►
From YouTube: CESM Workshop: Paleoclimate Working Group
Description
The 26th Annual CESM Workshop will be a virtual workshop with a modified schedule on its already scheduled date. Specifically, the virtual Workshop will begin with a full-day schedule on 14 June 2021 with presentations on the state of the CESM; by the award recipients; and three invited speakers in the morning, followed by order 15-minute highlight and progress presentations from each of the CESM Working Groups (WG) in the afternoon.
On 15-17 June 2021, working groups and cross working groups have half-day sessions, some with presentations and some that are discussion only.
A
B
C
A
To
go:
okay,
hello,
everyone
welcome
to
the
paleoclimate
working
group
session,
I'm
welcoming
you
on
behalf
of
myself,
esther
brady,
I'm
a
co-chair
and
arne
wingath
who's.
The
other
co-chair-
and
I
just
put
up
this
slide
just
to
give
us
a
few
logistics
about
the
meeting
today.
A
A
I
want
to
let
you
know
that
we're
recording
in
simulcasting
on
youtube
this
meeting
as
you
accepted
to
join
our
talks,
are
really
short:
they're
10
minutes
the
slots,
and
so
it'd
be
great.
If
you
could
keep
your
presentations
to
about
eight
minutes,
so
we
have
time
for
maybe
one
question
and
for
some
transitioning
and
then
we
can
utilize
the
chat
box
for
discussion
and
questions,
and
the
speakers
can
dialogue
there
with
the
rest
of
the
participants.
A
Please
use
the
raise
your
hand
button
to
ask
a
question
and
the
chat
is
going
to
be
saved.
A
All
discussions,
public
and
perhaps
private,
although
I
haven't,
found
that
to
be
the
case-
may
appear
in
the
saved
text
file.
A
Please
unshare
your
screen
at
the
end
of
your
talk,
so
we
can
transition
fast
and
keep
your
microphone
muted
when
you're
not
presenting
or
asking
a
question,
and
then
I'd
like
to
let
you
know
that,
following
this
meeting
we're
going
to
have
a
informal
meetup
networking
session
stephanie
who
will
be
facilitating
us
behind
the
scenes
here?
Stephanie
stephanie,
shearer
she'll
leave
us
at
that
point.
A
So
if
you
have
any
need
some
troubleshooting
done,
you
can
contact
stephanie
through
the
chat.
I
guess
and.
A
Okay,
all
right,
so
one
thing
I'd
like
to
make
you
aware
of
is
thursday
afternoons
alternate
earth
sessions.
A
We
have
a
participant
today,
colin
goldblatt,
a
professor
at
the
university
of
victoria
who's,
giving
a
talk,
an
invited,
talk
there,
and
it
should
be
a
really
interesting
session
to
tune
into.
So
I
wanted
to
make
sure
that
you
are
aware
of
this
and
we
did
have
a
speaker
who
wasn't
able
to
give
their
presentation
today
due
to
some
unforeseen
circumstances.
A
So
I
have
a
little
bit
of
time
and
I
thought
I
would
kind
of
give
a
brief
update
on
some
of
the
things
that
we're
doing
behind
the
scenes
here
at
ncar
in
the
paleoclimate
working
group.
So
I
have
a
couple
of
slides.
A
Here
first,
I
just
want
to
let
you
know
of
what
we're
working
on
currently
and
what
we
have
plans
for
with
our
current
csl
allocation
that
started
back
in
november.
A
I
did
put
the
link
to
the
csl
proposal
in
the
at
the
top
of
the
chat
and,
if
you're
curious,
and
want
to
read
a
little
bit
more
about
what
we're
doing
in
case
you're
interested
in
getting
involved
in
some
way.
A
You
can
check
out
that
link
and
go
there
and
read
more
about
our
plans,
but
so
first
is
the
paleo
calibrated
cesm2
work
that
jean
will
talk
a
little
bit
more
about
today,
but
he
also
gave
a
talk
in
the
atmosphere
model
working
group
session
yesterday,
and
then
we
have
two
transient
multi-millennial
simulations
that
we
are
have
and
will
be
performing
with
cesm2,
coupled
to
the
schism
to
ice
sheet
model.
The
first
is
the
last
interglacial
climate
ice
sheet
project
that
we
talked
a
little
bit
about
in
our
updates.
A
That
is
a
joint
landice,
paleoclimate
working
group
project
and
then
the
next
is
another
joint
landice
working
group
project
to
simulate
the
transient
holocene
starting
at
9
000
years
ago.
With
the
csm2
system
2
ice
sheet,
we
have
a
control
going
and
we're
just
about
to
get
a
transient
run,
started
we're
also
starting
to
work
on
running
or
simulating
the
last
glacial
maximum,
with
cesm2
coupled
to
the
wacom
6
middle
atmosphere
model,
and
this
is
something
that
john
zhu
has
also
started.
A
Working
on
using
the
paleo
calibrated
mg2
modifications
that
he
made
and
we're
working
on
this
jointly
with
the
whole
atmosphere,
working
group
members
who
who
are
at
in
car
and
then
lastly,
we're
doing
some
idealized
melt
water
simulation
for
the
heinrich
stadial
11
event,
and
that
is
planned
to
be
about
4
000
years
long,
but
we're
at
year
1000.
Currently
in
the
future.
A
We
want
to
get
going
with
these
high
resolution,
paleo
climates
with
cesm
1.3
running
cam
at
the
quarter
degree
in
the
sc
dykor,
with
the
couple
to
the
one
degree
pot
model.
A
We
also
are
planning
to
use
the
csm2
wacom
6
configuration
for
a
warm
paleo
climate
like
the
eocene,
marcus
laverstrom,
is
with
the
landice
working
group.
We
have
a
project
to
start
a
glacial
inception
simulation
starting
at
the
end
of
that
transient.
A
Lig
run
and
then
later
on,
towards
the
end
of
the
allocation,
we
hope
to
get
started
using
a
mon
coupled
to
cam
to
run
the
lgm,
which
will
be
a
good
test
for
that
configuration
and
then
last
we
are
still
planning
to
get
watered
isotope,
tracers
and
clm,
which
is
the
component
currently
that
it's
not
configured
in
at
the
moment.
So.
A
Working
on
it's
taking
quite
a
long
time,
because
it's
a
lot
of
data
to
move
but
we're
migrating,
our
older
cases
off
the
hpss
to
a
tape
archive
system,
that's
going
away
at
ncar
to
the
glade
campaign,
storage
space,
which
is
disk
disk
space
and
the
the
hpss
goes
away
in
october,
and
so
we're
and
it's
very
slow
to
move
data
off
of
it,
especially
old
data.
A
So
we've
been
betty
and
I
and
nan
rosenbloom
and
christine
shields
have
been
taking
a
look
at
our
at
what
we
have
on
the
hpss
and
moving
it
to
campaign
space
for
the
community.
So
this
is
just
a
list
of
what
we've
been
working
on
so
far.
We
have
simulations
from
earlier
pmf
projects,
the
pmf
2
with
the
ccsm3
and
the
pimax3
with
the
ccsm4.
A
We
have
additional
simulations
that
we
ran
with
paleo
working
group
time,
the
cc
with
ccsm3
and
ccsm4.
There's
a
number
of
blast.
Interglacial
runs
through
the
last
interglacial,
the
ccsm3
there's
some
melt.
Water
sensitivity
runs
starting
from
a
base
state
of
the
last
glacial
maximum
and
then
a
mid-holocene
case
for
the
8.2
k
event.
A
So
there's
the
cesm1
collections
of
idealized,
four
scenes
from
michael
herb
sensitivity
runs
and
then
clay,
tabor's,
n
member,
forcing
simulations
with
the
isotope
enabled
cesm2
1.2,
and
then
there
are
the
ccsm3
and
the
new
cesm
1.3
simulations
of
the
de-glacial
period
with
the
trace
the
lewin
betty's
trace
project.
A
So
the
one
point,
the
cesm
1.3
version
was
run
with
the
isotope
enabled
models,
and
then
we
have
cesm
1
collections
for
the
last
millennium
ensemble
and
then
the
mid
pliocene
sensitivity
runs
that
ren
fung
has
contributed.
Many
sensitivity
runs
too.
She
ran
most
of
these
runs.
F
I
was
just
wondering:
are
there
plans
to
have
like
a
carbon
cycle
coupled
csm
version?
For
you
know,
carbon
cycle
applications
ready.
A
Well,
I
in
my
update
talk,
I
mentioned
some
work:
that's
going
on
outside
to
put
the
carbon
isotope
tracers
in
cam,
that's
being
done
by
the
university
of
bern
researchers
altogether
and
fortunate
yost
and
we
hope
to
get
those
into
cam
and
then
at
that
point
I
believe
it
would
mostly
be
possible
to
get
of
the
carbon
the
couple
carbon
cycle
going,
I'm
not
sure
if
it's
in
the
runoff
model,
yet
I
think
they're
still
potentially
using
data
sets
to
prescribe
the
carbon
isotope
fluxes
to
the
ocean
and
the
carbon
fluxes
to
the
ocean
as
well.
A
So
that
would
be
one
remaining
model
that
I'm
aware
of
that
potentially
wouldn't
have
them
for
a
full
coupled
system
but
yeah.
I
think
it's
something
that
it's
in
our
research
priorities
for
you
know
down
into
the
future
and-
and
I
did
give
well-
I
can
show
it.
Let
me
show-
I
don't
know
if
we
have
time.
No,
we
don't
but
anyway,
it
is.
A
It
is
something
that
that
is
in
our
mind
for
the
future,
but
I
don't
think
we're
quite
there
yet
with
the
work
that
we
need
to
bring
in
from
the
external
collaborators.
So
yeah,
okay,
so
is
our
first
speaker
is
jen.
Kaye
is
john
k
here,
yes,
hi
jen
k
is
a
professor
at
the
university
of
colorado
here
in
boulder,
and
she
is
speaking
on
what
causes
20
more
two
times:
co2
global
warming
than
a
half-time
co2
global
cooling
in
the
community
earth
system
model.
