
a
morgue
as
well
as
other
ocean
processes
like
the
gel,
as
well
as
the
Eggman
transport.
Most
of
them
are
windy,
driven
circulations.
So,
but
you
know
so,
the
SST
actually
physically
is
a
fed
by
both
atmosphere
and
oceanic
processes.
Then
we
can
to
ask
so
what
is
role
of
the
atmosphere
in
the
North
Atlantic
warming
hole
so
recently,
so
there
is
ancient
study
by
the
arrows
they
use.
A
They
just
focus
on
the
local
events
and
it
neglect
the
ocean
heat
transport
in
the
in
the
northlandhole
regions
and
they
use
a
simple
model
to
analyze.
The
temperature
change
in
the
North
Atlantic
warming
hole
regions,
so
they
found
that
various.
You
know
surface
heat
flux.
It
is
decreasing
over
the
north
anticomical
regions.
Therefore,
in
this
case,
this
decreasing
of
the
surface
heat
flux
could
like
produce
50
of
the
cooling
in
the
warming
hole,
and
then
they
add
the
increments
of
the
mixed
layer
and
again
we
could.
A
It
could
produce
another
50
of
the
cooling,
so
in
total
it
seems
like
the
local
environments
could
produce
the
whole
warming
hole
without
the
change
of
the
ocean
heat
transport.
Then
this
comes
to
our
motivation.
So
first
is
so.
The
arrow
is
using
a
simple
energy
balance
model.
Then
what?
If
you
know
GCM
like
a
slow
motion
model
in
a
slab
ocean
model?
We
don't
have
Dynamic
ocean
currents
and
the
second
is
so.
A
So
this
comes
to
our
method,
so
we
are
using
the
CSM
slab
ocean
model
and
we
run
nine-member
historical
simulations
from
1920
to
2005,
and
the
full
signal
simulation
is
the
same
as
the
cs1
large
Ensemble.
So
we
will
compare
all
simulation
with
the
Cs
monologue
Ensemble
as
well
as
observation.
So
in
this
case
from
to
the
following
Capital
model,
we
have
a
model
hierarchy.
We
could
understand
the
row
of
the
atmosphere.
A
So
here
is
the
results.
What
I
show
here?
It
is
the
acetic
Trend
in
the
observation
and
the
csm1
large
Ensemble
from
1920
to
2005..
So
what
we
see
here,
actually
the
phony
company
model
recently
reproduces
warming
hole
in
both
the
intensity,
as
well
as
the
shape
of
the
warming
hole
and
what
I
showed
here
at
the
bottom.
It
is
the
warming
SSD
Trend
in
the
slab
ocean
model.
What
we
could
see
is
here.
There
is
also
a
warming
hole
in
the
slab
ocean
model.
A
However,
there
are
some
differences
between
these
slab
Ocean
Models
in
meru
mijo
and
the
observation
at
polycarbon
model.
The
first
is,
the
intensity
is
different.
The
intensity,
the
cooling
industration
model
seems
like
only
50
of
the
40,
copper
model
and
observation,
and
the
second
one
is
that
the
shape
of
the
wormhill
is
quite
different
in
the
slab
motion
model,
so
the
shape
is
more
like
bullets,
but
in
the
40
carbon
CS
and
observation,
so
they
looks
like
a
diamond
or
like
a
fingerprint,
so
we
will
return
to
this
pattern
difference
at
the
end.
A
So,
even
though
the
absolute
cooling,
it
is
just
50
in
the
slab
ocean
compared
to
the
observations,
however,
this
warming
hole
is
actually
relative
to
the
global
warming.
Therefore,
we
Define
the
warming
hole
index
as
the
warming
hole,
temperature
minus
the
global
SST.
Then
we
plot
this
warming
ho
index
in
the
observation,
as
well
as
in
the
models
now,
we
could
see
actually
right
now.
A
The
revolution
model
is
quite
close
to
observation
and
both
of
them
like
have
a
0.8
or
0.9
degrees
percent
Cooling,
and
it
is
because,
even
though
the
warming
hole,
temperature
is
kind
of
50,
but
the
global
warming
in
the
slab
ocean
is
much
stronger,
because
the
ocean
in
the
slab
ocean
model
is
shallow.
Therefore,
so
the
warming
is
much
stronger
and,
as
a
result,
it
produces
quite
consists
of
warming
hole
index
in
the
slab
ocean
model.
A
A
So
then,
the
question
is:
how
does
the
surface
heat
fluxes,
mechanically,
exciting
or
Central
warming
hole?
So,
like
I
said
before
the
SST
in
the
slow
motion
models
is
purely
driven
by
this
net
surface
heat
fluxes
and
for
this
net
surface
heat
plugs.
We
are
able
to
decompose
it
into
different
physical
components.
So,
as
a
result,
we
are
able
to
understand
how
these
physical
components
change
necessary.
Ziplocs
then
change
the
csub's
temperature
trend
so
for
the
latent
and
sensible
heat.
It
is
easy
because
we
have
this
old
formula,
so
we
basically
analyzes
for
formulas.
A
So
we
are
able
to
know
what
kind
of
components
are
changing:
the
latent
heat
and
sensible
heat,
but
for
long
wave
and
short
wave.
So
these
relative
calculations
in
the
model
is
complicated,
so
there
is
no
expect
formula
there
so,
but
if
we
think
about
this
linearization
process,
actually
it
is
a
first
order,
Taylor
information.
So
basically
it
is
the
more
linear
regression.
So
then
we
just
regress
this
long,
wave
and
short
wave
onto
the
of
different
physical
variables
here.
A
A
So
here
is
the
fourth
decomposition
of
the
SSD
Trend
in
the
zone.
So
this
is
most
can
be
part
in
my
talk,
so
please
pay
attention.
So
we
first
focus
on
the
first
row
in
the
first
row:
I
decompose,
the
SSD
Trend
into
different
components,
different
surface
heat
fluxes,
then
from
second
to
the
bottom,
I
decompose,
the
surface
heat
fluxes,
the
SLE
Trend
in
these
surface
heat
fluxes.
A
So
so
that
means,
if
we
add
from
second
row
to
the
bottom,
then
we
get
the
first
row
and
if
we
add
the
second
column
to
the
rightmost
column,
then
we
get
the
First
Column.
So
here
we
see
these
six
different
variables
actually
lead
to
very
different
SSD
Trends.
And
if
we
add
them
together,
we
could
we
get
this
total
and
among
the
66
variables.
Actually,
there
are
three
groups
here.
The
first
one
is
the
cooling
group,
because
it
is
mostly
Cooling
and
it
is
a
social
ways.
A
The
service
ring
stress
and
the
LC
temperature
difference
and
the
second
one
is
the
neutral
group.
You
call
it
because
when
we
average
the
average
in
the
region,
so
it
mostly
becomes
zero
and
it
is
associated
with
relative
humidity
and
cloud
cover.
And
finally,
it
is
the
warming
group.
It
is
due
to
the
SSD
itself,
as
well
as
the
greenhouse
gases,
so
in
the
next
few,
slides
I
will
walk
through
these,
these
groups
to
industrating
magnets
behind
them.
A
The
first
one
is
the
role
of
the
wind
strings,
so
the
wind
strands
need
to
occurring
mostly
in
the
East
Basin,
and
then
the
cooling
extends
into
the
interior
ocean.
We
investigate
the
wind
Trend
in
this
pronus
land
region
weapon.
Actually
in
this
region,
the
wind
stress
is
increasing
so
critically.
As
a
result,
an
increase
in
the
wind
stress
will
enhance
the
sensible
and
latent
heat.
So
in
this
case
it
will
try
to
act.
A
A
The
second
one
is
the
LC
temperature
difference,
so
the
LC
temperature
difference
leads
to
a
cooling
that
mostly
in
the
west
boundary
and
then
extend
to
the
interior
ocean,
and
we
found
this
LC
temperature
difference
is
actually
also
related
to
the
surface
wind.
What
I
show
on
the
left
is
the
correlation
between
LC
temperature
difference
and
the
surface
temperature
adaption.
So
we
see
here
they
are
highly
correlated
in
this
warming
hole
regions
physically.
A
It
is
because
the
surface
basically
intensifies,
so
it
will
advance
more
cold
air
from
high
latitude
or
nearby
continent
to
the
open
ocean.
In
this
case
it
will
Amplified
LC
temperature
difference
here,
then,
the
Amplified
LC
temperature
difference
will
increase
the
sensible
heat
and
the
latent
into
export
energy
automation
then
lead
to
this
cooling.
So
again,
this
one
is
related
to
the
surface
wing.
A
Now,
let's
come
to
the
neutral
group,
this
in
this
neutral
group,
so
the
cloud
and
the
red
humidity
lead
to
a
kind
of
a
zonally
dipole
response
and
if
we
add
them
together,
so
we
largely
get
a
neutral
response.
So
we
found
this
cloud
and
relative
humidity
are
also
related
to
the
Wesley,
but
it
is
at
high
level
what
I
show
on
the
left
here.
The
shading
is
the
climatology
for
the
Jets
and
the
counters
are
the
chain
of
the
jet
stream.
A
So
we
see
during
the
simulation
period
the
jet
shifts
northward
and
also
elongates
only
as
a
result
in
the
Easter
Basin,
so
the
jet
increase
and
the
storm
track
increase
and
in
the
west
spacing
it
decrease
and
the
storm
track
decreases.
So
the
cloud
cover
actually
changed
in
the
same
similar
way,
so
the
cloud
cover
increase
in
the
East
and
decrease
in
the
west.
So,
as
a
result,
it
perturbs
long
wave
and
shorter
wave
therefore
lead
to
this
dipole
response.
Also,
the
relatively
humidity
changed
in
a
similar
way.
A
Then
it
leads
to
this
opposite
response
due
to
the
latent
heat.
Therefore,
they
cancel
each
other
and
lead
to
a
largely
neutral
response,
and
finally,
is
this
warming
group
for
this
warming
group.
Actually,
it
is
very
intuitive
because
the
net
effect
of
the
processes
I
discussed
above
audience
to
a
cooling
response.
Therefore,
this
TS
will
reduce
in
the
latent
heat.
However,
this
acts
as
a
negative
feedback,
because
a
TS
reducing
the
latent
heat,
then
it
will
try
to
reduce
the
latent
heat
out
of
the
ocean.
A
So
the
summary
here
is
that
we
find
the
atmospheric
circulation
lead
to
this
Northland
warming
hole
in
our
slab,
Ocean
Models.
The
change
of
the
high-level
Wesley
observes
the
color
cover
and
affects
the
shorter
wave
and
the
long
wave,
but
the
most
important
is
the
circumstancy.
It
change
the
LC
temperature
difference
as
well
as
a
surface
wind
intensity
therefore
leads
to
a
cooling
response.
However,
this
cooling
response
is
partly
Exempted
by
the
SSC
itself,
as
well
as
the
CO2
in
the
long
wave.
A
Therefore,
overall,
we
have
a
warming
hole
in
the
slow
motion
models
and,
finally,
one
discussion
is
I
mentioned
before
that
the
ship
between
the
as
an
abortion
model
and
the
polycarbon
model
are
very
different.
We
hypothesis,
it
is
because
we
don't
have
the
eye,
commands
and
transport
in
our
establishing
model.
We
could
offline
calculate
the
Eggman
heat
Divergence
in
our
celebration
model
a
refund,
so
there
is
a
key
Divergence
at
the
source
of
the
warming
hole.
A
Therefore,
we
hypothesis,
if
we
have
our
own
argument,
so
it
will
make
it-
may
make
the
warming
hole
more
closer
to
the
observation,
as
well
as
the
polycarbon
model.
Right
now
we
are
actually
simulating
the
historical
climate
change.
Building
the
new
ocean
model
hierarchy,
different
internal,
so
yeah
I
think
I'll
stop
here
either
and
I'm
happy
to
take
any
questions
here.
Yeah
thanks.
A
A
Go
ahead:
ask
your
question
great
talk
in
multivariate
regression.
A
A
This
is
a
really
good
question
so
yeah.
Basically,
this
is
the
culinary
co-linearity
in
the
regressions.
So
we
verified
almost
more
linear
regression
with
in
the
latency
and
sensible
heat,
and
the
results
are
quite
consistent
with
this
linearized
linear
linearization,
based
on
the
Block
formula,
so
yeah,
but
we
didn't
check
the
10
series
for
different
different
variables
here:
okay,
okay,
I,
see.
Thank
you.
A
A
Hi,
thank
you.
I've
been
looking
at
the
warming
hole
in
cesm2
and
in
cesm2.
The
direction
of
the
wind
trend
is
opposite,
so
it's
a
weakening
and
sort
of
a
strengthening
of
the
wind
Trend
in
this
region.
So
I
was
just
wondering
what
what
you
think
about
that
is
the
esm2
just
really
weird,
or
do
you
think
that
this
would
maybe
be
different
in
a
cesm2
slab
model
instead
of
cesm1
well
good
question
so
yeah?
This
is
definitely
interesting
thing
there.
A
A
A
You
guys
see
the
sides
yep,
that's
good,
okay,
so
I'll
go
ahead
and
start
so
I'm
Tyler,
my
PhD
student
or
PhD
candidate
at
the
University
of
Miami,
working
with
Amy
Clement
and
today
I'm,
going
to
talk
about
North
Pacific
climate
variability
in
a
new
cesm
motion
model
hierarchy,
which
very
conveniently
team
now
just
presented
on
a
couple
sessions
ago.
So
these
talks
are
all
very
related.
His
talk
was
great
and
covered
the
whole
modeling
framework.
We're
using
so
I
definitely
recommend
checking
that
out.
A
If
you
didn't
catch
it
I'll
skip
over
a
lot
of
the
stuff
that
he
talked
about,
which
is
going
to
mostly
be
summed
up
with
this.
So
taniel
covered
this
very
well
and
extensively,
so
I'll
go
ahead
and
just
move
on
from
there,
but
basically
we
have
the
the
four
the
four
members
that
taniel
is
describing,
where
you
have
the
fully
coupled
model,
the
Ekman
mixed
layer
or
the
Imam
is
we're
calling
it
The
mixed
layer
or
the
MLM,
and
then
the
slab
Ocean
or
the
the
song.
A
So
these
are
the
the
four
members
we're
working
with,
and
here
we
wanted
to
kind
of
dive
into
what
North
Pacific
climate
variability
looks
like
in
this
hierarchy.
So
just
to
begin
really
simply.
This
is
a
filtered
SST
variants
over
the
North
Pacific
in
the
top
left.
Here.
This
is
the
observed
pattern,
so
you
get
a
pretty
Standard
Pacific,
decadal,
oscillation
or
PDO
like
shape,
and
then
from
here.
A
This
is
the
fully
coupled
there's
some
kind
of
polar
Coastal
signal
here
that
may
be
ice
related,
although
it
extends
out
into
the
the
koe
region
too,
so
could
be
something
more
there
and
then
the
three
hierarchy,
members
here
also
show
the
the
Ekman
mix
layer
shows
a
pretty
similar
PDO
shape
as
well,
but
we
don't
really
see
this
quite
as
much
in
the
mixed
layer
or
the
slab
ocean
model,
and
so
these
percentages
here
are
the
correlations
with
the
observed
values
here.
So
this
is
the
2D
correlation.
A
Basically,
how
similar
this
map
looks
to
this
map,
and
so
you
can
actually
see
that
surprisingly,
the
fully
coupled
is
doing
the
worst
with
only
27
and
the
Ekman
mixed
layer
is
doing
by
far
the
best
at
replicating.
The
observed
SST
variants
at
about
60
percent,
so
this
is
just
a
really
kind
of
broad
look
and
we'll
dive
into
more
of
the
specific
North
Pacific
climate
variability
modes.
A
Next
next,
so
the
the
most
common
of
these,
the
most
well
known,
is
the
Pacific
decadal
oscillation,
which
I
mentioned
on
the
last
slide.
A
So
here
I've
shown
the
this
is
the
the
first
eof
of
North
Pacific
Sea
surface
temperatures,
so
this
top
column
or
this
top
row
shows
the
spatial
patterns
across
the
four
hierarchy.
Members,
all
of
them
look
pretty
fairly
similar.
If
you
look
at
their
shape
generally
I.
Don't
have
them
shown
here,
but
they're
correlations
with
the
observed
values
are
all
relatively
High.
A
The
slab
ocean
actually
has
the
highest,
but
it's
not
that
much
higher
than
the
rest
to
really
be
kind
of
too
notable
or
anything,
and
so
the
spatial
patterns
look
fairly
similar.
A
The
variants
explained
of
the
first
eof
is
all
relatively
similar
right
around
35
percent
for
all
of
these
the
time
series
and
do
note
that
the
the
time
periods
here
are
a
little
bit
different,
which
is
why
this
fully
coupled
time
series
looks
different
than
the
rest,
but
for
the
most
part,
these
have
fairly
similar
characteristics
as
well,
so
we're
seeing
at
least
in
the
PDO
a
lot
of
similarities
across
all
of
the
hierarchy.
A
These
are
all
also
very
similar,
so
so
far,
at
least
with
the
PDO
we're
not
seeing
too
much
of
a
difference
across
the
different
hierarchy.
Members
and
I'll
get
into
what
that
means
at
the
end.
A
So
we're
just
going
to
go
through
a
couple
other
or
a
few
other
North
Pacific
modes
do
kind
of
the
same
thing.
This
is
still.
This
work
is
very
much
kind
of
in
the
exploratory
stage.
We're
still
working
on
a
lot
of
this
right
now.
