►
From YouTube: Paleoclimate Working Group - 2022 CESM Workshop Day 3
Description
The 27th Annual CESM Workshop will be a virtual event. Specifically, the Workshop will begin with a full-day schedule on 13 June 2022 with presentations on the state of the CESM; by the award recipients; and two presentations from our invited speakers in the morning, followed by order 15-minute highlight and progress presentations from each of the CESM Working Groups (WG) in the afternoon.
To learn more:
https://www.cesm.ucar.edu/events/workshops/2022/
B
A
A
A
D
B
B
B
D
A
A
A
D
B
Okay,
good
morning,
everyone
I'm
jungkook
together
with
co-chairs
samantha
stephenson
and
the
ryan
thumb
will
welcome
you
to
the
2022
csm
workshop
paleoclimate
working
group
meeting.
So
today
we
have
before
we
get
started.
I
want
to
remind
people.
The
code
of
conduct
are
shown
on
the
left.
So
by
joining
this
meeting,
you
agree
to
follow
the
code
of
conduct.
B
B
E
Thanks
john
thanks,
everyone
for
participating
on,
so
we
have
a
set
of
seven
talks
during
the
first
session.
I
will
give
everybody
a
note
when
you
have
a
two
minutes
left
we,
if,
if
we
have
more
questions
after
each
of
the
talk,
please
save
them
for
the
general
discussion
or
you
can
ask
them
in
the
chat
all
right.
Our
first
talk
is
given
by
ut
merkel.
The
first
three
talks
are
going
to
be
on
our
palma
initiative,
feel
free
to
share
the
screen
and
begin.
C
C
C
Very
much
okay.
So
thank
you
very
much
for
inviting
me
to
give
an
introduction
to
the
falmouth
initiative
at
this
meeting.
Thank
you
to
the
organizers
of
this
workshop.
So
I
would
like
to
introduce
you
to
the
palmut
initiative,
which
is
a
german
climate
modeling
initiative.
My
name
is
ultimaker.
I'm
working
at
marum
the
center
for
marine
environmental
sciences
at
bremen,
university
in
germany
and
the
long
title
of
pyromode
is
from
the
last
integration
to
the
anthropocene
modeling
a
complete
glacial
cycle.
C
By
using
these
sophisticated
esm's
that
we
tested
for
the
past,
primord
is
structured
basically
into
three
different
working
groups
and
there's
one
working
group
focusing
on
the
physical
system,
one
on
the
biogeochemistry
system
and
one
mainly
focusing
on
data
synthesis.
There
are
also
cross-cutting
activities
where
the
focus
is
on
model.
Coupling
runtime
optimization,
but
there's
also
activities
dealing
with
data
management
and
with
the
comparison
between
the
model,
results
and
proxy
data,
and
we
have
one
emic
on
board
of
this
whole
initiative.
C
That's
the
climber
x
model,
which
is
mainly
developed
at
the
potsdam
institute,
dick
in
the
group
of
angry
ganopolsky
and
pyromode,
is
and
along
with
this
emic.
We
also,
of
course,
have
the
three
main
esm's
on
board,
which
are
the
esm
of
the
alfred
weekner
institute
in
brainerd
and
the
max
blanc
institute
in
hamburg
massachusetts
for
meteorology
and,
of
course,
which
is
of
major
interest
for
this
meeting.
The
cesm
1.2
model
that
we
are
mainly
using
at
tomorrow
and
braeman
in
the
next
couple
of
minutes.
C
In
my
my
talk,
I
will
give
you
some
introduction
to
my
own
work
that
I
did
as
part
of
the
working
group
one
dealing
with
some
aspects
of
marine
isotope
stage.
Three
and
after
my
talk,
you
will
hear
two
more
palmer
talks
which
are
given
by
a
takazumi
ko
hashimakamura,
whose
work
is
from
working
with
two
dealing
with
the
marine
carbon
cycle
and
afterwards
tamash
kobach
from
working
group
3,
who
presents
some
work
with
the
isotope,
enabled
cesm
1.2
model
from
the
forward
modeling
work
package.
C
So
this
brings
me
already
to
my
scientific
part,
dealing
with
climate
variability
on
centennial
to
millennial,
scales
and
cesm,
and
I
think
I
don't
have
to
make
a
long
introduction
why
we
are
interested
in
this
topic.
So
we
all
know
that
especially
marine
isotope
stage.
3
is
characterized
by
very
pronounced
millennial,
scale,
oscillations
and
very
rapid
changes
between
cold,
stable
states
and
and
warm
interstadial
states.
C
So
there
are
several
candidates
out
there,
which
may
play
an
important
role
for
determining
millennial,
scale,
vulnerability,
and
one
of
our
palmer
exercises
is
to
test
boundary
conditions
which
may
contribute
to
to
the
density
window,
which
may
make
up
the
window
and
one
of
our
mini
model.
Intercomparison
projects
within
pile
mode
is
to
explore
the
parameter,
space
and
use
different
combinations
of
greenhouse
gases
and
ice
sheets.
C
Here
we
focus
on
those
of
the
time
slices
for
21
and
38k
to
capture
mostly
the
the
upper
and
lower
ends
of
the
greenhouse
gas
range
here
on
the
left
side
and
some
high
and
lower
and
ice
sheet
volume
for
marine
isotope
stage.
Three,
and
this
whole
exercise
is
done
with
cesm
1.2
using
the
comparatively
higher
resolution
setup,
given
the
long
simulations
that
we
are
going
to
do,
and
we
run
it
at
two
degrees
in
the
atmosphere
and
one
degree
in
the
ocean
and
the
simulations
are
done
at
age,
11
in
berlin
and
gotten.
C
One
of
the
first
results,
for
example.
If
we
look
at
the
atlantic
meridional
overturning
circulation
and
the
simulations
is
that
we
get
a
reduction
of
the
amount
compared
to
pre-industrial
levels,
and
on
top
of
that
lowering
of
this
of
the
amox
strength,
we
also
get
some
centennial
scale
or
multicentennial
scale
oscillation
at
the
lgm.
C
C
C
Yet
this
nicely
fits
the
the
bipolar
system
concept,
so
the
temperature
changes
in
these
millennial
scale
oscillations
are
quite
large
in
over
greenland.
So
it's
it's
more
than
10
degrees
within
a
comparatively
short
period
of
time,
but
it
nicely
fits
to
what
has
been
reconstructed
from
the
n-grid
temperature
ice
core
reconstruction,
for
example
like
india,
where
they
find
that
that,
during
marine
isotope
stage
3
the
temperature
changes
can
actually
be
quite
large
on
the
order
of
5
to
15
degrees
celsius
within
a
relatively
short
period
of
time.
C
C
C
Matthias
has
already
started
to
look
a
bit
more
closely
into
the
multi-centennial
mode
that
we
find
that
at
the
lgm
he
looked
at.
So
that
was
what
I
mentioned
already
earlier,
that
we
find
some
some
moderate
aim
of
variability
on
these
multi-centennial
time
scales
under
on
the
lgm
conditions,
and
what
matthias
found
so
far
is
that
there
is
quite
some
nice
co-variability
as
well,
between
amok
and
and
green
and
surface
air
temperatures
and
some
anti-correlation
to
the
to
the
arctic
sea
of
this
area.
C
On
these
time
scales
and
when
digging
a
bit
more
into
depth.
Regarding
the
mechanisms
you
found
that
there
is
quite
an
important
role
of
freshwater
transports
which
which
are
operating
in
this
centennial
mode
and
which
nicely
showed
that
that's
there
is
some
some
back
and
forth
between
strong
and
weak
amok
along
with
the
freshwater
transports
in
the
north
atlantic,
and
that
the
freshening
may
lead
to
to
impact
on
on
the
convection
or
the
northward
salt.
C
C
It
seems
that
that
our
cesm
1.2
simulations
confirmed
the
existence
of
the
density
of
window
with
respect
to
the
intermediate
ice
sheet
height,
so
the
mechanisms
behind
that
still
need
to
be
explored
a
bit
more
in
detail.
But
at
least
it's
already
quite
quite
interesting
that
the
millennials
have
climate
oscillations
we
get
in.
These
simulations
so
show
a
similar
green
and
temperature
range.
As
as
the
n-grid
record,
we
see
from
these
simulations
that
that
the
american,
green
and
temperature
are
fluctuating
in
concert.
C
So
preliminary
results
show
that
for
the
high
and
the
low
aim
of
phases
in
these
millennial
scale
oscillations
the
convection
sites
shift
quite
drastically
and
along
with
the
sea
ice
distribution
and
these
simulations
and
we're
currently
looking
a
bit
more
in
detail
into
that.
So
there
are
lit
efforts
going
on
within
filemod.
E
E
F
C
At
the
moment,
I
I
only
can
hypothesize,
so
I
cannot
give
a
definite
answer
yet,
but
what
we,
what
we
definitely
will
have
to
look
at
is
also
the
deep
ocean
so
from
from
simpler
models,
there
has
been
proposed
also
a
deep
decoupling
mechanism
so
which,
which
also
involves
the
deep
ocean.
So
I
think,
on
these
long
time,
scales
it.
E
Actually,
I
have
a
question
what
else
wondering
so
you
mentioned
that
the
simulations
are
initialized
with
21
ka
state
and
when
you
prescribe
the
38ka
ice
sheet,
you
get
this
very
super
interesting
sawtooth,
behavior
of
the
amok
and
the
greenland
temperature
I
was
wondering.
Does
that
have
also
have
something
to
do
with
how
you
initialize
the
run?
Let's
say
if
you
initialize,
with
38
ka
equilibrium,
run
and
a
prescribed
38k
sheet.
Would
you
also
get
the
same
oscillation
condition.
C
That's
an
interesting
question,
which
would
require
quite
some
additional
computational
efforts
to
to
really
test
that
at
this
point,
all
these
simulations
have
been
have
been
started
from
exactly
the
same
initial
state.
So
in
that
respect
we
we
cannot
say
anything
about
whether
the
initialization
would
make
a
difference.
C
So
it's
just
that
that
we
can
compare
these
this
set
of
experiments
and
know
that
they
all
have
been
initialized
from
the
same
state.
Thank
you
thank
you,
but
you're
definitely
right.
It
would
be
very
interesting
and
worth
to
look
more
into
that
to
to
see
how
how
different
initial
states
might
might
have
an
impact,
and
if
I
remember
correctly,
then
that
matthias
looked
more
closely
into
into
different
initial
states
for
the
lgm
and
there
it
it
may
actually
have
an
impact
on
what
what
oscillatory
behavior
you
actually
get.
E
G
E
H
So
I'd
like
to
I'd
like
to
say
a
big
thank
you
for
invitation.
I
really
appreciate
this
nice
opportunity
and
would
I
quickly,
as
would
I
quickly
introduced?
I
belong
to
the
working
group,
two
of
the
primal
project,
which
is
taking
in
charge
of
the
biogeochemical
part
of
the
project.
So
in
the
framework
of
the
project
recently
we
made
some
time
slice
simulations
aiming
at
miss
three
and
today
I
will
introduce
some
some
results
from
the
recent
work.
H
H
H
And
they
found
so
another
characteristic
of
the
two
kinds
of
two
different
types
of
stereos
and
in
the
long
one
with
hand
heavens.
So
in
the
number
one,
there
was
a
remarkable
increase
of
field
level
in
the
atmosphere
like
this,
but
on
the
other
hand,
in
the
short
studios,
there
were
only
minor
change
of
co2
level
like
this,
except
for
a
very
long
time,
scale
decreasing
trend.
H
But
in
this
studies,
we'd
like
to
suggest
actually,
there
is
another
difference
between
the
two
long,
these
pink
stereos
between
this,
this
stereo
and
the
distance,
this
stereo,
and
that
is
the
the
flux
of
aryan
dust
into
the
antarctic
region.
This
is
the
time
series
of
a
proxy
of
aerial
dust
flux
into
the
antarctic
regions
and,
during
the
past.
H
H
E
H
We
can
we
can
categorize
we'd
like
to
suggest
there
are
three
types
of
stereos
during
a
mystery,
depending
on
the
the
heindic
event
or
not,
and
depending
on
the
dust
input,
for
example,
in
this
type
1.
So
it
is
accompanied
by
the
hanukkah
event
and
it
has
low
dust
input
from
the
atmosphere
and
this
motivated
us
to
make
nine
simulations
in
total
and
three
simulations
for
each
time
period,
including
30,
aka,
29k
and
21ka,
and
each
of
each
of
the
three
time
slices.
H
H
H
We
gave
a
different
amount
of
additional
fresh
water,
forcing
onto
the
north,
the
high
latitude
of
north
atlantic
to
to
manipulate
the
ocean
saturation
global
ocean
saturation
and
for
the
dust
forcing
we
used.
The
modern
flux
for
the
low
dash
input
and
we
use
the
lgm
dust
input
for
the
high
dust
condition
to
mimic
them.
H
So
in
this
vertical
axis
you
see
the
different
three
time
slices
and
this
or
in
this
horizontal
justin
you,
you
see
the
the
different
additional
fresh
water
forcing
onto
the
north
atlantic
so
in
the
baseline
experiments.
We
we've
got
this
quite
strong
and
deep
amerk
structure
like
this
and
as
you
expected.
H
H
H
Okay,
yeah
in
the
next
exercise,
I
will
discuss
the
processes
in
the
simulations
first
one.
I
I
show
the
effect
of
dust
input.
The
dust
input
is
the
in,
in
short,
the
iron
fertilization
effect
to
to
strengthen
the
biological
pump
to
draw
down
the
co2
level.
So
for
that
comparison
we
I
picked
up
the
two
simulations
one
from
3080
and
one
from
29
ka,
which
have
very
similar
aim
of
structure
like
this,
but
a
different
co2
level
and
with
high
dust
input
in
antarctic
region
in
this
experiment.
H
Greater
consumption
of
macronutrients
at
the
surface
and
shown
at
this
stage
here
the
next
one
is,
I,
I
discussed
the
the
effects
of
the
freshwater
input.
The
freshwater
input
has
two
effects,
so
one
effected
change,
sst
distribution
and
the
other
one
is
a
more
stagnant,
deep
water
to
for
more
efficient
carbon
storage,
but
in
total
and
the
effects
of
ssd
change
is
slightly
higher
than
this
more
efficient
storage
of
carbon
by
the
stagnant
water.
H
And
this
is
a
kind
of
summary
diagram
showing
that
the
temperature
showing
the
relationship
between
the
temperature
change
in
the
untucked
region
in
the
one
dance
guard,
er
sugar
event
and
the
co2
change
water
level
change.
Also
in
the
same
dashboard,
the
same
event
and
the
circles
show
the
result
of
our
study.
H
H
And
unfortunately,
we
not
find
a
conclusive
differences
between
type
2
and
v,
at
least
in
the
framework
of
the
study
and
possibly
the
first
water.
Perturbation
on
of
a
different
magnitude
might
be
explain
the
difference,
but
the
relationship
between
the
lengths
of
time,
length
of
stereos
and
the
amount
of
freshwater
input
should
be
further
examined
to
it.
H
E
I
All
right
so
welcome
to
the
third
and
then
last
of
the
related
talk.