B
Yeah
good
morning,
thanks
for
the
opportunity
and
to
speak
briefly
about
this
work
that
I've
done
in
collaboration
with
the
folks
listed
on
the
slide.
I
want
to
especially
call
out
jason
chalmers
who's,
now
a
graduate
student
at
the
university
of
california,
santa
barbara
who
did
much
of
this
work
during
the
pandemic
last
summer
and
so
really
kudos
to
him
for
getting
that
off.
This
project
really
off
the
ground,
and
the
question
is
really
big.
B
Why
the
climate
and
cloud
responses
to
idealized,
greenhouse
warming
and
cooling
differ,
and
the
reason
we
ask
this
question
is
because
it's
a
fun
big
picture
question
to
work
on
with
an
undergraduate
student
over
the
summer
for
sure,
but
also
because
these
idealized
greenhouse
warming
and
cooling
experiments
have
really
been
called
out
as
being
important
as
a
constraint
on
equilibrium.
B
Climate
sensitivity,
in
particular
the
amount
of
cooling
that
you
get,
could
be
a
constraint
on
how
high
equilibrium
climate
sensitivity
might
go,
and
so
these
experiments
are
rather
idealized,
so
they're
a
little
easier
to
understand
than,
for
example,
the
least
last
glacial
maximum,
where
a
lot
more
things
have
changed
so
just
quickly
to
introduce
the
experiments,
we
basically
have
run
the
fully
coupled
one
degree,
ces
m1
version
for
150
years.
B
So
this
was
kind
of
a
first
sort
of
result.
That's
that's
kind
of
interesting
to
see.
A
lot
of
this
response
happens
in
the
first
couple
of
decades.
That's
consistent
with
these
fast
time,
scales
that
are
important,
and
then
this
long
tail
you
can
see
these
such
simulations
are
way
out
of
equilibrium.
They've
only
been
run
for
150
years,
so
the
the
ocean
circulation
is
is
not
quite
an
ocean
here.
Tick
is
not
quite
equilibrated,
so
really
basic.
Why
are
we
seeing
more
global
warming
than
global
cooling?
B
We
diagnosed
the
effective
radiative
forcing
and
we
found
that
is
about
10
larger
for
carbon
dioxide
doubling
than
for
a
carbon
dioxide
having
in
cesm1.
B
We
also
looked
at
radiative
feedbacks
globally
and
unsurprisingly,
cloud
feedbacks
end
up
being
very
important
for
some
of
the
differences
that
we're
seeing
under
warming
and
cooling
and,
in
particular,
the
shortwave
cloud
feedback
being
important
for
more
warming
than
cooling,
with
some
compensation
from
the
lapse
rate
feedback.
B
So
if
we
look
at
maps
of
what
this
looks
like
so
here's
just
the
warming
and
the
cooling,
we
see
a
lot
of
similarities
in
the
patterns,
stronger
responses
at
high
latitudes
and
over
land,
even
some
of
the
patterns
in
the
tropical
oceans
look
sort
of
similar
when
you
have
and
when
you
cool
just
opposite
in
sign.
So
if
we
look
at
the
difference
and
then
on
the
bottom,
the
pattern
difference
which
is
just
normalized
by
the
response
globally
response,
we
can
see.
B
You
know
some
interesting
features
that
relate
to.
Why
there's
this
asymmetry,
we
see
differences
in
the
high
latitudes,
of
course,
where
sea
ice
has
been
lost
and
gained
so
strong
differences
associated
with
that,
but
then
also
some
other
more
intriguing
differences,
for
example,
more
warming
over
land
than
cooling
over
land.
So
I'm
not
sure
entirely
sure.
What's
behind
that
and
then
also
some
some
pattern
differences,
especially
in
the
tropical
pacific,
we
can
see
this
interesting
cooling
feature
off
the
coast
of
north
america.
B
So
we
looked
at
the
maps
of
the
radiative
feedbacks
to
help,
explain
the
surface
pattern,
differences
and
so
you'll
probably
be
most
familiar
with
what
the
feedbacks
look
like
in
response
to
a
carbon
dioxide
doubling
here,
we
just
also
show
the
difference
between
the
doubling
and
the
having,
and
you
can
see
you
know
the
high
latitudes
it's
these
coupled
surface
albedo
and
lapse
rate,
feedbacks
that
are
really
amplifying
the
surface
temperature
response
associated
with
cs
loss
and
then
the
cloud
feedback
difference.
You
know
more
positive
cloud
feedback
over
much
of
the
globe.
B
There
is
an
exception
again
off
the
coast
of
north
america
here
where
we
actually
see
an
opposite
sign
where
so
the
cooling
is.
The
feedback
is
somewhat
stronger,
so
yeah
just
some
here.
Some
notes
also
important
to
note
that
the
southern
ocean
and
the
mid-latitudes
and
the
land
cloud
feedbacks
appear
more
positive
under
doubling
than
under
having
in
the
tropical
pacific.
We
see
many
similarities
associated
with
the
surface
wind
response
that
look
similar
under
doubling
and
having.
B
But
when
we
look
at
the
pattern
difference
between
doubling
and
having,
we
see
a
strong
role
for
the
cloud
feedbacks.
In
particular,
the
cloud
feedbacks
are
important
for
sort
of
amplifying
this
stronger
warming
in
the
western
tropical
pacific
and
then
for
the
the
sort
of
rich
spatial
structure
in
the
eastern
tropical
pacific,
which
tends
to
be
somewhat
delayed
in
its
response
due
to
the
the
ocean
circulation.
B
So
we
can
look
at
this
kind
of
slow
time
scale
response,
so
here
we're
just
looking
at
how
things
are
evolving
over
the
sort
of
the
last
50
years
of
these
simulations.
It's
not
particularly
slow
for
the
paleo
climate
working
group,
but
it
is
a
slow
time
scale
in
these
particular
experiments,
and
we
see
some
really
interesting
contrasts,
for
example,
over
the
southern
ocean,
where
warming
tends
to
be
somewhat
delayed.
B
So
we
have
a
relatively
large
slow
fraction
under
having
the
the
cooling
actually
appears
to
be
relatively
fast,
and
some
of
that
could
be
due
to
the
advancing
sea
ice
edge,
but
also
differences
in
ocean
stratification
playing
a
role
there.
B
B
So
the
ocean
dynamics
and
the
cloud
feedbacks
are
really
important
for
explaining
the
tropical
sst
response,
time,
evolution
and
maybe
just
for
the
the
sake
of
time,
I'm
just
going
to
skip
through
this,
but
just
to
say
that
the
cloud
feedbacks
again
are
playing
a
really
important
role
in
shaping
these,
both
slow
and
fast
time
scales
and
our
experiments.
B
Okay,
so
now
moving
on
to
something
kind
of
fun.
Actually,
this
month's
nature,
climate
change
cover,
features
cloud
feedbacks.
Whether
clouds
will
warm
or
cool
the
planet
under
climate
change
is,
of
course,
very
uncertain
and
there's
two
studies
in
this
issue
that
are
quite
relevant
to
cloud
feedbacks.
One
of
them
in
particular,
I
want
to
call
out,
has
to
do
with
how
warm
and
cold
clouds
precipitate
and
what
this
study
found
moment
shot
at
all.
B
Is
that
basically
we're
missing
a
change
in
precipitation
efficiency
and
that's
sort
of
a
missing
negative
feedback,
and
so
it
really
calls
into
question,
for
example,
these
really
large
equilibrium
climate
sensitivities
that
we're
seeing
in
the
latest
generation
of
cmm6
models,
including
cesm2
and
says
these-
are
not
the
final
word
on
this
there's
plenty
of
reasons
to
think
that
there
are
other
processes
that
are
quite
relevant.
B
That
could
lead
to
negative
feedbacks
that
would
make
these
equilibrium
climate
sensitivity
values
come
down,
so
we
were
interested
in
repeating
our
experiments
without
the
cloud
feedbacks,
because
cloud
feedbacks
tend
to
be
a
real
source
of
confusion
and
just
a
lot
of
different
things
going
on
so
one
more
minute,
yeah
I'll
be
pretty
quick
here.
So
we
just
we
disabled
the
cloud
radiative
feedbacks
just
to
point
out
that,
even
after
we
disabled
the
cloud-rated
feedbacks,
we
still
have
20
more
global
warming
than
global
cooling.
B
So
there's
some
interesting
compensation
going
on
there,
but
that
number
seems
to
be
rather
robust
globally
and
that
we
see
differences
in
sort
of
this
cloud
influence,
which
is
the
difference
between
the
control
simulations
and
the
cloud
block.
Simulations
and
you
know,
for
example,
clouds
are
amplifying
responses
at
high
latitudes
when
they're
enabled,
whereas
we
don't
see
any
evidence
for
that
when
we
look
at
these
diagnostic
cloud
feedbacks.
B
And
finally,
I
think
it's
really
interesting
that
this
cloud
influence
affects
the
response
magnitude
more
than
the
response
pattern.
So
it
really
emphasizes
that
clouds
are
coupled
to
the
circulation
to
other
feedbacks,
and
so
you
know
just
diagnosing
cloud-rated
feedbacks
offline.
After
the
fact
it
gives
you
a
really
different
story
than
what
you
learn
when
you
actively
disable
them
in
the
model.
B
A
Thank
you,
jen.
We
might
have
questions
in
the
chat.
I
think
just
to
keep
us
on
time.
That
was
really
interesting.
B
A
So
our
next
speaker
is
marcus
yoko,
a
professor
at
the
niels
bohr
institute.
H
H
H
So
what
we
want
to
understand
are
the
millennials
k,
millennial,
scale,
co2
fluctuations
that
we
see
in
the
ice
core
record
and
I'll
show
some
parts
of
the
ice
core
direct
right
here
shown
here
are
50
000
years
of
temperature
in
blue,
you
see
every
couple
of
thousands.
We
have
very
strong
warming.
These
are
the
transfer
ashkar
events
followed
by
a
slight
cooling
preceding
the
warming,
is
what
is
called
as
a
stadial,
so
those
are
the
cold
periods
within
the
glacials
and
associated
with
these
schedules.