So
this
next
mode
is
the
Victoria
mode.
It's
also
known
as
the
North
Pacific
gyre
oscillation.
This
is
the
second
eof
North
Pacific
ssts
and
again
you
end
up
with
a
pretty
similar
kind
of
story,
as
you
had
with
the
PTO.
A
Although
you
start
to
see
more
of
this
variability,
the
spatial
patterns
change
a
bit.
There's
there's
a
you
know
a
general
shape,
but
it's
kind
of
morphing
a
bit
across
the
hierarchy.
You've
got
them
these
two
one
or
two
kind
of
relative
maximums,
and
these
seem
to
move
around
a
bit
depending
on
the
the
member
you're
in,
but
that
also
could
just
be
attributable
to
just
noise.
These
aren't
particularly
long
simulations.
A
We
have
about
250
200
years
of
data
for
each
that
we
can
use,
but
generally
again,
there's
nothing
here.
That's
really
standing
out
as,
like
you
know,
wow.
This
is
really
different
across
the
members
and
so
with
the
linear
regressions
on
the
Victorian
mode.
You
get
again
something
similar,
as
we
saw
with
the
PDO.
A
All
of
them
are
relatively
close,
close
enough
to
not
really
sound
off
any
alarms
or
anything
like
that,
and
the
the
same
same
story
with
the
Power
Spectrum,
so
I
think
there's
a
pretty
clear
Trend
here
with
the
the
sea
surface
temperature
modes.
A
A
So
in
the
fully
coupled
you
see
this
pretty
clear,
simple,
gyre,
just
one
one
spot,
but
it
starts
to
kind
of
expand
from
its
its
isolated
spot,
just
south
of
Alaska.
Here
it's
where
it's
now
kind
of
spreading
across
the
whole
basin.
So
this
is
something
that's
interesting.
We're
not
really
sure
we
don't
have
a
good
answer
yet
as
to
why
or
what
exactly
this
means.
A
But
that's
one
of
the
areas
we're
working
on
now
so
I'll
skip
through
these
just
to
save
some
time
but
again
with
with
these
sea
level
pressure
modes,
it's
again
a
fairly
similar
story
actually
with
the
SST
regressions,
where
you
get
again
you're
getting
very
similar
patterns
across
all
of
the
members
and
then
finally
we'll
look
at
one
more,
which
is
the
North
Pacific,
oscillation
or
I'm.
Sorry,
this
this
NPO
is
the
first
eof
of
sea
level
pressure.
The
other
one
is
the
the
NPI.
A
Is
the
area
mean,
but
this
North
Pacific
oscillation,
the
first
dof
is
the
atmospheric
counterpart
to
the
Victoria
mode
or
North
Pacific
gyro
oscillation,
and
these
all
look
very,
very
similar,
almost
exactly
the
same
and
they're
all
the
correlations
with
the
observations
are
all
very
close
to
each
other
and
quite
High.
A
So
same
story
again
with
all
of
this,
so
I'll
go
ahead
and
move
on
to
the
takeaways.
B
The
leading
sea,
surface
temperature
and
sea
level
pressure
modes
in
the
North
Pacific
roughly
look
the
same
across
all
of
these
different
ocean
hierarchy
members.
So
whether
you.
C
One
of
the
primary
drivers
of
the
PDO
or
you
have
the
Ekman
one
stress
included.
It
doesn't
really
seem
to
affect
how
the
the
North
Pacific
variability
is
looking
overall
and
so
one
of
the
ways
we're
interpreting
this
is
that
it
seems
to
suggest
a
more
minor
role
or
lesser
role
for
ocean
processes
than
in
the
North
Pacific
variability.
A
Than
may
have
been
previously
thought
within
this
see.
B
Esm1
framework
as
well,
and
all
of
this
is
without
historical,
forcing
it's
well.
Everything
we've
seen
is
just
the
pre-industrial
control
arms
like
chunk.
They
had
mentioned
in
the
last
talk,
we're
running
some
historical
simulations
for
this
hierarchy,
but
we're
still
in
the
process
of
getting
those
simulations
completed,
and
so
I
didn't
want
to
share
any
of
those
results
until
that
was
done,
so
I
think
they'll
end
there
and
take
any
questions.
B
Nice
talk,
Tyler
I.
Think
you
can
hear
me.
Have
you
compared
with
observations?
Would?
Is
it
correct
to
say
that
you're
fully
coupled
patterns
are
more
similar
to
observations
in
all
cases
or
are
there
somewhere?
One
of
your
other
kind
of
hierarchical
runs
is,
is
more
I,
don't
know
yeah.
So
that's
a
that's
a
good
question.
B
The
one
the
the
of
the
four
members,
the
one
that
actually
ends
up
having
the
highest
correlations
with
observations
is
the
slab
ocean
so
that
one
consistently
across
all
of
the
the
modes
we
saw
has
the
highest
correlation.
But
it's
it's
a
very
modestly
above
the
fully
coupled.
So
it's
like
0.7
to
0.65
or
something
like
that,
and
we
see
it.
So
we
see
that
Trend
pretty
clearly
in
all
of
them,
where
it's
the
slab
ocean
doing
the
best,
then
the
fully
coupled
and
then
the
Ekman
and
mixed
layer.
B
Each
of
those
two
are
a
little
bit
below
that,
but
they're
all
very,
very
close,
not
to
nothing
to
the
point
where
I
would
it
really
stands
out,
I
think
yeah.
That's
a
good
question.
B
Yeah
I
guess
it's
more
of
a
comment
than
a
question,
but
I
was
kind
of
interested
in
your
NPI
plots,
because
I
didn't
give
that
sort
of
a
proxy
for
illusion
low
variability.
Yes
to
me,
it's
interesting
that
all
of
the
sea
level
pressure
anomaly
patterns-
you
point
out,
are
different-
that
the
SST
patterns
still
look
so
similar.
B
So
it
was
kind
of
interesting
to
me
that
that
sort
of
Southward
elongation
of
your
you
know
illusion,
though,
essentially
that
that
wouldn't
impact
the
winds
and
then
the
RC
heat
fluxes
and
then
the
SST,
so
I
just
thought
that
was
kind
of
interesting
that
that
didn't
seem
to
be
the
conclusion
that
you
you've
all
Drew
from
looking
at
the
SST
patterns
that
look
so
similar,
yeah
yeah,
that's
a
that's
a
good
point
and
I
yeah
I
think
it's
definitely
still
open
to
interpretation
or
the
the
conclusions
we
made.
B
Are
we
haven't
written
it
up
or
anything
yet
so
I
think
you
can
kind
of
yeah.
Take
that
for
what
you
will
and
a
lot
of
this
we
put
kind
of
a
big
asterisk
on,
because
we
want
the
historical
simulations.
We
are
pretty
confident
that
those
are
going
to
give
us
a
better
I
guess
overall
view
of
all
of
this,
but
unfortunately
the
timing
didn't
work
out
to
have
those
simulations
ready.
B
I
think
there's
some
discussion
going
on
about
the
other,
there's
one
in
the
room:
okay,
one
more
question:
yeah
I!
Guess:
I,
don't
know
if
this
this
question
wouldn't
matter
anyway,
because
the
I
know
this
model
has
a
poor
tropical
North,
Pacific
relationship.
But
but
given
that,
do
you
see
any
differences
in
enso
with
all
these
different
configurations?
B
Is
that
having?
Yes?
So
is
it
just
that
the
tropics
is
so
disconnected
from
the
North
Pacific
in
this
model
that
it
doesn't
matter
right?
So
in
the
only
member
that
should
have
any
kind
of
in-so
variability
is
the
fully
coupled
the
slab
ocean?
Obviously
won't
in
the
mixed
layer
model
won't
said.
B
I
think
I
I
think
there
is
some
at
least
there
can
be
and
so
variability
in
the
slab
kind
of
a
slab
ocean
in
the
tropics,
yeah,
so
I
think
there
can
be
but
I
think.
Theoretically
there
there
shouldn't
be
any
since
the
mechanism-
yes
yeah
I
know
she
has
a.
She
has
a
paper
on
on
the
southern
oscillation
in
the
in
the
slab
ocean,
so
yeah,
but
generally
we're
not
supposed
to
be
seeing
in
so
on.
B
Much
of
these,
so
yeah
I
don't
know
if
that
really
helps
too
much,
but
I
haven't
looked
at
it
specifically.
I
haven't
like
looked
at
the
inso
time
series
or
anything,
but.
B
B
So
I
would
like
to
talk
about
the
CSM
pen
solution
model
and,
as
you
can
see,
from
a
large
number
of
courses,
kcj
collaborated
with
effort
across
the
broad
range
of
the
group.
So
what
is
the
pencil
model?
I
would
like
to
use
the
schematics
of
the
ocean
component
in
CSM
to
introduce
what
the
pencil
ocean
model
is
also
called
one
Department
color
motion
model,
so
mixed
layer
or
some
others
in
different
literatures.
B
B
B
So
we
can
compare
the
full
ocean
case
and
the
pen
solution
case
and
learn
about
the
role
of
ocean,
Network
processes
and
climate
variability
and
change,
or
we
can
compare
celebration
and
pencil
ocean.
You
can
explicitly
diagnose
the
rural
vertical
mixing
and
mixed
layer
depths
in
chamber
and
the
chairman
and
the
re-emergence.
B
And
decision
I
see
new,
as
you
already
heard,
from
multiple
excellent
talks
in
this
session.
Already
there's
been
multiple
studies
using
similar
1D
Ocean
Models.
The
series
of
papers
listed
here,
including
my
2011
paper.
We
use
so-called
Mike
Alexander's
mixoliation
model
using
gas
per
mix
layer
skin.
It
calculates
the
prognostic
mix
layer
scheme,
but
the
mix
list
can
be
used
in
this
mix.
Layer
or
1D
ocean
model
is
not
consistent
with
the
Quest
pointing
full
ocean
model,
so
that
was
the
limitation.
B
The
mix
layer,
depths
spatial
and
temporal
variability
prescribed,
so
it's
not
prognostically
calculated
during
the
model.
Integration
there's
another
similar
one
demanded
by
you
can
UK
metal
fits
group.
They
use
the
kpp
parameterization,
underneath
their
metal
fish,
unified
atmosphere,
model
kpp
is
used
in
most
of
the
ocean
generated
circulation
model
these
days.
So
it
could
be
a
nice
setup,
but
they
did
not
have
full
awesome
model
in
this
case.
B
In
this
case
we
use
the
3D
flux
correction
for
temperature
salinity
and
we
use
the
target
climatology
from
the
fully
corporate
simulation,
so
the
ones
can
be
consistently
compared
something
new
in
the
CSM
ocean,
10
solution
module.
We
also
apply
the
flux
equation
for
the
U
and
B
in
the
ocean,
because
the
kpp
needs
the
foreign
cashier.
B
So
this
is
governing
equation
for
pen
solution
model
and
the
terms
in
Black
are
the
only
terms
that's
pronatically,
calculated
y.
The
model
integration
did
direct
terms.
We
call
G
terms
are
the
terms-
that's
not
represented
explicitly
in
the
when
the
processes,
so
they
are
the
lateral
processes
in
the
ocean.
B
So
what
do
we
do
about
that?
How
do
we
prescribe
that?
So
we
need
a
two-step
processes.
The
step
one
is
the
to
calculate
the
geotoms,
so
we
do
the
prognostic
restoring.
So
we
run
some
preliminary
rights
that
get
the
target
reference
climatology
and
at
each
time
that
we
not
the
temperature
or
salinity
or
velocity
towards
the
target
climatology
and
we
calculate
the
G
term
throughout
the
stimulation.
A
D
B
C
B
Locks
convergence
and
salinity
flux,
convergence,
the
salinity
side
you
can
see
urja
is
along
the
coast,
especially
Where.
The
River
mouth
is
because,
unlike
the
heat,
the
temperature,
the
dam
actively
damped
by
heat
flux,
the
surface
flux
for
the
freshwater
does
not
activate
them
the
salinity,
so
we
have
those
apologize
which
need
to
be
taken
care
of.
B
F
B
A
A
D
A
H
H
H
What's
interesting,
is
that
the
mixer
layer-
that's
fluctuation,
is
much
smaller
compared
to
the
the
full
3d
ocean
model
result.
This
could
be
due
to
the
lack
of
lateral
processes,
but
also
there's
some
slight
chance
that
maybe
we
are
constraining,
mix
layer,
depths
artificially
too
much
so
student
investigating
how
much
this
stage
just
do
today
missing
letter
process.
H
This
is
Southern.
Ocean
plot
now
receive
better
strong,
salinity
drift
in
this
case,
but
the
other
barrier
with
behaving
okay
for
100
years.
H
So
this
is
such
a
quick
introduction
of
the
pencil
ocean
model
in
CSM,
so
implementation
is
ongoing
and
we
still
have
some
outstanding
issue,
but
we
think
we
are
close
to
completion
of
the
implementation
and
the
goal
is
to
have
multi-century
of
2000
years
of
Pi
control
configuration
csm2
coupled
with
pencil
ocean
model
simulation
made
available
through
this
working
group,
and
also
after
that
we
will
add
ecma
transport
to
the
pencil
ocean
configuration
so
hopefully,
in
about
a
year.
The
hierarchy
of
motion
models
available
in
CSM
framework
will
be
like
this.
H
Thank
you
for
the
talk.
I
think
you
were
answering
my
question
just
at
the
end
here,
but
I'll
just
ask
for
further
clarification.
The
current
iteration
of
the
pencil
model
doesn't
deal
with
upwelling
at
all,
correct.
That's
what
the
Eckman
transport
version
is
going
to
do.
H
H
H
I
was
just
wondering
how
computationally
expensive
the
pencil
model
is
compared
to
the
full
model
or
the
slab
ocean,
so
it's
actually
just
a
little
bit
of
over
it
today,
atmosphere
and
land
component,
so
it's
relatively
cheap.
Of
course
the
atmosphere
does
takes
most
of
the
computational
resources
for
full
CSM
too
anyway,
but
it's
substantially
cheaper
than
the
full
ocean.
H
H
H
Yes,
everything
looks
good:
okay,
thanks
for
joining
everybody,
I'm
K,
mcmonagle
I'm,
a
postdoc
at
NC
State,
working
with
Sarah
Larson
right
now,
but
I'm,
starting
at
University
of
Alaska
Fairbanks
in
August
and
I'm,
going
to
talk
about
changes
to
wind,
driven
ocean
circulation
and
how
they
can
amplify
externally
forced
warming
in
cesm2.
H
So
I
want
to
start
by
showing
this
observed
surface
temperature
warming
pattern
over
1979
to
2014,
and
what
we
can
see
no
surprise
is
that
the
globe
has
warmed
due
to
the
increase
in
greenhouse
gases,
plus
some
effect
of
aerosols
and
internal
variability
projecting
onto
the
trend,
and
so
we
have
some
kind
of
warming
pattern.
Here.
That's
been
observed
and
I
kind
of
want
to
point
out,
particularly
that
there's
very
little
warming,
or
maybe
even
a
little
bit
of
a
cooling
in
the
Southern
Ocean
and
the
East
tropical
Pacific.
H
In
the
observations
we
can
look
at
something
similar
in
the
cesm2
large
Ensemble.
So
if
we
look
at
the
ensemble
mean
or
the
externally
forced
component
in
cesm2
same
time
period,
this
is
what
we
see
so
clearly.
These
two
warming
patterns
are
different,
which
maybe
isn't
too
surprising,
because
there
is
internal
variability
in
the
observations
and
we've
tried
to
remove
that
in
the
by
taking
the
ensemble
mean
and
see
esm2.
But
we
want
to
understand
the
warming
pattern.
H
What
is
setting
this
warming
pattern,
both
in
the
real
world
and
in
the
model
and
the
understanding,
the
warming
of
or
sorry
understanding?
The
pattern
of
the
warming
is
really
important,
because
radiative
feedbacks
depend
upon
the
pattern
of
the
warming.
This
is
commonly
called
the
pattern
effect,
so
in
other
words,
if
we
have
a
different
warming
pattern,
we're
going
to
get
a
different
globally
average
warming
rate
and
a
different
rate
of
things
like
equilibrium,
climate
sensitivity.
H
So
we
know
that
the
pattern
of
warming
can
be
altered
by
ocean
circulation
changes
and
there's
also
evidence
that
the
ocean
circulation
has
changed
over
the
past
several
decades.
So
we
think
that
ocean
circulation
could
play
some
kind
of
role
in
this
warming
pattern
and
we
want
to
better
understand
how
these
changes
to
the
ocean
circulation
have
impacted,
both
the
pattern
and
the
rate
of
global
warming
and
broadly,
we
could
divide
these
ocean
circulation
changes
into
two
different
categories.
H
So
one
is
the
Atlantic
Primal
overturning
circulation,
which
we're
going
to
think
of
as
a
buoyancy,
driven
circulation
and
then
the
other
is
wind
driven
ocean
circulation,
and
so
there
have
been
quite
a
few
studies
showing
that
amoc
redistributes
heat
and
the
role
of
that
on
climate,
and
so
there
have
been
several
studies
that
quantify
how
a
change
of
amoc
would
alter
the
rate
of
global
warming.
H
These
tend
to
be
kind
of
future
climate
projection
type
runs,
so
something
like
a
doubling
or
a
quadrupling
of
CO2
amok
looks
something
like
this.
So
there's
warm
Waters
moving
northward
through
the
Atlantic,
deep
water
formation
and
the
high
latitude
North
Atlantic,
and
then
cooler
Waters
moving
Southward.