The
following,
slides
were
presented
in
a
project
meeting
about
a
month
ago
by
my
colleague,
martin
verna.
Now,
unfortunately,
he
is
busy
today
and
cannot
make
it,
but
instead
I'm
very
happy
to
be
here
and
present
some
of
our
results
from
from
people
listed
here
from
the
arthur
degeneres
institute
and
at
marum
the
university
of
bremen.
I
Now
what
I
will
present
here
today,
luckily
uttar
has
already
showed
this
structure
of
pamus.
I
don't
need
to
go
too
deep
into
it.
I'm
gonna
present
results
of
the
working
group,
three
or
specifically
3.3
over
here.
The
focus
of
this
group
is
an
integrative
forward
proxy
modeling,
with
the
aim
to
bridge
the
gap
between
data
assembled
and
models
used
within
paramount.
I
A
key
question
in
palmer
is
how
much
climate
variability
depends
on
the
mean
climate
state
and
also
whether
our
models
can
simulate
it
correctly.
Of
course,
capturing
the
mean
climate
is
very
important
for
the
past
and
the
future,
but
so
is
the
range
of
climate
variability
and
also
as
an
outlook.
If
the
models
can
do
that
in
the
past,
then
what
can
we
expect
from
the
future,
how
it
looks
like
and
and
what?
I
What
can
we
learn
from
that
for
studying
climate
variability,
we're
using
the
data
compilation,
product
palmut,
130k
assembled
by
our
colleagues,
lucas
junckers
at
all,
within
the
power
mode?
Here
you
can
see
a
map
of
data
coverage
for
for
oxygen,
18
records
from
marine
sediments
within
this
product
and
we're
also
using
the
first
long
transient
simulation
within
palmore
low
resolution,
mpism
simulation
with
prescribed
ice
sheets
from
our
colleagues
in
hamburg.
I
If
you
look
at
the
spectrum
here,
this
is
what
a
true
climate
spectrum
would
be.
This
is
power
density
versus
frequency,
and
what
we
can
actually
observe
in
proxies
is
the
blue
line
is
different,
because
so
that's
the
that's.
The
observed
proxy
stratum
and
it's
different
because,
of
course,
there's
noise,
noise
from
measurement
error
and
data
aliasing.
I
Our
colleagues,
andrew
dorman
and
tom
lepla
came
up
with
an
estimated
proxy
spectrum
that
eliminates
the
effect
of
this
noise
and
they
developed
a
new
autocorrelation
function,
methods
for
gapi
time
series,
and
with
this,
for
the
first
time,
we
can
actually
compare
the
true
proxy
spectrum
to
the
spectrum
of
the
simulated
climate
variability
and
results.
Some
results
of
these
efforts
are
shown
here
for
the
that
hit.
You
know
of
calcite
in
the
ocean
and
andrew
and
tom
found
that
the
climate
variability
is
quite
state
dependent.
I
This
is
a
really
valuable
study,
but
it
still
has
one
caveat,
namely
that
this
simulation
with
the
mpsm
has
no
isotopes
in
it.
So
we
are
just
modeling.
The
data
180
from
temperature
and
sanity
results
based
on
the
transfer
function
and,
of
course
the
question
is
how
stable
is
that?
Can
we
apply,
apply
the
same
transfer
function
over
time
and
or
it
changes,
and
if
yes,
how?
How
does
it
change?
I
We
are
trying
to
find
answers
for
this
with
the
use
of
isotope
enabled
simulations.
These
are
very
valuable
because
they
can
directly
simulate
the
isotopic
composition
of
water
in
the
entire
hydrological
cycle,
and
it's
useful
because
this
way
we
can
use
them
to
validate
and
calibrate
the
transfer
function
plus.
Of
course,
they
can
also
provide
a
better
data
for
model
proxy
compression.
I
So,
looking
at
some
of
these
and
some
of
the
the
time
size
simulations,
we
can
see
here
that,
for
example,
for
the
lgm,
the
lgm
minus
vic
surface
temperature
anomalies
see
icsm.
Results
are
in
the
top
of
esm
in
the
bottom.
What
we
can
see
here
is
that
icsm
simulates
simulates
a
much
larger
cooling
than
of
esm.
I
That
gets
the
global
mean
ssd
normally
remarkably.
Well,
if
you
compare
it
to
the
gloom
up,
reconstruction
recently
produced
within
powered.
Looking
at
regional
differences
and
comparing
them
to
the
temperature
reconstructions,
these
are
shown
as
these
colored
dots
here.
Icsm
does
a
does
a
good
job
in
simulating
the
north
atlantic
cooling,
but
shows
too
much
cooling
at
low
and
mid
latitudes.
That's
where
mostly
this,
this
larger
anomaly
comes
from,
and
also
the
southern
ocean
is
too
cold
in
icsmd
and
not
at
least
too
large
there
something
that's
not
uncommon
for
a
couple
model
simulations.
I
On
the
other
hand,
lvsm
gets
the
low
and
mid
latitudes
much
better,
but
shows
too
little
north
atlantic
cooling.
Overall,
looking
at
the
reaction
slopes
between
reconstructed
and
simulated
ssd
anomalies,
the
models
well
perform
modest.
The
best
suggesting
many
reasons
original
characteristics
are
not
particularly
well
captured.
I
That
is
why
it
is
particularly
interesting
that
if
you
compare
the
simulated
delta
oxygen
18
in
calcite
to
the
reconstructions,
with
the
get,
we
get
a
much
better
agreement
with
the
data.
So,
for
some
reason,
the
model
is
able
to
produce
a
much
better
fit
than
for
ssd
for
both
models,
and
so
this,
this
kind
of
suggests
that
the
conversion
between
oxygen,
18
and
ssd
for
the
data
products
or
model
analysis
that
we
use
might
need
a
little
bit
more
more
attention
again.
I
Also
in
terms
of
the
results,
we
see
a
bias
in
a
sudden
ocean,
a
two
large
enrichment
in
this
case,
actually
for
both
models
more
so
in
icsm,
and
also
the
models
do
not
agree
on
the
on
the
that
hit.
You
know
in
the
arctic
surface
water.
This
is
unfortunately,
a
bit
difficult
to
evaluate
due
to
the
lack
of
data
points
to
compare
there.
I
I
Well,
these
are
these
brownish
triangles,
so
the
fit
is
quite
bad
for
those,
but
this
is
also
a
common
long-term
problem
for
isotope
models
for
the
arctic
ice
cores
they're
in
a
blue.
The
ibsm
does
a
much
better
job
a
better
fit,
but
it
actually
shows
a
too
small
depletion
over
and
around
greenland.
That's
the
purple
squares
here
than
icsm
that
reproduces
data
in
the
high
scores
significantly
better
with
icsm.
We
also
looked
at
misc
at
38ka.
I
I
You
know
in
precipitation
for
model
viewpoints
over
greenland,
that's
again
in
purple
and
antarctica
and
blue
for
different
climate
states,
and
what
we
have
found
is
that
our
icsm
simulations
suggest
different
slopes
for
the
pi
lgm
and
mis3,
especially
over
greenland.
So
comparing
these
numbers
here
in
purple.
I
However,
if
we
make
use
of
the
fact
that
we
have
two
isotope-enabled
models
in
talmud
and
do
the
same
analysis
for
other
esm,
we
get
that
the
models
do
not
agree.
The
relationship
is
much
more
stable
for
obesm
and
its
slopes
are
actually
also
closer
to
the
real
ones
inferred
from
ice
core
data
for
the
pre-industrial,
also
for
greenland
and
antarctica
too.
I
Finally,
if
we
look
as
a
little
outlook,
I'd
like
to
give
a
bit
of
outlook
also
on
the
on
the
warm
climate
simulations,
these
are
with
avi
esm
for
the
late
and
early
holocene
and
on
the
right
side
for
the
last
interglacial
at
127
k.
These
are
normal
leads
of
that
18
precipitation
on
the
top
and
ocean
water
surface
in
the
bottom.
I
There's
still
some
questions
to
be
figured
out
like
the
precise
application
of
isotopic
melt,
water,
fluxes
and
also,
at
the
same
time,
at
modern
we
are
preparing.
The
transient
runs
for
icsm,
okay,
so
a
few
take
home
messages
in
the
end,
based
on
our
climate
variability
studies,
the
glacial
climate
was
much
more
variable
than
in
the
holocene
and
our
transient
simulations
show
realistic
viability
for
the
glacial
transition
periods,
but
they
appear
to
be
too
stable
in
the
holocene,
especially
in
the
late
holocene.
I
The
simulations
show
a
very
good
agreement
of
similarity:
surface
water
does
180
anomaly
for
the
rgm
and
a
fair
to
go
to
very
good
agreement
or
values
in
a
precipitation
for
the
lgm
and
the
late
hall
scene
and
the
last
interglacial,
our
icsm
results
suggest
the
different
conversion
factors
for
isotopes
reconstructions
in
antarctica
and
greenland
for
different
climate
states.
At
the
same
time,
the
obvi
esm
results
do
not,
of
course,
the
question
here
is
why
so
that's
for
future
work
again.
I
Finally,
our
transient
first
transient
isotope
simulations
are
being
either
in
progress
or
in
planning
now,
and
they
will
follow
soon
so
expect
some
exciting
results
about
that
too
in
the
near
future.
Yeah
with
this,
I
would
like
to
thank
you
for
your
attention
and
I
hope
I
left
some
time
for
a
few
questions.
E
I
E
F
Hi
good
morning,
everyone
do
you.
Do
you
see
my
slides?
Yes,
yes,
okay!
So
so
I'm
fanny
kellerman,
I'm
from
the
gyote
university
in
frankfurt
and
my
topic
is
the
partitioning
of
meridianal
heat
transport
in
the
early
using
climatic
optimum
in
deep
model
simulations
and
I'm
working
in
a
project
called
fever
with
my
colleagues
and
this
project
is
investigating
past
worm
periods,
which
could
be
analogs
for
a
high
co2
future.
F
It
is
also
known
that
the
meridional
heat
transport
is
mainly
determined
by
the
earth's
sun
geometry
and
the
mean
albedo,
and
it
is
very
quite
insensitive
to
the
structure
of
the
and
the
dynamics
of
the
atmosphere
ocean
system.
Nevertheless,
if
we
are
partitioning
the
meridian
and
heat
transport
into
different
processes,
this
helps
us
understand
better
that
climate.
What
is
happening
there
with
the
different
processes?
F
F
And
if
you
look
at
the
annual
meridional
heat
transport
here,
you
can
see
the
three
models
we
just
choose
now
and
the
black
lines
representing
the
total
meridian
and
heat
transport,
the
red
ones,
the
atmosphere
and
then
the
blue.
The
ocean,
the
solid
lines
are
the
control
simulations
and
then
the
dashed
lines
are
the
different
polar
simulations.
F
And
if
the
value
is
positive,
then
it
means
northward
transport.
If
it's
negative,
it
means
southward
transferred,
and
what
you
can
see
is
that
indeed,
the
original
heat
transport
is
similar
in
the
different
models
and
stable
among
the
different
co2
values
and
the
contribution
of
the
par
to
the
polar
amplification
from
the
models
is
not
coherent.
F
But
if
you
look
at
the
moist
and
the
dry
part
of
the
meridian
overturning
circulation,
which
is
the
light
blue
and
the
yellow
yellow
lines,
you
can
see
that
the
circulation
in
the
tropics
between
30
degree
and
0
is
indeed
increased
in
the
in
the
higher
co2
simulations,
which
is
the
dashed
lines.
But
in
the
end
the
whole
meridional
overturning
circulation.
Transport
is
not
changing.
So
much
in
the
subtropics,
which
is
interesting
is
that
we
found
an
increase
in
polar
moist
transport
by
stationary
and
this,
which
is
means
the
monsoon
systems.
F
So
in
this
figure
you
can
see
the
change
in
transport
via
stationary
at
these
between,
in
this
time,
the
csm9
times
co2
and
one
times
co2
simulation.
So
these
peaks
here
at
the
do,
you
see
my
cursor
at
the
30
degrees
and
if
we
look
into
the
non-seed
to
effect.
So
if
we
compare
the
control
simulation
to
the
one
time
co2
poly
simulation.
So
this
means
that
we
are
comparing
the
effect
of
fellow
geography,
the
changed
vegetation,
aerosols
and
that
there
is
no
con
continental
ice
sheets
in
the
polar
simulation.
F
F
And
what
I
found
very
interesting
is
that
at
the
mid-latitudes
we
find
an
increase
in
polar
transport
by
transient
at
these
and
a
decrease
by
a
stationary
at
this.
So
this
means
that
there
is
probably
more
cyclones
and
less
anticyclones
over
the
northern
hemisphere,
mainly
during
winter,
and
to
look
into
more
that
what
causes
this
I
plotted,
also
the
mean
sea
level
pressure
values
from
from
csm,
and
what
I've
found
is
that
there's
more
semi-permanent
low
pressure
systems
due
to
parallel
geography
in
the
paleo
setup.
F
So
we
have
the
icelandic
law,
which
is
in
current
climate,
the
the
alternating
law
so
to
say,
and
there's
also
one
in
the
gulf
of
alaska
and
also
the
eurasian
low
or
siberian
silo.
So
that's
probably
the
reason
for
for
the
increased
number
of
cyclones
or
transport
by
a
traditional,
no
transient
at
these.
I
don't
know
yet,
if
there's
more
cyclones,
but
that's
what
I
want
to
find
out.
F
So
the
conclusions
is
that
the
role
of
transporting
for
amplification
in
the
equal
signations
is
not
clear,
but
the
partitioning
of
marijuana
heat
transport
shows
the
changes
in
the
different
processes
and
the
compensating
values
also
and
what
we
found
that
high
co2
values
affect
the
circulation
of
the
heavy
cell.
It
becomes
more
intense
and
also
the
monsoon
systems,
and
that
in
the
palau
setup
set
up
is
changing
them:
energy
transported
by
military,
cyclones
and
blockings,
mainly
in
the
winthrop
period.
D
E
F
E
J
Okay.
So
the
broad
motivation
for
this
work
is
the
fact
that
several
generations
of
climate
models
have
predicted
drier
central
american
climate
by
the
end
of
the
century,
as
we
can
see
on
the
left,
precipitation
reductions
by
as
much
as
40
percent
perhaps-
and
this
is
likely
due
to
a
southward
shift
in
the
pacific
to
atlantic
itcz,
which
is
pulling
precipitation
away
from
central
america
and
further
south
towards
the
tropical
pacific.
J
So
there's
still
quite
a
bit
of
uncertainty
in
these
projections,
so
we
want
to
better
understand
hydroclimate
in
central
america
and
a
great
way
to
do.
That
is
to
look
towards
the
past
and
specifically
here
at
the
last
glacial
maximum,
which
is
a
good
time
period
for
looking
at
central
america,
because
it
had
a
similarly
southward
shift
in
the
pacific
to
atlantic
itcz.
J
But
it's
very
important
to
note
that,
rather
than
being
forced
by
greenhouse
gases,
it
was
forced
by
northern
hemisphere
ice
cover.