H
We
have
antarctic
warming
so
shown
here,
and
so
this
has
been
well
explored
now
also
with
the
use
of
ccsm,
and
what
is
less
understood,
though,
is
the
connection
between
these
signals
here
and
the
co2
variant.
So
we
see
that
whenever
the
template,
whenever
there's
a
stedial
and
the
temperature,
is
increasing
antarctica,
we
see
it
increase
in
the
co2.
H
This
gave
rise
to
a
little
hypothesis
that
the
co2
is
driven
or
the
c2
signal
is
driven
by
the
warming
in
an
article.
Although
we
should
note
that
the
increase
in
co2
here
lasts
longer
than
the
increase
in
temperature,
so
the
cooling
starts
and
then
the
co2
tends
to
increase.
H
What
we
did
is
we
used
ccsm,
it's
this
by
three
set
up,
so
we
get
something
like
a
hundred
years
per
day
in
our
cluster.
It
is
a
setup
guido
where
the
retti
has
published
much
of
this
on
the
physical
component.
What
we
did
is
we
added
biogeochemistry
to
this
physical
setup
yeah,
so
I
won't
have
time
to
go
through
all
the
details,
but
we
use
this
setup
and,
first
of
all,
we
want
to
check
if
it
reproduces
the
observations.
H
This
is
shown
here
in
the
upper
left
corner.
We
have
here
like
a
3000
year
segment
of
a
9000
year,
integration
that
we
did.
It
features
two
schedules.
We
look
here
at
the
black
line
that
shows
the
temperature
over
greenland
museum
is
very,
very
rapid,
warping
and
like
in
the
observations
that
has
an
amplitude
of
some
50
degrees
celsius.
H
During
the
schedules
the
antarctic
temperature
increases.
I
hope
you
can
see
my
arrow
it's
a
bit
smaller.
I
should
have
increased
it,
but
the
blue
line
shows
the
temperature
increase
in
an
arctic
term
again
in
line
with
your
observations.
It
also
is
about
a
100
year.
Time
lag
between
the
onset
of
the
stadial
interstellar
and
the
cooling
in
an
article
which
has
been
is
very
similar
to
the
records
from
inferred
from
volcano
records
recently.
H
As
soon
as
the
schedule
begins,
we
get
a
increase
in
atmospheric
co2
concentration
of
about
1
ppm
per
100
years,
so
this
is
the
same
rate
that
we
have
in
the
observations
our
stereos
are
rather
short
in
this
particular
experiment.
It's
only
four
hundred
years,
so
we
get
a
co2
response
of
three
to
four
ppm,
which
is
in
the
lower
end.
But
again,
our
schedules-
a
bit
shorter
associated
with
these
schedules,
is
an
aimoc
response
which
is
shown
here
which
has
been
documented
before
and
on
the
lower
left.
H
An
important
response
is
the
precipitation
response,
so
there
are
large
precipitation
shifts
associated
with
the
which
is
shown
in
red,
and
one
important
part
is
a
drying
over
africa,
for
example,
and
there
are
slight
changes
in
the
wind
stress
over
the
southern
ocean,
there's
about
a
five
percent
reduction
in
general,
and
this
helps
us
then
to
explain
the
stated
anomalies
of
the
surface-to-air
carbon
fluxes,
which
are
shown
here
on
the
right
side.
H
H
What
we
decided
to
do
is
to
lock
together
the
north
pacific,
the
north
atlantic
response
together,
so
they're
highly
correlated
to
the
amok.
The
basic
effects
here
are
you
have
an
increased
sea
ice
cover
which
reduces
the
uptake?
We
have
a
cooling
here
which
increases
the
uptake,
so
those
are
compensating
effects
on
both.
H
Also,
reduction
of
the
amore
is
aggressors
earlier
study
showcase
as
well
associated
with
the
collapse
of
the
marine
productivity,
because
the
nutrients
are
just
simply
lacking.
H
So
those
are
the
three
major
responses
in
the
northern
hemisphere
and
the
southerners
hemisphere.
The
total
response
is
a
bit
smaller
and
again
we
have
a
compensating
effect
of
basically
less
subduction
of
carbon-rich
water
here
in
the
southern
ocean,
which
gives
rise
to
an
atmospheric
co2
signal,
and
then
we
have
less
outgassing
here
in
the
tropical
pacific,
mostly
but
throughout
the
tropical
ocean.
H
So
we
lob
this
together,
then
in
three
different
components.
We
have
a
land
component
caused
here
by
the
drying
and
the
two
different
ocean
components,
and
we
can
integrate
these
carbon
flux
anomalies
to
try
to
explain
what
explains
the
co2
response
that
we
see
again.
These
black
lines
here
indicate
the
transfer
events
or
the
sudden
warming
in
greenland,
preceded
by
them
other
stadials,
and
we
see
that
most
of
the
stereo
response
in
co2
comes
from
the
land.
H
There's
a
little
contribution
of
the
northern
hemisphere
of
the
north
atlantic
in
the
north,
pacific
and
the
southern
ocean,
and
the
tropics
act
to
counterbalance
this
now.
The
interesting
part
here
is
this
large
overshoot
here
that
as
soon
as
the
northern
hemisphere
warms
again,
there
is
a
massive
outgassing
co2,
which
then
explains
the
delayed
response
that
we
saw
earlier
that
the
co2
is
rising,
even
though
the
temperature
has
equilibrated
it
or
has
returned
to
a
normal
state.
H
Our
signals
are
a
bit
smaller.
I
think
we
have
the
right
increase
of
co2
in
the
right
race,
but
our
squash
events
are
a
bit
short,
so
we
helped
us
out
with
some
housing
and
shown
here.
Is
this
co2
response
giving
a
done
squarespace?
The
blue
one
is
the
one
that
we
discussed
before.
We
have
a
three
to
four
ppm
increase.
H
We
can
prolong
these
stereos
by
hosing,
so
in
the
middle
of
the
schedules,
we
simulate
our
hydrogen
events
by
adding
slight
hosing
to
the
north
atlantic.
This
prolongs
the
stereos
for
another
800
years,
which
gives
us
a
larger
co2
signal,
and
the
nice
thing
is
that
the
rate
of
co2
increase
does
not
change,
and
this
is
very
much
in
line
with
observations
where
high
next
stereos
and
normal
stations
do
not
have
a
different
increase
in
co2.
H
One
third
experiment:
we
did
this:
we
reduced
the
iron
deposition
or
the
aleona
flux
of
iron.
Some
of
the
ideas
related
to
the
southern
ocean
suggest
that,
since
it's
warmer,
there
should
be
less
dust
deposition
in
the
southern
ocean,
which
would
reduce
the
productivity
and
increase
the
atmospheric
co2
concentration.
This
was
something
we
investigate
here
with
our
reduced
iron
flux,
experiment.
H
H
The
blue
here
shows
an
increase
or
reduced
uptake
in
the
southern
in
the
north
atlantic
and
pacific,
which
we
expect
through
a
hosing
experiment.
But
one
of
the
surprising
things
is
that
the
biggest
signal
that
we
get
from
the
reduced
iron
experiment
does
not
happen
in
the
southern
ocean.
There's
very
little
response
in
the
southern
ocean,
but
we
do
get
this
in
the
north
pacific
and
it
helps
to
look
at
the
spatial
pattern
of
our
differences.
So
this
is
the
difference
in
the
stable
carbon
fluxes
between
iron
and
hose.
H
The
blue
line
over
lane
here
shows
the
pattern
of
the
urine
duster
position
and
we
see
in
areas
where
we
have
large
dust
deposition.
We
do
have
indeed
a
stronger
response,
but
wherever
we
have
increased
a
reduced
uptake
of
iron,
we
also
have
reduced
outgassing.
So
there
are
large
compensating
effects
at
work
that
we
get
very
little
net
contribution
from
the
southern
ocean
and
the
tropics
again,
but
the
main
net
contribution
happens
to
be
from
the
north
pacific.
H
H
A
Thank
you,
marcus.
That
was
very
interesting.
I
don't
think
we
have
time
for
questions
I'll
have
to
be
done
in
the
chat
and
or
we
should
get
going
with
our
next
speaker,
renfeng,
an
assistant
professor
at
the
university
of.
C
Connecticut,
hey
I'll,
get
started
thanks
everyone
for
joining
on
today,
and
thank
you
to
esther
and
arne
for
organizing
us.
So
I
was
set
out
to
talk
about
the
source
for
the
enhanced
earth
system,
sensitivity
in
the
cesm2,
but
I
end
up
having
more
questions
and
answers
that
I
have
on.
So
I
thought
I
would
share
with
you
what
I'm
working
on
and
what
are
the
directions
that
I'm
thinking
about
going
with
this
today.
C
Before
I
start
talking
about
our
system
sensitivity,
I
just
want
to
remind
everybody
about
the
definitions
and
distinctions
between
the
equilibrium,
climate
sensitivity
and
earth
system
sensitivity,
so
equilibrium,
climate
sensitivities.
This
is
the
ecs
that
we're
most
familiar
with
this
is
the
global
mean
surface
warming,
air
warming
to
a
doubling
of
co2.
C
So
usually
it
was
calculated
by
holding
the
ice
sheet
vegetation,
amsterdam,
composition
as
prescribed
and
using
the
step
ocean
model.
So
the
ocean
circulations
also
assumed
to
be
constant.
C
So,
on
this
time,
scale
chart
made
a
couple
years
ago
in
the
pages
news
by
gary
schmidt,
so
ecs
incorporate
feedbacks
happening
from
decades
to
maybe
up
to
a
century
or
so,
but
when
we're
really
looking
at
payload
climate
time
scale
we're
dealing
with
our
system
sensitivity.
This
is
the
warming,
the
amount
of
surface
warming
by
incorporating
the
long-term
feedbacks.
C
C
There
is
a
international
modeling
project
going
on
trying
to
simulate
this
time
period.
It's
called
plexing
modeling
the
comparison
project
and
we're
in
the
phase
two
of
the
project
to
the
top.
Here
shows
what
are
the
best
estimates
for
the
biome
and
ice
sheet
distributions
for
this
time
interval,
and
I
want
to
highlight
a
couple
places.