So
this
is
a
net
northward
heat
transport
in
these
climate
projection
runs
amok
weakens
that
then
leads
to
a
cooling
of
the
high
latitude
North
Atlantic,
which
increases
the
deep
ocean
heat
uptake,
and
then
that
reduces
the
global
surface
warming
rate.
H
So
what
this
looks
like
in
one
of
those
previous
studies
from
Winton
at
all
2013,
they
compare
two
models:
one
with
fixed
ocean
currents
and
one
with
free
ocean
currents,
and
this
is
the
SST
Trend
in
those
models
and
you
can
see
in
the
free
current
scenario.
The
high
latitude
North
Atlantic
is
much
cooler
and
that
leads
to
a
cooling
overall.
So
in
other
words,
projections
of
amoc,
where
you
have
an
amoc
weakening,
show
that
that
weakening
of
amok
actually
leads
to
less
global
warming.
H
But
if
we're
thinking
back
towards
this
kind
of
historical
warming
pattern,
it's
really
debatable
whether
or
not
amok
has
changed
in
recent
decades,
and
so
we
don't
think
that
amok
probably
played
a
very
big
role
over
the
historical
period.
Maybe
it's
more
important
for
projections,
but
so
far
we
haven't
seen
a
large
weakening
of
amok.
H
So,
instead,
what
we
looked
at
is
the
wind
driven
ocean
circulation.
The
one
driven
ocean
circulation
also
redistributes
heat,
so
I'm
thinking
things
like
the
sub
Pole
or
the
subtropical
gyres
here
and
many
of
the
observed
ocean
circulation
changes
seem
likely
to
be
wind,
driven
and
therefore
connected
to
atmospheric
circulation
changes,
so
in
particular,
I'm
thinking
of
the
strengthening
and
shifting
of
the
Southern
Hemisphere
westerlies,
that's
been
seen
in
observations
and
also
the
stronger
Pacific
Trade
Winds,
which
you
could
also
think
of
as
a
strengthening
of
the
Walker
circulation.
H
H
So
one
Ensemble
that
I'm
going
to
call
the
fully
coupled
model
or
fcm
has
all
of
the
changes
all
of
the
physics
that
we
think
are
relevant
for
the
real
world.
So
it
contains
these
changes
to
wind-driven
Ocean
circulation,
and
if
we
take
the
ensemble
mean
of
that,
we
we
get
the
total
externally
forced
signal.
H
Our
second
Ensemble
has
all
of
the
same
processes,
except
for
wind,
driven
ocean
circulation,
changes
and
I'm,
going
to
call
this
the
mechanically
decoupled
model
or
MDM.
So
if
we
take
the
ensemble
mean
there,
we
get
the
externally
forced
signal
from
everything
except
wind,
driven
ocean
circulation
changes.
H
So
if
we
take
the
difference
of
those
two,
what
we're
left
with
is
the
impact
of
wind
driven
ocean
circulation
changes
and
I'm,
going
to
focus
on
the
time
period
of
1979
to
2014.
H
a
few
more
details
about
these
ensembles.
So
these
were
forced
by
historical,
forcing
with
smoothed
biomass
burning
for
the
fully
coupled
model
we
use
50,
Ensemble
members
and
for
the
mechanically
decoupled
model.
We
use
20
Ensemble
members,
where
the
wind
stress
forcing
on
the
ocean
is
fixed
to
a
six
hourly
climatology
from
a
pre-industrial
run
of
the
fully
coupled
model.
H
So
this
is
one
way
to
visualize.
What's
going
on
so
in
each
of
the
model
ensembles
we
have
pop2
our
ocean
model
and
cam
6
the
atmosphere
model
and
these
exchange
different
fluxes.
So
if
we
look
first
at
the
fully
coupled
model,
cam
6
and
pop2
are
exchanging
time.
Varying
buoyancy,
fluxes
and
time
varying
wind
stress
or
momentum,
fluxes
in
the
mechanically
decoupled
model,
that
wind
stress
or
momentum,
flux
is
just
a
climatology
and
there's
no
deviations
from
that.
H
Okay,
so
we
want
to
focus
on
the
Wind
driven
ocean
circulation.
So
the
first
thing
to
check
is
to
make
sure
that
amok
is
pretty
similar
in
this
model
so
that
we
can
isolate
everything
except
for
amoc
in
terms
of
the
ocean
Trend.
H
So
this
is
showing
amok
Max
in
a
Time
series
over
the
two
models.
So
our
thick
lines
are
The,
Ensemble
means
red
is
the
fully
coupled
and
blue
is
the
mechanically
decoupled
and
the
shading
shows
two
standard
deviations
around
those
Ensemble
means
so
in
cesm2,
there's
an
increase
in
amok
three
to
about
1980
and
then
a
decline
after
that
and
over
to
the
right
is
a
histogram
of
the
trend
over
1979
to
2014
in
all
of
the
different
Ensemble
members.
H
So
in
both
ensembles,
the
amok
Trends
are
similar,
showing
that
the
simulated
amok
decline
is
primarily
due
to
changes
in
buoyancy,
not
winds,
and
that's
what
we
would
expect.
H
Next,
we
can
look
at
the
trend
in
the
wind
driven
ocean
circulation,
so
this
is
the
ensemble
mean
of
the
fully
coupled
model
minus
the
ensemble
mean
of
the
mechanically
decoupled
model
and
I'm,
showing
the
barotrope
extreme
function
here.
So
Reds
are
clockwise
and
blues
are
counterclockwise
and
the
stippling
is
where
this
is
statistically
different
from
zero,
and
so
there's
a
lot
of
features
on
here.
H
But
the
big
things
that
we
can
see
are
a
strengthening
and
shifting
of
the
Antarctic
circumpolar
current,
which
is
tied
to
changes
in
the
southern
hemisphere
westerlies
and
in
general.
These
changes
in
the
southern
hemisphere
westerlies
in
the
model,
are
consistent
with
the
reanalysis
wind
Trend.
H
So
we
can
find
Ensemble
members
that
agree
with
the
era5
one
stress
trend,
and
then
we
also
see
a
weakening
of
the
tropical
and
North
Pacific
circulations
in
the
model
and
that's
tied
to
a
weakening
of
the
trade
winds
in
cesm2,
and
this
is
inconsistent
with
the
era5
wind
stress
Trend.
So
there
aren't
any
Ensemble
members
that
are
able
to
reproduce
something
that
looks
like
the
trend
that
we
see
in
Era
5..
H
We
can
then
ask
what
role
that
has
in
terms
of
the
surface
temperature
Trends.
So
this
is
the
air
surface.
Air
temperature
Trend
in
the
two
model,
Ensemble
means
and
both
of.
D
Them
show
warming
nearly
everywhere.
There
is
a
North
Atlantic
warming
hole
in
both
model,
Ensemble
means
and
they're
fairly,
similar
in
strength
and
in
shape.
If
we
take
the
difference
now
between
these
two,
we
can
see
the
role
of
wind
driven
ocean
circulation
onto
the
externally
forced
warming.
So.
H
Red
blue
dipoles,
so
some
kind
of
Shifting
of
the
warming
pattern.
We
can
also
look
at
this
in
a
global
average
sense.
Oh
sorry,
going
back
to
highlighting
our
two
regions
that
I
talked
about
earlier.
We
can
see
that
wind
driven
ocean
circulation
seems
to
be
enhancing
warming
in
the
East
tropical
Pacific
and
in
the
Southern
Ocean.
H
Then,
if
we
look
at
what's
happening
to
the
global
average
warming
rate,
we
can
look
at
this
over
time.
So
the
similar
plot
is
the
amok
plot
before
our
thick
lines
are
The,
Ensemble
means
and
I
added
on
observations
here
as
the
black
line.
So
we
can
see
that
our
two
model
Ensemble
means
warm
at
very
similar
rates
until
about
the
1990s
and
then
you
can
see
the
lines
begin
to
diverge,
and
that
leads
to
an
a
difference
of
17
in
the
warming
rate
between
1979
and
2014
between
these
two
models.
H
Looking
over
at
the
histogram.
This
black
line
here
is
observations
and
also,
interestingly,
it
looks
like
observations
if
anything
maybe
agree
better
with
the
mechanically
decoupled
model
than
the
fully
coupled
model,
even
though
the
mechanically
decoupled
model
is
certainly
the
less
realistic
model,
so
we
can
think
about
why
we
see
this
Amplified
warming
due
to
changes
in
the
wind-driven
ocean
circulation.
H
One
thing
that
we
could
think
leads
to
a
different
warming
rate
is
ocean
heat
uptake,
but
the
fully
coupled
model
warms
more
than
the
mechanically
decoupled
model,
and
it
also
has
greater
ocean
heat
uptake.
So
the
difference
in
Ocean
heat
uptake
cannot
explain
this
difference
of
global
average
warming
rate.
So
instead,
what
we
think
is
happening
is
tied.
D
G
A
Sea
surface
temperature
Trend
due
to
changes
in
the
wind
driven
ocean
circulation
and
this
Amplified
warming
in
the
East
tropical
Pacific
has
been
shown
in
previous
studies
to
lead
to
a
reduction
in.
H
Low
clouds,
which
can
then
increase
the
downward
shortwave
radiation
towards
Earth
and
enhance
global
warming,
and
so
we
think
that's
what
we're
seeing
in
the
model.
We
see
things
that
are
consistent
with
this
in
terms
of
the
type
of
atmosphere
radiation,
as
well
as
the
cloud
changes.
H
So
in
conclusion,
changes
in
the
wind
driven
ocean
circulation
alter
the
warming
pattern
and
amplify
the
rate
of
global
average
surface
forming
by
17
percent
over.
A
1979
to
2014
and
cesm2
some
implications
of
this.
It's
likely
that
the
wind
driven
ocean
circulation
changes
in
the
real
world
have
impacted
the
surface
warming
rate
and
pattern.
I
What's
happened
in
cesm2
is
probably
not
likely,
but
in
the
real
world
we
do
expect
something
like
this
could
be
happening
and
then
also
atmospheric
circulation
changes,
maybe
a
significant
source
of
bias
and
historical
and
future
climate
projections.
I
So
getting
these
wind
stress,
Trends
right
as
well
as
the
ocean's
response
to
them,
would
be
one
way
potentially
to
improve
our
climate
projections
and
so
large
scale.
I
think
we
really
need
to
be
adequately
understanding
and
simulating
the
wind-driven
ocean
circulation
to
improve
our
climate
projections,
so,
in
other
words,
amok
is
not
the
only
part
of
ocean
circulation
that
has
these
global
climate
feedbacks
and
thank
you,
I'll,
take
any
questions.
I
Have
you
also
looked
at
Trends
in
SST
variability
between
the
two
different
runs,
and
is
there
a
difference
between
the
two?
If
you
have
Trends
and
variability
I,
we
haven't
looked
at
Trends
and
variability.
We
do
also
have
pre-industrial
versions
of
the
mechanically
decoupled
model.
So
we've
looked
some
at
like
internal
variability,
but
I
haven't
looked
at
Trends
in
the
variability
okay
yeah.
It
would
be
interesting
to
know
if
variance
changes
in
one
versus
the
other
yeah
agreed.
I
Great
talk,
Kay
I
was
noticing
in
your
wind
driven
ocean
circulation
map.
You
add
the
like
meridiano
tripol
in
the
North
Pacific
that
you
had
the
Box
around,
but
then
also
in
the
North
Atlantic,
and
also
in
this
plot
yeah
that
like
near
like
Cape,
Cod
and
Canada
area,
like
what
is
going
on
there
is
that,
like
maybe
like
Gulf
Stream
related
or
what
do
you
think
yeah?
That's
a
good
question.
That's
I'm,
trying
to
take
a
deeper
look
into
what's
happening
in
the
North
Atlantic
now,
but
I.
I
Yeah
again,
oh
very
nice,
tall,
Kay
I,
just
wonder
whether
you
can
really
disentangle
winged
and
the
heat
flux
in
the
tropics.
Have
you
looked
at
the
as
that?
Compare
the
ssts
or
in
in
the
tropics
I
just
things
is
Wayne
is
strongly
coupled
to
heat
flux
and
the
trumpets
can
really
separate
those.
I
So
the
wind
variability
in
the
heat
fluxes
is
still
there.
I
I'm,
not
totally
sure,
if
that's
what
you're
asking,
but
so
in
our
mechanically
decoupled
model,
if
the
wind
is
stronger
that
can
still
impact
the
heat
fluxes.
It's
just
the
wind
stress
that
is
a
climatology.
That's
passed
from
the
atmosphere
to
the
ocean.
Does
that
answer
your
question?
Yes,
okay,
but
I'm.
Also
Cur
I
was
curious.
I
Whether
you
can
look
at
what
happened
in
the
big
couple
runs
in
the
SST
in
terms
of
SST
tropics
like
in
the
mean
or
Trend
everything,
I
mean
or
trained
yeah
we're
looking
into
it,
I
mean
so
yeah
and
the
trend.
It's
there's.
Certainly
the
differences
here
in
the
mean,
there's
also
some
differences,
but
in
general
the
mean
State
between
the
fully
coupled
model
and
the
mechanically
decoupled
are.
Are
fairly
similar,
thank
you
any
other
question.
I
I
I
For
this
previous
session
was
again
conversations
with
many
individuals
in
attendance
and
also
Amy
Clement,
who
couldn't
be
here,
and
we
were
talking
about
how
there
seems
to
be
a
lot
of
you-
know,
development
sort
of
in
parallel
by
different
groups
across
you
know
our
community
about
you
know
these
different
either
idealized
ocean
runs
or
sort
of
switching
on
and
off
different
capabilities
like
as
Kate
was
describing
the
wind
driven
ocean
variability
and
I've
seen
some
other
groups.
You
know
kind
of
playing
around
with
the
RC
Heat
fluxes.
I
Like
what's
the
easiest
thing
and
then
ultimately
to
potentially
gauge
interest
either
from
individuals
or
the
whole
group
as
a
whole
about
whether
sort
of
a
greater
you
know,
Workshop
sort
of
focused
on
what
groups
are
doing
and
and
sort
of
big
picture
where
we
think
we
should
go
as
a
group
in
this
ocean
hierarchy
question
that
posed
as
like
a
a
greater
Workshop.
So
so
I
guess
I'll.
Just
it's
a
little
hard
for
me
to
know.
I
What's
going
on
in
the
room,
so
I
guess
I
shoe
and
Isla
I'll
sort
of
lean
on
you
guys
in
there.
But
you
know,
based
on
what
we've
seen
there's
there's.
You
know
the
stochastic
model,
the
mixed
layer
model
with
the
seasonally
varied
mixed
layer,
depth
the
slab
adding
in
the
echman
capabilities
case
version.
Where
he's
going
from
a
fully
coupled
model
and
removing
complexity.
I
You
know,
apart
from
those
sort
of
what
gaps
do
we
think
we
see
in
the
current
hierarchy?
And
what
would
we
like
to
see
moving
forward
and
I'll
just
open
it
up
to
anyone
really.
I
Now
this
is
I,
guess
yeah
I,
don't
really
work
in
this
area,
but
I
feel
like
one
thing.
I
learned
from
all
of
these
talks
is
that
it's
very
easy
to
get
the
right
answer
for
the
wrong
reasons
in
some
cases
or
or
get
the
wrong
answer,
because
you're
just
missing
some
parking
like
the
amok
and
so
I'm
wondering,
if
is
there
some
kind
of
like
standard
practice
for
some
kind
of
I,
don't
know
like
yeah
best
practices
for
ensuring
that
what
you're
seeing
is
reasonable
I
think
it
was
Chang
faded.
I
I
In
addition
to
you
know,
some
of
these
models
have
an
AMAC
and
some
don't
I,
think
Matt
Newman
brought
up
that
enso
is
different,
so
things
like
the
slab
ocean
have
like
this
thermodynamic
kind
of
insole-like
feature,
but
it's
different,
but
fully
couple
has
enso
and
how
does
that
sort
of
project
onto
what
we're
looking
at
too
I
think
is
a
is
a
big
question
and
being
aware
of
that,
when
we
do
these
comparisons
is
really
important,
but
I
don't
think
there
are
best
practices.
I
I
think
we're
still
figuring
out
what
what
everyone's
doing
and
and
understanding
everyone's
development
efforts
I
mean.
It
seems
like
having
this
broader
hierarchy
is
going
to
help
that
we
can
start
to
see
across
the
hierarchy
where
we
get
things
right
for
the
wrong
reasons
or
we're.
Adding
in
a
different
process
helps
a
lot.
I
Yeah
I
will
come
in
here,
so
maybe
right
now,
although
for
this
Authority
we're
trying
to
simplify
the
motion
model,
our
first
team
of
six
there's
another
set
of
simulation
called
fafmik,
it
flux
a
zombie,
14
experiment
in
those
experiments.
We
basically
try
to
isolate
what
is
the
temperature
in
the
circulation
change
caused
by
the
internal
protein,
how
that
will
contribute
to
a
local
and
Global
changes
in
in
a
climate?
That's
why
we
just
need
to
build
some
other
capability
into
the
3D
model
to
isolate
those
processes
while
the
model
is
running.
I
I
Yeah
yeah
I
just
want
to
quickly
share
my
thought
on
this,
so
it
seems
to
me
that
that
there
are
currently
two
parallel
Pathways
to
build
the
model
hierarchy,
so
one
is
making
the
slab
ocean
model
more
and
more
complicated
and
the
other
is
making
the
fully
coupled
model
on
the
modern,
simpler
and
simpler,
so
I'm,
not
sure
on
the
weather.