So
this
means
that
the
lgm
is
a
good
time
period
to
try
to
separate
some
of
these
large-scale
climate
forcings
and
better
understand
some
of
the
ways
that
hydroclimate
in
central
america
responds
to
these
changes.
J
Now,
looking
at
paleoclimate
in
central
america,
it's
a
very
under-researched
region,
so
there
are
very
few
proxy
records
that
exist,
telling
us
about
climate
in
central
america,
and
even
so,
some
of
the
proxy
records
that
exist
tend
to
contradict
each
other.
So
we
don't
have
a
great
grasp
on
lgm
hydroclimate
in
central
america.
So
we
can't
so
far
answer
some
very
simple
questions
such
as
was
central
america
wetter
or
drier
during
the
lgm.
J
So
in
these
simulations
we
have
water
tagging
capabilities,
and
here
I
am
showing
the
terrestrial
and
marine
water
tag,
regions
that
we
have
to
look
at
rainfall
and
isotopic
composition,
and
for
those
of
you
who
maybe
haven't
looked
at
water
tagging,
results
aren't
as
familiar
with
it.
Here's
an
example
of
what
we
could
look
at
from
tag
15
here
in
the
north
atlantic.
J
So
essentially,
what
this
plot
is
showing
is
for
the
pre-industrial.
Any
moisture
that
is
evaporated
from
the
north
atlantic
surface
is
then
tracked
through
the
model,
and
then
we
can
quantify
where
that
moisture
ends
up
raining
out.
So
we
can
do
something
like
take
the
yucatan
peninsula
region
of
central
america
here
and
quantify
the
amount
of
rainfall
that
was
actually
derived
from
the
north
atlantic.
J
So
that's
exactly
what
we're
doing
in
these
simulations
and
we're
taking
this
yucatan
peninsula
that
tiny
little
red
box.
If
you
can
see
on
the
maps-
and
I
just
want
to
point
out
for
the
remainder
of
this
presentation-
I'm
going
to
narrow
down
to
the
regions
of
greatest
importance
for
the
yucatan,
so
we
don't
have
so
many
distracting
regions
we're
just
looking
at
15
different
water
tags.
J
So
looking
in
our
pre-industrial
simulation,
we
can
see
that
rainfall
in
central
america
is
largely
following
the
movement
of
the
idcz
and
is
high
during
the
boreal
summer
and
looking
at
the
vectors
here
are
integrated,
vapor
transport.
We
can
see
that
the
caribbean
low-level
jet,
which
is
this
easterly
flow
throughout
the
caribbean,
is
primarily
delivering
moisture
from
the
caribbean
and
atlantic
to
the
east.
J
E
J
J
J
This
way
how
this
change
in
rainfall
actually
manifests
throughout
the
climate
system,
so
a
great
way
to
do
this
analysis
is
to
look
at
a
water
tagging
analysis,
and
here
specifically,
I
am
looking
at
changes
in
rainfall
for
the
sake
of
time
for
this
presentation,
but
I
want
to
mention
that
I'm
also
looking
at
the
changes
in
isotopic
composition.
So
if
anyone
is
interested
in
these
types
of
results,
I'd
be
happy
to
discuss
those
after
the
presentation.
J
So,
first
of
all,
looking
at
the
tropical
atlantic
and
the
caribbean,
we
can
see
the
change
in
the
percentage
of
rainfall.
So
this
is
essentially
telling
us
how
important
this
region
is
to
contributing
rainfall
to
central
america,
and
this
is
the
percent
change
due
to
the
addition
of
northern
hemisphere
ice
sheets.
J
And
here
on
the
left,
I
am
showing
the
integrated
vapor
transport
and
the
evaporative
flux.
So
I'm
still
digging
into
these
results,
and
I
think
they're
very
interesting.
Perhaps
thinking
about
the
southward
itzz
shift
that
ice
sheets
force
here
we
have
some
circulation
changes
downstream
of
this
or
or
that
are
related.
So
I'm
still
digging
into
this.
But
I
think
this
is
a
very
interesting
finding
so
far.
J
We
can
also
see
that
the
north
american
land
surface
becomes
a
much
greater
moisture
source,
a
more
important,
moisture
source
to
central
america
during
when
we
have
lgm
ice
sheets
and
even
during
the
winter
we
have
a
greater
increase
in
this
rainfall
per
percentage.
So
perhaps
ice
sheets
are
changing
types
of
circulation
that
lead
to.
J
So
you
know,
I
think,
it's
very
interesting
to
see
that
perhaps
there
are
some
circulation
changes
in
the
north
atlantic
subtropical
high
that
are
leading
to
more
evaporated,
moisture
and
transport
into
central
america.
At
the
same
time,
in
the
colder
climate
with
northern
hemisphere
ice
sheets,
we
perhaps
see
a
shift
towards
more
proximate
regions
of
of
moisture
sources.
J
These
records
are
currently
being
produced
by
organic
geochemists
from
washu
in
saint
louis
notre
dame
and
pitt
as
a
part
of
this
collaborative
research
grant.
So
you
know
this.
This
is
hopefully
going
to
lead
to
some
great
model
data
intercomparison
in
the
future,
and
you
know
potentially
we're
going
to
look
at
not
just
the
lgm
but
different
glacial
interglacial
cycles
throughout
these
records.
E
E
F
A
A
J
J
E
L
B
L
E
M
M
Great
okay,
so
let
me
get
started
so
I'm
presenting
a
eustatic
sea
level
rise
compatible
model
for
the
younger
dryas.
This
work
has
been
done
with
deepak
chandan
and
dick
pelche
at
the
university
of
toronto.
I
presented
some
earlier
work
of
this
at
the
last
paleoclimate
working
group
meeting,
so
I'm
gonna
move
somewhat
quickly
through
the
introduction
so
that
I
can
get
to
the
kind
of
new
results
so
introduction.
M
We
all,
I
think,
know
what
the
younger
dryas
is
in
our
model,
we're
we're
kind
of
pegging
it
at
12.8
thousand
years
ago
to
11.5,
which
is
pretty
standard.
It's
known
to
be
caused
by
release
of
fresh
water
from
pro-glacial
lakes,
on
the
edge
of
the
laurentide
into
the
arctic
ocean.
Through
the
mackenzie
river
outlet
and
the
outflow
fresh
water
seems
to
have
subdued
or
shut
down.
Depending
on
the
terminology,
you
want
to
use
the
the
amoc
for
about
a
millennium.
M
M
So
there
have
been
lots
of
prior
modeling
attempts
to
try
to
get
a
climate
response
from
a
gcm
by
injecting
fresh
water
into
the
ocean.
A
recent
paper
by
love
at
all
summarized
a
number
of
these
prior
modeling
attempts
and
identified
a
few
key
areas
of
improvement
or
areas
where
the
models
weren't
quite
accurate.
These
were
forcing
strength
forcing
location
and
boundary
conditions.
M
Many
models
apply
extremely
strong,
forcing
to
the
north
atlantic
with
lgm
boundary
conditions.
We're
talking
something
like
you
know:
five
sv
for
a
hundred
years
straight
ups,
I
mean
so
not
only
is
this
forcing
too
strong,
but
because
it's
being
applied
for
a
long
time,
the
total
water
volume
is
incompatible
with
sea
level,
result
record.
M
So
an
example
of
this
is,
if
you
look
at
a
recent
modeling
by
hey
at
all
in
2021,
they
apply
quite
a
large
amount
of
forcing
over
the
entirety
of
the
younger
dryas,
which
leads
to
an
unrealistically
large
increase
in
sealab
with
the
boundary
conditions.
Part
of
the
issue
is
that
if
you
use
lgm
boundary
conditions,
they're
quite
different
from
those
that
existed
at
the
younger
driest
time,
there's
some
debate
as
to
whether
or
not
the
bearing
straight
was
open
or
closed
at
this
time.
M
And
of
course,
just
the
the
the
this
ice
sheets
are
going
to
be
in
different
configurations.
These
sea
levels
are
significantly
higher
at
younger
driest
time
and
ice
coverage
is
much
more
limited.
So
in
terms
of
some
of
the
proxies
that
we're
trying
to
use
here
kygren,
it
all
found
two
irony
layers
in
the
beaufort
sea,
one
at
fourteen
and
a
half
thousand
years
ago,
and
one
at
younger
driest
time.
M
M
So
in
teresa
and
peltier
they
argued
that
the
peak
flow
rate
couldn't
be
much
more
than
about
0.2
squared
drops,
but
there
that's
there's
some
debate
about
that
and
in
terms
of
what
we're
using
to
constrain
of
what
the
proxies
that
we're
using
to
constrain.
There's
the
barbados
record,
which
constrains
rsl
rise
to
be
somewhere
in
the
vicinity
of
five
meters
during
this
time.
M
That's
not
a
very
tight
constraint
later
on
we'll
talk
a
bit
about
some
of
the
constraints
imposed
by
lake
volume
analysis
and
just
as
a
caveat
that
this
increase
includes
all
rsl
contributions
across
the
world.
So
if
there's
melting
going
on
in
antarctica
or
greenland
or
fenescandia,
then
that's
going
to
be
picked
up
by
the
barbados
sea
level
record
and
can't
be
attributed
directly
to
the
younger
dryas.
M
So
some
quick
numerical
facts
we
use
csm1
constant
background,
dipigmo,
diffusivity,
tidal,
mixing
and
overtimes
are
turned
off
for
orbital
parameters,
trace
gases
and
oreography
the
thermometry.
That's
all
set
to
13k
the
one
caveat
from
perfect
boundary
conditions
for
younger
drives
is
that
the
land
sea
mask
is
left
at
lgm.
M
We
decided
to
keep
the
bearing
straight
closed
and
we
integrated
for
2
300
years
in
until
the
amok,
stabilized
and
then
started
doing
our
foresight
runs
so
in
terms
of
forcing
runs.
We
have
two
main
types:
low
and
slow,
where
we
force
at
a
very
low
amount
for
the
entirety
of
the
younger
dress
for
about
a
thousand
years
and
then
hard
and
fast.
M
So
we
started
to
spend
more
time
looking
at
these
types
of
simulations,
so
here
we
have
all
the
different
simulated
collapse
scenarios,
as
you
can
see,
we've
run
quite
a
number
of
them
I'll
get
into
some
more
details
about
specific
ones,
but
the
what
you
see
here
is
runs
that
differ
by
forcing
amount
and
duration
of
forcing
one
of
the
key
facts
here.
Is
that
once
a
run
collapses,
it
recovers
exactly
the
same?
M
So
here
are
three
extreme
runs.
There's
our
quickest
collapse,
which
was
hosing
for
one
year
at
quite
a
high
value
of
six
sv,
our
lowest,
forcing
collapse,
which
is
the
lowest
total
volume
where
we
forced
for
five
years
at
one
sv
and
our
borderline
collapse,
which
was
the
the
this
the
weakest,
forcing
that
we
could
find
a
collapse
for
which
was
one
1.25
sv
for
115
years.
M
These
are
all
summarized
here.
These
are
all
of
our
various
forcing
runs
that
collapsed.
There's
some
more
that
collapse,
but
they're,
very
similar,
and
so
we're
comparing
this
against
an
estimate
of
glacial
lake
volume
from
around
the
8.2
kilo
year
event.
It's
it's
somewhat
difficult
to
find
good
estimates
of
exactly
how
much
the
lake
levels
would
have
dropped
in
lake
agassi
lakes,
mcconnell
and
other
glacial
lakes.
M
So
it
looks
like
we
have
a
few
different
types
of
runs
here
at
various,
forcing
levels
and
durations
that
could
produce
the
collapse.
We're
looking
for
so
over
here.
We
see
a
bunch
of
runs
that
didn't
quite
collapse,
so,
as
we
were,
testing
runs
we'd,
get
to
a
point
in
for
a
given,
forcing
where
our
forcing
duration
was
too
short
to
create
a
collapse,
and
we
get
an
immediate
recovery.
M
So
once
we
noticed
that
this
was
happening,
our
next
step
was
to
try
and
fill
out
a
phase
diagram
of
duration
versus
forcing
strength
to
see
what
that
might
look
like.
So
we
have
a
figure
here
where
the
red
dots
are
collapsed
in
blue
or
no
collapse,
and
you
can
see
that
there's
this
kind
of
this
relationship
where,
as
you
go
to
shorter
duration,
you
need
more
volume,
and
this
is
a
log
scale.
So
we
can
show
it
on
a
on
a
nice
graph
here
without
being
blown
out.
M
But
what
might
be
more
instructive
is
to
look
at
the
total
volume
versus
duration,
where
we
see
an
almost
linear
relationship
as
we
force
for
shorter
amounts
of
time.
We
need
less
total
volume,
which
is
interesting,
because
there
is
some
evidence
that
the
younger
dress
was
a
very
quick
forcing
event,
and
so
the
quicker
we
make
it
the
less
volume
we
need.
M
So
in
terms
of
what
the
recovery
looks
like
when
it
recovers
it
recovers
through
the
opening
of
a
polyneia
in
the
irma
c.
This
happens
in
february
of
it
depends
on
the
model
year
of
the
model
we're
running.
This
is
a
very
similar
mechanism
to
what's
discussed
in
peltier
2020
with
respect
to
dan's
grid,
oscar
oscillations
and
it's
very
sensitive
to
the
forcing
strength
right.
If
we're
just
a
little
bit
too
slow,
we
don't
get
this
type
of
opening.
M
So
in
terms
of
comparison,
we
wanted
to
compare
this
to
the
8.2
kilo-year
event.
This
is
where
we,
instead
of
forcing
in
the
arctic
we
forced
in
the
north
atlantic.
We
did
three
different
runs
here:
one
with
a
six
sv
run
for
a
year,
a
0.6
sv
run
for
10
years
and
then
a
point,
a
0.2,
for
I
believe
it
was
100
years
there
and,
as
you
can
see
in
all
of
these
cases,
the
minute
the
force
is
turned
off.
M
We
get
an
immediate
recovery,
so
what
this
shows
us
is
that
the
boundary
conditions,
because
for
the
8.2
kilo
year
event,
we
used
appropriate
boundary
conditions
that
whether
or
not
the
amox
stays
turned
off
is
extremely
sensitive
to
those
boundary
conditions.
So
that's
very
interesting
not
showing
here,
but
we
actually
did
try
forcing
the
north
atlantic
with
the
boundary
conditions
of
the
of
yd,
and
we
did
find
that
that
state
collapsed.
So
this
is
this
is
not
location-based.
M
M
What
we're
going
with
this
is
we're
working
on
writing
up
the
results
right
now,
because
we've
done
a
very
large
number
of
sensitivity,
tests
to
see
what
that
curve
looks
like,
but
yeah.
So
that's
what
we're
where
we're
at
right
now,
and
I
guess
as
a
kind
of
conclusion
if
anyone
in
the
audience
knows
of
some
good
estimates
of
volume
of
water
involved
in
the
younger
dryas.
That
would
be
interesting
for
us
to
see
or
if
anyone
has
any
specific
information
on
based
on
kind
of
geologic
features.
M
M
Yeah,
so
I'll
go
back
to
the
palinya
feature,
so
this
is
what
the
sea
ice
looks
like.