C
This
is
what
the
greenland
ice
sheet
look
like
during
this
time.
It's
very
much
limited
to
the
eastern
side
of
the
island,
and
this
is
what
the
antarctic
ice
sheet
looked
like.
So
you
see
the
loss,
the
massive
loss
of
west
antarctic
ice
sheet,
and
generally
you
see
the
northward
shift
in
biome
types.
C
So
when
models
are
prescribed
with
co2
and
weighs
those
biome
distributions-
surprisingly,
not
just
that
this
simulates
or
not
just
that
the
model
have
different
xcs,
but
they
also
shows
different
amount
of
amplification
of
warming
due
to
prescribed
vegetation
ice
sheets.
C
So
due
to
the
radiative
perturbations
from
those
slow
responding
systems,
and
so
this
column
shows
the
ess
to
ecs
ratio
which
quantifies
the
warming
amplification
in
those
models
due
to
vegetation
ice
sheet-
and
you
can
see
it
varies
from
not
much
in
some
models
to
a
whole
bunch
in
some
other
simulations
and
it's
uncorrelated
with
ecs,
I
set
up
a
series
always
set
up
a
series
simulations
to
look
at
what,
if
we
just
look
at
one
aspect
of
changes
in
forcing
conditions
from
vegetation
combined
with
aichi,
so
this
long-term
earth
system
feedback
or
just
the
co2
or
just
the
geography
and
topography
changes.
C
And
surprisingly,
the
vegetation
actually
actually
contributes
the
most
warming
for
in
the
midpiece
in
simulations.
C
The
good
news
is
geography
and
topography
has
felt
very
little
effect,
so
that
reduces
the
complexity
a
little
bit
and
this
warming
primarily
happens
in
the
high
issued
in
the
arctic.
C
In
the
canadian
arctic,
greenland
and
as
well
as
the
western
antarctic,
it
also
has
some
pretty
unique
continental
warming
pattern
going
on
in
the
across
eurasia
continent
and
north
africa.
C
If
you
look
at
precipitation
the
vegetation
ice
sheet,
forcing
actually
has
really
strong
precipitation
response,
especially
compared
to
co2
and
geography
and
topography,
and
there's
a
really
strong
response
in
the
north
and
subtropics
from
sahel
to
the
southeast
asia.
C
We're
actually
analyzing
those
results
in
conjunction
with
the
rest
of
the
plymouth
2
models
and
we're
having
a
paper
in
review
discussing
this
response,
because
this
is
actually
seeing
in
the
proxy
records.
The
proxy
records
can
actually
validate
this
response.
C
So
we're
curious
so
how
much
radiative
perturbations
we're
actually
getting
from
those
different
forcing
conditions
and
similar
to
what
jen
stock
has
shown.
I
did
a
a
series
of
prescribed,
sst
and
cis
experiment
to
constrain
this
effective
radiative
four,
so
this
is
top
of
the
troposphere
relieve
perturbation
introduced
by
forcing
agent.
So
erf
does
include
rapid
adjustment
from
like
water,
vapor
depth
rate
happening
at
weeks
or
months
or
even
the
first
few
years,
and
clouds
and
a
trick
is
when
we
do
the
prescribed
ssdncs
simulations
we
didn't
see.
C
There
is
a
surface
temperature
response,
so
we
do
have
to
which
compensate
the
radiated
forcing
so
we
do
have
to
adjust
for
that,
and
this
has
been
done
in
the
radio,
forcing
map
for
stem
up
six
and
as
well
as
in
john's
paper,
which
did
it
for
the
lgm
just
to
see
how
much
temperature
compensation
we're
looking
at
from
those
runs
from
co2.
This
is
very
little
about
1.1
degrees
and
prescribed
geography.
C
Topography
has
a
slight
cooling
effect,
because
there
is
a
little
bit
of
burning
street
closure,
land
exposure
and
there's
a
bit
of
a
shelf
tropicals
on
the
shelf.
Indonesian
seashell
shoring
for
vegetarian
ice
sheet,
the
crack,
the
woman,
it's
a
little
bigger
and
combined
geography;
vegetation
again,
there's
a
little
warming.
If
you
look
at
erf,
this
is
unadjusted
for
co2.
It's
about
2
watts,
per
meter
square
1.96
for
geography
and
photography.
It's
about
1,
minus
0.1,
virtual
ice
sheet.
C
I
C
Okay,
yeah,
I'm
almost
done
so.
If
we
I
get
the
simplest.
The
adjustment,
using
the
the
amount
of
four
things
from
a
doubling
co2
and
divided
by
the
equilibrium
climate
sensitivity
from
andrew
gettleman's
paper
numbers
are
from
andrew
generman's
paper
and
for
co2.
This
is
after
the
adjustment
it's
about
2.04
watts
per
meter
square.
This
is,
I
would
say,
consistent
with
this
radiate
a
radio
forcing
map
estimate
for
1.4
times
co2
1.5
to
4
times
this
397
ppm,
while
we're
having
400
ppm
right
for
geography
type
topography.
C
This
is
about
minus
0.21
watts
per
meter
squared.
So
for
the
large
part,
this
is
negligible
for
vegetation
ice
sheet.
This
is
about
1.51
watts
per
meter.
Squared,
so
here
are
some
unanswered
questions
that
I'm
really
interested
in
exploring
at
the
moment.
So
what
is
the
stores
for
this
1.51
watts
per
meter
square
url
from
vegetation
actually
changes?
C
Is
it
rapid
adjustment
of
albedo
water,
vapor
clouds,
and
how
dependent
is
this
estimated
on
model
versions?
I'm
also
working
on
setting
up
ccs74
or
cam
4
for
this,
and
why
is
this
forcing
so
effective?
I
did
a
calculation
of
efficacy
for
the
registration.
Actually
for
saying
it's
40
more
effective
than
co2,
so
that's
kind
of
interesting
to
me
and
perhaps
we
could
call
the
paleo
radiative
forcing
map
to
really
help
understand
those
paleoclimate
simulations
among
multiple
models.
Yeah.
So
that's
all.
C
A
A
Okay,
our
next
speaker
is
jean-ju
a
project
scientist
here
at
ncar
and
he
is
speaking
on
an
lgm
calibrated
cesm2
for
paleoclimate
studies.
Okay,
thank
you.
Esther.
J
J
So
many
of
you
may
know
that
in
a
recent
study
we
performed
lgm
simulation
using
the
csm2
cam6.
What
we
found
was
that
the
simulated
lgm
global
cooling
is
close
to
12
degrees
c,
and
it
is
way
outside
the
proxy
suggested
range.
That's
the
latest
tyranny.
At
all
2020
estimation,
and
based
on
that
we
concluded
the
high
ecs
of
csm2
is
very
likely
unrealistic,
and
this
is
not
ideal.
It
prevented
us
from
doing
exciting
research
using
the
new
capabilities
of
csm2.
J
J
So
the
first
key
ingredient
of
this
paleo
configuration
is
to
remove
an
inappropriate
limiter,
which
is
called
ni
max.
So
basically,
an
imax
sets
the
maximum
cloud
ice
number.
It
was
designed
to
work
with
the
older
ice
nucleation
scheme
with
a
longer
time
step,
but
it
was
not
modified
accordingly
when
they
developed
the
new
schemes.
J
Therefore,
we
need
to
explore
its
impact
on
climate
sensitivity,
so
the
basic
process,
level
understanding
is
that
when
you
remove
this
limiter,
you
have
a
higher
cloud
ice
number
and
a
more
cloud
ice
mass,
which
implies
a
more
negative
cloud
phase
and
a
lifetime
feedback,
so
weaker
overall
weaker
cloud
feedback.
Therefore,
a
smaller
temperature
response.
J
J
So
when
you
increase
10
step,
it
improves
the
simulation
of
the
cloud
ice
number.
Secondly,
it's
motivated
by
this
paper.
Basically,
they
show
that
mg2
contains
several
processes
that
are
fully
resolved
at
a
default
time
step
of
300
seconds,
and
we
found
that
when
you're
shortening
the
macrophagical
time
step,
you
have
a
overall
weaker
cloud
feedback
that
also
helps
in
giving
us
this
overall
realistic,
lgm
simulation,
and
we
also
did
extensive
exploration
and
we
found
that
aid
sub
step
is
sufficient
for
a
converging
solution
in
cloud
feedback.
J
J
J
I
also
quantified
the
aerosol
cloud
interaction
which
is
20
percent
weaker.
The
right
figure
shows
the
global
warming
in
the
historical
period.
The
blue
is
our
paleo
calibrated
configuration
and
you'll
see
that
overall,
it
well
captures
the
20th
century.
Warming,
women
and
its
performance
is
as
good
as
the
other
csm2
chem
6
simulations.
J
We
think
what
is
going
on
is
like
this.
So
in
this
paleo
calibrated
configuration,
we
have
less
greenhouse
gas
induced
global
warming
at
the
same
time,
less
aerosol-induced
global
cooling.
Because
of
this
cancellation
we
still
get
overall,
realistic,
20th
century
warming,
and
this
plot
shows
the
lgm
cooling
in
surface
temperature
compared
with
the
compilation
in
tlu-2020.
J
This
is
csm2
paleocalibrated
configuration
and
csm1,
and
you
see
that
this
is
certainly
much
warmer
than
csm2
and
is
closer
to
csm1.
But
there
are
some
differences
which
was
further
looking
to,
for
example,
over
the
topics
at
the
table.
I'm
listing
the
root
mean
squared
error
in
different
simulations
compared
with
different
proxies.
So
it's
big
improvement.
You
save
over
the
csm2,
but
it
still
is
not
as
good
as
csm1.
J
So
this
plot
shows
a
power
spectrum
of
enzo.
The
black
line
is
observation
and
the
red
is
default.
Configuration
pre-industrial
simulation
you'll
see
that
the
model
overestimates
the
enzo
variability
and
the
blue
is
our
paleo
calibrated
pre-industrial
simulation,
and
you
see
that
we
have
somewhat
improvement
in
terms
of
the
annual
variability.