Those
two
Pathways
will
at
some
point
on
the
converge
or
not
because
technically
there,
these
are
very
different
approaches.
I
So,
for
example,
the
slab
ocean
model
always
confuse
confusingly
in
the
sense
that
in
that
model
the
ocean
circulation
is
not
affixed,
but
rather
it's
the
ocean
heat
transport
is
is
fixed
right.
So
if
we
think
about
underclimate
change
using
the
slab
ocean
model,
so
we
are
saying
that,
let's,
first
of
all
assume
the
ocean,
heater
transfer
doesn't
change.
So
what
the
climate
Global
combat
would
look
like.
I
So
this
is
what
slapper
ocean
tells
you
so
then,
on
top
of
this
slab
ocean
model,
we
start
to
add
in
the
components
like
the
Eggman
transport,
for
example.
So
in
that
way
we
are
adding
back
the
oceanic
processes
like
an
ocean
circulation.
So
it
seems
to
me
that
those
two
approaches
do
not
fit
each
other,
because
one
is
focusing
on
Ocean
heat
transport.
The
other
is
ocean
circulation,
but
it
maybe
it's
just
a
meat
that
have
haven't
fully
understood
this
approach.
Yet
so
I
just
want
to
hear
the
thoughts
from
the
audience.
I
Like
I
guess,
I
can
share
a
little
bit
of
my
thought
because
I
guess,
because
I'm
I
think
what
you're
saying
is
top
Town
versus
button
up
kind
of
development
and
I
think
my
is
I
I
developed
that
Ekman
mix
layer
model,
but
on
top
of
everyone
else,
but
but
I
think
that's
more
like
a
button
up
process
and
the
bottom
process.
I
I
think
the
the
pros
is
that
you
always
understand
what
you're
exactly
doing,
because
all
the
processes
you
are
you
just
name
it
and
and
when
you
name
it
you
actually,
if
you're
able
to
name
a
process,
that
means
you
are
able
to
quantify
it
right
like.
But
when
I
was
developing,
that
I
I
realized.
That
say
how
is
it
possible
to
decouple
Ekman
and
the
geotrophic
flow,
because
I
mean
you
physically
do
actually
the
same
thing
like
you
actually
have
the
you.
I
I
So
what
what
I
came
across
is
the
the
hard.
The
problem
is
about
interpretation,
if,
if
you
go
from
bottom
up
and
that's
where
kind
of
I
find
the
top
down
the
beauty
of
top
down,
because
the
beauty
of.
J
I
Right-
and
that
is
not
really
a
thing
when
you
use
the
simplified
model,
but
you
if
you
want
to
connect
to
the
say
the
ogcm
and
the
ogcm,
just
let
the
model
resolve
itself,
they
don't
necessarily
Define
such
a
layer.
I
So
I
think
that's
where
that's
probably
kind
of
that
falls
into
the
same
category
of
the
question
you.
K
I
I,
eventually
I
think
it's
still
very
problem
dependent,
but
it's
necessary
to
kind
of
think
about
this
when
anyone
wants
to
study
a
problem
using
a
hierarchy.
It's
like
your
interpretation
is.
I
A
L
L
L
L
This
might
be
heretical
to
say
at
a
CSM
working
group
meeting
but
I'm
wondering
does
anybody
happen
to
know
if
any
efforts
are
being
made
at
places
like
gfdl
to
do
something
similar
just
noticing
the
differences
between
csm1
and
csm2
I'm,
just
I'm
worried
that
we're
going
to
learn
a
lot
of
useful
things
in
csm2
that
are
going
to
be
very
model
dependent
and
I'm,
just
wondering
if
it's
worth
the
effort
of
bringing
along
other
modeling
centers,
while
while
you're
developing
an
ocean
model
hierarchy,
so
that
we
can
look
across
models
as
well
as
within
models,
yeah,
Dylan,
I,
think
you're.
L
I'm
assume
that's
still
and
I,
think
yeah,
okay,
I,
think
you're
really
hitting
on
sort
of
the
motivation
for
like
the
question
for
which
would
be
yeah,
bringing
not
just
the
esm
model
users,
but
getting
the
more
of
the
community
involved
to
get
a
better
sense
of
like
model
dependence.
L
I
know
there
is
some
version
of
the
gfdl
model
where
they,
you
know,
Force
the
ocean
with
climatological
wind,
stress
and
I,
think
you
use
jeans
papers
so
that
there
is
some
stuff
I
know
from
the
the
top-down
approach,
but
I
don't
know
about
the
bottom
up,
and
so
so
it
sounds
like
if
we
did
try
to
move
forward
because
it
sounds
like
there
are
a
lot
of
pretty
big
questions
that
we
all
need
to
kind
of
get
in
a
room
together
and
figure
out.
L
You
know
where
we're
going
to
meet
in
the
middle
or
if
it's
all,
just
going
to
be
Science
question
based.
You
know
how
we
decide
to
develop
these
things,
but
it
does
sound
like
we
need
other
modeling
groups.
You
know
part
of
this
too
right.
M
L
It
working
yes,
everything
looks
good
excellent.
Let's
start
good
afternoon,
everyone
from.
L
A
Forcing
agents
leads
to
substantial
changes
in
our
global
climate,
so
here
I'm
showing
an
example
about
this
climate
effect
for
the
sea
surface
temperature
chance
from
the
simplified
multi-monamine.
A
N
N
On
the
right
hand,
side
I'm,
showing
the
sea
surface
temperature
chance
from
the
csm1
large
Ensemble.
This
one
is
from
cloud
desert
paper
in
2020
and
they
have
shown
the
long-term
Trend
50-year
Trend
during
different
period,
and
they
have
found
the
aerosol
aspergamic,
aerosols
and
Warehouse
gas
are
leading
forcing
agents
during
different
periods.
N
And
a
lot
of
previous
studies
have
focused
on
Surface
and
atmospherical
responses
to
the
asymogenic
aerosol,
forcing
including
the
surface
temperature
I
have
mentioned
in
the
previous
slide,
and
also
the
idcc
shift
associated
with
the
hemispheric
energy
imbalance
and
also
the
sea
level
pressure
changes
such
as
the
channels
of
illusional
and
here
importantly,
we
found
the
aerosol
forcing
can
also
give
rise
to
substantial
changes
within
the
ocean.
So
here
I'm
showing
the
example
about
that.
N
This
is
the
a
mark
responses
to
different
external
forces
such
as
the
aerosol
single,
forcing
and
also
the
warehouse
gas
forcing,
and
we
can
see
a
strong
effects
of
asymogenic
aerosols
on
the
a
mark
anomalies.
N
However,
in
this
talk,
I'm
not
going
to
talk
about
the
changes
in
the
Atlantic
Basin
I
will
focus
on
the
subsurface
temperature
changes
in
the
non-specific,
because
we
can
found
a
very
strong
signal
in
the
non-specific
so
to
show
this
point
clearly
I'm,
showing
this
plot
about
the
UF
patterns
of
sea,
surface
temperature
from
the
csm1
aerosol,
forcing
ensemble.
So
here
I'm
showing
the
first
two
UF
modes
and
all
of
these
Smo
members
are
coming
calculated
in
time.
N
As
we
can
see
the
first
mode,
the
time
series
is
quite
noisy
and
the
special
pattern
is
associated
with
IPO
seems
to
reflect
the
internal
variability
in
the
second
mode.
We
can
find
a
very
a
tremendous
cooling
signal
in
the
non-specific
and
this
one
is
likely
to
reflect
the
effect
from
the
aerosol
4C
and
based
on
the
time
series
we
can
see
it
increased
before
the
1980s
and
decreased
in
future
scenarios.
N
So
it
seems
not
specific
could
be
a
good
place
for
us
to
find
aerosol
signals
within
the
ocean.
However,
if
we
also
look
at
the
uf1,
the
Pacific
also
is
Rich
of
a
lot
of
strong
internal
variabilities
and
its
complex.
So
another
question
is
whether
the
fingerprint
driven
by
the
asymogenic
aerosols
identifiable
from
the
background
noise,
such
as
internal
variability,
and
we
found
a
large
sum
of
simulations,
are
important
and
very
helpful
to
answer
these
questions.
N
So
next
I
will
show
the
two
models
with
large
Ensemble
simulations
used
in
this
study.
The
first
model
we
use
is
the
csm-1
large
Ensemble
simulation
on
the
first
set
of
larger
assemble
is
the
old
forcing
runs.
It
is
well
used
widely
used
with
40
members
and
ethnical
reinforcing
and
the
different
initial
conditions
and
another
set
is
CES
and
one
all,
but
one
forcing
runs.
So
it
is
same
with
this
all
14
runs,
but
without
the
industrial
average
of
forcing
with
20
members.
So
we
can
use
this
equation
to
excuse
me.
N
We
can
use
this
equation
to
derive
the
isometric
aerosol
single,
forcing
s
number
so
to
test
whether
our
result
is
robust
across
models.
We
also
look
at
the
result
from
the
Canadian
model,
the
Canadian
esm5.
It
is
also
a
large
Ensemble
simulation.
So
in
this
model
the
model
designed
about
this
single
forcing
is
a
little
bit
different.
So
the
aerosols
are
the
only
time
where
enforcing
agents-
and
this
slide
Ensemble
includes
50
members.
N
So
although
they
have
different
experimental
design
about
this
single
forcing
of
aerosol-
and
we
will
show
that
I
will
show
you
later
that
the
result
from
these
two
models
are
quite
consistent
in
terms
of
the
aerosol
individual
responses
and
the
region
we
focus
on
is
the
upper
1000
meter.
Temperature
change
in
the
non-specific
and
the
period
we
are
looking
at
is
from
1950s
to
2014.
N
and,
furthermore,
we
use
a
signal
to
noise
maximizing
pattern
analysis
to
find
the
pattern
that
maximize
the
signal
to
noise
ratio.
So
in
this
message,
the
these
extracted
patterns
are
associated
with
the
maximization
of
the
ratio
of
signal,
which
means
that
the
variance
of
the
enzymomine
to
the
total
variance
from
all
the
assembly
members
and
previous
studies,
such
as
the
paper
from
Rod
Wheels,
have
fun.
This
method
is
very
full,
very
useful
for
us
to
capture
extract
the
false
response,
especially
given
the
limited
Ensemble
member
and
below
that
I'm,
showing
the
result.
N
The
the
signal
to
noise
ratio
in
these
two
modes
are
higher
than
one
essentially
indicate
these
two
moles
associated
with
the
fourth
response,
from
the
csm1
aerosol
signal
forcing
sample
and
for
for
the
first
first
mode
we
can
see
it.
There
is
a
broad
Cooling
in
the
non-specific
and
the
corresponding
time
series
for
that
is
quite
linear.
So
we
can
regard
this
mode
as
a
linear
mode
induced
by
the
aerosol
faucet,
and
the
second
mode
is
more
interesting.
N
It
has
power,
has
a
very
unique
domain
pattern
in
the
non-specific,
for
example,
of
theory,
is
a
broad
cooling
at
a
surface
with
a
polar
work,
intensification,
especially
in
the
subpolar
regions,
and
we
can
also
found
the
cooling
signals
penetrate
into
the
interior
ocean
around
the
ventilated
thermocol
and
the
two
source
of
that
on
the
equator
word
flank
of
the
subtropical
gyrus.
We
can
find
a
warming
spot
right
here
and
the
corresponding
time
series
for
the
second
mode.
N
Fourth
mode
is
non-linear
or
non-monotonic:
it
increases
first
until
the
1980s
and
after
that
there
is
a
decrease
for
this
time
of
evolution,
and
we
can
compare
this
result
from
csm1,
with
the
result
from
the
Canadian
model
and
again,
the
first
mode
and
the
second
mode.
Surprisingly,
the
first
node
there.
There
are
quite
some
discrepancies
between
these
two
models,
especially
the
magnitude
of
these
cooling
signals.
However,
the
time
series
seems
to
telling
the
same
thing
these
first
nodes
both
represent
in
the
linear
mode
and
the
second
in
the
second
mode.
N
So
in
the
next
step,
we
will
focus
on
the
second
fourth
mode,
which
is
associated
with
the
non-linear
effect
now
monotonic
effect
of
the
aerosol,
forcing
and
and
the
first
mode
is
quite
linear.
It
seems
this
mode
will
be
partly
compensated
or
even
overwhelmed
by
the
warehouse
gas
warming
effect,
which
is
also
linear.
N
So
it
seems
it
is
possible
for
us
to
detect
this
non-linear
non-monitonic
effect
in
reality,
and
then
we
can
do
some
regression
analysis
on
this
time
series
from
the
second
mode,
so
here
I'm,
showing
the
regression
pattern
of
aerosol
Optical
depths
on
this
time
series.
So
we
can
use
this
to
find
the
primary
sources
of
aerosol
forcing
as
we
can
see.
Most
of
these
sources
came
from
North,
America
and
Europe.
N
It
seems
to
us
so
this
time
series
it
seems
to
indicate
the
aerosol
emission
due
to
the
Industrial
Development
and
after
that
the
decrease
seems
to
associate
with
emission
regulation
in
these
two
regions,
and
we
can
also
found
this
the
aerosol
sources
located
mainly
to
the
north
of
30
degree
nodes.
That
seems
to
be
why
we
can
find
a
polar
word:
intensification
of
surface
Cooling.
N
So
in
the
simulated
in
the
simulations
we
mentioned
earlier,
we
regard
noise
as
internal
variabilities
because
they
have
a
plenty
of
noises
associated
with
internal
variability
and
in
reality
these
observation
data
sets.
They
have
identical,
Force
response
and
The
Identical
internal
variability,
so
the
Noise
We
analyze
in
this
method
is
the
uncertainty
due
to
the
method
they
produce.
These
data
sets
such
as
the
Gap
filling
method
and
the
correction
of
the
biases
of
xpt,
and
here,
on
the
left
hand,
side
I'm,
showing
the
result
based
on
these
three
data
sets.
N
So
all
of
these
results
are
detrended
before
the
patent
recognition,
as
we
can
see
for
the
first
extracted
pattern
right
here
we
can
found
a
broad
warming
at
the
surface
and
this
woman
penetrate
into
the
interior
ocean
following
the
thermocline,
and
we
can
to
the
source
of
that.
We
can
also
found
a
negative
change
of
subsurface
temperature
on
the
southern
flag
of
this
subtropical
gyre
and
for
the
time
series
for
this
first
extracted
pattern.
N
It
is
also
non-linear
the
the
pin
curve
is
Anime
and
the
red
curve
represent
the
five
year
running
and
for
the
second
mode.
The
extracted
pattern
we
can
see
is
has
a
triple
pattern
right
here,
and
then
we
got
the
the
patterns
from
the
observations
we
can
compare
these
patterns
with
the
ones
we
have
shown
earlier,
such
as
the
fourth
pattern,
two
from
the
aerosol
single
forcing
runs
like
this,
and
in
addition,
we
also
show
the
the
pattern
associated
with
the
internal
variability.
N
For
example,
we
are
showing
the
PDO
related
pattern,
which
is
derived
from
the
pre-industrial
control
run,
so
we
can
clearly
see
the
difference
between
this
signal
and
the
noise
associated
with
the
internal
variability.
So
in
the
next
step
we
calculate
the
correlate
pattern
correlation
between
the
observed
and
simulated
pattern.
For
example,
the
left
in
the
first
table.
D
N
N
A
G
K
The
North
Pacific,
rather
than
in
the
North
Atlantic,
because
North
Atlantic
is
much
more
closer
to
the
its
emission
regions,
so
yeah
yeah
for
the
first
question.
You
mentioned
that
there
is
a
hemispherical
symmetry
in
the
sea.
Surface
temperature
induced
by
Warehouse
gas.
Is
that
right?
Yes,
it
seems
not
that
clear,
but
for
the
aerosol
effects.
Hemispheric
symmetry
is
much
more
clear
because
the
sources
of
aerosol
forcing
is
mainly
distributed
in
the
northern
hemisphere,
but
for
the
grass
gas
It
is
Well
mixed.
K
So
it's
hard
to
say
there
is
a
clear
asymmetry
in
the
sea:
surface
temperature
change,
the
service
temperature
change
induced
by
Warehouse
gas,
and
is
that
good
to
answer
your
third
question
because
I'm
mainly
focused
on
the
Arrow
so
effects,
not
the
greenhouse
gas?
The
effects
in
this
talk
so
yeah
for
your
next
question
about
error.
So
can
you
repeat
that
again?
So
my
question
is
why
the
fingerprint
is
likely,
mostly
in
the
north
specific
role
of
event,
just
in
the
Atlantic,
so
as
a
possible
reason,
is
the
location
of
these
houses.
K
So
it's
right.
There
is
an
open
ocean
wide
open
ocean
in
the
non-specific,
but
for
the
North
Atlantic
Basin
is
smaller
and
we
can
find
there
is
also
on
not
a
warming
hole,
as
you
mentioned,
but
aerosol
can
induce
a
cooling
hole,
it's
warming,
so
it
may
affect.
We
cannot
find
a
a
a
very
strong,
negative
change
of
temperature
at
the
surface
induced
by
aerosols.
K
K
Yeah,
good
okay,
this
is
Kevin.
Can
you
hear
me?
Yes,
he's
kind
of
low
hello,
yeah,
that's
good
yeah!
Well,
I
am
you
know
in
car.
Won't.
Allow
me
to
share
my
screen
and
so
somewhere
you've
got
my
presentation.
There
I
hope
and
it's
all
degraded.