You
can
see
it's
up
to
100
right
down
to
like
newfoundland
in
the
in
the
west
and
almost
you
know
past
what
is
that
the
bay
of
biscay
on
the
right
so
yeah?
We
have
quite
extensive
sea
ice
once
we've
applied
the
fresh
water,
okay.
H
M
M
E
E
E
Before
I
jump
to
on
this
topic,
I
would
like
to
give
you
an
overview
of
the
suite
of
play,
application
comparison
project
to
experiments
that
we
have,
and
so
we
completed
and
published
the
applicant
model
in
comparison
project,
two
climbing
two
baseline
experiments.
We
used
the
one
degree
nominal
resolution
for
all
model
components
and
the
simulation
were
around
four
three
generations
of
cesm
from
ccsm4
to
csm,
1.2
and
csm2.
E
So
simulations
are
runways
targeting
mid-prior
transient
with
400
ppm,
co2
and
midplacing
boundary
conditions
from
then.
We
actually
now
have
a
suite
of
single,
forcing
experiments
when
we
force
the
model
with
either
appliance
vegetation,
ic
changes
or
plysing,
topography
of
symmetry
or
just
the
co2
only,
and
we
have
published
some
of
the
results
using
those
simulations
to
understand
the
hydroclimate
across
asia
and
africa
and
if
you're
interested
in
exploring
aspect
of
lake,
placing
with
any
of
those
simulations
a
lot
of
data.
E
E
What
do
I
mean
by
this?
If
you
compare
the
showing
the
top
panel
here?
If
you
compare
the
surface
temperature
change
between
400
ppm
co2
and
the
pre-industrial
level
of
co2
weighs
pre-industrial
boundary
conditions
to
the
simulated
warming
for
the
same
amount
of
co2
increase
with
pricing
boundary
conditions,
the
warming
waste
price
in
boundary
conditions
is
actually
18
greater,
and
this
translates
to
about
0.8
kelvin
difference
in
equilibrium,
climate
sensitivity,
whether
you
estimate
it
with
this
warming
or
this
warming.
E
E
On
the
other
side,
I
prescribed
plexing
boundary
conditions
and
ocean
heat
fluxes,
and
I
was
three
different
levels
of
co2
and
those
splicing
boundary
conditions.
Oh,
this
is
the
prescribed
co2
level
in
the
pricing
mixed
layer.
Similar
ocean
simulations
are
also
scaled
to
reflect
the
increase
of
co2,
forcing
equivalent
to
one
watts
per
meter
square
from
pre-industrial
to
to
three
watts
per
minute
square.
E
So
you
can
see
that
the
simulated
amount
of
global
warming
per
one
watts
per
meter
square
of
forcing
shows
a
linear
relationship
or
increases
always
background
global
mean
temperature.
The
simulated
results
are
showing
those
red
dots.
What
you
would
expect
if
there
is
no
background
dependency,
is
this
gray
dot?
So
you
would
not
have
different
amount
of
warming
corresponding
to
one
watts
per
meter
square,
forcing
regardless
of
background
global
mean
temperature
and
the
blue.
E
E
We
can
carry
out
taylor,
expansion
of
radiative
response
as
a
function
of
surface
warming,
delta
r
equals
the
first
derivative
of
delta
r
as
a
function
of
surface
temperature
change
times.
The
temperature
change
plus
second
derivative,
divided
by
two
times
the
surface
temperature
change
to
the
second
so
and
so
forth.
Typically,
we
keep
the
first
order
term
of
the
taylor
series,
which
is
delta
r
toa
is
approximated
by
the
first
derivative
times
the
amount
of
temperature
change.
E
If
we
plug
in
this
expression
in
our
simple
energy
balance
model,
we
get
the
toa
relief
imbalance
equals
14
minus
first
derivative
of
delta
r
times
the
surface
temperature
change,
and
this
is
the
equation
used
by
the
gregory
method,
to
infer
the
equilibrium
climate
sensitivity,
delta
ts
for
a
doubling
of
co2
experiment
right
at
equilibrium
after
wco2,
forcing
is
balanced
by
the
delta
r
first
derivative
times.
The
temperature
change
and
temperature
change
is
the
equilibrium
climate
sensitivity.
E
E
This
is
showing
our
linear
forcing
designed
for
a
mixed
layer,
ocean
stimulation
from
one
watts
per
meter
square
to
three
and
four
watts
per
meter
squared,
and
you
can
see
this
red
dots
corresponding
to
the
simulated
amount
of
warming
corresponding
to
those
linear
forcing,
and
you
can
see
the
distance
between
the
red
dots
along
the
x-axis
actually
increases
with
the
amount
of
forcing
and
this
non-linearity
result
in
a
greater
warming.
If
we're
following
the
true
or
better
non-linear
model
compared
to.
E
If
we
follow
a
linear
fit
and
at
five
watts
per
meter
square,
the
linear
model
would
suggest
7.35
kelvin
warming.
However,
following
the
non-linear
model,
the
second
water
polynomial
model,
you
actually
would
get
9.15
kelvin
warming
and
notice
that
the
coefficient
for
the
second
order,
polynomial
term,
is
actually
negative.
E
However,
in
this
case,
radio
response
is
approximated
by
the
first
derivative
times
delta
s,
but
also
added
with
the
second
derivative,
divided
by
two
and
times
delta
ts
to
the
second.
This
is
based
on
our
second
order:
polynomial
fit
following
the
definition
of
net
response
parameter,
so
we're
going
to
take
the
response
and
divided
by
the
amount
of
surface
warming.
E
In
this
case,
you
can
see
that
this
response
parameter
is
a
function
of
the
amount
of
surface
warming
has
incurred
and
for
csm2,
like
I
mentioned
this,
the
coefficient
for
the
second
water
polynomial.
This
term
is
actually
negative,
which
suggests
that
the
response
parameter
actually
decreases
with
warming.
So,
given
that
the
forcing
is
balanced
by
response
parameter
at
equilibrium
times,
the
amount
of
warming
increasing
background
warmth
as
a
result
would
increase
climate
sensitivity.
E
I
don't
have
a
lot
of
results,
but
I
want
to
show
a
first,
a
quick
first
order
diagnostics
to
see.
Where
is
this
rise
of
response
parameter,
non-linearity
from
for
every
watts
per
meter
square
linear
increase
of
forcing
I
calculated
the
amount
of
changes
in
the
on
top
of
ensure
reflected
shortwave
radiation,
and
I
calculated
the
changes
in
the
planetary
emissivity.
E
The
results
are
plotted
here
again.
X-Axis
is
the
linear
increase
of
forcing
one
watts
towards
30
watts
and
two
four
watts
per
meter
square
from
the
pre-industrial
on
the
y-axis
shows
the
changes
in
the
reflected
short
wave
radiation,
and
you
can
see
the
range.
The
relationship
between
the
changes
in
the
shortwave
reflected
shortwave
radiation
and
the
fourth
thing
is
nonlinear,
so
there
is
a
reduction.
E
E
So
here
are
some
quick
thoughts
about
the
implications
to
paleoclimate,
so
those
results
seem
seem
to
suggest
that
estimating
equilibrium
climate
sensitivity
from
past
womb
climate
really
have
to
consider
the
background
warmth.
If
so,
csm
csm2
results
are
proven
to
be
true
and
it's
perhaps
less
useful
than
estimating
the
net
response
parameter
in
terms
of
understanding
what's
happening
in
the
system.
E
Hopefully,
I've
demonstrated
I've
showed
you
that
ecs
very
substantially
with
background
warmth,
and
this
is
tight.
This
is
tied
back
to
non-linearity
in
both
short
wave,
long
wave,
toa
responses
and
of
overarching
or
a
little
bit
of
speculation.
Is
this
perhaps
offers
a
explanation
for
warm
climates
with
moderate,
with
moderately
elevated
co2
such
as
myosin,
regardless
of
what
co2
reconstruction
you're
looking
at
it's
less
generally
less
than
two
times
co2,
yet
the
global
mean
temperature
is
five
to
sixty
five
to
six
degrees,
warmer,
hey
thanks!
G
J
E
B
E
B
A
K
K
K
B
K
K
L
K
K
All
right
well
yeah.
I
guess
we
can
stay
on
time,
then
so
we
can
get
started
with
our
second
session
of
this
morning.
So
yeah,
our
first
speaker
in
the
second
session,
is
going
to
be
pedro
de
nisio.
Speaking
about
an
ensemble
of
hosing
simulations
to
help
predict
changes
future
changes
in
tropical
climate
so
picture.
Take
it
away,
you
can.
You
can
share
your
screen
all.
O
N
N
So
I
thought
it
would
be
time
to
just
share
with
you
the
final
results.
We
all
know
that
the
abrupt
events
that
punctuated
the
last
glacial
period
produced
large
changes
in
the
tropics.
So
here
we
have
records
that
I'm
sure
everyone
is
familiar
with
greenland
temperature
on
the
carieco
basin
record,
attracts
the
rainfall
or
hydroclimate
associated
with
the
atlantic
idcc
and
in
red
in
the
middle
and
a
record
from
the
indian
monsoon.
N
And
by
the
time
this
record
from
the
indian
monsoon
was
produced,
we
all
realized
that
abrupt
events
had
a
global
influence
in
the
tropics,
and
you
can
see
here,
henrik
sterile
one
with
cooling
over
greenland
and
pronounced
drying
over
that's
northern
south
america,
where
the
spacing
is
located
and
also
drying
of
the
indian
monsoon.
So
our
understanding
of
these
mechanisms,
the
mechanisms
that
drive
this
global
response
is
far
from
complete,
there's
a
long
story
of
hosting
simulations
which
are
used
to
weaken
the
a
mog.
N
We
had
a
lot
of
presentations
today
about
that,
but
there's
a
lack
of
a
coordinated
protocol
or
a
multi-modal
approach
model
data
evaluations
have
been
limited
until
now.
There's
a
few
regional
evaluations
or
using
just
one
single
model,
and
some
of
some
other
ones
have
inconsistencies
on
how
the
hierarchies
responds
to
a
change
in
the
a
mock
or
during
the
abrupt
event,
is
isolated
relative
to
the
rapid,
the
glacial
changes
that
happened
alongside
these.
N
So
you
can
see
highly
coherent
patterns
across
the
global
tropics
and
I'm
gonna
get
into
more
detail
on
what
we
think
they
mean
later
on,
but
there's
an
abundance
of
records
from
the
tropics.
Now
we
should
be
amazed
by
all
this
amazing
data.
Each
one
of
these
records
is
just
someone
going
to
the
field
producing
a
record.
N
N
This
is
not
a
coordinated
ensemble,
so
it's
an
example
of
opportunity.
It's
just
all
the
simulations.
We
could
get
our
hands
on
they're
based
on
the
kariyama
2013
study,
so
we
kept
nine
simulations
from
that
ensemble
and
we
added
another
eight
with
newer
models.
They're
all
run
relative
to
glacial
conditions.
N
The
bars
are
the
number
of
sides
that
are
drier
in
different
latitude
bands,
which
you
could
you
could
have
infer
this
from
the
map.
I
show
you
before,
but
if
you
want
to
be
quantitative,
we
have
a
predominance
of
sites
in
the
northern
hemisphere,
showing
drier
conditions
and
a
predominance
of
sites
in
the
southern
hemisphere,
showing
weather
conditions
during
heinrich
ster1,
and
then
the
solid
line
is
the
ensemble
mean
annual,
mean
mean
rainfall,
change
simulated
by
the
models
and
the
spread
is
the
min
max
range.
N
So
you
see
that
almost
simulated
sonar
means
shift
in
the
atcc
there's
this
little
kink.
That
is
very
interesting.
Some
models
do
not
show
show
weather
conditions
in
in
the
northern
hemisphere
here,
but
on
on
average,
we
see
the
sun
almond
response,
but
when
we
compute
the
agreement
and
if
you've
been
following
our
research,
we
use
this
coins.
N
Capa
statistic
to
estimate
agreement
between
patterns
simulated
by
models
and
proxies
in
categorical
form
when,
when
it's
zero,
it
means
no
agreement
when
it's
one,
it
means
perfect
agreement
here
you
have
all
the
simulations
all
17
and
they
show
a
wide
range
of
agreement
with
the
proxies
from
close
to
zero
to
pretty
high
up
here.
Point
six:
it's
super
high.
The
coins
copy
statistic
normalizes
for
random
agreements.
It's
really
difficult
to
get
to
perfect
agreement.
Every
single
side
will
have
to
agree
with
the
sim
web
simulation
too.
So
what
is
driving
this
diversity
of
agreement?
N
N
Where
has
the
largest
influence,
and
I
would
argue,
we
don't
see
a
clear
connection
models
simulate,
a
lot
of
cooling,
others
simulate
very
little
cooling
and
the
last
the
last
quantity
we
find
a
bit
of
a
more
of
a
consistent
relationship
with
the
magnitude
of
the
cooling
over
the
tropical
atlantic.
Although
it's
not
a
clear
cut,
you
have
exceptions
but
models
that
simulate
moderate
cooling.
So
these
models
are
here
towards
the
right,
I'm
showing
them
with
my
pointer,
oh
and
I
circle
them,
those
show
cooling
over
the
tropical
atlantic.
N
So
this
is
the
caribbean
and
the
western
tropical
atlantic
of
roughly
two
or
three
degrees
and
those
show
are
the
ones
that
show
the
highest
agreement.
Although
we
have
two
models
here
that
don't
show
such
high
agreement
but
there's
a
tendency,
then
the
stronger,
the
cooling
in
this
region,
the
higher
the
agreement,
so
we're
going
to
take
an
ensemble
mean
approach
to
look
where
this
agreement
is
coming
from.
N
But
before
we
do
that,
I
just
want
to
show
you
what
the
patterns
of
cooling
look
in
the
atlantic,
so
we
group
the
malls
into
three
different
groups
based
on
the
magnitude
of
the
cooling
over
this
domain
in
the
caribbean
and
the
western
tropical
atlantic.
The
domain
in
red,
moderate,
strong
and
muted,
and
you
can
see
that
averaging
the
responses
of
all
the
models
you
can
see
distinct
patterns
there.
N
So,
first
of
all,
I
want
to
show
you
the
models
that
show
the
best
agreement,
the
ones
the
ones
with
moderate
tropical
cooling,
and
I
think
this
is
mind-blowing
that
you
have
all
these
sites
with
records
that
are
rainfall
that
are
used,
proxies,
are
sensitive
to
rainfall
and
the
patterns
agree
with
such
detail.
I
mean
there's
so
many
places
you
could
look
at,
but
on
this
group
of
models,
explain
or
agree
with
the
process
almost
all
over
the
topics
so.
N
Let's
look
into
the
details
so
on
the
left,
I'm
showing
the
three
different
groups
of
models
showing
the
annual
mean
the
annual
mean
ensemble
mean
rainfall
changes.
So
on
top
the
model,
cooling
ones,
the
middle,
the
strong
tropical
cooling
there's
three
models.
I
show
a
lot
of
cooling
in
the
tropical
atlantic
and
then
on
the
bottom,
the
ones
with
muted
tropical
hunting
cooling.