J
The
green
is
the
payload
calibrated
lgm
simulation
so
consistent
with
what
we
had
in
csm1.
So
we
still
see
this
reduction
in
enzo
variability
in
the
lgm
climate.
Oh,
so
here
is
a
quick
summary.
We
have
developed
this
paleo
calibrated
csm2
cam6,
it's
very
important
that
we
still
use
most
of
all
the
advanced
chem
6
feed
physics
in
this
configuration,
and
this
configuration
simulates,
realistic,
lgm,
realistic,
20th
century
warming
has
a
low
ecs
with
weaker
cloud
feedback,
and
with
this
configuration
we
planned
some
community
research.
A
Thank
you,
john,
very
interesting
talk
and,
let's
see,
do
we
have
time
arnie
for
yeah.
I
think
we
we.
K
J
J
Is
two
degree
atmosphere
because
it's
cheaper
to
run
and
we
expect
similar
behavior
in
every
one,
but
that
need
to
be
explored
thanks
for
the
question.
Yeah.
A
A
Okay,
our
next
speaker
is
elaine
kuo
from
at
the
university
of
toronto.
Yuling
is
a
student
there.
M
Okay,
thank
you
for
the
introduction
I'm
eating,
and
so
today
my
talk
will
be
about
the
dynamically
downscaled
simulations.
We
did
in
our
group
to
study
the
medical
scene
also
in
south
and
southeast
asia,
and
also
to
investigate
the
influence
the
teleconnections
between
the
green
sahara
and
the
asian
museum
system.
M
So
generally
proxy
data
shows
that
the
male
holocene
was
generally
warmer
than
today
during
summer
and
cooler
in
winter.
Due
to
the
different
earth's
orbital
parameters
from
today
and
the
complex
orography
of
south
and
southeast
asia
renders
the
regional
climate
complex,
and
here
we
use
that
we
constructed
a
series
of
regional
climate
simulations
using
dynamic
downscaling
to
downscale
global
global
metal,
holocene
simulations.
M
We
have
also
taken
care
to
incorporate
the
greens
and
her
boundary
conditions
to
investigate
the
interactions
between
the
vegetations
over
north
africa
and
the
asian
muslim
system.
Okay.
So
the
regional
couple
modern
system
here
consists
of
original
crime
models
and
worth
model
and
the
3d
ocean
model
and
the
enter
gcm,
which
is
the
university
version,
a
university
of
toronto
version
of
the
ccsm4
and
the
the
worth
model
we
use
in
the
version
4.1
at
30
kilometer
resolution.
M
M
The
ocean
domain
is
the
white
rectangular
here
that
covers
most
of
the
ocean
areas
in
the
world
domain
and
the
the
wharf
croco.
The
coupled
with
croco
model,
is
driven
by
three
different
gcm
simulations
one
for
the
pre-industrial
and
two
for
the
middle
policy
and
the
two
metaholic
simulations
one
use
the
which
is
called
one
of
them
is
called
reference
with
hosting
it
to
use
the
same
vegetation
distribution
as
the
pre-industrial
and
the
other
one
is
called
mhgs.
M
It
incorporates
the
rings
and
horizontal
conditions,
meaning
it
includes
a
vegetated
north
africa,
oh
yeah,
okay.
So
let's
first
look
at
the
temperature
anomalies
during
the
middle
housing
here,
so
this
figure
shows
the
ssd
and
the
continental
temperature
changes
in
the
gcm
and
in
overdone
skill
simulation.
M
The
sst
distribution
is
pretty
similar,
except
that
over
around
the
southern
tip
of
india,
our
regional
ocean
model
simulates
a
warmer
anomaly
while
as
the
gcm
simulates
cold
city
anomalies
over
the
entire
ocean
domain
and
the
over
the
land
they're
during
meta
harvesting,
their
temperature
is
actually
warmer
by
up
to
2
degrees.
M
Northward
of
the
tibetan
plateau
and
the
largest
warming
center
occurs
around
the
40
degrees
and
the
intensity
and
special
extent
of
this
warming
center
is
larger.
In
the
downscaled
simulations
than
the
gcm
and
over
south
asia.
We
see
a
small
cooling
and
over
southeast
asia
we
see
actually
small
opposite
temperature
anomalies
in
western
and
eastern
regions.
M
As
a
result,
the
land
see
thermal
contrast,
increases
due
to
the
specially
averaged
increase
in
jgs
temperature
over
land
and
is
decreased
over
in
south
in
sst,
which
intensifies
the
moisture
transport
towards
the
land
and
yeah
okay.
So
now,
let's
look
at
the
annual
mean
precipitation
changes
in
the
referencement
housing
simulation
most
of
the
south.
M
South
east
asia
actually
experience
weather
climate
compared
to
the
pre-industrial,
especially
over
the
nozzle
part,
and
the
gcm
stimulus
drying
over
the
eastern
himalayas
and
southern
india,
whereas
the
downside
of
simulation
shows
an
increase
in
any
rainfall
compared
to
proxy
data
compared
to
two
pollen-based
reconstructions
here
showing
circles
and
diamonds
and
overall
there
is
qualitative
agreement
with
reconstructions
with
weather
conditions
in
the
northern
region
of
southeast
asia
and
along
the
himalayas.
M
And,
however,
they
disagree
with
the
reconstructions
in
eastern
tibetan
plateau
and
southeastern
china
and
and
over
the
regions
north
of
the
tibetan
plateau.
And
overall,
the
downside
simulation
displays
a
slightly
greater
similarity
in
pattern
to
reconstruction
than
the
gcm
in
order
to
quantify
this
fit
to
reconstructions.
We
use
this
mean
relative
air
using
this
formula
here
and
considering
all
the
points
with
prox
data
in
the
work
domain.
M
Dynamical
downscaling
reduced
the
mre
by
about
ten
percent
and
since
all
of
the,
since
both
model
fails
to
capture
the
right
sign
of
precipitation,
normally,
four
points
north
of
the
like
40
degrees.
So
if
we
only
consider
points
that
south
of
the
40
degrees,
the
mre
of
the
downscale
simulation
is
actually
about
30
percent
smaller
than
the
gcm
and
the
other
proxy
data,
however,
suggests
much
larger
precipitation
enhancement
and
therefore
it
is
less
consistent
with
spouse
models
and
the
jjs.
M
The
summer
precipitation
normally
patterns
pretty
similar
to
that
of
the
enemy,
except
that
the
increased
amplitude
is
greater,
and
that
also
contributes
to
an
increase
in
cloudness
and
evaporation,
which
may
explain
the
lower
temperature
over
south
asia
and
southeast
asia
we've
seen
before.
And
if
this
two
figures
shows
the
annual
cycle
of
precipitation
over
south
asia
and
southeast
asia
and
the
precipitation
during
the
early
stage
in
in
june.
M
Here,
an
early
stage
of
the
monsoon
season
in
june
here
is
actually
weaker,
which
is
which
suggests
that
the
monsoon
actually
starts
later
in
the
housing
than
in
the
pre-industrial,
and
also
the
maximal
precipitation
increase
over
these
two
regions
occurs
at
the
end
of
the
museum
season
in
september,
and
also
there's
small
precipitation
increase
in
october
suggests
that
the
withdrawal
of
the
summer
as
soon
as
postponed
and
finally
all
of
these
increased
monsoon
changes
discussed
above,
are
stronger
in
the
downskill
simulations
than
in
the
gcm
okay.
M
Now,
let's
then
look
at
the
impacts
of
the
green
sahara,
boundary
conditions.
So,
first,
let's
look
at
the
temperature
changes,
so
the
inclusion
of
green
sahara,
surface
boundary
conditions
leads
to
an
enhanced
gjs
temperature,
normally
over
the
majority
of
the
world
domain
and
with
the
largest
increase
over
the
tibetan
plateau
in
the
east
asia.
So
this
one
showed
the
difference
between
these
two
mid
harvesting
simulations
and
the.
K
M
And
there
compared
to
the
reference
compared
to
the
reference
methodology
simulation,
the
sst
is
actually
warmer
in
the
mhgs
simulation
and
this
favors
more
evaporation
of
the
ocean
and
thus
contributing
to
the
ocean
monsoon
precipitation
increase.
M
So
now,
let's
look
at
the
new
precipitation
normally
in
the
mhgs
simulation,
so
in
closure
of
a
green
sahara,
significantly
reinforced
the
precipitation
army
over
south
asia
and
south
and
southeast
asia,
and
the
interest
is
particularly
large
over
the
northern
regions
and
it
also
enhanced
the
precipitation
around
the
himalayas
in
tibetan
plateau
and
the
northern
regions,
north
china
and
the
region's
north
of
the
tibetan
plateau
and
significantly
improving
the
proxy
data
comparison
and
it
actually.
Inclusion
of
a
green
sahara
leads
to
a
about
10
reduction
in
the
mre
yeah.
M
Okay,
let's
jump
to
the
conclusions.
Okay,
so
we
constructed
a
series
of
regional
climate
simulations
using
the
coupled
warf
qualcomm
model,
driven
by
the
output
from
the
gcm
uft
and
sn4,
and
we
also
incorporate
a
set
of
downscale
simulations
to
include
the
green
sahara
boundary
conditions
to
access
the
remote
response
of
the
south
and
southeast
asian
museum
system
to
to
the
green.
M
Bringing
of
the
sahara
compared
to
the
pre-industrial
the
reference
meta
horizon
simulation,
it
shows
a
strength
in
the
summer
monsoon
intensity
over
both
south
and
southeast
asia,
especially
over
the
northern
region
and
a
late
monsoon
state,
and
a
delay
in
the
monsoon
destroyer
over
both
regions
are
also
found,
and
overall,
the
worth
results
have
been
better
agreement
with
prostate
data
and
the
gcm
really
of
the
sahara,
also
amplifies
the
response
of
the
two
monsoon
systems
to
the
and
the
leads
to
a
further
increase
in
the
mid
heart
signals,
precipitation
and
the
model.