Unfortunately,
it
has
a
lot
of
actions
on
it.
So
I'll
just
have
to
say
next.
Can
you
pull
up
the
presentation?
Please.
K
Also
I'm
a
bit
light
here
on
on
modeling.
This
is
mainly
observations,
but
that's
one
of
the
deficiencies
of
a
lot
of
the
work
that
goes
on.
There's
there's
very
little
regard
for
observations.
I
I
was
pleased
to
see
some
talk
of
observations
in
the
last
in
the
last
presentation.
K
K
Next,
so
you
know
conventionally,
we
look
at
the
at
global
warming
through
the
global
means,
surface
temperature
record
next
and
and
the
relationship
with
carbon
dioxide.
So
this
is
the
up-to-date
record
through
2022
next
and
it's
quite
noisy
next,
and
so
here.
What
I'm,
showing
is
the
ocean
heat
content
down
to
2000
meters
on
the
bottom
and
the
global
mean
surface
temperature
at
the
top,
and
you
can
see
that
the
ocean
heat
content
is
a
much
more
reliable
indicator
that
that
global
warming
is
actually
happening.
K
Some
of
the
fluctuations
from
year
to
year
in
the
ocean
heat
content
are
are
real.
Some
of
them
are
probably
in
the
noise
level,
but
the
Relentless
increases
are
are
certainly
real.
Next.
K
And
so
these
the
observational
base
for
the
for
the
various
records
from
gfdl
or
from
the
IAP
record,
which
is
mainly
what
I've
been
using
and
you
can
see
on
the
right
hand,
side
the
changes
over
time
and
it's
so
it's
only
after
about
2006
that
you
can
really
go
down
to
2000
meters,
but
by
using
that
record,
we've
been
able
to
reconstruct
the
record
reliably
back
to
about
1958
next,
and
so
this
is
this.
K
This
is
not
quite
up
to
date,
but
here's
the
overall
record
up
above
and
what
I
wanted
to
show
here
is
the
record
of
warming
as
stripes
for
the
500
meter
layers
from
from
the
surface
down
to
2000
meters,
and
so
that
you
can
see
the
warming
in
the
top
500
meters
showing
up
in
the
in
the
late
1970s
and
then
in
the
500,
to
1000
meter
layer
somewhere
around
1990
and
then
in
the
below
a
thousand
meters,
a
thousand
fifteen
hundred
the
late
1990s
and
then
after
about
2005
down
below
1500
meters
and
next,
and
so
that
this
heat
penetration.
K
The
studies
we've
done
looking
at
the
data
suggests
that
the
models
have
incorrect
stratification
in
the
top
layers
and
and
they
have
too
much
heat
penetrating
with
depth,
and
so
there's
a
lot
more
use
of
the
observations
that
are
needed
in
order
to
validate
models.
Next.
K
So
this
is
this
is
the
record
up
to
date
through
2022
and
there's
the
2022
anomaly
in
the
lower
left
down
to
2000
meters.
Next,
okay-
and
this
is
down
to
100
meters-
you
can
see
some
similarity,
but
you
can
see
the
La
Nina
signature
much
more
pronounced
in
the
top
hundred
meters.
K
You
can
see
it
also
to
some
extent
down
to
2000
meters
and
next,
and
so
you
know,
here's
the
La
Nina
a
good
part
of
the
La
Nina
signature,
and
you
can
see
the
warmth
in
the
Western
Pacific
and
especially
down
in
the
Coral
Sea.
And,
of
course,
that's
the
next
El
Nino
that's
coming
next
and
in
the
very
pronounced
warming
in
the
North
Pacific
and
throughout
the
South.
The
Pacific,
and
especially
around
where
I
am
now
in
in
New.
K
K
And
so
next
yeah,
and
so
if
we
look
at
this
region
here,
you
can
see
this
is
below
the
surface
as
to
you
know,
what's
happening
pretty
much
up
to
date
and
you
can
see
the
penetration
of
this,
the
warmth
along
the
Equator
out
into
the
Central
Pacific,
but
there's
this
huge
volume
of
warm
water
in
the
Coral
Sea,
just
south
of
the
Equator
that
is,
is
surely
going
to
play
a
strong
role
as
we
move
on
next.
K
K
The
South
Atlantic
did
a
little
as
well,
and
that
doesn't
match
quite
as
much
with
the
climatology
down
below,
but
you
can
see,
there
is
a
resemblance
between
these
two
and
so
in
general.
This
pattern,
which
we've
shown
for
the
general
Trends
in
salinity
in
the
ocean,
is
that
the
salty
areas
are
getting
saltier
and
the
fresher
areas
are
getting
fresher,
which
means
that
the
wet
areas
are
getting
wetter
and
the
dry
areas
are
getting
drier,
and
so
this
is
a
key
part
of
the
signature
and,
of
course,
that
extends
over
land.
K
So
so
the
dry
areas,
the
drought
areas
are
more
prone
and
the
wet
areas
are
are
flooding
that
that
does
depend
upon
connections
to
water
sources.
Next,
this
is
the
sort
of
regional
breakdown
and
there's
some
unusual
profiles
here
in
the
Indian
Ocean
and
the
North
Pacific
Mediterranean
Sea,
relatively
abrupt
increases
after
about
1990
or
2000,
much
steadier
larger
increases
in
the
Southern
Ocean
and
in
the
Atlantic
as
a
whole.
K
So
that's
where
the
strongest
warming
is,
and,
of
course,
this
is
very
much
related
to
the
role
of
enso
in
the
Pacific
is
to
keeping
the
overall
Pacific
cooler
as
a
consequence.
Next.
K
And
so
there
are
these
hot
spots
that
develop
as
a
part
of
the
overall
global
warming.
Some
sometimes
they
if
they
last
long
enough,
they're
called
Marine
heat
waves,
but
they
tend
many
of
them
tend
to
self-destruct
because
they
attract
activity
in
the
atmosphere
and
then
they
lose
heat
through
evaporative
cooling
and
they
fuel
the
storms.
And
so
this
is
one
of
what
is
giving
rise
to
all
of
the
increases
in
Extremes
in
the
weather
and
climate
around
the
world.
Next,
so
a
few
examples,
I'll
just
mention
here:
the
California
floods.
K
Next,
the
Pakistan
floods
last
year
and
and
many
Australian
floods,
rather
eventful,
which
is
typical,
of
course,
of
La
Nina
in
Australia
next
and
we've
been
bombarded
here
in
New
Zealand
I
want
to
I
want
to
show
you
a
little
bit
of
that
just
recently
and
and
next
the
just
recently
there
was
two
days
ago,
there's
23
inches
of
rain
in
the
Sao
Paulo
area
of
Brazil
in
24
hours,
just
incredible
flooding
with
rain
bombs.
K
So
next
and
so
in
Pakistan,
there
was
record
heat
in
in
March
through
May
of
2022,
and
you
know
there's
a
lot
of
information
on
the
slide
which
I
don't
have
time
to
dwell
on,
but
you
can
see
especially
hot
in
the
Indus
River
floodplain
region,
and
then
next
this
is
March
through
May
and
then
July
August,
the
Indus
flood
plain
was
bombarded
with
the
rainfall
and
there
was
extensive
flooding
with
over
1700
dead
and
the
crops
were
devastated
and
that's
still
in
the
in
that
state
today-
and
this
includes
things
like
cotton-
the
livelihood
of
of
the
farmers
in
Pakistan,
wiped
out
by
by
this,
but
very
much
related
to
the
warm
Indian
Ocean
Northern
Indian
Ocean,
that
that
was
related
to
the
heat
that
we
saw
in
March
through
much
in
much
room
May
earlier
next
I'm,
just
showing
one
slide.
A
2022,
which
led
to
huge
amounts
of
flooding.
O
And
790
millimeters
of
rain
in
a
week
and
over
20
deaths
a
lot
of
a
lot.
K
Of
this
was
relatively
local,
but
still
devastating.
Next
and
here's
some
scenes
from
the
atmospheric
rivers
that
we're
feeding
off
of
the
warm
sea.
O
Temperatures
ocean
temperatures
in
the
North
Pacific
coming
into.
Q
I
am
you
can
see
in
the
second
slide,
where
my
location
is,
we
had
280
millimeters
of
rain
in
a
day,
that's
11
inches
and
the
City
of
Auckland,
a
little
further
south
was
had
over
10
inches
and
all
through
there.
All
of
those
red
spots
had
very
extensive
frames,
very
much
related
to
this
La
Nina
pattern
and
it's
essentially
a
rain
bomb
that
came
down.
There's
a
subtropical.
G
K
Was
was
flooded
and
major
highways
were
flooded
and
so
on.
Next.
R
Where
I'm
located,
you
can
see
the
sea
level
pressure
map
down
the
bottom
960
68
millibars
of
pressure
was
recorded
right
near
where
I
am,
and
this
is
a
hybrid
system
now
and
and
caused
extensive
flooding
and
and
damage.
The
rain
and
error
on
this
occasion
was
about.
K
And
so
there's
the
actual
rain
Trace
there
in
this
in
this,
in
this
case,
right
where,
where
we
are
and
and
so
this
is
a
again
a
consequence
of
the
very
warm
ocean
temperatures
in
the
initially
in
the
Coral
Sea
and
throughout
the.
R
The
southwestern
Pacific
in
particular,
as
a
concept,
was
one
of
these
hot
spots
of
marine
heat
waves
associated
with
the
ocean
warming.
Next.
K
There's
still
over
a
thousand
people
missing
in
New
Zealand,
there
are
huge
areas
that
are
just
completely
covered
with
silt
and
and
and
slides,
houses
and
and
cars
have
been
buried
and.
A
Are
cut
off
and
they're
they're
also
cut
off
from
even
cell
phone
coverage.
So
there's
a
lot
of
recovery
work
going
on
right
now
and
we
don't
know
what
the
death
toll
is
next
and.
C
So
ocean
warming
is
continuing
ocean
heat
content
is
going
up,
but
we
get
these
hot
spots
Marine
heat
waves-
some
of
them
are
transient,
but
overall
the
ocean
is
warming
over
90
percent
of
the
the
heat
from
global
warming
goes
into
the
ocean,
and
one
of
the
consequences
is
that
we
have
a
lot
of
activity
in
the
atmosphere,
especially
with
the
extra
water
vapor
that
leads
to
flooding
and
and
there's
a
tremendous
amount
of
evidence
for
that
and-
and
the
question
is
how
well
is
all
of
this
kind
of
stuff
simulated
in
models
next.
C
Yes,
I
have
actually
a
question
for
Kevin.
Thank
you
so
much
Kevin.
Are
you
a
very
nice
presentation
nice
to
see
this
well?
Nice
I
mean
it's
worth
this
song
but
nice
to
see
your
analysis
of
the
subsurface
seat
content.
So
a
quick
question
I
have,
is
you
seem
to
be
relating
this
extreme
floods
in
Western
in
eastern
Australia
and
New
Zealand
to
to
this
increased
heat?
To
what
extent
do
you
think
they
are
due
to
this
prolonged
La,
Nina
that
we
have
seen
during
the
past
few
years?
C
Well,
yes,
that's
obviously
played
a
role,
and-
and
the
thing
is
the
first
year
of
a
la
nina-
inherits
the
EC
temperatures
in
the
extra
Tropics
from
the
the
previous
El,
Nino
and
and
they're
quite
different
in
the
extra
Tropics
and
over
time.
C
We
we
get
this
buildup
of
heat
in
the
North
Pacific
and
in
the
South
Pacific,
and
so
the
second
year
of
the
La
Nina
is
rather
different
than
the
first
and
the
third
year
compounds
it
even
further,
and-
and
so
this
buildup
of
heat,
especially
in
in
the
Coral
Sea
and
and
all
around
the
Tasman,
Sea
and
and
the
New
Zealand
area
is,
is
very
much
a
a
part
of
the
La
Nina
signature
and
the
Pawan
nature
of
it
has
has
meant
that
this
has
got
not
just
a
surface
signature,
but
quite
a
substantial
signature
below
the
surface,
so
that
so
that
I
think
in
New
Zealand
we
have
had
six
or
seven
of
these
tropical
subtropical
storms
there.
C
D
A
hurricane
that
they
call
it
a
cyclone
in
this
area
that
became
a
hybrid
storm
and,
and
so
it
had
some
baroclinic
aspects
to
it
as
well,
that
that
was
a
fairly
rare
event,
but
it's
caused
a
tremendous
amount
of
damage
as
a
consequence,
so
I
I
think
it's
a
combination
of
global
warming
and
and
the
La
Nina
sounds
good.
Thank
you.
Thank
you.
Kevin
next
talk
will
be
given
by
paiju.
C
Like
everyone
else,
I'm
interested
on
understanding
why
the
Pacific
is
cooling,
while
the
rest
of
the
planet
is
warming
so
a
few
months
ago,
I
had
this
intuition
that,
and
it
was
simply
an
intuition
I
could
tell
you
more
how
I
thought
about
it,
but
then
did
someone
involved
having
higher
resolution
simulation
of
ocean
processes
in
the
Southern
Ocean
to
produce
the
right
amount
of
cooling
in
the
Southern
Ocean.
P
D
D
A
So
here
I'm
using
I'm,
going
to
be
showing
results
from
a
comparison
of
high
resolution
versus
standard
research
and
simulations
performed
with
csm1.
A
These
runs
with
the
high
resolution.
Csm
will
run
under
historical
and
RCP
8.5,
forcing
from
1920
to
2100
very
much
like
in
the
large
ensemble
there's
40
members
of
the
large
Ensemble
I'm
going
to
be
looking
at
three
runs
using
the
high
resolution.
Csm.
J
Runs
I'll
be
focusing
on
two
analysis
intervals,
one
spanning
the
entire
21st
century,
where
forcing
dominates
so
I
can
use.
These
three
runs
to
isolate
externally
for
signals
and
compare
them
with
the
large
Ensemble
and
then
I'm
also
going
to
take
a
look
at
the
last
40
years,
1980
to
20
to
2021,
which
is
the
period
in
which
the
Pacific
has
been
Cooling
and
the
Southern
Ocean
has
been
Cooling
all
right.
J
So
first
results
first
things:
first,
if
we
look
at
the
changes
between
2120,
21,
2001
and
2100,
so
the
entire
21st
century
in
Sea
surface
temperatures
in
these
three
runs
which
I'm
calling
round
one
two
and
three
from
the
CSM
high
resolution.
J
J
These
three
blogs
show
us
that
the
changes
are
very
robust
to
internal
variability
and
they're,
largely
and
internally
and
externally
forced
response
to
increasing
greenhouse
gases,
which
is
the
dominant.
Forcing
over
this
interval,
we
can
convert
with
the
ensemble
mean
so
I'm
going
to
take
a
naive
three-member,
ensemble
mean
over
this
interval,
so
these
are
linear,
Trends
least
squares,
linear,
Trends
and
I'm
plotting
the
change
over
represented
by
this
trend
over
this
Hundred
Year
interval,
which
you
can
see
in
the
color
scale
and
I'm,
also
exploring
the
same.
G
H
Ensemble
to
the
left-
and
you
can
see
that
many
of
the
buttons
are
similar,
but
we
see
a
lot
less
cooling
over
the
Southern
Ocean.
Actually,
you
can
compute
the
difference,
but
and
that's
what
I'm
showing
on
the
right,
but
you
don't
have
to
compute
the
difference
to
see
that
the
high
resolution
version
of
CSM
produces
actual
Cooling
in
response
to
greenhouse
gas,
forcing
whereas
the
lens
produce
less
warming,
but
nowhere
in
the
Southern
Ocean
it
was
cooling.
H
So
it's
a
much
stronger
response
or
or
relative
cooling,
depending
how
you
want
to
see
it
in
many
of
these
areas
in
excess
of
two
degrees
different
between
what
the
lens
does
and
what
the
high
resolution
three
member
shows.
H
So
we
can
ask
ourselves:
can
this
response
explain
the
cooling
over
the
past
four
years?
Well,
things
get
a
bit
more
complicated
because
internal
variability
starts
to
play
a
role.
So
we
can
do
the
same
thing.
Do
my
naive
three
member
ensemble
mean
to
the
right
and
the
40-member
mean
from
the
lens
which
we
trust
is
capturing,
the
externally
Force
response,
and
we
can
see
that
the
lens
produces
a
pattern
of
enhanced
equatorial
warming
that
a
lot
of
us
have
described
in
the
past.
H
Less
warming
in
the
Southern
Ocean,
but
no
signs
of
actual
cooling
like
observed,
our
three-member,
mean,
shows
a
Trends
on
resemble
more
what
we
have
observed.
But
these
are
only
three
members.
So
let's
take
a
look
at
each
individual
member,
and
here
you
can
start
seeing
differences
and
the
differences
are
attributable
to
internal
variability,
which
can
be
strong
in
the
Pacific
over
40
year
periods
like
the
one
shown
here,
but
you
can
also
see
that
there's
some
patterns
that
are
similar,
particularly
the
strong
Cooling
in
the
Southern
Ocean.
H
Well,
we
can
look
at
the
lens
and
see
if
the
runs
that
look
like
my
three
runs
from
ihas
so
on
here
on
this
top
row,
you
have
the
in
the
trends
over
the
last
40
years,
simulated
by
CSM,
high
resolution
and
I
identify
through
pattern
correlation
over
the
southern
hemisphere
in
the
Pacific
and
identify
three
runs
from
the
lenses
I
Look,
to
resemble
the
trends
simulated
by
by
high
hasp,
and
you
can
see
that
despite
is
simulating
Stronger
Cooling
in
the
Southern
Ocean
is.