N
So
you
see
that
all
of
them
have
a
shift
of
the
itcc
in
the
atlantic
with
different
intensities,
but
it's
quite
coherent
and
not
that
different,
but
the
mod
when
we
compute
the
kappa
values
so
the
model
proxy
agreement
for
the
ensemble
mean
patterns
now,
so
we're
not
looking
at
individual
simulations.
Just
the
pattern
simulated
by
all
these
the
simulation
scene
is
in
each
group
average.
Together
we
see
what
we
saw
in
the
previous
figure
that
the
models
with
tropical
cooling
show
higher
agreement
over
the
entire
domain,
this
domain
in
the
global
tropics.
N
So
where
is
this
higher
agreement
coming
from?
Well,
it's
all
coming
from
the
atlantic
and
you
can
clearly
see
with
the
naked
eye
that
all
three
types
of
simulations
capture
the
patterns
in
the
atlantic
domain,
so
the
atlantic
domain
is
sites
in
land,
masses
surrounding
the
tropical
atlantic
and
the
difference
is
actually
coming
from
the
rest
of
the
troubles.
N
So
you
can
see
a
much
higher
agreement
for
the
models
we
tropic
with
other
cooling
and
a
bit
less
for
the
ones
with
strong
cooling.
We
think
that
those
models
overdo
the
response
and
at
some
point
they
start.
They
start
not
agreeing
in
some
of
the
boundaries
of
the
responses
and
the
ones
with
muted
cooling
show
much
weaker
agreement
which
less
statistical
significance.
So
I'm
shading
these
bars
based
on
the
statistical
significance
of
the
kappa
values.
N
So
let's
look
at
the
differences
in
the
rest
of
the
tropics,
so
we
divided
the
domains,
our
global
domain
into
different
regions
and
computed
the
kappa
values
regionally.
So
we
have
in
caribbean
and
eastern
pacific
itcc
domain.
So
it's
all
these
sites
along
the
eastern
coast
of
the
americas
and
on
that
domain.
N
Our
models,
with
muted
with
moderate
cooling,
are
the
ones
that
best
explain
the
paleo
data.
Although
the
strong
pulling
also
simulates
patterns
that
explain
the
data-
and
here
we
have
the
highest
value.
0.7
is
a
lot
almost
perfect
agreement-
muted
cooling,
that
that
not
that's
not
explains
the
patterns,
although
it
shows
some
agreement
because
you
can
see
here
some,
but
those
models
still
produce
some
drying
over
central
america.
N
N
Okay,
perfect,
so
we
also
see
that
the
moderate
cooling
gives
us
a
much
higher
agreement,
although
not
with
the
same
levels
of
statistical
significance,
because
the
proxies
are
a
lot
more
noisier
there
and
then,
if
we
focus
on
the
sites
around
the
indian
ocean
which
capture
changes
in
the
east
african,
indian
and
australian
monsoons,
we
see
also
a
similar
feature
with
the
models
with
moderate
cooling
best
explaining
the
paleo
data.
So
I'm
going
to
jump
into
conclusions.
N
N
And
the
magnitude
of
the
a-maxious
or
the
highlighted
cooling
is
not
as
important
and
if
you
recall
the
review
paper
in
2008
by
clement
and
peterson,
they
proposed
two
mechanisms:
the
sauna
linkage
to
the
tropic,
with
mostly
sunlight
asymmetric
responses
triggered
by
the
tropical
atlantic,
and
that's
essentially
what
we
find
that
this
mechanism
is
the
one
that
explains
the
data
and
this
horizontally.
Symmetric
adjustment
does
not
explain
the
data,
and
this
is
what
we
see
in
the
models
with
muted
pulling
in
the
tropical
atlantic.
More
of
a
zonally
symmetric
shift
in
the
idcc.
N
You
can
focus
here
on
the
caribbean
and
the
amazon,
but
there's
a
lot
of
other
regions
where
the
discrepancies
are
huge
and
it
could
be
a
main
source
of
uncertainty
in
future
predictions.
What
how
the
tropical
atlantic
responds
to
changes
in
the
a
mod
in
the
future?
So
I'll
stop
there,
hopefully
there's
time
for
a
question.
K
K
Okay,
great
well
yeah
thanks
peter,
that's
really
interesting
yeah!
If
people
do
have
questions
feel
free
to
put
them
in
the
chat.
We
can
have
discussion
that
way
as
well,
but
in
the
meantime,
to
kevan
chen.
Who
is
our
next
speaker
who
is
going
to
be
speaking
about
drought
and
volcanic
eruptions,
influencing
the
history
of
china
in
the
context
of
the
ming
dynasty
mega
drive.
G
Hello:
everyone,
my
name,
is
kovan
chung
from
nanjingnum
university
and
I'm
a
visiting
student
of
the
ohio
state
university.
My
advisors
are
professor
genny
liu
and
professor
yang
ning.
Today,
my
topic
is
one
job
and
one
volcanic
eruption
influenced
his
history
of
china.
The
main
dynasty
method-
this
is
the
outline
first,
is
the
motivation.
A
G
I
can't
go
down
okay,
many
efforts
have
been
devoted
to
studying
the
mechanisms
behind
civil
drought
over
eastern
china,
like
the
ming
dynasty
metros.
Previous
findings
confirmed
that
decadal
metros
in
eastern
china
are
primarily
caused
by
internal
reliability
such
as
pdo
and
ammo.
Changes
on
cons
of
this
external
forcings,
such
as
volcanic
eruptions,
can
also
influence
precipitation
changes
in
eastern
china
in
multiple
time
scales,
for
example,
on
decatural
time
scale.
G
G
G
So
in
this
study
we
investigated
the
mechanisms
behind
the
mean
dynasty
method
from
the
combined
effects
of
internal
reliability
and
volcanic.
Forcing
in
this
work
we
use
the
reconstructed
drywall
index
and
precipitation
in
china.
We
also
used
13,
cesm
lne
or
forcing
simulations
we
produce
them.
We
reproduce
the
ming
dynasty
electrode
by
volcanic
sensitivity,
experiments
using
csm
and
in
this
paper
we
define
draws
as
negative
presentation
anomalies
lasting
for
five
consecutive
years
with
maximum
precipitation
anomalies
lower
than
one
standard
deviation,
and
this
is
our
word
research
region.
G
G
G
According
to
the
the
two
reconstructions,
there
are
no
strong
external
forcings
at
the
beginning
of
the
mechanics
at
1637,
which
means
these
drugs
may
be
triggered
by
internal
reliability.
First,
the
ring
for
minimum
occurred
at
the
middle
of
the
chart
around
1641,
which
coincides
with
the
eruption
of
monson
parker
in
the
same
year.
G
The
dry
conditions
persisted
about
four
years
after
the
election.
As
a
consequence
of
this,
we
hypothesized
that
this
drug
may
be
initially
triggered
by
internal
reliability
in
1637
and
was
then
amplified
and
extended
by
the
eruption
of
mountain
parker
in
1641,
generating
a
seven
year.
Backdrop
event
verify
our
hypothesis:
we
study
the
impacts
of
internal
reliability,
volcanic
forcing
and
their
combinations
respectively.
G
G
We
counted
the
frequency
of
drugs
with
seven
year
length
and
about
three
sigma
intensity,
which
is
the
same
with
the
ming
dynasty
metros
in
each
of
the
2000
times
series,
and
then
we
draw
the
probability
distributions
of
draw
frequencies
on
the
right
hand,
side
in
this
favor
the
x,
coordinate,
is
drop.
Frequency
and
y
coordinate
is
the
probability
of
the
jaw
frequency.
G
We
can
see
that,
if
the
mean
dynasty
electrode
is
a
natural
event,
the
probability
of
its
occurrence
is
only
10
percent,
which
is
very
small.
Therefore,
the
mean
dynasty
method
has
a
small
occurring
probability
if
it
is
nearly
caused
by
internal
climate
changes,
the
superposition
of
external
forcing
can
improve
its
occurring
probability.
G
Then
our
superposition
mechanisms
can
be
examined
by
comparing
draw
events
with
and
without
volcanic
influences.
In
this
favor,
the
blue
lines
show
the
ensemble
main
joint
events
from
the
certain
authors
in
csm
rnac
simulations
in
the
past
millennium.
In
the
first
figure,
the
jaws
are
influenced
merely
by
volcanic
eruptions
without
a
natural
charge.
Job
occurred
before
the
eruption
and
we
can
see
that
the
drugs
persisted
nearly
two
years
after
volcanism.
G
G
G
Firstly,
we
selected
15
drug
events
induced
by
internal
reliability
from
a
500
year,
control
simulation
each
drought
persisted
3
to
5
years,
which
equals
the
first
half
period
of
the
ming
dynasty
map
drugs.
Then
we
added
volcanic
forcing
of
mountain
parker
into
the
last
year
of
each
of
the
15
drug
cases.
G
This
is
the
ensemble
mean
precipitation
anomalies
in
a
control
simulation
and
volcanic
sensitivity
simulation.
We
can
be
seeing
that
without
volcanic,
forcing
the
last
for
about
five
years
ending
with
a
wet
event.
However,
volcanic
eruptions
suppressed
these
otherwise
plant
events
and
prolonged
drugs
up
to
three
years.
This
is
consistent
with
observations
and
highlights
the
roles
of
internal
reliability
and
volcanic
forcing
in
triggering
and
extending
drugs
respectively.
G
Then
we
further
investigated
whether
volcanic
impacts
remains
the
same
when
the
eruption
occurred
in
different
job
phases.
Here
we
carried
out
another
two
groups
of
volcanic
sensitivity,
experiments
with
volcanic
aerosols
added
in
the
late
and
early
stages
of
the
15
shot
cases
in
this
figure.
The
red
and
blue
lines
are
ensemble
mean
precipitation
anomalies
with
volcanic
forces
added
to
the
last
year
and
the
second
year
of
the
natural
charge
figures
on
the
top
are
control
simulations
because
in
the
middle
are
volcanic
sensitivity,
experiments
and
figures
at
the
bottom
are
differences
between
sensitivity
and
control
simulations.
G
G
G
However,
the
weakening
of
land,
seasonal
contrast
and
the
retreating
of
west
pacific
subtracted
high
are
consistent
in
late
phase
and
early
phase
sensitivity,
experiments
which
cannot
expand
the
nonlinear
response
of
charts,
but
the
soil
moisture
are
different
in
these
two
experiments
in
this
figure.
The
red
and
blue
lines
are
soil,
moisture
changes
in
late
and
early
phase
experiments.
G
G
So,
in
the
late
phase
volcanic
eruptions,
the
accumulated
cellular,
moisture
deficiencies
before
volcanism
can
lead
to
drier
prey
conditions
and
less
evaporations,
which
give
rise
to
larger
small
or
larger
lingual
decreases
after
volcanism.
However,
in
the
early
basic
experiments,
the
dry
preconditioning
of
soil,
moisture
and
its
feedback
from
precipitation
are
frequent
so
leading
to
smaller
short
responses.
G
G
G
F
K
K
P
P
P
So
the
green
sahara
was
a
period
time
when
the
sahara
was
much
greener
than
today.
So
it
wasn't
like
this,
it
was
it
was
inhabited
by
vegetation
and
it
had
lots
of
lakes
and
wetlands
and
rivers,
and
it
was,
I
am,
and
it
was
inhabited
by
people
and
animals
and
there's
a
lot
of
evidence
for
the
presence
of
queen
sarah
from
all
of
these
different
lines
that
I've
listed
here
and
but
this
holocene
green
sahara
was
not
the
only
green
sahara
out
there.
P
So
let
me
go
back
to
this
question
about
greece,
and
so,
like
I
mentioned,
we
want
to
be
able
to
do
accurate
climate
modeling
for
some
of
these
time
periods
to
understand
how
climate
would
have
impacted
these
civilizations,
these
societal
changes,
and
so
we
want
to
understand
the
land,
surface,
boundary
condition
which
is
the
green
sahara,
and
this
question
I'm
going
to
break
down
into
three
different
questions.
How
does
the
green
sahara
begin?
P
P
So,
okay,
so
how
does
the
greece
arab
begin?
We
know
now
for
a
long
time
that
the
primary
the
primary
external
influence
was
the
rising
solar
insulation
and,
during
this
last,
during
this
most
recent
green
sahara,
and
this
last
jinger's
last
interglacial,
but
over
the
past
decade
worked
a
couple
of
work.
Interesting
work
has
shown
that
the
while
the
rising
insulation
is
important.
There
is
a
a
very
important
influence
of
a
millennial
scale.
P
Events
in
the
north
atlantic
that
tend
to
synchronize
the
the
start
of
greenstar
over
africa
and
and
and
and
control
their
special
features,
so
medium
skill
events
are
critically
important.
So
in
the
last
deglaciation
the
younger
dryers
was
one
such
millennial
event
and
the
ultimate
is
the
glaciation.
The
highway
stadium
11
was
one
such
there's
such
an
event,
and
this
is
a
figure
from
mandela
at
all.
That
recently
came
out.
How
is
the
green
sahara
sustained?
P
This
is
obviously,
this
has
been
a
very
important
area
of
research
and
over
the
last
four
years
or
so.
A
lot
of
work
has
been
done
in
this
regard,
showing
the
influence
of
the
feedbacks
that
that
can
be
that
are
present
from
the
land
surface,
such
as
feedback
from
vegetation,
soil,
surface
water
feedback
from
the
ocean
itself,
for
example
that
lead
to
the
enhancement
of
the
initial
effect
of
higher
summer
insulation
and
in
the
middle
policy.
So
this
is
a
very.
P
Here,
I'm
showing
you
on
the
left-hand
side,
the
enhancement
to
precipitation
during
the
metaholic
holocene
that
is
obtained
just
from
change
and
just
from
the
higher
summary
insulation
during
the
mid
holocene,
and
you
can
see
that
there
is
some
enhancement,
and
this
here
is
a
zonal
mean
of
the
precipitation
and
that's
the
blue
curve.
Here.
The
black
is
the
present
day,
and
this
is
the
anomaly.
So
you
can
see
that
there
is
some
enhancement
just
in
the
tropics,
but
it's
not
enough
and
it
certainly
doesn't
cover
the
most
of
africa
and
it's
not.
P
The
magnitude
is
not
high
enough
to
to
agree
with
these
proxies,
which
are
shown
here
in
in
in
circular
markers.
But
then
we
try
to
incorporate
the
green
side
changes
and
the
way
I
do
that
is
by
changing
the
land
surface
conditions
such
as
the
vegetation
distribution,
the
presence
of
lakes
and
the
soil
properties.
I'm
just
showing
you
here
the
changes
to
vegetation
and
lakes
here
so
in
each
grid
cell
in
clm.
P
These
are
the
dominant
currently,
for
the
present
day,
these
are
the
dominant
plant,
functional
types,
and
I
changed
them
to
this
thing
here.
In
fact,
in
the
changes
that
are
prescribed,
there
are
multiple
plant
function,
types
in
each
grid
cell,
just
like
they
are
in
present
day
as
well,
but
I'm
just
showing
you
instead
of
that
17-dimensional
figure,
I'm
showing
you
just
the
dominant
and
functional
types.
P
So
once
I
apply
these
changes
to
the
model,
then
I
have
this
new
figure
here
on
the
metallic
precipitation
anomaly.