M
Data
comparison
also
shows
significant
improvements
in
simulating
precipitation,
especially
of
the
tibetan
plateau
in
the
northern
regions,
and
these
results
emphasize
the
importance
of
high
resolution
modelling
over
regions
with
complex
land
surface
and
highlight
the
climate
sensitivity
of
saharan
vegetation
changes,
the
ocean
atmosphere
and
teleconnections
on
the
asian
motion
system.
Yeah.
That's
all
thank
you.
Everyone
and.
M
A
Thank
you,
ewing,
yeah.
Your
last
statement
drives
home
the
point
that
we
need
to
do
more
high
resolution,
paleo
yeah
yeah.
Thank
you.
I
think
we'll
have
to
take
questions
in
the
chat
and
go
move
on
to
our
next
speaker,
william
rush,
a
phd
student
at
uc,
santa
cruz,
he's
speaking
today
on
regional
variability
of
the
south
asian
monsoon
through
the
petm.
N
Great
thank
you
and
yeah.
It's
a
great
talk
to
follow
up
with
on
that
previous
one,
looking
more
at
high
resolution
models
of
the
south
asian
one.
Soon,
thanks
give
me
the
opportunity
to
present
on
this.
This
is
a
side
project.
I've
been
working
on
for
a
couple
months
now
and
I
like
to
start
out
all
my
talks
talking
about,
like
my
motivations,
for
doing
this
sort
of
thing,
the
south
asian
monsoon
supplies
eighty
percent
of
the
water
supply
in
south
asia,
which
is
the
principal
supply
for
well
over
a
billion
people.
N
I'm
going
to
be
going
a
bit
further
back
in
time
than
our
previous
talk
looking
at
the
petm,
and
this
is
the
world
of
the
petm,
and
although
we
have
a
slightly
different
continental
configuration
than
the
current
day,
particularly
in
that
the
indian
subcontinent
had
not
yet
reached
south
asia,
we
can
you
still
use
this
to
draw
some
parallels
to
modern
day
warming.
N
What's
particularly
interesting
about
this
is
in
the
past
few
years
there
have
been
several
studies
that
have
come
out
looking
at
these
petm
sections
in
the
region
of
the
himalayas
and
in
particular,
even
though
these
are
in
close
proximity
today,
they
can
provide
us
with
insight
into
hydroclimate
changes
that
happened
at
fairly
fairly
separate
from
each
other
at
this
point
in
time,
namely
on
the
northern
and
southern
tethos.
N
N
So
to
describe
these
changes
that
have
been
taking
place
in
the
northern
tethys.
This
was
around
15
to
20
degrees,
north
or
so
the
pre-petm.
The
sediments
that
were
deposited
at
this
point
in
time
are
largely
of
a
carbonate
ramp
variety
during
the
petm.
N
This
shifts
to
a
solicoclastic
dominated
system
which
then
recovers
into
the
carbonate
ramp
faces
that
were
being
deposited
prior
to
the
petm
afterwards,
and
the
authors
of
this
study
found
that
there
was
no
change
in
the
provenance
of
the
salicylastic
material
that
was
being
deposited
either
before
or
during
the
petm,
and
so
therefore
they're,
arguing
that
the
changes
that
are
being
observed
are
not
due
to
a
change
in
the
tectonic
regime.
N
That's
happening
at
this
time,
but
rather
just
a
an
increased
solicitoclastic
flux
due
to
the
intensified
more
extreme
precipitation
resulting
from
the
seasonal
monsoon
and
the
southern
tethys
and
india.
This
is
a
more
equatorial
location
and
we
see
a
similar
trend.
You
go
from
these
carbonate
type
deposits
that
shifts
into
a
conglomerate
during
the
peak
of
the
petm
and
again,
this
is
being
this
has
been
interpreted
in
the
literature
as
this
rapid
deposition
from
a
this
fluviolic
system,
indicating
an
intensified
hydrologic
cycle
occurring
at
this
location.
N
At
this
point
in
time,
so
to
understand
what
exactly
is
going
on
with
the
hydrology
here
and
see
how
our
models
compare?
I
utilized
the
cam
output
from
the
cam5
model,
so
app
purely
atmospheric
model
run
at
a
quarter
degree,
spatial
resolution.
This
was
run
for
20
years
with
a
six
hour
time
step
for
the
ocean
boundary
conditions.
We
used
cesm
1.2
the
fully
coupled
model
ocean
atmospheric
components
with
a
2
degree
resolution
that
was
run
for
1800
years
with
a
one
month.
N
Time
step
between
these
two
runs,
I'm
going
to
refer
to
them
as
lp
for
late
paleocene
and
the
petm
methane
was
held
constant.
The
paleocene
co2
is
at
680
parts,
per
million,
petm
was
1590
and
due
to
increasing
evidence
of
how
orbit
can
actually
influence
was
influencing
climate.
N
During
these
eosine
hyperthermals,
we
also
looked
at
the
influence
of
orbit
by
utilizing
both
a
circular
orbit,
which
is
to
say
that
the
insulation
during
northern
hemisphere
summer
was
the
same
as
during
southern
hemisphere
summer,
as
well
as
an
elliptical
orbit
that
maximized
insulation
over
the
northern
hemisphere
during
the
boreal
summer,
and
I'm
going
to
refer
to
this
more
elliptical
orbit
as
the
orb
max
run
in
all
of
my
model
output.
N
So
this
is
the
first
of
the
output
I'm
going
to
be
showing
you,
so
this
is
for
the
tibetan
location.
So
the
north
side
of
the
tethys,
I'm
going
to
be
using
the
same
color
scheme
here
for
all
of
the
different
model,
runs
wherein
the
black
is
going
to
be.
The
late
paleozine
gray
is
going
to
be
late.
Paleocene
with
that
orb
max
configuration
blue
is
the
ptm,
so
the
high
co2
and
red
is
the
combining
high
co2
and
this
more
eccentric
orbit
and
there's
a
couple
things
that
stand
out
here.
N
One
is
when
we
hold
hold
the
orbits
completely
circular,
there's
very
little
change
in
the
strength
of
the
monsoon.
When
you
apply
this
co2
forcing
within
this
model,
so
this
is
looking
at
the
total,
the
average
monthly
rainfall
that
occurred.
N
However,
when
we
apply
this
enhanced
eccentricity
to
the
model
that
maximizes
insulation
during
the
northern
hemisphere
summer
in
the
late
paleocene,
we
have
almost
a
quadrupling
of
the
intensity
of
the
monsoon
from
about
a
hundred
millimeters
within
the
month
of
august,
at
its
peak
in
the
late
paleocene,
going
to
close
to
400
millimeters
in
august
during
the
orb
max
configuration.
N
However,
when
we
apply
the
co2
forcing
to
this
same
run,
we
see
a
reduction
of
about
half
of
the
total
strength
of
the
monsoonal
signal
going
from
a
peak
of
400
millimeters
a
month
to
about
200,
and
this
seems
contradictory
to
the
observations
that
seem
to
suggest
enhanced
seasonal
precipitation
during
the
petm
at
this
location.
N
So
again,
the
late
paleo
scene
are
in
black
and
gray,
and
the
petm
are
in
red
and
blue,
and
what
we
see
here
is
a
enhanced
seasonal
signal.
So
in
the
winter
months
so
january
through
march,
it
is
there's
there
are
drier
conditions
and
then,
in
the
summer
months
there
are
wetter
conditions
during
the
petm
runs.
N
So
not
all
I
look
not
only
at
the
seasonality
of
precipitation,
but
also
the
precipitation
extremes.
So
in
this
plot,
I'm
showing
exceedance
frequency.
So
this
is
a
log
log
plot.
The
x-axis
is
showing
the
rate
of
precipitation
predicted
within
the
model.
The
y-axis
is
showing
the
frequency
with
which
the
the
model
exceeded
that
a
given
precipitation
rate
and
so
basically
the
higher
and
further
to
the
right.
N
The
line
is
that's
showing
more
extreme
precipitation,
and
what
you
can
see
is
that
the
similar
to
the
seasonality,
the
late
paleocene
or
max
configuration
results
in
typically
the
most
extreme
precipitation,
with
a
few
outliers
towards
the
end
of
the
the
petm
or
max
run
the
petmore
max
results
in
the
second
most
extreme
precipitation
and
then
the
similar
to
the
seasonality.
Again,
there's
very
little
shift
in
the
precipitation
extremes
between
the
late
paleocene
run
with
the
circular
orbit
and
the
and
the
petm
run
with
the
circular
orbit.
N
Looking
at
the
precipitation
rate
at
the
southern
location,
it's
again
more
in
line
with
what
we
would
expect,
where
co2
forcing
results
in
an
overall
increase
in
the
intensity
of
the
extreme
precipitation.
N
So,
to
conclude,
there
are
different:
there
are
obviously
different
mechanisms
that
are
driving
the
hydrologic
change
in
the
northern
and
southern
tethys,
and
there
is
an
agreement
between
the
interpretation
of
the
sediment
record
and
the
model
output
in
the
southern
texas.
However,
there
is
a
conflict
between
the
interpretation
of
the
sediment
record
and
the
model
output
in
the
northern
tethys.
So
I
don't
entirely
know
why
this
is
yet
it
could.
Could
it
be.
These
changes
are
actually
tectonically
driven.
N
It
would
make
it
hard
to
explain
the
recovery,
though,
could
it
be
that
there's
some
sort
of
problem
with
our
model?
Could
it
be
due
to
our
paleogeographic
configuration
that
we're
using
it
could
be
different
parameters?
I
should
use
to
look
at
this
or
a
different
orbital
configuration
that
I'm
not
getting,
but
if
we
trust
the
model,
the
influence
of
co2
forcing
on
the
south
asian
monsoon
appears
to
be
highly
dependent
upon
the
orbital
configuration
and
so
going
further.
N
I
want
to
look
dive
more
into
the
dynamics
behind
this
and
there
should
be
further
analysis
of
the
sediment
changes
to
try
to
disentangle
tectonic
and
climatic
effects
and
there's
a
few
other
locations.
I'd
like
to
look
at,
but
since
I'm
out
of
time,
I'm
just
going
to
wrap
up
and
say
thank
you
and
I'll.
Take
any
questions.