I
H
Strong
as
absurd
to
the
the
bigger
blood
shows,
The
observed
Trends
in
the
same
color
scale
over
on
the
same
interval,
and
it
shows
much
stronger
Cooling
in
the
Southern
Ocean.
We
can
compare
pick
a
region
of
a
the
Pacific
sector
in
the
Southern
Ocean
and
we
can
compare
the
magnitude
of
the
cooling
there
over
this
40
plus
year
period.
The.
A
And
then,
in
light
blue,
you
see
the
distribution,
the
histogram
of
the
values
simulated
by
the
individual
members
of
the
lens,
the
40
members
and
the
three
blue
lines
are
the
three
new
runs
with
the
high
resolution
CSM,
and
you
can
see
that
they're
still
within
the
distribution
of
the
lens,
but
very
unlikely
to
be
explained
by
the
letter.
In
fact,
there's
two
of
these
runs
in
the
high
resolution.
A
Csm
can
only
remind
you
of
cooling,
can
only
be
explained
by
only
one
member
in
the
lens,
so
there's
only
one
member
that
simulates
slight
cooling
over
this
interval,
so
there's
something
going
on
in
this
higher
resolution
version
of
CSM
that
produces
more
Cooling
in
the
Southern
Ocean
I'll
be
eat
not
as
strong
as
observed.
So
I
want
to
share
some
speculation
and
mechanisms.
It's
quite
unsatisfactory.
I've
started
looking
at
this
a
few
months
ago,
but
I
thought
I
would
share
these
results
here
to
motivate
some
discussion.
So
this
is
the
same
plot.
A
F
O
So
we
can
start
looking
at
other
responses
in
the
ocean
here,
I'm
showing
the
changes
in
sea
surface
height,
which
reflect
changes
in
the
upper
in
the
thermal
structure
of
the
upper
ocean
and
they're
related
to
changes
in
winds
and
an
ocean
currents.
And
we
see
that
when
mesoscale
80s
are
resolved
in
the
high
resolution,
CSM
the.
C
Changes
in
sea
surface
height
are
stronger.
The
pattern
is
quite
similar.
Both
types
of
models-
simulate
reduction-
is
subscribe
to
Polar
from
The
Polar
front.
So
these
two,
this
area
and
this
area
are
a
reduction
in
C
Surface
high
power
of
the
Polar
front.
So
this
means
that
the
polar
front
is
becoming
stronger
and
arguably
a
reduction
is
a
Surface
guide.
It's
going
to
be
mirrored
By
changes
in
the
thermal
gland
and
upwelling
and
it
will
produce
more
Cooling.
C
If
we
look
at
sonal
current,
we
also
see
that
there's
a
lot
of
Meandering.
If
you
notice
in
the
previous
plots
the
changes
in
sea
surface
temperatures
show
a
lot
of
Meandering,
which
is
the
same
between
all
three
brands,
so
resolving
it
is
and
meanders
in
the
Antarctic
second
polar
current.
A
Influences
ssds
are
very
small
scales,
so
here
are
the
changes
in
zone
and
velocity
over
these
hundred
year
period,
and
we
also
see
the
same
same
features
like
stronger
velocity
change,
with
an
accelerating
or
an
increase
in
zonal,
win
velocity
in
the
higher
resolution
version
of
CSM
in
that
area
in
the
Pacific
sector,
where
the
cooling
is
more
pronounced.
A
If
you
notice
these
areas
of
stronger
currents
coincides
where
the
area
where
the
gradient,
where
there's
a
gradient
in
the
changes
in
surface
height,
consisting
with
geography
between
surface
height
and
the
currents,
if
we
look
a
little
deeper-
and
this
is
the
last
result
that
I
have
if
we
look
a
little
deeper
into
what's
going
on
with
the
currents,
we
see
that
there's
actually
a
shift
in
the
polar
front
equator
world.
So
this
is
quite
counterintuitive
and
I.
Don't
have
an
answer
for
why
this
happens
because
the
vessel
is
over.
A
This
interval
increase
in
intensity
and
shift
forward,
yet
the
polar
front
moves
equator,
world
and
I'll
show
you
where
you
can
see
that
these
orange
curves
are
the
sea
surface
height
in
these
two
in
two
intervals,
over
the
21st
century.
The
dotted
line
is
the
first
half
and
the
full
line
is
in
the
second
half,
and
this
is
a
way
to
show
what
this
changes
in
sea
surface
height
represent.
They
actually
represent
an
equatorial
shift
in
the
later
half
of
the
21st
century
of
the
Polar
front.
A
Here
characterized
by
this
pronounced
gradient
in
Sea
surface
height
across
the
Southern
Ocean,
it
coincides
with
the
area
of
stronger
zonal
currency,
Antarctic
Circle,
polar
current,
which
also
becomes
stronger
towards
the
power
front.
You
can
see
that
this
dotted
line
is
the
first
half
of
the
21st
century,
and
these
solid
line
is
the
second
half
I
just
realized
that
I
inverted.
The
order
of
high
resolution,
so
high
resolution
now
is
on
the
left,
not
on
the
right.
A
If
we
go
to
the
to
the
right,
we
see
the
same
changes
very
similar
changes
in
the
lower
resolution
version
of
CSM,
but
not
us
strong
and
not
representative
of
a
shift
in
the
polar
front.
So
we
can
imagine
that
if
the
polar
front
is
Shifting
equator
world,
it's
bringing
colder
Waters
colder
polar
Waters
equator
y
Equator
was
producing
the
cooling,
so
I
know.
This
is
a
very
incomplete
explanation
and
yet
I
haven't
explained
why
the
Aries
play
a
role.
A
Arguably
ocean
Aries
play
a
role
in
this
distinct
Oceanic
responses
to
changes
in
the
west
service,
so
increased
resolution
leads
to
simulation
of
Southern
Ocean
cooling
under
greenhouse
gas.
Forcing
the
cooling
over
the
South
Pacific
sector
involves
an
equatorial
shift
in
the
polar
front.
It's
likely
due
to
resolving
ocean
Edis,
but
recall
that
the
atmosphere
model
is
also
higher
resolution.
Although
I
checked
and
the
changes
in
the
Westerly
winds
are
comparable
all
in
magnitude
between
the
two
versions
of
the
model.
The
full
mechanism
is
still
now
clear.
A
These
Cooling
in
the
Southern
Ocean
in
these
simulations
emerges
in
21st
century
relative
to
internal
viability
during
the
current
last
four
year,
observational
period,
it's
not
as
strong
to
be
distinguishable
from
internal
variability,
but
it
could
explain
the
Absurd
cooling
Trend
recall
the
magnitude
of
cooling
that
higher
resolution
version
of
CSM
simulates
over
the
last
40
years
can
only
be
explained
by
only
one
member
of
the
large
Ensemble.
So
we
have
two
new
runs
that
can
only
be
explained
by
one
member.
A
Because
the
timing,
we
just
take
one
question,
go
ahead:
if
you
have
any
other
questions,
just
ask
about
your
offline,
all
right,
Pedro!
Thank
you.
For
the
nice
talk,
I'm,
just
curious,
whether
you
have
look
at
the
email.
Compare
our
the
the
the
the
the
the
temperature
Trend
difference.
Whether
and
let's
see
whether
it's
connected
to
the
e-mark,
it's
a
good
question.
I
haven't
thought
about
it,
but
a
weaker
a
Mark
should
produce
warming
in
the
Southern
Ocean
right
so
and
in
both
simulations
I,
know,
I.
A
Think
the
the
high
resolution
I
haven't
checked
the
King
Charles
paper,
but
I
know
I.
Remember!
There's
that
comparison
of
high
harassment,
looking
at
the
lack
of
spatial
coherence
in
the
high
resolution,
assimilations
and
I
wonder
whether
the
the
amount
Behavior
can
be
quite
different
in
the
high-res
yeah.
It's
it's.
A
group
I'm
gonna
have
to
check
thanks
for
pointing
that
out.
G
G
That
work
is
in
collaboration
with
Richard
Seeger
and
then
I
will
briefly
mention
my
earlier
work
about
the
consequence
of
Indian
Ocean
warming
with
Alexei
and
also
another
word.
Recent
work
led
by
my
students,
each
in
Tien
and
in
collaboration
with
the
Clara
desert
about
the
cause
of
Indian
Ocean
warming.
G
India
ocean
is
unique.
It
sits
in
the
Deep
tropics
and
the
entire
Indian
Ocean
is
characterized
by
warm
sea
surface
temperature,
that's
close
to
the
convective
threshold
and
which
means
that,
even
if
you
have
a
a
small
increase
in
SST
over
the
Indian
Ocean,
it
can
potentially
be
translated
into
an
increase
of
convective
rainfall
and
therefore
affects
the
global
climate.
G
So
now,
if
you
look
at
the
Indian
Ocean
SST
as
a
function
of
the
time,
you
will
find
it
closely
tracks.
The
global
mean
surface
temperature
right,
so
it
has
been
persistent,
persistently
warming,
especially
after
1950s,
and
since
then
it
has
been
warming
by
about
one
degree.
So
now,
in
terms
of
a
spatial
structure,
the
almost
the
entire
Indian
Ocean
has
been
warming
on
the
robustly
with
the
peak
magnitude
along
the
Equator
and
so
across.
G
Different
data
sets,
even
though
there
is
some
disagreement
in
the
tropical
Pacific,
but
it
seems
that
they
all
agree
on
the
fact
that
the
Indian
Ocean
has
been
warming
faster
than
the
tropical
ocean
on
average.
G
So,
given
of
the
strong
Indian
Ocean
warming,
there
have
been
a
lot
of
studies
trying
to
understand
the
impacts
of
Indian,
Ocean
warming,
and
here
I
listed
a
few
examples
in
the
next
few.
Slides
I
will
show
you
two
examples
in
particular,
so,
first
of
all
not
Atlantic
oscillation.
G
So
in
the
second
half
of
the
20th
century
there
was
a
upward
Trend
in
the
Nao
which
has
drawn
a
lot
of
attention
because
it
has
influences
on
the
European
Hydro
climate.
So
the
hauling
out
all
basically
use
the
agcm
simulations
and
demonstrated
that
the
upward
Trend
in
the
Mao
index
within
this
time
period
and
it
was
primarily
driven
by
the
Indian
Ocean
warming
in
the
next
example.
G
That's
a
how
so
again
in
the
second
half
of
the
20th
century,
the
Sahara
region
has
experienced
a
multi-decado
drying
Trend
right
so
and
there
are
a
couple
of
studies
using
agcm
simulations
with
different
models,
and
they
all
find
that
the
Indian
Ocean
warming
can
potentially
cause
the
drought
of
the
cell
region
through
the
modification
of
the
worker
circulation.
G
So
a
few
years
ago,
I
did
some
work
rather
than
the
work
along
the
same
line,
but
we
argue
that
the
impact
of
Indian
Ocean
is
actually
Beyond
on
the
atmosphere.
It
can
influence
the
Atlantic
Ocean
through
the
arc.
Interaction,
so
on
depend
on
different
time
scales.
The
impacts
are
different,
so
on
short
time
scales
on
the
Indian
Ocean
can
induce
a
positive
AO
associated
with
the
Wesley
wind
anomalies,
south
of
Greenland,
so
acting
on
top
of
the
climatological
Wesley
gel
stream.
G
This
Westerly
wind
Trend
can
potentially
lead
to
a
local
cooling
through
the
Eggman
transport
and
also
the
enhanced
turbulent
heat
flux.
So
we
actually
find
that
this
mechanism
can
potentially
explain
about
90
percent
of
observed
nor
satanic
warming
hole.
G
So
our
longer
time
scales,
we
find
that
India
ocean
warming
can
modify
the
tropical
worker
circulation
and
suppress
the
rainfall
over
the
Tropical
Atlantic
and
the
Sahel
and
therefore
increase
the
salinity
over
the
Tropical
Atlantic.
So
then,
the
tropical.
Suddenly,
the
anomaly
can
be
transported
northward
by
by
the
climatological
ocean
circulation
and
eventually
the
North
Atlantic
becomes
the
South.
Here
the
a
mark
becomes
stronger
so
that
process
takes
several
decades
to
complete.
G
So
but
all
the
mechanisms
I
propose.
They
all
started
from
the
fact
that
the
Indian
Ocean
has
been
warming
faster
and
the
cause
in
the
local
rainfall
change
and
affecting
the
global
teleconnections.
But
that
idea
has
actually
been
challenged
20
years
ago
by
multiple
studies,
so
here's
one
example
from
Claire's
earlier
work
by
analyzing,
the
co-eds
ship-based
data
and
then
looking
at
the
difference
between
the
earlier
period
and
the
latter
period.
So
if
you
are
thinking
about
the
long-term
trends,
you
need
to
flip
the
the
sign
over
here.
G
So
the
top
panel
shows
the
cloudiness
change
and
basically
in
the
Indian
Ocean,
you
see
the
on
the
cloudiness
increase
in
the
western
part.
But
you
see
the
decreased
signal
over
the
eastern
part,
so
on
average,
maybe
it's
around
zero
and
interestingly,
the
the
most
robust
increase
of
cloudiness
is
seen
over
the
tropical
Pacific.
G
So
then
they
also
look
at
the
sea
level
pressure
data
also
from
the
ship-based
measurements
and
basically
it's
consistent
with
the
cloudiness
change
and
where
you
see
the
cloudiness
increase
of
the
tropical
Pacific
It's
associated
with
the
sea
level,
pressure
decrease
and
over
the
India
ocean
sea
level.
Pressure
actually
increases.
A
S
Indeed,
affecting
the
global
atmosphere,
so
why
we
couldn't
see
consistent
signal
in
the
atmosphere
right
on
over
the
Indian
Ocean
right
by
the
sea
level.
Pressure
increases
so,
but
one
thing
I
need
to
mention
is
that
that
works
were
completed
20
years
ago.
S
So
at
that
point
we
don't
have
sufficient
data
for
the
really
for
the
long-term
insurance,
and
this
is
the
the
India
ocean,
relative
sea,
surface
temperature
and
basically
with
the
tropical
mean
SST
subtracted,
and
so
that's
that's
because
in
the
tropics
it's
the
relative
SST,
that's
most
relevant
to
the
rainfall
change,
which
is
the
so-called
warmer
get
sweater
mechanism.
S
So
now
we
have
more
than
20
years
of
more
data
so
that
we
can
take
a
revisit
this
problem,
looking
at
the
long-term
trend
on
in
the
relative
India
ocean
warming
and
its
impact
on
the
overlying
atmosphere.
S
So
here
we
are
doing
similar
type
of
analysis,
and
this
is
from
the
Ico
ads
ship
data,
which
is
updated
version
of
the
coets
and
the
first
I
think
I'll
show
you
is
the
cloudiness
long-term
Trend
over
the
past
seven
decades
and
over
the
Indian
Ocean.
You
see
the
general
cloudiness
increase
in
the
western
part
and
southern
part
and
there's
some
cloudiness
decrease
over
here
and
on
average.
Now
it's
positive,
but
the
general
structure
is
very
similar
to
Clara's
earlier
results
and
in
the
tropical
Pacific.
S
Surprisingly,
we
still
see
quite
a
strong
increase
of
cloudiness
in
the
tropical
Pacific,
which
is
also
looking
similar
to
the
results
earlier
right.
But
there
are
two
cavities
I
should
imagine
about
the
tropical
Pacific
and
first
of
all,
the
tropical
Pacific.
There
is
a
strong
internal
variability
that
can
potentially
contaminate
the
long-term
trend
on
in
observation,
so
I
took
account
for
that.
S
We
normalized
this
long-term
trends
with
the
local
standard
deviation
so
and
basically,
the
middle
panel
shows
the
signal
to
noise
ratio
in
terms
of
accounting
extract
right,
and
in
that
case
you
see
more
robust,
a
signal
now
over
the
Indian
Ocean
and
the
less
robust
of
the
Pacific
compared
to
the
top
panel.
So
but
another
calculator
is
about
the
data
availability.
The
bottom
panel
on
the
shows
the
how
much
fraction
of
the
70
years
of
the
ship
data
is
available
at
each
grader
point.
S
So
where
you
see
red
color,
it
means
a
hundred
percent,
so
the
data
is
complete
and
the
the
interesting
thing
is
that
the
the
area
where
you
see
the
strongest
cloudiness
increase
over
the
tropical
Pacific
is
collocated
with
the
area
with
not
much
data
available.
So
where
you
see
blue
colors,
it
means
that
there
is
a
less
than
50
data
available
to
compute
the
long-term
Trend,
and
so
we
really
need
to
be
careful
for
the
results
from
those
blue
area
and
from
the
India
ocean.
S
S
We
also
look
at
the
sea
level
pressure
change
again
from
the
Ico
as
a
ship
data
and
consistent
with
the
previous
results.
We
see
the
sea
level
pressure
increase
instead
of
decrease
already
in
the
ocean
and
over
the
tropical
Pacific.
We
see
the
sea
level
pressure
decrease,
also
consistent
with
the
result
20
years
ago
and
about
on
top
of
this
panel
I
highlighted
those
crosses
highlighting
the
areas
with
the
more
than
30
percent
of
the
data
is
missing
so
and,
interestingly,
those
areas
co-located
with
the
area.
S
So
that's
why
we
took
a
look
at
another
data
set
on
the
sea
level,
pressure
product
on
the
developed
by
the
Headless
Center,
and
in
this
case
we
are
looking
at
the
same
time
frame
the
long-term
trends
in
sea
level
pressure
and
over
the
India
ocean.