With
respect
to
british
industrial,
you
can
see
that
the
magnitude
of
precipitation
increases
considerably
more
and
then
there
is
a
lot
of
precipitation
that
in
that
that
extends
deeper
into
the
sahara.
The
zonal
mean
here
again
is
shown
here.
P
We
do
have
a
little
bit
of
overprediction
here.
That
tends
to
be
the
case
for
many
models,
but
otherwise
we're
doing
a
pretty
good
job.
So
again,
a
modern
couple
kind
of
model
can
can
reconcile
the
available
proxies
simply
on
the
basis
of
land,
surface
feedbacks
and
a
key
feature
here
is
that
the
precipitation
that
we
get
here
all
over
africa
is
actually
sufficient
to
maintain
the
vegetation
that
we
prescribed
as
opposed
to
what
happened
in
some
of
the
older
studies,
where
the
vegetation
that
is
prescribed
leads
to
an
improvement
in
precipitation.
P
P
But
now
I'm
going
to
talk
about
how
green
sarah
ends
and
just
this
this
is
going
to
focus
a
little
bit
on
the
work
that
currently
is
in
the
review,
and
so
the
key
feature
of
one
of
the
interesting
features
of
the
termination
of
green
sahara.
Is
this
tri-time
transgressive
termination,
where
several
studies
have
shown
that
the
sahara
terminates
earlier
in
the
north
and
earlier
in
the
east?
So
there
are
various
results
over
the
years
that
have
talked
about
this.
P
This
is
one
of
the
newer
figures
here
results
from
china
and
based
on
this.
This
figure
shows
here
the
time
of
the
onset
of
green
style
termination
and
the
and
the
markers
are
color
coded.
According
to
this,
this
color
bar
here
and
you
can
see
that
they
sort
of
started
earlier
in
the
east
and
north,
and
there
is
a
as
you
go
westwards
southwards.
P
They
they
terminated
later,
okay,
so
we
try
to.
I
try
to
understand
that
why
this
happened
why
this
might
be
happening
in
in
in
the
real
system,
so
we
found
that
this
is
the
result
of
a
feedback
that
is
that
arises
from
two
different
two
different
processes
that
are
taking
so
the
first
one
is
the
effect
of
land
changing
the
land
surface.
P
So
if
I
have
a
land
surface
that
has
an
east
west
difference,
so
actually
I
should
have
put
this
thing
something
here.
I
forgot
to
put
this
thing,
but
let
me
just
remind
you
that
my
land
surface,
initially
in
the
first
paper,
was
very
zonal
uniform
changes,
but
this
one
I'm
talking
about
when
I
do
have
an
east-west
asymmetry.
So
it's
not
generally
uniform.
I
don't.
I
forgot
to
put
a
figure
here.
P
I
should
have
so
okay,
so
when
there
is
an
east-west
difference
in
the
amount
of
vegetation,
this
leads
to
an
east-west
difference
in
the
simulated
moisture
that
is
available
for
this
region.
So
this
is,
these
are
all
anomalies
with
respect
to
pre-industrial,
so
here
I'm
showing
for
the
pre
ambient
holocene.
So
it's
wetter
on
the
west
and
it's
drier
in
the
east,
and
I
do
the
same
thing
for
the
early
holidays
in
which
here
is
9ka
and
in
early
holistic
as
well.
Everything
else
being
the
same.
P
So
this
this
influence
of
a
difference
in
the
east-west
vegetation
is
actually
more
important
in
the
mid
holocene
and
is
weaker
in
the
early
holiday.
So
the
key
phase
is.
The
key
fact
is
that
this
is
important
during
the
remaining
phase
of
the
green
sahara,
not
in
the
early
phase
when
green
sahara
was
more
prolific.
So
that's
one
thing
to
keep
in
mind.
P
The
second
thing
to
keep
in
mind
is
the
effect
of
change
in
now,
okay
yeah.
So
the
reason
why
this
happens
is
because
there
is
this
land
surface
change
leads
to
an
anomalous
of
the
tropospheric
change
in
the
water
circulation.
You
have
an
anomalous
ascent
in
in
the
west,
which
leads
to
more
convective
precipitation,
and
you
have
descent
over
the
western
parts
of
eastern
part
of
sahara,
which
leads
to
drawing
so.
The
second
effect
is
the
effect
from
the
change
in
insulation.
P
So
here
I'm
showing
you
these
three
figures
that
are
all
with
holocene
minus
early
holocene
with.
But
what
is
different
between
each
of
these
is
the
vegetation
that
has
been
prescribed
on
the
surface.
So
if
we
have
only
a
pre-industrial
vegetation,
then
going
from
the
early
holocene
to
the
mid
holocene
leads
to
a
uniform
drawing
all
across
africa.
P
So
if
you
have
in
the
reference
case
where
I'm
only
having
a
pi
vegetation,
one
minute
warning
right
again,
I'm
just
going
to
wrap
up
here,
there's
this
continuous
descent
over
over
sahara.
But
when
you
have
a
mid
holocene,
a
green
silence
simulation,
then
you
have
a
change
in
water
circulation.
Releasing
this
difference
so
those
two
effects.
P
The
change
in
change
in
insulation
leads
to
a
change
in
the
ground
surface
and
the
change
in
the
stress
greening
then
itself
leads
to
a
change
in
the
change
in
the
precipitation
response
leads
to
a
total
response.
That
is
that
is
shown
here
on
the
right.
Okay,
so
you
have
more
drying
in
the
east
and
you
have
actually
a
greater
amount
of
precipitation
in
the
west
between
early
holistic
and
obesity.
P
So
what
this
does
is
that
coming
up
from
9k
to
6k,
your
desertification
would
have
started
in
the
iraq
in
the
east,
but
it
would
have
been
effectively
delayed
in
the
east
here,
and
the
key
impact
is
that
this
this
feedback
actually
weakens
later
in
the
holocene.
So
if
you
look
at
the
around
three
ka
versus
mid
holocene,
then
after
the
after
embed
holocene
this
this,
this
certification
continues
in
a
zonally
uniform
way,
which
is
exactly
what
we
observe
in
proxies.
P
We
observed
that
the
onset
there's
a
difference
in
onset
for
some
time
and
then
later
in
the
in
the
desertification
process.
The
desertification
process
proceeds
as
well
in
the
informal
way.
Okay,
so
quickly,
just
very
simple
summaries.
Blank
surface
feedbacks
are
helpful
to
are
important
to
sustaining
green
sahara,
but
they
are
also
important
during
determination
of
the
green
sahara
and
it's
the
interaction
of
the
land
surface,
with
the
changing
insulation
during
the
holocene
that
delays,
desertification
in
the
west
and
the
answer
to
the
certification.
The
lesson
she
said.
K
Okay,
well,
yeah.
I
think
we
do
need
to
move
on.
Unfortunately,
in
the
interest
of
time,
but
again,
please
feel
free
to
put
questions
in
the
chat.
If
you
do
think
of
anything,
yeah
super
interesting
again,
all
right.
So
I
guess
our
next
speaker
is
theodore
meyer.
Theater.
Are
you
here,
yes,
perfect
I'll,
be
speaking
about
water,
isotopic
imprints
on
the
pacific,
walker
circulation,
so
yeah?
Take
it
away.
Q
K
Q
Q
So
yeah
I'm
going
to
talk
about
the
applied
specific
work
circulation
and
just
to
give
some
background.
Specific
rock
circulation
is
a
model
of
air
flow
in
the
tropical
pacific,
and
it's
characterized
by
the
racing
air
masses
of
the
west
pacific,
warm
pool
and
substance
in
the
east
and
understanding
how
the
circulation
is
changing.
Q
To
respond
to
the
response
of
current
and
future
climate
through
co2
increase,
it's
important
as
it
has
been
shown
that
changes
influence
the
global
bond
potential
and
tropical
hydrochloric
plants
greatly
and
one
of
the
key
predictions
of
global
circulation.
Climate
models
is
the
weakening
of
this
circulation.
Q
However,
international
observations
over
the
past
few
decades
have
been
inconsistent
and
do
not
support
this
clear
signal
and
reasons
might
be
strong
internal
variability
and
antibiotic
aerosol
forcings,
which
makes
analysis
more
difficult,
but
nevertheless
it's
like
it's
it's
very.
It's
apparent
that
there's
a
need
for
new
and
model
prediction
assessment,
and
for
that
we
propose
to
use
the
water
stoplog
archive
of
the
actual
pacific.
Q
So
why
I'm
looking
at
the
what
is
topology
well?
Recent
research
has
shown
that
there
exists
a
strong
linkage
between
the
circulation
and
precipitation
pattern
changes,
for
example,
this
last
figure
from
chongadao,
where
you
can
see
the
strong
correlation
between
the
vertical
velocity
and
station
of
the
last
few
decades
and
other
research
from
deidol
on
the
right
here,
which,
as
we
can
see,
clears
topic
difference
between
eastern
and
western
pacific
during
different
pacific
october.
Q
States
like
the
linear
and
the
linear
conditions,
and
the
iso
composition
is
depends
on
the
the
incoming
moisture
from
different
regions
and
also
phase
change
processes
and
all
these
combined
effects
I'll
get
influenced
by
changing
of
this
work
circulation.
So
these
first
results
here
show
that
there's
a
international
time
scale.
Q
As
you
can
see
here,
the
black
bar
and
because
of
the
possibility
of
obtaining
leaf
excellent
theorem
records
through
my
collaborators.
I'm
currently
like
all
the
figures
you
see
here
are
focusing
on
simulated
precision
deteriorate
and
this
connection
with
the
circulation
strength
changes
my
experiment.
I
used
computer
systems
version
1.2
with
iso
tracking
enabled
this
allows
for
the
tagging
of
water
and
wireless
isotopologs
originating
from
user-friend
regions
throughout
the
reinhardt
and
transport
history,
and
also
important
note
that
I'm
using
a
mixed
layer,
ocean
component
with
sims
adhesive
physics
and
I'm
in
total.
Q
Q
Q
So
that's
what
we
get
on
the
left
right
axis
is
the
esector
pacific
and
the
east.
The
blue
right
acts
as
the
western
side,
and
we
have
two
figures
for
the
pre-dos
and
hysteria
and
binations
as
you
can
see.
As
the
co2
increases
on
the
x-axis,
we
have
an
increase
or
like
an
enrichment
of
the
deuterium
on
both
sides
of
pacific,
and
this
can
be
explained
by
the
weaker
fractionation
higher
temperature.
So
it's
a
direct
warming
response
of
the
hydroclimatic
pattern.
Q
Q
Now
here
I
looked
at
the
decreasing
difference
between
east
and
west
we've
just
seen,
and
that
correlates
pretty
well
with
the
weakening
specific
work
circulation
as
co2
increases.
So
again,
the
left
we
have
the
difference
tier
and
on
the
x
axis,
the
mass
overturning
which
are
used
as
a
direct
measure
of
circulation,
strength,
which
I
calculated
from
the
top
right
equation
up
here.
Q
So
that's
very
promising
we
can
see
like
okay,
at
least
for
like
a
larger
co2
changes.
We
can
see
some
the
deuterium
change,
which
might
be
useful
for
a
specific
rocket
strength.
Proxy,
like
the
difference
between
specific
the
eastern
western
pacific,
might
be
useful
for
that,
and
just
for
completeness
sake.
Q
I'm
only
like
going
to
scratch
the
surface
here.
It's
just.
Some
first
looks
at
it,
but
we
can
also
do
some
interesting
stuff,
so
the
water
tagging,
as
I
mentioned
before,
allows
to
track
water
throughout
the
hydrological
transport
and
we
look
at
the
western
pacific
again
as
sink
regions,
and
I'm
only
looking
at
the
moisture
recycling
this
time.
Q
So
here
we
see
the
change
of
fraction
presentation
from
western
east
pacific
and
it's
petitioned
in
the
local
recycled
preservation,
common
region
itself
and
to
remote
moisture
from
the
rest
of
the
world
and
and
despite
absolute
increase
to
the
warming,
there's
reduction
in
the
locally
sourced
well-depreciation
at
both
sides.
Pacific
right-
and
this
is
compensated
by
the
increase
from
remote
regions.
Q
This
is
very
interesting
because
remote
sources
are
more
like
colder
and
thus
more
depleted
in
the
ethereum
or,
like
other
any
isotopes,
and
we
think
that
the
increase
in
moisture
is
muting,
the
normal
enrichment
rate
and
because
effects
large
in
the
east.
We
have
this
slow
rate
of
the
team
enrichment.
You
have
seen
being
four
compared
to
the
west,
so
let's
just
kind
of
run
outlook
by
change.
K
K
O
Right,
everyone
see
my
screen:
okay,
perfect,
okay,
all
right,
hi,
everyone.
My
name
is
mary
grace
albright,
and
I
am
just
finishing
up
my
first
year
as
a
phd
student
at
the
university
of
connecticut
and
I'm
working
with
rand
fing
on
a
project
called
mid-pleisine,
north
american
monsoon
and
weather
resolving
coupled
simulations.
O
O
They
were
about
1.8
to
3.6
degrees
warmer
than
pre-industrial,
which
you
can
see
about
right
here,
and
these
temperatures
are
similar
to
predicted
warming
in
the
future,
based
on
intermediate
warming
scenarios,
which
is
highlighted
here
in
the
light,
blue
and
so
on.
The
north
american
monsoon
itself
is
an
atmospheric
circulation
feature
of
hydroclimate
and
the
north
american
southwest.
O
So
it
starts
to
develop
in
mexico
in
june
and
then
moves
up
into
the
u.s
southwest
in
july
until
about
mid-september,
and
the
north
american
southwest
really
depends
on
its
rainfall
as
a
source
of
water,
with
mexico
alone
receiving
upwards
of
75
of
its
average
annual
precipitation
during
this
monsoon
season.
So
this
means
that
future
changes
would
have
a
really
big
impact
on
the
people
living
here,
which
makes
it
all
the
more
important
for
us
to
understand
how
it
might
change
going
forward.
O
However,
more
recently,
a
new
study
by
booze
and
pascal
demonstrated
that
the
mechanical,
forcing
by
the
sierra
monterey
occidental
mountains,
is
perhaps
more
important
than
a
thermal
forcing
here.
So
they
did
this
by
running
two
different
simulations,
one
with
modern
orography
present
and
one
with
flattened
orography,
and
we
can
see
that
in
the
figures
down
here.
O
So
the
one
with
the
orography
present
has
a
clear
band
of
summer,
precipitation
in
our
monsoon
region
that
we're
concerned
about,
and
the
red
contours
are
near
surface
eastward
wind.
So
here
we
have
the
wind
converging
pulling
air
up
and
with
the
mountains,
there's
moisture
convergence
to
the
western
side,
which
we
leads
to
locally
enhanced
precipitation
formation.
O
Meanwhile,
with
the
in
the
simulation
with
the
flattened
horography,
there's,
essentially
no
monsoon
ring
band
in
our
region
that
we're
concerned
about,
and
the
wind
contours
stay
to
the
north
and
south
here
now
within
these
simulations
high
horizontal
resolution
is
really
important
for
capturing
some
of
the
key
features
of
this
monsoon.