A
Thank
you
will
I
believe
you
have
a
question
already
in
the
chat,
so
I
believe
we
should
move
on
here.
G
Thank
you
esther.
I
want
to
give
you
an
update
of
long
term.
Csm
1.2
simulation,
acknowledge
my
courses
and
also
the
sponsors
in
support
of
the
study.
As
we
heard
in
the
previous
talk,
the
palestinian
eocene
summer.
Maximum
is
of
great
interest
because
it's
a
suitable
analog
for
future
climate
change,
but
it's
also
of
interest
for
large
changes
in
the
carbon
cycle.
G
G
So
the
the
carbon
cycle
is
controlled
by
the
particular
organic
carbon
flux
controlled
by
photosynthesis,
which
itself
is
controlled
by
a
light
temperature,
nutrient
availability
and
some
mixing
processes,
and
so
one
can
now
predict
the
productivity
or
heinkers
the
productivity
of
the
past
by
using
marine
bio
by
right,
fluxes.
G
And
so
these
fluxes
are
somehow
correlated
to
productivity,
which
is
shown
in
the
next
slide
on
the
left,
where
one
has
estimated
the
barod
fluxes
for
the
pre-ptm
or
the
late
palestine
and
the
ptm,
and
that
suggested
that
the
buried
fluxes
increase
so
the
productivity
during
the
ptm.
However,
modeling
results
suggested
the
contra,
the
the
opposite,
because
I
want
to
expect
actually
more
stratification,
less
upwelding
and
those
reduced
upwarding.
G
So
we
are
using
here
a
series
of
deep
time,
modeling
comparison,
project
simulations
which
I
have
described
before
and
then
using
also
box
modeling
results
to
better
estimate
these
productivity
estimates
and
also
carbon
fluxes.
G
So
we
have
then
said
several
simulations,
and
so
we
have
now
been
able
to
run
this
fully
carbon
cycle,
the
late
palestinian
simulation,
which
is
equivalent
to
the
three
times
co2
simulation
for
two
thousand
years
with
carbon
cycle
and
the
six
times
co2
simulation
which,
as
the
analog
for
the
ptm
simulation,
is
still
running
in
the
machine,
and
we
need
to
complete
at
least
700
more
years.
G
So
the
associated
climate
changes
and
the
assimilation
indicate
a
substantial
warming
towards
the
ptm
and
also
a
stronger
rise
in
temperature
of
our
land
compared
to
the
ocean
warming
of
high
latitudes
and
reduction
of
the
poultry
equator
temperature
gradient.
G
If
we
compare
the
pgm
simulation,
the
six-time
co2
simulation
with
that
of
the
proxies,
we
see
reasonable
agreement,
but
we
still
see
some
biases
in
the
higher
altitudes,
particularly
in
the
arctic
region,
and
one
can
argue
either
it's
a
higher
co2
level,
which
would
be
a
bit
better
fit
or,
on
the
other
hand,
maybe
processes
via
lacking
like
cloud
both
as
discussed
earlier
so
the
overturning
circulation.
G
Here's
a
show
on
the
global
mark
is
dominated
from
the
southern
hemisphere,
so
we
have
a
southern
hemisphere
dominated
mark
which
has
been
supported
by
a
paper
of
michelin
over
and
myorima
in
the
90s
as
suggesting.
If
we
closing
the
direct
passage
and
open
the
central
american
seaway,
we
are
shifting
the
mark
to
the
southern
hemisphere,
southern
dominant
hemisphere
and
then
the
mock
declines.
The
global
mock
declines,
with
increased
co2,
radiative
sourcing,
so
consistent
with
multiple
model
studies.
G
G
So
the
productivity
patterns
are
shown
here
quite
similar
to
the
ccs
m3
simulation
with
reduced
upwelling,
decreased,
overturning
and
associate
decline
in
the
nutrient
simulat
supply,
and
so
those
we
do
have
a
reduced
productivity
shown
also
at
the
lower
part.
Here
is
the
changes
in
upwelling
in
particular,
that
is
reduced
at
the
equatorial
region
and
there's
also
changes
in
the
ecosystem
dynamics,
which
is
will
be
explored
as
part
of
a
phd
thesis,
so
we're
somehow
supporting
the
previous
results.
G
We
are
now
able
also
to
better
predict
the
oxygen
with
these
long-term
simulations,
and
we
notice
that
the
deep
water
formation
takes
place
mostly
in
the
waters
here,
and
so
we
are
ventilating
the
atlantic
from
the
battle
c
northwest
and
we
see
then
a
decline
in
the
oxygen
at
the
bottom
water
and
if
we
increase
the
co2
levels
so
from
the
late
paleocene
to
the
petm.
G
We
would
already
have
a
lower
dissolved
oxygen
in
the
bottom
water
because
of
the
warming
and
the
lower
solubility,
and
then
we
have,
on
top
of
that,
a
decline
in
the
deep
water
source
and
the
water
see
and
increase
stratification,
which
leads
a
further
decline
into
oxygen,
which
does
provide
a
stress
in
the
foramini
for
the
foraminifera.
G
So
we
looked
also
as
as
a
measure
for
the
biological
pump
in
the
apparent
oxygen
utilization,
which
is
determined
by
the
oxygen
saturation
minus
the
stimulated
oxygen
concentration
and
for
the
late
paleocene
simulation.
We
see
a
strong,
apparently
oxygen
utilization
below
the
equatorial
upwelling
zone
in
the
atlantic,
and
that
is
due
to
the
high
amount
of
criminalization
of
particular
organic
carbon
at
the
stone,
which
is
also
observed
in
the
present
day.
G
Record
now
of
interest
is
now:
how
does
the
apparent
oxygen
utilization
changes
if
we
are
transitioning
from
the
late
palestine
to
the
ptm
and
what
we
notice
is
actually
an
increase
in
the
oxygen
utilization
in
the
twilight
zone?
So
don
penman
has
looked
in
and
and
box
models
of
the
lost
car
models,
and
so
the
motivation
were
first
looking
in
other
times
like
the
kp
fpg
boundary,
one
notice
and
enhanced
remineralization.
G
So
if
we
have
an
increase
in
apparent
oxygen
utilization,
that
may
be
a
favorable
foreign
enhancement
monetization,
and
so
we
carried
out
the
simulation
in
a
recently
published
paper
of
liz,
griffiths
and
company,
and
so
with
this
box
model
we
can
actually
protect
an
increase
and
and
and
the
pco2.
So
we
can
actually
simulate
positive
feedbacks
in
the
climate
science
due
to
the
warming
of
the
paleozoic
in
summer.
G
Maximum
put
up
the
conclusion-
and
I
I
think
my
time
is
already
quite
progress
and
I'm
happy
to
answer
any
questions
in
the
chat.
A
Thank
you,
arnie
we
have
here,
maybe
have
a
minute
for
a
question
if
there's
one,
otherwise
we
could
take
questions
in
the
chat
and
looking
for
a
hand,
don't
see
one.
Why
don't
we
take
some
questions,
then
in
the
chat,
and
we
can
start
we're
now
going
to
start
our
discussion.
We
have
a
brief
discussion
focused
on
some
questions
that
we
thought
were
kind
of
emerged
from
some
of
the
talks
today.
Let
me
share
my
slides.
A
A
Yeah,
because
when
I
run
the
slideshow,
it
obliterates
everything
else
on
my
screen
and
I
kind
of
want
to
see
what's
going
on,
so
we
thought
we
would
have
two
short
discussions
with
some
focused
questions
for
us
to
think
about.
The
first
is:
what
does
paleoclimate
modeling
tell
us
about
sensitivity
of
the
earth
system
to
forcings.
A
G
I
think
I
think
that
discussed
in
several
talks.
A
It
seems
like
the
cloud
feedbacks
are
the
biggest
uncertainty,
but
also
have
the
biggest
impact
to
climate
sensitivity,
and
so
that's
a
real
area
that
needs
further
research.
But
it's
one
that,
with
better
reconstructions
that
we
have
our
our
latest
reconstructions.
They
really
provide
a
much
closer
constraint
on
a
paleo
climate
that,
for
example,
what
john
showed
with
the
lgm
reconstruction
now
constraining,
helping
to
constrain
some
of
our
choices
for
parameterizations
in
the
model.
A
Oh,
I
do
okay
sure
go,
go
marcus.
H
We've
been
looking
at
the
ventilation
and
I
think
we're
not
the
only
ones
but
there's
a
very
poor
and
ugly
bottom
water
formation,
which
makes
it
difficult
to
you
know
to
get
lgm
carbon
budgets
and
other
interesting
questions.
But
that's,
I
think,
that's
an
order.
One
problem
is
this
being
addressed
or
are
people
addressing
this
right
now
good.
A
In
the
ocean,
no
it's
a
model.
I
think
it's
it's
been
an
issue
that
that
they
continually
have
been
trying
to
address
with
the
ocean
mixing
parameterizations,
I
think
they're
aware
it's
yeah.
A
One
bias
in
the
ocean
model
is
that
the
ventilation
in
the
north
pacific
is
very,
very
weak
and
I
think
we'll
see
some
examples
of
that
in
our
next
session.
A
A
You
know
where
the
oldest
water
is
found
is
it's
very,
very,
very
old
and
ideal
age,
and
some
of
the
other
tracers
like
the
carbon
isotope,
carbon,
14,
tracer,
so
yeah.
I
think
it's
something
that
keith
is
no
longer
here,
but
the
ocean
modelers
have
been
trying
to
work
on
over
the
successive
versions
of
pop
and
now
into
mom,
so
yeah.
I
think
that
is
a
an
issue
that
that
is
forefront
to
be
worked
on
getting
the
deep
ocean.
H
A
I
think
it's
mentioned
in
gokan's
paper.
I'd
have
to
check,
but
I
believe
it's
something
that
has
been
presented
in
some
of
the
working
group
presentations
but
yeah
yeah.
It
is
an
issue
I
don't
know
who's
next.
K
J
Have
you
checked
the
sst
pattern
effect
on
the
cloud
feedback,
so
that
has
been
discussed
a
lot
in
the
community
and
the
related
question
is:
oh,
what
do
you
think
about
additional
simulations
with
say,
sst,
plus
4k
and
sst
minus
4k?