It's
more
or
less
consistent.
You
see
the
sea
level
pressure
increase,
but
the
tropical
Pacific
is
very
different
and
you
even
see
the
opposite
sign.
So
if,
from
this
product
within
this
time
period,
you
see
the
sea
level
pressure
increase
over
the
tropical.
G
S
With
the
even
stronger
magnitude
than
the
Indian
Ocean,
then
putting
this
figure
in
a
broader
Global
context.
You
actually
see
the
top
Tropics
generally
has
a
sea
level
pressure
increase
and
the
higher
latitudes
have
a
sea
level.
Pressure
decrease
in
both
hemispheres
and
in
add,
in
other
words
over
the
past
seven
decades,
and
this
data
tells
you
that
the
air
mass
has
been
migrating
from
higher
latitudes
equate
award
towards
the
tropics
in
general
right
and
this
phenomena
has
been
reported
by
other
modeling
studies
as
well
on
the
global
warming.
S
So
the
bottom
line
here
is
that,
yes,
we
do
see
the
sea
level
pressure
increase
over
the
Indian
Ocean,
but
that
sea
level
pressure
increase
is
weaker
than
the,
for
example.
Tropical
Pacific.
So
then,
this
result
is
is
at
least
not
inconsistent
with
the
fact
that
that
Indian
Ocean
has
been
warming
faster.
S
So
then,
the
the
the
the
the
question
is:
what
caused
the
the
enhanced
historical
Indian
Ocean
warming
right.
So
this
work
is
led
by
my
PhD
student,
each
and
10
in
collaboration
with
Clara,
and
we
are
analyzing
the
CSM
one,
large
Ensemble
simulations
and
all
forcing
and
orbital
one
force,
experiments
and
basically
itching
found
that
the
bam
is
a
burning
aerosol
is
the
primary
driver
of
the
enhanced
Indian
Ocean
warming
scene
in
the
oil.
S
Forcing
experiments
and
I
should
have
mentioned
that
in
the
two
panels
shown
here
is
a
relative
SSD
Trend,
with
the
tropical
mean
being
subtracted.
So
the
red
color
means
that
the
Indian
Ocean
has
been
warming
faster
than
the
tropical
Main,
and
you
can
see
the
all-forcing
experiments
basically
show
consistent
results
with
observation
and
80
of
that.
S
Relative
warming
comes
from
the
biomass
burning,
aerosol
change,
so
the
bottom
panel
here
shows
the
long-term
Trend
in
the
BMB
areas
of
optical
depths
and,
as
you
can
see,
that
of
the
Indian
Ocean
has
experienced
a
long-term
reduction
in
the
Obama's
burning
aerosols,
in
contrast
with
the
tropical
Pacific
and
Atlantic,
where
the
biomass
burning
aerosols
have
been
increased
increasing
with
time.
S
So
as
a
result,
you
see
the
relative
in
the
ocean
warming.
So
now,
if
you
overlay
this
kind
of
a
pattern
on
top
of
a
uniform
warming
due
to
global
warming,
for
example,
you
will
start
to
see
the
enhanced
warming
over
the
Indian
Ocean.
S
So
I
don't
have
time
to
dig
into
details
about
this,
but
iching
is
planning
to
present
this
work
in
the
CSM
Workshop
in
June.
So,
if
you're
interested
please
come
to
that
talk
or
post
her.
S
So
this
is
my
final
main
slide
and
the
question
we
are
asked
is
to
what
extent
our
climate
models
can
reproduce.
This
enhance
in
the
ocean
warming
I
showed
the
results
from
csm1.
So
how
about
the
rest
of
the
models-
and
this
is
the
a
figure
I
grabbed
from
Lee
Jones-
resulted
three
years
ago
and
they
are
looking
at
the
cement
5
multi-model
main
on
the
sea,
surface
temperature
Trend
as
a
function
of
longitude
along
the
Equator
and
which
is
shown
as
a
red
line.
S
So
some
models
may
do
better
than
the
others,
but
on
average
multi-model
means
shows
that
almost
uniform
warming
along
the
Equator,
in
contrast
with
the
observed
the
trends
where
you
see
the
zonal
difference
clearly
so
more
recently,
we
look
at
the
same
F6
on
the
model
results
as
well,
and
basically
we
see
the
very
similar
on
the
response
and
the
Indian
Ocean
warming
didn't
really
stand
out
from
the
tropical
ocean
warming
in
general.
S
So
now,
if
you
look
at
the
inter
model
spread,
you
would
find
that
for
the
models
with
the
stronger
Indian
Ocean
warming,
they
tend
to
produce
a
stronger
tropical
ocean,
warm
in
general
as
well
so
and
those
different
models.
They
fall
onto
a
one-to-one
line
here,
I'm
highlighting
the
fact
that
the
relative
Indian
Ocean
warming
is
close
to
zero
for
all
the
models.
S
So
and
then,
if
you
compare
with
observations,
they
fall
on
the
lower
part
of
the
this
one
to
one
line,
which
means
that
Indian
Ocean
has
been
warming
faster
than
the
tropical
ocean.
So
that
raised
the
question
that
what
made
the
models
unable
to
reproduce
the
enhance
in
the
ocean
warming.
So
it
could
be
internal
variability.
S
So
you
know
in
The,
Observer
observe
the
trends
the
multi-decatal
internal
variability
could
potentially
contaminate
on
the
observe
the
trend,
and
then
it
could
also
be
due
to
the
model
bias
in
on
responding
to
the
external,
forcing
so,
for
example,
if
the
model
is
learnino-like
response
in
the
tropical
Pacific,
which
means
that
the
relative
Indian
Ocean
warming
would
be
weaker
because
of
the
warming
elsewhere.
S
Right
so
and
it
could
also
be
uncertainty
in
the
external
forcing
so
if
we
indeed
find
that
the
biome
is
a
burning
aerosol
is
the
main
driver
of
the
enhanced
Indian
Ocean
warming.
So,
as
we
all
know,
the
aerosol
forcing
is
not
well
constrained
in
the
historical
period.
So
if
those
kind
of
areas
of
forcing
is
underestimated,
then
we
will
be.
We
will
see
weaker,
like
enhanced
in
the
ocean
walking
signal
in
common
models.
So
in
the
next
few
years
we
plan
to
like
to
refine
our
what
is
exactly
the
cost
of
the
models.
S
So
it's
critical,
because
if
the
models
cannot
get
the
in-house
in
the
ocean
warming
right,
it
means
that
all
the
climate
impacts
are
related
to
the
enhanced
Indian.
Ocean
warming
won't
be
reproduced
by
climate
models.
So
here
is
just
a
quick
summary,
because
the
time
I
think
I
would
just
leave
those
take-home
messages
here
and
and
take
your
questions.
Thank
you
foreign.
S
S
Last
two
are
in
person,
so
Claire
and
I
get
the
end
in
person,
award,
I,
guess
perfect
attendance
right,
so
I
want
to
acknowledge
collaborators
on
this
work:
John
Sasha,
antonetta,
Julian,
Ishu,
and
since
we're
going
to
be
talking
about
effects
on
the
interdictable,
Pacific,
oscillation
or
IPO
connected
to
the
Australian
Bushfire
smoke,
let's
just
look
and
see
what
the
IPO
may
have
been
doing
over
the
last
few
years.
S
So,
of
course,
the
IPO
was
famously
in
its
negative
phase,
starting
at
about
2000,
and
that
was
associated
with
this
kind
of
slowdown
in
the
rate
of
global
mean
warming
has
been
written
about
quite
a
bit,
but
around
2015
or
2016.
There
were
a
couple
papers
that
were
written,
making
the
case
from
observations
that
maybe
the
IPO
had
transitioned
to
a
positive
phase
around
that
time
period.
In
fact
the
IPO
or
the.
S
If
it
wasn't
a
positive
phase,
it
was
consistent
with
this
more
rapid
increase
in
the
rate
of
global
warming,
because
that's
a
kind
of
a
well-known
connection
between
the
IPO
positive
phase,
you
get
more
rapid
rate,
score
warming,
IPO
negative
phase,
you
get
kind
of
slower
rates
of
global
warming,
and
so
we
looked
at
some
initialized
hindcast
here
from
the
dple.
This
is
one
example
from.
S
S
If
you
look
at
the
verifying
observations
average
from
2015
to
2019,
you
also
see
this
kind
of
positive
phase
of
the
IPO
compared
to
persistence,
which
was
persisting
the
negative
phase,
the
IPO,
so
it
indicate
there
was
some
indications
from
observations
and
from
these
initialized
hindcasts
that
perhaps
the
IPO
was
trying
to
transition
to
positive
around
that
time
period.
S
So
we
what
could
be
the
physical
processes
and
the
mechanisms
that
are
producing
this
kind
of
behavior
with
the
IPO
and
one
of
the
things
that
seems
to
be
important.
Are
these
officatorial
ocean
heat
content
anomalies
in
the
Western
Pacific?
So
if
you
look
in
this
upper
right
panel
here
you
can
see
these
two
boxes
are
drawn
where
we're
going
to
be
looking
at
these
officatorial
heat
content
anomalies
and
that
pattern
there
is
just
kind
of
a
generic
IPO
positive
phase.
S
These
are
as
temperatures
and
the
next
panel
down
there
in
the
middle
right
is
the
PC
time
series
that
often
is
used
as
a
index
for
the
IPO
and
the
fact
that
you
may
want
to
be
interested
in
trying
to
predict
something
like
this
is
indicated
by
that
lower
right
panel,
where
we
correlate
that
PC
time
series
with
observed,
precipitation,
you
can
see
there's
various
pattern,
anomalies,
precipitation
that
many
of
which
we've
been
seeing
over
the
last
five
to
ten
years,
and
so
there
is
interest
in
trying
to
do
a
good
job
in
analyzing
it.
S
So
when
you
look
at
the
observations
in
this
case,
this
is
a
representation
of
observations.
These
are
the
initial
States
from
the
dple
and
it's
also
consistent
with
other
re-analysis
ocean
observation
data
sets.
You
can
see
that
for
these
two
areas,
when
the
IPO
transitioned
from
negative
to
positive
here
in
the
70s,
you
started
seeing
the
official
heat
content
dropping
when
it
transitions
from
positive
to
negative
around
the
late
90s.
It
started
increasing
again
it
was
a
pretty
high
values
here.
S
Going
into
the
2000s,
and
then
we
got
the
2015-26
daily
Nino
and
started
to
drop
again
and
the
fact
that
we
think
there's
some
connection,
possibly
between
and
so
on,
and
your
annual
time
scales
and
the
IP
on
interdecatal
time
scales
is
that
this
transition
from
negative
to
positive
occurred
right
around
the
time
of
the
7273
big
El
Nino
event.
This
transition
from
positive
to
negative
occurred
around
the
time
of
the
two-year
La
Nina
from
98
to
2000,
and
then
this
most
recent
apparent
Triad
of
transition
was
the
2015-2016
El
Nino.
A
I
I
Looking
at
the
1800
year,
control
run
of
csm1
composite
IPO
Transitions,
and
for
this
time
period
we
got
about
50
cases
of
each
positive
negative
positive.
And
if
you
look
at
these
Composites
over
here
on
the
right,
you
can
see
that
the
this
is
the
composite
IPO
index.
Here,
the
blue
line
going
from
negative
to
positive
and
by
construction,
it's
across
a
zero
here
lag
zero.
S
S
S
S
These
things
could
actually
be
happening
in
the
climate
system,
so,
okay,
this
all
looked
pretty
good.
This
seemed
to
be
a
fairly
consistent
story.
We
had
this
these
kinds
of
big
drops
in
offer
control
heat
content
here
associated
with
this
2015-2016
El
Nino.
S
We
seem
to
have
some
evidence
that
the
IPO
was
trying
to
transition,
but
then
this
happened
so
the
officator
of
heat
content
rebounded
around
2019
and
you
can
see
the
increase
there
and
if
you
look
in
the
far
right
there,
you
can
see
that
the
equivalent
average
temperature
Trends
or
the
the
rate
of
warming
actually
flattened
out,
which
is
indicative
of
a
negative
phase
of
the
IPO
as
well.
So
what
what.
A
Well,
one
thing
that
happened
right
around
this
time
when
the,
when
this
all
seemed
to
kind
of
go
in
ways,
we.
T
T
And
this
animation
of
black
carbon
transport
from
NASA
here
on
the
lower
right
shows
that
the
the
smoke
was
actually
getting.
It
was
getting
entrained
in
these
various
synoptic
systems
going
across
the
subtropical
Pacific.
Basically-
and
you
see,
the
smoke
is
making
it
all
the
way
over
to
the
off
the
coast
of
South
America,
so
it's
least
making
its
way
across
the
Pacific.
T
T
So
to
look
at
this,
we're
going
to
look
at
two
sets
of
initialized
hind
casts
with
csm2.
Both
are
initialized
in
August
2019
right
around
the
time
the
Bushfire
smoke
started
running
from
three
years
to.
G
T
That
John
fasulo
published-
and
he
has
a
couple
papers
on
this
in
2021
and
2022
looking
at
this
kind
of
first
year,
and
so
the
plots
here
on
the
left
side
are
showing
the
aerosol
burden
in
the
model.
This
again
is
the
smoke
line
of
snow
smoke,
and
you
can
see
these
big
kind
of
orange
and
red
colors
here
coming
off
of
Australia,
and
you
see
those
orange
colors
going
across
the
Pacific
which.
T
See
from
the
observations
and
then
that
continued
on
to
December
2019
in
that
middle
panel,
lower
left
panel
by
February
2020,
the
Smoke's
starting
to
kind
of
Wane
and
by
March
2020,
the
wildfires
were
pretty
much
over
and
so
it's
kind
of
a
by
by
early
2020.
This
is
pretty
much
done.
T
The
fact
that
this
smoke
was
behaving
in
this
way,
there's
a
paper
by
lobidal
from
modus
observed
data
showing
that
that
is
not
far
off
from
what
the
model
experiment
is
trying
to
show,
and
if
you
get
these
smoke
particles
getting
all
the
way
across
and
getting
into
the
cloud
decks
off
the
coast
of
South
America.
You
can
expect
that
you're
going
to
have
increases
of
cloud
obedos,
and
you
see
these
orange
and
red
colors
here
off
the
coast
of
South
America
and
what's
happening.
Of
course,
these
smoke
particles
go
over
there.
T
They
provide
more
Cloud,
condensation
nuclei,
you
get
smaller
and
more
numerous
Cloud
droplets,
which
produces
brighter
clouds,
and
so
that's
what's
happening
in
the
model,
and
that
seemed
to
also
be
seen
in
the
series
data
from
satellite
observations
as
well.
T
Now
we
can
look
beyond
that
kind
of
first
year
and
this
is
going
to
have
an
effect
on
sea,
surface
temperatures
and
sure
enough.
It
does
because
if
you
look
here
at
the
first
year,
you
can
see
the
area
here
where
you
had
the
increased
Cloud,
albedos
you're,
getting
reduced
solar
at
the
surface
and
these
sea
surface
temperatures
are
cooling
off
here
in
this
model,
smoke
minus
no
smoke
difference.
T
If
you
go
to
the
next
year,
these
actually
are
growing
and
spreading
all
the
way
across
the
Pacific
and
by
the
third
year,
they've
actually
grown
even
more
and
now
they're
extending
from
the
southern
hemisphere,
almost
mid-latitudes
all
the
way
across
the
tropical
Pacific.
And
if
you
look
at
a
kind
of
a
comparable
period
in
the
lower
left
from
observations,
this
is
October
2020
to
September
2022.
You
see
a
similar
kind
of
pattern
of
a
huge
area
of
anomalously
cold
water
there
in
the
kind
of
the
Eastern
Pacific.
T
But
the
interesting
thing
about
this
is,
if
you
think
the
smoke
was
the
cause
of
this.
The
smoke
was
pretty
much
done
by
the
end
of
that
first
panel
up
there.
So
in
the
second
and
third
panels
lead
years,
two
and
three
were
these
SST
anomalies
are
actually
growing
in
area
and
intensity
and
the
Smoke's
not
causing
that
so
what's
causing
this,
this
kind
of
persistence
and
these
growth
of
these
of
these
anomalies.
T
And
indeed,
if
you
look
at
the
rainfall
anomalies
here
in
lead
years,
two
and
three
on
the
left
side,
you
can
see
that
over
these
cooler,
ssts
you're,
seeing
these
blue
colors
are
an
almost
the
or
negative
precipitation
anomalies
in
this
model.
Experiment
smoke
minus
no
smoke
over
the
area
where
you
get
kind
of
warmer
sea
surface
temperatures,
where
you
get
more
low-level
convergence,
you're,
getting
increased
precipitation
and
the
fact
this
could
have
been
happening
in
observations.
T
You
can
see
here
in
2020
2021,
you
get
this
increased
precipitation
with
the
maritime
continent,
the
decrease
over
the
Western
tropical
Pacific
and
the
same
in
2021
and
2022..
So
this
seemed
to
have
set
off,
at
least
or
probably
at
least
two
sets
of
coupled
feedbacks
that
were
able
to
maintain
and
grow
these
Sea
Service
temperature
anomalies.
The
first
has
to
do
with
this
kind
of
increase
in
precipitation
convection
over
the
maritime
continent,
and,
of
course,
when
you
start
doing
that,
you
affect
the
water
circulation,
you
get
a
stronger
descent
over
the
Eastern
Pacific.