So
these
right
here
are
two
of
the
same
models
run
at
different
resolutions.
O
So
you
can
see
in
the
low
resolution
the
ring
band
is
really
poorly
defined
and
you
can
only
really
start
to
see
the
clear
north
american
monsoon
ring
dependency
on
topography
and
the
high
resolution
simulations
now.
Also
within
these
simulations,
the
north
american
monsoon
is
predicted
to
weaken
in
response
to
future
warming.
O
So
there
is
general
model
agreement
of
drying
in
the
north
american
southwest,
which
we
can
see
in
this
figure
on
the
left,
where
the
the
red
box
is
highlighting
our
southwest
north
american
region,
and
so
this
is
showing
us
a
precipitation
decrease
of
10
to
40
percent
for
the
end
of
the
21st
century,
and
this
drawing
only
increases
in
magnitude
over
time
as
a
function
of
the
strength
of
radiative.
Forcing.
O
However,
this
agreement
is
not
necessarily
shown
in
all
proxy
records
for
the
pliocene,
so
this
figure
on
the
right
here
is
showing
us
a
southwest
north
american
proxy
reconstruction,
where
this
gray
band
is
the
mid-ply
scene.
So
I
wanted
to
highlight
the
middle
trend
line,
which
is
showing
a
more
enriched
delta
d
signal
for
precipitation,
which
could
be
indicative
of
an
increase
in
warm
season,
precipitation
or
tropical
storms
in
relation
to
today.
O
So
one
of
the
framing
questions
of
our
work
here
is
whether
we
can
use
high
resolution
simulations
to
start
to
bridge
this
gap
between
model
results
and
proxy
data,
and
so
to
do
this,
we
use
csm
1.3
with
the
spectral
element
dynamical
core
for
the
atmospheric
component,
then
for
sea
ice
we
used
community
ice
code
version
4
for
ocean.
We
used
parallel
ocean
program
version
2
and
for
land
we
used
community
land
model
version
four
and
for
comparison
purposes.
O
We
ran
this
model
in
two
different
resolutions,
both
run
on
a
cube
sphere,
so
they
both
use
the
typical
100
kilometer
resolution
for
ocean
and
sea
ice.
But
the
high
resolution
model
used
a
resolution
of
any
120
in
the
atmosphere
on
land
which
regrets
to
about
25
kilometer
lat
long
and
the
low
resolution
used
any
30
which
comes
out
to
be
100
kilometer
lot
long
and
our
motivation
behind
these
resolution.
O
Now,
when
we
set
up
our
high
resolution
simulations,
our
boundary
conditions
came
from
prison
4d,
which
maps
to
the
high
resolution,
topography
and
bathymetry,
and
the
initial
conditions
for
ocean
ice
land
and
terrestrial
carbon
cycle
come
from
a
1200
year.
Long
simulation
from
a
previous
mid-plyom
and
the
only
part
that
was
initialized
from
code
was
the
atmosphere
and
this
initialization
allowed
for
a
pretty
short
spin
up
time,
where
our
high
resolutions
are
high
resolution
experiment
equilibrated
within
about
the
first
five
years.
O
So
we
have
initial
diagnosis
for
the
first
30
years.
Here
the
2a
radiation
imbalance
and
we
do
actually
have
45
years
of
the
hires
run
now
and
we're
planning
to
ultimately
extend
this
to
70
years
now.
First,
I'm
going
to
show
the
added
benefit
of
high
resolution
to
the
mean
climate
state
by
just
looking
at
our
pre-industrial
simulations.
O
So
we
use
the
last
50
years
of
our
low
resolution
runs
to
compare
to
the
last
30
years
of
our
high
resolution
runs
now.
First,
looking
at
the
difference
in
elevation,
we
can
see
that
in
the
high
resolution
the
mountain
range
sits
at
a
higher
elevation
and
is
much
more
confined
to
the
western
side
of
mexico.
Meanwhile,
in
the
low
resolution,
this
mountain
belt
is
more
diffusive
and
it
covers
a
wider
area.
O
And
in
this
low
resolution
the
band
is
more
diffusive
and
it
runs
across
the
mountains,
and
we
think
that
this
difference
relates
to
the
enhanced
topographic
gradient,
which
would
cause
greater
upslope
compression
and
limit
the
amount
of
precipitation.
That's
developing,
and
the
high
high
resolution
simulations
so
now
I'll
show
the
comparison
between
the
ply
scene
and
pre-industrial
data
that
we
have.
O
So.
The
one
thing
that
I
do
want
to
point
out
before
we
move
on
is
that,
because
of
our
boundary
conditions
that
were
based
on
the
prism
40
reconstruction,
this
mountain
range
does
actually
sit
at
a
lower
elevation
in
the
pliocene
than
in
the
pre-industrial,
which
we
can
see
in
these
figures
up
here.
And
this
difference
part
down
here
where
everything
in
the
blue
sits
lower
in
the
pliocene.
O
And
this
difference
in
the
high
resolution
is
the
same
as
the
difference
in
the
low
resolution,
and
this
would
be
because
of
the
geodynamic
model
by
dalsada.
All
in
2016,
which
runs
to
adjust
topography
on
earth
due
to
the
loss
of
greenland
and
west
antarctic
ice
sheets.
So
we
can
see
that
in
their
reconstruction
right
here
where
this
is
our
north
american
southwest
region
and
the
blue
is
our
lowered
mountain
range
and
so
in
terms
of
what
this
means
for
the
annual
precipitation
differences.
O
We
can
see
consistent
bands
of
drying
in
both
of
these
resolutions,
and
we
can
see
that
the
lowered
elevation
in
the
plyocene
leads
to
anti-cyclonic
circulation
at
850
millibar
forming
in
front
of
the
mountain
in
both
resolutions.
So
this
takes
precipitation
away
from
the
region
resulting
in
drying
to
the
western
side
of
the
mountain,
and
we
said
that
this
would
be,
and
rather
due
to
the
lowered
mechanical,
forcing
which
is
consistent
with
results
by
booze
and
pascal
from
2021
that
I
talked
about
in
the
beginning.
O
However,
we
then
look
at
early
summer
precipitation
changes.
There
is
actually
more
of
an
enhancement
of
the
eastern
monsoon
rain
belt
and
the
high
resolution
run,
and
these
findings
are
consistent
with
the
proxy
evidence
that
I
also
talked
about
earlier
of
the
weather,
rather
than
drier
conditions.
O
O
O
However,
for
the
north
american
monsoon
region
and
southern
asia,
these
results
showed
a
pretty
good
match
with
observation
derived
mesoscopic
mesoscale
convective
system
numbers,
but
we're
also
working
on
applying
our
parameter
sets
with
er8
data
to
check.
If
our
model
parameters
can
reproduce
realistic,
mcs
counts
from
observations,
since
our
models
are
hydrostatically
balanced,
it's
possible
that
they
might
not
be
capturing
enough
intense
precipitation,
which
means
that
we
might
need
different
thresholds.
O
But
once
we
did
run
this
algorithm
on
the
pliocene
and
pre-industrial.
We
can
see
that
in
these
figures
here
that
there
is
an
overall
increase
in
mcs
frequency
in
the
northern
side
of
the
monsoon
region,
where
we
previously
saw
enhanced
precipitation
during
the
pliocene.
So
this
northernmost
area
right
here
and
I'm
here.
O
And
then
in
the
high
resolution
there
was
an
enhanced
eastern
monsoon
rain
belt
in
the
mid
pliocene
early
summer,
precipitation
and
last
week.
Lastly,
we
saw
that
preliminary
results
suggest
the
increase
in
summer.
Precipitation
is
associated
with
an
increase
in
mesoscale,
convective
system
frequency
and
now
just
some
future
direct
directions
that
we're
planning
on
taking.
K
A
O
K
O
O
O
A
J
O
K
L
L
K
L
L
Skills
like
this
figure
shows
the
precipitation
response
to
enso
variability
to
enso,
and
you
can
see
interesting
that
the
precipitation
response
is
not
a
spatially
coherent
change,
but
there
is
an
obvious
diaper
response,
with
contrast,
different
with
contrast,
change
between
the
west
central
amazon
and
the
eastern
coastal
coast
of
brazil,
and
it's
also
interesting
to
see
that
in
a
very
long
longer
time
scale
the
orbital
time
scale.
We
can
also
see
a
similar
dipole
response
in
both
data
18,
reconstructed
from
specialism
site
and
also
the
snapshot
climate
model
simulations.
L
L
Most
previous
studies
show
that
there
is
a
coherent
variability
of
mineral
variability
of
the
tau
18
over
these
sites,
and
if
we
look
closely
to
the
right
figure
within
the
last
25
000
years,
it's
very
clear
that
they
show
a
coherent
variability
of
the
tau
18
and
we
can
also
see
that
at
the
hs1
hierarchy
stage
1.
There
is
a
much
more
depleted.
The
tau
team,
which
shows
shows
a
wet
climate
conditions
and,
based
on
this
previous
work,
conclude
that
the
during
the
at
the
hs1,
the
south
american
summer,
monsoon
system
increased.
L
It
is
if
you
recall
that
we
mentioned
in
inter-annual
and
orbital
time
scales.
There
is
a
dipole
response
of
precipitation
and
deteriorating,
but
but
for
mineral
scale,
this
dipole
seems
to
disappear.
It
is
very
weird
right.
Well,
we
can,
if
we
note
that
the
location
of
these
sites,
most
of
them
located
in
the
eastern
coastal
brazil
region
and
some
of
them
at
the
western
amazon
and
in
this
region,
little
of
them
located
in
the
england
amazon
region.
So
we
would
question
this
the
conclusion
of
this
coherent
variation.
L
L
If
so,
what's
the
mechanism
that
induces
such
a
large
scale
dipole
response
of
the
mineral
hydroclimate
variability
and
if
there
is
a
dynamically
robust,
dipole
response
in
precipitation
and
the
toy
team,
what
is
the
proper
index
for
the
south
american
summer
monsoon
system?
And
what's
what?
How
does
this
system
respond
to
melt
water?
Forcing
so
to
answer
these
questions?
L
We
have.
We
did
three
parts
of
work.
Firstly,
we
will
do
the
model
proxy
data
comparison
to
validate
the
performance
of
the
our
simulation,
and
the
second
part
is
to
investigate
the
mechanism
mechanism
of
the
nano
ring
for
response,
and
the
third
part
is
the
mechanism
of
the
nano
data
rating
response.
L
Okay.
So,
firstly,
now
let's
look
at
the
model
proxy
data
comparison
so,
as
this
figure
shows
again
most
of
the
available
speciosum
cave
data
that
is
available
to
study
the
mineral
variability
during
the
last
degradation
located
in
concentrated
in
this
coastal
regions.
Only
one
period
also
is
in
amazon
region.
L
So
since
our
model,
since
the
icsm
model
is
enabled
the
water
isotopes,
so
we
can
directly
compare
the
reconstructed
proxy
calcite
data
18
with
our
model
chelsea
data
18.
So
the
black
line
is
for
proxy
and
the
green
line
is
for
the
model,
so
you
can
see
that
all
of
these
eastern
coastal
sites,
our
model
can
capture
a
very
good
variability
of
calcite
data
18
compared
with
the
proxy
data.
L
But
one
thing
is
that
model
tend
to
underestimate
the
magnitude
of
the
millennial
variability.
The
reason
for
this
we
don't
know
yet
now,
and
we
will
continue
to
study
that
in
the
future.
So
since
the
model
is
validated
with,
we
think
it's
okay
to
study
the
mechanisms
of
millennial
rainfall
variability
over
the
tropical
south
america.
L
So,
as
we
mentioned
before,
due
to
the
spatial
coverage,
limited
spatial
coverage
of
the
proxy
data,
we
cannot
capture
the
complete
a
complete
spatial
temporal
very
variability
of
tropical
south
american
monsoon.
But
without
with
this
icsm
model,
we
can
do
that
and
firstly,
we
perform
the
combined
eof
of
precipitation
in
the
tower
team,
and
this
is
this
figure
shows
the
leading
mode
of
the
c
c,
o
f,
and
you
can
see
that
the
pc
one
time
series,
the
black
line,
shows
a
very
clear
millennial
scale,
variability
variability
and
combined
with
the
blue
line.
L
L
If
we
look
at
the
spatial
pattern
of
this
ceof
mode,
we
can
see
a
very
clear
dipole
response
right.
The
contrast,
change
of
precipitation
and
natalie
team
between
the
western
and
mid
amazon
and
eastern
coast,
brazil,
yeah
so
and
our
model
results
is
also
consistent
by
the
way
it
is
the
transient
model
simulation.
L
So
another
question
we
want
to
know
is
that
how
will
the
south
american
summer
monsoon
intensity
change
at
the
hs1
or
younger
dryas
due
to
the
increased
melt
water?
Forcing
so
like?
The
figure
e
and
f
shown
the
the
increased
melt
water
forcing
produces?
The
itcz
shift,
northward
southward
and
this
produced
an
anomalous
ascending
motion
in
coastal
brazil,
but
a
anonymous
descending
motion
in
west
central
amazon
region,
and
it
is
also
associated
with
an
anonymous
cyclonic.
L
Anonymous
cycle
on
the
upper
troposphere,
which
decrease
the
climatology
bolivia
high
and
in
the
lower
troposphere
we
can
see,
there
is
an
anomalous
westerlies
which
decrease
the
climatology
low
level
westerly
so
based
on
this
climb
atmospheric
circulation
anomalies,
we,
which
we
conclude
that
the
south
american
summer
monsoon
intensity,
should
be
decreased
at
htf1
and
younger
dryers.
This
is
contrary.
Contrary
to
conclusion,
in
previous
studies
and
the
reason
of
that
again
comes
to
the
previous
argument
that
they
only
have
the
proxies
concentrated
in
the
eastern
part
eastern
brazil
part.
L
L
Yeah,
so
the
next
part
is
the
mechanism
of
many
millennial
deteriorating
over
the
tropical
south
american
region.
So
starting
from
1916
step,
stanceguard
pointed
out
that
that
the
tropical
dataway
team
mainly
represents
the
local
ring
for
amount.
In
fact,
and
that's-
and
this
is
widely
accepted
by
paleo
climate
communities,
and
they
also
think
that
the
look
the
change
of
data
rating
over
the
tropical
south
american
also
represents
the
local
change
of
precipitation
amount.
L
But
here,
in
our
model
in
manila
time
scale
simulations.
We
can
also
see
this
pattern,
like
the
increased
precipitation
is
associated
with
depleted
deteriorating
seems
like
yeah,
it's
the
amount
effect
dominant,
but
later
we
will
show
that
the
data
rating
change
cannot
indicate
the
local
rainfall
change.
L
It's
also
out
of
out
of
our
expectation
so
to
to
investigate
the
mechanisms
of
data
18.
We
did
the
water
tagging
technique,
that
is,
we
attacked
the
source
region,
o16
water,
o16
and
o18
from
the
evaporation
of
ocean
or
land
region,
and
they
become
the
water
vapor,
and
then
they
go
go
inland
into
the
land
region
and
from
the
source
region
to
the
sink
region.