T
T
This
be
it
vecting,
cooler
water
up
into
this
tropical
Pacific
is
indicated
by
these
papers
by
Zhang
and
collaborators.
The
other
one,
of
course,
is
the
Ws
the
West
feedback,
that's
part
of
the
South
Pacific
Mariano
mode.
They
could
also
get
these
cooler
sea
surface
temperatures
up
into
the
tropical
Pacific
okay.
So
that's
one
set
of
a
couple
feedbacks
that
would
be
associated
with
what
was
going
on
in
the
tropics.
T
Look
at
the
wind
stresses
which,
which
is
the
wind
stress
knowledge
of
the
colors
here
and
in
this
amps,
for
only
experiment.
Just
with
the
convective
heat
anomaly,
there,
the
negative
anomaly
you're,
seeing
you're
eating
stronger
easterlies
and
the
stronger
the
Westerly
anomalies
to
the
North
and
the
South,
and
what
that's
going
to
do
to
the
ocean,
of
course,
is
you're
going
to
get
Ekman
convergence
going
on
right
about
15,
North
and
15
South
and
Ekman
pumping
is
going
to
go
on
and
it's
not
effective
for
a
region.
G
T
Which
is,
of
course,
a
Hallmark
of
of
this
mechanism
that
we
think
is
important
for
the
sustaining
the
IPO
on
digital
time
scales.
If
you're
looking
at
the
smoke
minus
no
smoke
experiment
here
in
the
upper
left,
you
see
a
similar
kind
of
thing
happening
again.
This
is
UC
component
wind
stress
anomaly,
so
the
blue
colors
are
enhanced.
Easterlies
red
colors
are
anomalous.
T
Westerlies
get
the
skill
type
response
again,
because
we've
got
this
negative
convective
he
anomaly
over
here,
and
that
also
should
give
us
this
Ekman
pumping
going
on
at
15
North
and
15
South
and
increased
officatorial
ocean
heat
content.
So,
let's
see,
if
that's
actually
happening,
let's
go
down
here,
and
this
is
heat
content.
The
smoke
when
it's
no
smoke
experiment
the
red
colors,
are
buildups
of
heat
content
in
the
office
control,
Western
Pacific!
T
That's
going
on
as
a
result
of
this
kind
of
wind
stress
forcing
connected
these
convective
heat
anomalies
over
here
and
there's
some
evidence
of
that
going
on
in
the
observations
where
you
get
this,
especially
in
the
southern
hemisphere.
That
Kevin
was
just
talking
about
kind
of
big
buildups
of
Office
patrol
heat
content
here
and
some
in
the
officatorial
Western
Pacific
north
of
the
equator.
T
Okay,
so
that
all
went
by
kind
of
quick.
So
let's
try
to
summarize
this
kind
of
chain
of
processes
that
I
was
just
talking
about.
So
if
you
start
out
with
the
smoke
coming
off
of
Australia,
it's
evicted
across
the
Pacific,
this
is
happening
in
the
subtropics.
So
it's
a
lot
of
it's
not
rained
out
a
lot
of
it's
making
all
the
way
across
to
the
coast
of
South
America
and
subtropical
South
America
that
brightens
the
clouds
up
the
South
American
Coast
that
reduces
incoming
solar
at
the
surface.
T
That's
moved
up
into
the
equatorial
Eastern
Pacific
that
then
triggers
seedbacks
and
that
spreads
the
cool
water
right
across
the
Pacific
and
that
gets
reduced
sea
surface
temperatures
from
precip
in
the
western
equatorial
Pacific,
and
that
means
Sea,
Service
temperatures
and
precip
increase
over
the
maritime
continent
and
the
water
circulation
strengthens,
so
that
all
sets
up
these
two
sets
of
coupled
feedbacks
that
make
this
pattern
grow
and
persist
past
the
initial
smoke
impulse,
the
strength
and
Walker
circulation.
T
That
gives
even
more
northward
surface
wind
stress
in
the
Southeast
Pacific,
even
more
cool
water
gets
into
the
equatorial
Pacific
and
that
keeps
going
there's
nothing
that's
going
to
stop
that
until
something
happens
to
cut
off
this
convection
that's
going
on
and
that
Maritime
continent,
the
other
one,
is
this
negative
convective,
he
anomalies
produces
Gill
type
response
that
gives
you
the
offer:
control,
Wesley,
wind
stress,
anomalies,
get
the
Ekman
pumping
from
the
wind
stress
curled
down
into
and
putting
heat
content
into
the
officer,
Western
Pacific
near
15,
North
and
15
South,
and
that
sustains
the
negative
IPO
and
also
contributes
to
this
reduced
rate
of
global
warming
that
we've
been
seeing.
T
T
Thanks
Jerry
appreciate
the
talk:
did
you
look
at
subsurface
ocean
content?
I
know
that
you're
invoking
a
lot
of
the
heat
content
mechanisms,
but
I'm
wondering
if
you
had
cross-sections
of
smoke
minus
no
smoke
today,
yeah
we
just
looked
at
kind
of
the
average
heat
content,
especially
for
that
Zhang
magic
mechanism,
because
his
this
advection
mechanism
he's
looking.
T
T
Okay,
can
you
hear
me
all
right,
I'll
keep
it
brief.
Thank
you
for
sticking
around
Now
for
Something
Completely
Different,
we're
gonna,
look
at
Marine
heat
waves
and
cold
waves
and
specifically
investigate
how
will
the
intensity
and
duration
of
these
Marine
heat
and
cold
waves
change
in
the
future
Adam
and
a
whole
bunch
of
NOAA
people
are
collaborators
on
this
project:
okay,
whoops,
okay,
that
works
okay.
T
So
of
course
there
are
two
components
to
future
climate
change.
One
is
background.
Warming
and
one
is
changes
in
variability.
So
it's
a
no-brainer
that
as
the
oceans
warm
Marine
heat
waves
will
become
more
intense
and
warmer.
So
that's
the
what's.
T
So
we
have
used
seven
different
initial
condition:
large
ensembles
and
we're
looking
at
the
time
period
1970
to
2100
for
this
study.
We've
picked
four
cmip6
model,
large
ensembles
and
three
of
the
cement
five
generation,
and
they
have
between
30
and
100
members,
each
so
I'm
just
going
to
outline
our
procedure
that
we
used
to
identify
Marine
heat
waves
and
cold
waves
and
then
to
look
at
their
future
changes.
U
All
the
members
of
a
given
large
Ensemble
and
then
we
identify
the
seasonally
varying
thresholds.
In
this
case
we
use
the
10th
and
90th
percentile.
We've
also
used
the
fifth,
the
95th
doesn't
matter
too
much,
and
we
compute
these
thresholds
for
separately
for
these
three
31-year
periods,
just
to
get
an
idea
of,
of
course,
how
things
change
over
time.
T
We
have,
according
to
this
recipe
over
almost
well
37
100
samples
of
marine
heat
waves
and
the
same
four
Marine
cold
waves
in
each
period.
So
we
have
really
robust
sampling
with
which
to
look
at
forced
changes
in
the
variability
okay,
so
I'm
going
to
illustrate
things
first
with
the
csm2,
large
Ensemble
and
then
I'll
show
you
the
intermodel
comparison
in
a
moment.
So
this
is
the
composite
Marine
Heat
Wave
intensity
for
these
three
different
time
periods.
T
According
to
the
protocol
that
I
just
described-
and
you
can
see,
of
course,
hot
spots,
equatorial
Pacific
and
then
along
the
curricio
extension
and
along
the
Gulf
Stream
in
all
three
time
periods.
And
then
you
can
definitely
see
that
as
time
goes
on
that,
at
least
in
the
equatorial
Pacific.
There
is
some
evident
decrease
in
the
strength
of
these
composite
Marine
heat
waves,
so
to
show
that
more
clearly
we
can
look.
A
V
So
mid-century
changes
there
are
certainly
patches
of
significant
increases
in
Marine
Heat
Wave
intensity
on
the
order
of
a
couple
of
tenths
of
a
degree
C
in
the
composite
and
then
by
the
end
of
the
century.
You
can
see
for
this
model,
there's
a
decrease
in
the
intensity
in
the
tropical
Indian
Ocean,
the
equatorial
Pacific,
the
parts
of
the
Tropical
Atlantic
and
parts
of
the
Southern
Ocean.
So
it's
a
messy
pattern.
One
might
expect
in
the
future.
V
V
So
again,
remember
we're
using
monthly
mean
ssts,
so
you
have
to
we're
using
a
course
a
course
metric
here,
but
this
is
these:
are
the
results
for
the
duration
of
marine
heat
waves
in
these
three
different
time
periods,
and
here
by
I,
you
can
really
see
that
there's
a
a
reduction
in
the
duration
by
the
end
of
the
century
in
the
tropical
Eastern
Pacific
and
then
here
are
the
difference.
Maps
again
gray,
shading
denotes
insignificant
changes
and
mid-century
not
too
much
going
on,
but
by
the
end
of
the
century.
V
There's
a
lot
of
blue,
so
Marine
heat
waves
are
becoming
less
long-lived.
The
North
Atlantic
is
a
notable
exception,
so
that's
the
csn2
large
Ensemble.
Let's
now
we're
going
to
look
at
that
late
end
of
century
difference,
map
and
I'll
show
it
for
all
seven
of
the
model.
Large
ensembles.
So
that's
just
easy
way
to
do
the
comparison.
L
V
In
the
intensity
of
marine
heat
waves
due
to
internal
variability
and
the
models
are,
there
is
no
single
answer
here.
There's
a
lot
of
model
dependents
in
the
pattern
and
the
magnitude
and
the
amount
of
area
that
has
a
significant
change,
but
if
we
sort
of
step
back,
maybe
there
are
some
general
themes
that
there's
tends
to
be
in
the
exotropics
in
some
of
the
models,
an
increase
in
the
intensity
and
in
the
equatorial
Pacific
in
some
of
the
models,
a
decrease
in
the
intensity.
V
So
now
I'll
show
you
the
results
for
duration
and
again
you
saw
csm2
in
the
upper
left
before
and
now
here
are
all
the
other
models
and
I
would
say
here:
there's
more
blue
than
yellow,
so
on
average
more
the
models
in
many
regions
do
project
a
short
shortening
of
the
duration
of
marine
heat
waves.
There
are
exceptions
in
the
Arctic
where
there's
a
lot
of
model
uncertainty,
okay,
so
nothing
profound
at
all,
but
I
should
say:
people
haven't
really
isolated.
This
component
of
of
changes
in
Marine
heat
waves
before
I
think
they.
V
It
was
just
waiting
to
be
done
and
that's
what
we've
done
here,
but
the
thing
I'm
most
interested
in
is
actually
to
show
you
and
then
wrap
up
is
whether
there's
a
role
for
changes
in
enso.
V
Obviously,
you
saw
a
lot
of
things
going
on
in
the
equatorial
Pacific,
so
it
raises
the
question-
and
we
know
from
so
many
studies
that
enso
is
projected
to
change
in
the
future.
So
we
decided
to
repeat
this
whole
analysis.
Just
for
enso,
neutral
years
and
I
can
give
you
the
details.
We've
we've
explored
different
ways
to
Define
and
so
neutral
months.
V
In
what
I'm
going
to
show
you
here
is
that
we,
if
we
identify
a
marine
heat
wave
at
a
particular
grid
box
and
month
and
year,
we
will
put
it
in
the
end
so
neutral
bin,
if
the
preceding
five
six
months
and
the
concurrent
month
are
all
and
so
neutral.
According
to
the
the
leading
two
PCS
of
tropical
Pacific
ssds,
so
I
think
it
works
pretty
well,
okay,
so
this
was
the
result.
V
So
it's
pretty
much
all
gray
like
it
goes
away
and
I
just
think.
We
haven't
recognized
that
enso
changes,
whether
they
are
you
know
credible
or
not.
In
the
model
projections,
they
are
going
to
exert
a
huge
influence
globally
and
I
think
this
is
just
an
important
thing
to
keep
in
mind
that,
and
so
you
know,
is
we
really
have
to
Grapple
with
understanding
how
credible
our
enso
projections
are
for
things
like
this
consequences
for
marine
heat
waves.
So
it's
very
dramatic
for
duration.
Now,
I'll
just
show
you
the
last
result
which.
V
V
So
I
think
that
that
is
it
and
I'll
give
you
my
conclusions
so
initial
condition
large
ensembles
provide
a
robust
way
to
quantify
future
changes
in
any
variability,
but
in
this
case
Marine
heat
and
cold
waves
that
arise
from
from
changes
in
variability
and
in
general.
That
is
the
operating
word
here.
V
Changes
in
enso
account
for
almost
all
of
the
changes
in
duration
and
most
of
the
changes
in
intensity
outside
of
the
Arctic
and
North
Atlantic
that
we
noted
in
point
number
two
and
then
finally,
just
to
put
this
step
back
and
put
this
in
context.
There
are
projected
changes
in
variability,
but
they
are
small
compared
to
what
the
mean
state
background
warming
does.
V
So
we
have
maps
of
how
large
the
contribution
of
changes
in
internal
variability
is
to
the
total
change
in
Marine,
Heat,
Wave
intensity
and
duration,
and
it's
generally
under
10
percent,
except
in
the
Arctic
and
near
Antarctica.
So
if
organisms
in
the
ocean
can
adapt
slowly
to
climate
change,
then
they're
only
going
to
be
seeing
these
changes
in
variability,
but
if
they
can't
adapt
to
the
slow
background
warming,
then
that's
going
to
be
the
main,
the
main
factor
that
will
affect
them
in
the
future.
Thank
you
for
sticking
around.
Thank
you
very
much,
foreign.
V
Distribution
to
the
moments
of
the
distribution
and
so
yeah
I
think
I
think
there's
some.
It
could
probably
related
to
percentile
based
metrics
you're,
looking
at
to
more
things,
simple
things
like
SST
variants,
and
it
did
stand
out
in
some
of
your
figures
that
these
look
like
a
lot
of
like
what
I've
seen
for
like
the
SST
variants,
yeah.
So
I
guess
you
kind
of
answered
it
here,
but
if,
if
that
was
a
case
that
it's
due
to
variance
you'd
expect
the
Marine
cold
waves
to
have
kind
of
a
mirror
image.
V
Is
that
what
you
found
exactly?
And
we
didn't
want
to
presuppose
that
and
I?
Think
a
lot
of
sudden
have
just
yeah
looked
at
you
know
as
if
it
was
semantic
distribution,
we're
about
to
quantify
If.
There
really
is
any
skeenness
or
asymmetry
in
the
cold
and
warm
Tales
haven't
done
that.
Yet
so,
but
to
my
eye
they
look
very
symmetric,
except
in
the
Far
Western
equatorial
Pacific,
where
Marine
cold
waves.
You
know
they
can
extend
farther
west
just
like
and
farther
west.
V
V
So
thanks
for
that,
Rob
give
you
one
more
question
Jerry.
Yes,
sir,
it
seems
like
you're
you're,
the
duration
changes
were
the
one
of
the
kind
of
more
notable
results
and,
if
they're
connected
to
enso,
as
you
pointed
out,
so
try
and
figure
out
physically
what?
What
does
that
mean
if
the
duration
of
a
marine
Heat
Wave
gets
shorter
in
a
warmer
climate
is?
Is
it?
Is
it
related
to
the
definition
or
is
there
something
physically?
That
would
make
it
shorter?
V
Do
you
think,
well,
I,
think
in
this
case
we're
just
demonstrating
that
as
I
mean
we've
seen
it
in
the
csn2
reference
large
Ensemble
reference
paper
that
the
Enzo
cycle
itself
becomes
near
the
the
spectral
Peak
becomes
narrower
and
short,
and
somewhat
and
I
think
that's
actually
due
to
that
La
Nina
events
tend
to
not
last
more
than
one
year
in
the
future,
so
enso
is
shortening
its
life
cycle
and
I.
Think
that's
what
we're
picking
up
in
this
change
in
living,
Heatwave
creation.
V
Can
I
can
I
ask
a
question:
you
may
Kevin
yeah
I've
had
my
hand
up
for
some
time:
yeah
hi,
Clara,
I
I,
wonder
if
SST
is
the
is
the
proper
way
to
look
at
this
because
I
want
you
shouldn't
be
looking
at
something
like
specific
saturation
specific
humidity,
because
if
you
have
a
one
degree
anomaly
on
top
of,
say,
27
degree,
water
versus
25
degree
water,
these
saturation
specific
humidity
is,
you
know,
15,
more
or
20
percent
more
and
the
rainfall
and
the
consequences
in
the
atmosphere
are
much
greater.
V
V
Thank
you.
Now
we
are
very
close
to
five,
so
we
just
end
our
meeting
for
the
day
and
for
the
people
are
still
here.
We
are
going
to
go
to
understand
if
you
need
a
ride.
Just
let
us
know
wait.
I
have
a
I
have
a
car,
so
we
have
quite
a
few
others
have
come
yeah.
We
don't
actually
have
a
reservation
under
the
Sun.
So
if
you're
there
first
I
have
a
table,
I
have
space
as
well,
and
then
tomorrow
we
have
the
joint
session
with
the
Earth
system
prediction
working
group
at
the
moment.
V
I
think
the
plan
is
still
to
come
here,
but
maybe
keep
an
eye
on
your
emails
in
case
and
card
sites
have
a
late
opening.
So
thanks
I'm,
sorry
for
the
technical,
difficult
at
what
time
are
you
going
to
start
tomorrow
morning,
nine
o'clock
tomorrow,
nine
o'clock,
okay,
thanks
48,
inches,
I,
think
about
48.