L
L
We
can
just
take
the
difference
between
the
total
data
rating
in
precipitation
between
hs1
and
lgm,
and
this
can
be
this
total
change
can
be
further
decomposited
into
the
data
waiting
composition
and
the
relative
moisture
source
contribution,
and
so,
let's
first
look
at
this
right
panel.
It
shows
that
the
total
data
rating.
L
Okay,
would
you
agree
that
I'll
be
quicker
yeah
thanks,
so
you
will
see
that
the
total
data
rating
change
between
hs1,
l
and
lgm,
the
dipole
pattern,
is
mainly
dominated
by
the
relative
source.
Moisture
contribution
this.
This
means
that
I
will
give
you
an
example
that,
like
in
eastern
brazil
in
lgm
the
blue
line,
the
large
contribution
of
moisture
is
from
south
atlantic
ocean,
but
at
the
hs1,
the
red
bar
for
the
red
bar
the
con.
L
L
We
mainly
focus
on
the
variability
mining
availability
of
tropical
south
american
american
summer
monsoon
and
they
are
highly
connected
to
the
itcc
shift,
my
southwest
antecedent
shift
and
yeah
and
connected
to
the
atmospheric
circulation
which
combined
together
to
produce
this
diaper
response
of
both
precipitation
and
natalie
team
yeah.
That's
all
for
my
presentation.
K
Great
thanks
so
much
yay
all
right,
so
I
don't.
Unfortunately
I
don't
think
we'd
have
time
for
questions,
but
again
there's
always
the
chat.
Okay.
So
I
think
our
schedule
had
us
taking
a
break
over
the
past
10
minutes
and
coming
back
at
12
20.
So,
just
as
a
suggestion,
you
want
to
take
a
five
minute
break
and
reconvene
and
start
again
with
our
csl
discussion.
Does
that
sound
reasonable
john
yeah.
B
K
B
A
B
B
B
B
I
E
E
I
E
Yeah
I
was
on
it.
So
are
those
sea
surface
temperature
all
from
calcite,
my
understandings
are
from
a
variety
of
sources.
I
guess
my
question
is
more
like.
If
you
pick
out
the
sea
surface
temperature
from
calcite
just
do
that
comparison,
while
the
model
proxy,
the
comparison
better
than
what
it
looks
like
now,.
I
B
I
Yeah,
maybe
I
can
just
very
quickly
for
a
second
show,
an
additional
slide
for
that.
If
that's
okay
here
that,
I
did
not
show
in
the
presentation
so
yeah,
depending
of
course
which
what
data
set
you
compare
it
to
so
this
is
an
extended
table
from
from
what
I
showed
and
if
you
compare
it
to
most
of
them,
are,
are
smaller,
but
you
also
find
published
values
that
are
a
lot
larger,
for
example,
from
tienetta
recently
with
icsm,
so
that
is
a
much
closer
mesh
to
those.
I
I
picked
the
most
recent
one
from
from
reconstructed
data
which
that
was
also
compiled
within
palmod,
and
for
that
the
vsm
was
a
better
match,
but
yet
so
it's
it's
pretty
broad.
I
mean
it's
after
all
these
years,
if
you're
not
sure,
especially
the
lower
latitudes.
What
was
the
what
was
the
rgm
anomaly
so
whether
that
was
actually
a
slight
warming
or
slight
cooling?
It's
it's
not!
We
don't
know
for
sure.
B
N
N
That's
our
main
hypothesis.
That
is
something
that
the
mothers
do,
that
some
cool
the
tropical
countries
and
others
don't
because
when
you
look
at
the
agreement,
it
goes
by
families
of
models
like
the
the
myroc
model
cannot
call
the
tropical
atlantic,
no
matter
how
you
force
it
and
the
young
models
tend
to
so
I
think
we
need
more
models
and
to
be
able
to
get
more
of
those
responses.
N
But
having
said
that,
I
have
a
hunch
that
the
glacial
boundary
conditions,
probably
matter
for
that
connection
between
the
tropical
atlantic
and
the
high
latitude
north
atlantic
to
propagate
that
cooling.
I
have
a
feeling
that
the
ice
sheets
are
important,
but
I'm
not
gonna
be
asking
to
run.
You
know
the
different.
You
know
different
lgm
simulations
and
then
host
them.
That
would
be
expensive,
but
it's
something
to
think
about.
N
If,
if
we
want
to
be
more
systematic
about
that,
and
so
I
think
we
need
to
buy
more
models
too,
but
it's
difficult
right,
because
you
need
an
lgm
simulation
which
are
expensive
and
not
every
modeling
center
can
do
glacial
simulation,
so
you're
kind
of
stuck
in
there.
I
wonder
what
other
people
think
about
that.
K
I
mean,
I
think,
we're
going
to
have
trouble
actually
getting
to
the
point
of
making
statements
about
which
model
physical
differences
are
important
for
these
types
of
things,
unless
we've
run
simulations
with
all
these
different
models,
so
yeah
tend
to
agree
how
to
do.
It
is
another
challenge
right.
N
Yeah,
probably
we
need
more
research
and
we
plan
to
do
it
understanding
this
connection.
It
seems
like
it's
a
structural
issue
in
models.
You
can
see
it
for
the
amo
there's
a
paper
by
marketing
and
others.
That
shows
that,
even
if
you
look
at
the
modern
day
and
amo
in
industrial
runs
models
also
separate
into
two
groups,
so.
N
N
N
B
Usually,
we
use
the
ice
6g
ice
5g
and
in
this
morning
we
saw
quite
a
few
talk
from
especially
from
the
paramount
group
looks
like
like
they
use
the
pterosaur
ice
sheet
reconstruction
and
in
their
studies
I'm
wondering
we
can
have
some
added
like
value
of.
If
we
explode
this
uncertainty
from
the
high
sheet
boundary
condition.
N
B
N
N
K
It's
like,
if
we're
talking
about
kind
of
big
picture,
uncertainties
particularly
like
you
know,
set
up,
maybe
like
use
of
csl
allocation
resources.
Then
it
seems
like
ice
coupling
is
one
or
you
know
water
hosing
sensitivities,
but
also,
I
think
the
resolution
thing
is
something
that
we
should
explore
in
a
lot
more
detail
too.
Like
you
know,
the
types
of
things
that
mary
grace
was
talking
about
were
super
interesting.
B
So
this
is
supposed
to
be
informal,
and
please
stop
me
if
you
have
any
questions,
so
I
will
quickly
do
a
very
brief
introduction
about
the
csl
proposal
and
the
upcoming
one
will
be
for
november.
First,
this
year
to
october
31st,
2024
and
the
deal
date
is
a
in
late
august.
I
need
to
double
check
that
and
based
on
the
current
machine
situation,
looks
like
that
location
might
be
partly
on
cheyenne
and
partly
on
the
ratio.
B
We
are
also
working
on
that,
so
we
proposed
10
million
hours
for
that
and
also
the
lgm
simulation
using
the
high
top
atmosphere
is
done
and
we
are
in
the
process
of
publishing
the
data
and
the
henrik
11
mild
water
event.
Using
csm2
is
also
done
and
ryan
led
a
small
location
performing
files
and
simulation
using
icsm
1.2.
B
J
J
A
A
In
our
lab
could
help
test
some
of
this,
but
using
you
know,
the
cam6
physics,
I
think,
would
be
interesting
if
we
could
pull
over
pop
and
then
whether
we
need
to
update
with
the
asteroid
box
model.
That
could
be
a
bit
of
a
challenge,
but
that
would
be
something
I'd
be
interested
in
seeing
and
possibly
helping
out
with
thanks.
B
D
Just
to
expand
on
that,
it's
the
land
model
that
I
don't
think
we
have
any
hope
of.
Unless
we
can
come
up
with
a
simpler,
you
know
bucket
model
or
something
we've
discussed
this
with
jessie,
but
so
far
it
hasn't
moved
forward.
So
that
would
be
great
if
clay,
for
example
or
others
want
to
kind
of
brainstorm
with
us,
but
so
far,
that's
the
land
model,
more
than
anything,
cam6
has
it,
but
they
haven't
been
tested.
J
B
Okay,
I
can
try
to
answer
your
question,
so
you
run
and
make
risk
simul
a
presentation.
You
already
see
that
they
are
working
on
high
resolution
files
and
simulation
and
we
have
a
separate
project
in
which
we
propose
to
do
high
resolution
so
roughly
a
quarter
degree
atmosphere
and
a
10th
degree
ocean,
high
resolution
simulation
of
clouds.
So
this
time
fully
coupled
and
early
yields
and
as
well
as
the
last
commercial
maxima
and
we
may
consider
to
add
a
metahorsen
if
we
have
the
computing
resources.
A
K
Okay,
sure
yeah.
I
guess
my
I'm
not
sure
if
this
is
a
comment
or
a
question
or
both,
but
it
follows
nicely
from
what
fanny
was
just
saying.
I'm
really
interested
in
the
high
resolution
aspect
as
well,
because
I
think
that
that's
really
necessary
for
doing
more
detailed
comparisons
of
models
and
proxies.
D
D
We
figure
50
years
at
that
resolution
will
cost
10
million
core
hours
on
the
ratios.
So
maybe
they'll
give
us
a
lot
of
time,
but
I
agree
with
you.
It
would
be
great.
I
should
also
comment
we
patterned
these
after
the
ihisp,
so
they
do
have
an
historical,
not
with
isotopes
and
a
future
scenario
run
that
we
could
take
advantage
of.
But
I
agree
I
just
we
need
to
be
proactive
and
maybe
there's
other
computing
systems
like
the
university
of
texas.
D
N
N
Those
are
probably
visible
with
a
few
decades
worth
of
computer
time,
and
I
think,
there's
so
much
evidence
that
high
resolution
matters
in
the
mean
latitudes
for
earth
interactions
over
western
boundary
currents
and
those
are
the
things
maybe
think
of
a
high
latitude
volcano
with
a
tropical
volcano
or
something
a
bit
not
idealized,
but
but
still
maybe
pave
the
way
to
one
day
do
high-res
simulations
of
the
last
million,
and
so
that's
one
comment
and
I'm
happy
to
continue
elaborating
on
this.
If
everyone
thinks
it's
a
good
idea.
N
B
N
D
D
N
D
That
said,
I
know,
there's
quite
a
bit
of
support
from
the
community
to
have
maybe
a
two
degree
ocean,
maybe
not
a
three
degree
but
a
two
degree,
and
maybe
with
you
know,
performance,
I'm
not
sure
duratio
is
going
to
be
processor
faster,
but
it's
going
to
have
more
and
maybe
work
on
that
by
the
software
engineers.
Maybe
you
know
so.
N
D
Yeah,
you
know
nsf
better
than
we
do.
I
mean
the
paleo
group
cannot
submit
to
the
paleo
program
at
nsf,
just
at
least
that
dave
perrardo
will
not.
But
if
the
and
my
sense
is
talking
to
the
model
developers
it's
time
to
get
hard
to
get
money
for
model
development,
but
with
a
good
science
emphasis,
I
don't
see
why
you
couldn't,
and
so
I
agree,
I
I
think
you've.
D
B
K
I
B
K
Sure
yeah,
I
guess
I
was
pondering
kind
of
similar
things
and
I
wonder
if
there's
an
opportunity
here,
I'm
not
sure
which
version
of
the
model
I'm
talking
about
exactly
maybe
zsm2
or
3
but
like
if
we
can
think
about
constructing
perturbed
physics
simulations
in
in
paleo
contexts.
That
would
correspond
to
things
that
we
expect
might
be
at
least
pre-production
versions
of
the
new
models
to
see.
K
If
there's
emergent,
behaviors
that
corresponds
to
you
know
21st
century
simulations,
maybe
we
can
kind
of
get
ahead
of
the
inevitable
development
cycle,
where
paleo
kind
of
gets
put
to
the
side
by
thinking
about
likely
parameters
that
are
being.
You
know,
under
development
now
and
pick
those
for
some
strategic
time
slice
experiments
which
I
think
would
relates
to
what
pedro
was
saying,
but
it
would
depend
on
the
science
question.
B
D
So
I'm
just
thinking
for
the
csl
proposal
you're
doing
that
for
cam.
Maybe
it
makes
sense
to-
and
I
know
they
have
one
for
the
land
to
you
know
maybe
include
that
in
the
next
proposal.
I
don't
know
if
it's
possible,
maybe
esther
or
jean
you
know
to
do
the
same
for
the
ocean
or
does
it
make
sense
with
mom.
K
A
Know
I
haven't
heard
of
any
talk
of
such
a
an
effort
in
mom
and
kind
of
sounds
a
little
scary
to
me,
actually
knowing
the
oceanographers
involved.
So
but
I
mean
never
say
never.
I
guess
I
haven't
heard
of
them
at
all
talking
about
this
I
mean
there
was
a
talk,
but
not
relative
to
mom
by
laura
zanna
on
using
machine
learning
to
look
at
ocean
parametrizations.
B
D
Thank
you,
john
it'd
also
be
useful
if
you
and
ren
and
samantha,
or
one
of
you
tried
to
find
out
from
gokan
about
what
the
maybe
he
doesn't
know
yet,
but
it's
really
hard
to
plan
what
to
do.
Unless
you
know
it,
and
I
should
say
in
the
past,
he
tells
us
what
our
allocation
is
and
then
we
fit
what
we
can
fit
in
it.
So
you
know
it.
D
You
know
the
high
recipe
great,
not
just
for
less
millennium,
for
but
for
deeper
time
you
know,
but
but
yeah,
I
just
don't
know
what
he's
going
to
give
us
so
be
great.
If
you
could,
if
he
has
any
idea
yet,
and
maybe
he
doesn't
because
the
ratio
keeps
getting
postponed
in
terms
of
its
delivery
and
then
release
to
to
the
community.
B
E
I
I
guess
the
other
question
I
had
is:
are
there
any
progress
on
like
developing
faster
spin-ups?
I
remember
ocean
model.
Well,
obviously,
we
are
all
tied
to
the
ocean
models
clean
up
for
any
of
the
deep
time,
paleo
or
even
quaternary
simulations.
So
I
recall
back
in
the
days
hearing
like
a
lot
of
talks
about
the
accelerators
thing
up
for
the
ocean
model.
So
is
there
any
development
on
mom
and
perhaps
that's
something
we
could
try
as
well.
B
D
D
E
Yeah,
I
was
many
worrying
about
john
mentioned.
The
new
model
like
csm3,
is
going
to
be
20
times
more
expensive
and
based
on
our
past
experience.
If
the
ocean
model
is
the
one
is
the
reason
that
we
have
to
run
the
simulation
for
a
long
time
to
reach
equilibrate.
Perhaps
we
need
to
think
about.
There
are
some
intermediate
ways.
We
can
spin
up
ocean
faster.
A
Solver
is
only
works
right
now
for
passive
tracers,
not
for
like
temperature
and
saline,
which
are
the
dynamic
density,
go
into
computing.
The
density
in
the
ocean
model
so
and
I'm
not
even
sure
they
work
for
all
of
the
passive
tracers
currently,
but
he's
been
using
it
up
using
it
to
spin
up
the
carbon
pools
in
the
ocean
bgc
which
take
a
really
long
time
to
come
to
equilibrium.
So.