Will
that
be
helpful?
How
it
just
opens
another
box
of
questions.
Thank
you.
B
Yeah,
I'm
here
those
are
great
questions.
Jenny,
I'm
happy
to
chat
more
offline
as
well.
I
think
the
pattern
effect
for
a
doubling
and
having
there's
a
lot
of
similarities,
in
particular,
there's
more
warming
in
the
tropical
western
pacific
on
this
kind
of
fast
time
scales
and
then
on
the
slow
time
scales
like
you
get
the
the
kind
of
response
in
the
eastern
tropical
pacific.
B
But
there
are
some
nuances
that
are
different
in
the
patterns
of
the
clouds
under
cooling
than
under
warming,
and
I
think
it'd
be
fun
to
try
to
understand
those
and
diagnose
them
more,
and
I
guess
just
as
long
as
I
have
the
floor.
I'll
just
briefly
say
that
I
think
it's
really
fun
to
disable
the
cloud
feedbacks
I
mean.
B
I
think
it's
great
that
you
know
we
can
make
these
minor
tweaks
to
the
model
like
what
gang
suggested
and
it's
kind
of
shocking,
that
we
can
reduce
the
equilibrium
climate
sensitivity
so
much
with
so
few
little
changes
in
the
cloud
parameterizations
and
while
I
think
that
does
lead
to
a
more
reasonable
equilibrium,
climate
sensitivity,
I'm
not
really
sure
you're
getting
the
right
answer
for
the
right
reasons,
and
so
it's
there's
a
lot
to
think
about
there,
and
so
I
think,
simulations
with
the
cloud
feedback's
disabled
might
be
helpful
for
kind
of
removing
this
layer
of
uncertainty.
B
C
Thank
you
sure.
I
would
just
like
to
echo
some
of
what
john
engine
said
about
cloud
feedbacks
as
well
as
looking
into
the
equilibrium
climate
sensitivity
a
little
bit
more
magnetically,
especially
there
are
a
lot
of
effort
in
proxy
community
which
are
very
interested
in
using
this,
almost
like
immersion
constraint,
idea
to
constrain
the
equilibrium
climate
sensitivity
from
those
sst
proxies,
or
even
some
of
the
terrestrial
proxies.
C
But
the
a
big
question,
a
big
problem,
especially
for
well
even
in
quaternary,
but
especially
for
like
pricing,
medicine
or
you're
saying,
is
there's
no
real
serious
thinking
on
the
fourth
thing.
So
a
lot
of
people
thought
it's
just
co2,
so
I'm
just
gonna
take
the
co2
as
the
fourth
thing,
and
it's
very
much
casually
done
a
lot
of
times
like
citing
maybe
one
model
simulation,
while
the
changes
in
boundary
conditions
cost.
Maybe
just
this
much
warming
and
then
that
that
number
has
populated
through
literature.
C
So
I
think
that's
kind
of
a
at
least
from
what
I
see
that's
kind
of
a
gap
in
pilot
climate
modeling
effort
in
terms
of
helping
tell
the
sensitivity
of
our
system
of
from
the
proxy
records
is.
We
can
provide
better
information
on
the
fourth
and
the
non-co2
related
responses.
Yeah.
F
Hi
hi,
I
was
wondering
you
know
after
marcos's
talk,
which
was
really
interesting
in
in
looking
at
these
co2
responses
and
suggests,
perhaps
that
iron
fertilization
played
a
role
there
for
these
millennial
variations,
and
we
have
found
also
in
a
recent
study
that
iron
fertilization
is
probably
has
probably
been
really
important
for
glacial
interglacial
co2
variability.
F
So
if
that
is
the
case,
that
means
that
an
interactive
dust
model
would
be
really
good,
and
so
I
was
just
going
to
throw
this
out.
So
if,
if
there's
any
plans
or
so
to
include
like
an
interactive
dust
model,
I
know
that
natalie
and
others
have
done
like
lgm
and
snapshot
simulations,
but
it
would
be
nice
to
have
it
in
a
model
so
that
you
could
do
transient
simulations.
G
G
A
I
do
know
that
the
dust
in
a
cam
is
a
prognostic
quantity.
I
mean
dust,
so
we
do
have,
although
it's
kind
of
something
that
like
when
we,
when
we
did
our
mid
holocene
and
last
interglacial
cement
6
runs
it.
We
natalie
tuned
the
dust
a
little
bit
so
because
we
had
to
remove
some
vegetation
feedbacks
because
our
vegetation
wasn't
quite
right
for
those
time
periods.
So
we
kind
of
took
that
out,
but
it's
yeah
I
mean
I
think
dust
is
important.
I
know
for
the
lgm
simulation.
A
Gokan
has
questioned
the
role
of
dust
in
the
lgm
in
in
our
simulation,
so.
G
Would
it
be
difficult
to
have
these
tuning
what
natalie
did
is
kind
of
so
you
you
mentioned
that
depends
on
the
vegetation
that
we
can
somehow
include
that
actively.
So,
if
you're
changing
the
vegetation
type
that
we
have
also
different
dust
fluxes
so
that
we
don't
need
to
compute
that
always
offline.
A
Yeah,
I'm
not
that
knowledgeable
about
dust
and
cam
myself.
I
know
that
there's
a
dust
emission
factor
that
can
be
used
to
tune
dust.
I
think
it
depends
on
the
surface
conditions.
You
know
the
soils
and
the
land
surface.
The
tlai,
I
think,
is
one
of
the
factors
it's
dependent
upon.
A
Do
you
have
any
any
other
thing
to
say
jean
is
perhaps
more
knowledgeable
about
dust
and
cam.
I
guess.
J
This
partly
brings
back
to
the
discussion
of
vegetation,
because
the
dust
evasion
of
course
depends
on
the
vegetation
in
the
current
lgm
setup.
We
are
using
the
pre-industrial
vegetation
and
when
we
incorporate
the
phenology
feedback,
which
means
we
have
lei
changes
and
but
no
vegetation
type
change.
We
can
see
a
big
contribution
in
the
lgm
global
cooling.
So
I
think
the
point
is
that
vegetation
itself
is
already
very
complicated
and
it
impacts
dust.
A
Right
betty
added
in
the
chat
that
it's
a
good
question
for
the
land
model
working
group-
I
think
you
know
we-
we
could
perhaps
do
more
collaborations
with
the
land
model
working
group.
On
some
of
these
questions.
A
In
the
future
I
mean
they
provide
us
transient,
vegetation
surface
data
sets,
but
I
think
we
could
do
more
joint
projects
closer
with
the
land
model
working
group
members,
so
that
might
be
a
good,
a
good
place
to
betty.
K
Has
our
hand
raised
so
just
to
add,
maybe
john
can
say
something.
The
terrestrial
sciences
section
of
the
land
model
working
group
are
doing
some
picture,
parameter,
ensembles
of
with
clm5
and
they're
quite
interested
in
perturbing
or
using
scenarios
with
lower
co2,
so
not
just
looking
at
present
day
or
high,
but
also
low
co2.
H
A
Can
you
say
that
again
marcus?
I
know
peter.
H
I
think
I
think
permafrost
is
something
is
another
thing
I
mean
my
efforts
underway
to
actually
add
an
interactive
permafrost
model
on
top
of
the
land
model,
which
I
think
the
land
model
actually
is
quite
nice.
But
if
you
do
go
into
agm
and
want
to
look
at
the
transients.
K
I
just
wanted
to
to
back
up
what
betty
and-
and
you
were
saying
yesterday-
that
the
land
model
and
the
learn
model
working
group
should
be
very
closely
related
with
what's
going
on
paleo,
it's
obviously
a
lot
of
overlap
and
it's
really
important
to
test
the
model
outside
of.
Like
you
know,
our
current
observed
climate,
forcing
this
is
one
of
the
key
areas.
So,
yes,
the
perturbed
ensemble
is
looking
at
like
low
co2
levels.
So
can
we
run
our
model
in
in
the
sort
of
in
sort
of
paleo,
so
situations
does
that?
K
Is
it
out
of
sample
still
working
very
well?
The
other
component
is
yes,
we
would
be
very
excited
about.
How
do
we
better
bring
together
vegetation
dynamics
like
what's
happening,
with
fates
being
able
to
reproduce
like
paleo
distributions
of
vegetation?
So
this
is
an
exciting
area
that
we'd
like
to
be
involved
with,
and
obviously
we've
been
involved
in
the
last
place
of
maxima,
not
the
last
sorry,
the
transit
holocene
and
the
last
millennia.
So
we
could
somehow
create
a
project
where
we
look
at
well.
K
How
do
we
do
dynamic
vegetation
going
back
in
time
and
it's
just
in
a
systematic
way
that
works
well,
and
I
thought
there
was
a
number
of
points
that
were
brought
up
about.
How
do
you
constrain
the
four
things
that
are
put
into
paleo
runs
across
models
and
those
sort
of
comparisons
are
really
interesting.
So,
yes,
it's
really
important
for
us
in
the
development
of
clm
or
ctsm
now
to
have
these
out
of
sample
tests
to
understand.
K
If
the
model
performs
it's
one
thing
to
say:
okay,
we
can
get
the
current
day
right
or
we
get
the
historical
period
right,
but
if
we
don't
get
paleo
situations
right,
that
also
tells
us
about
what
it
looks
like
out
of
sample
going
into
the
future.
You
know
if
we,
if
our
model
has
a
certain
co2
response,
that
it
can't
be
supplied
by
the
paleo
record.
Does
that
tell
us
something
about
how
we
projected
forward
under
higher
c2
and
higher
climate
scenarios
so
yeah?
A
Great,
can
you
speak
more
directly
about
permafrost
in
ctsm?
Are
there.
K
Oh
sure,
so
I
mean
one
of
the
big
things
that
we
did
in
stage
well.
Clm5
was.
We
now
have
50
meter,
deep
soils
that
go
down
to
bedrock,
and
that
was
a
big
part
of
that
was
about
developing
frozen
layers
and
being
able
to
do
permafrost
representations.
Dave
lawrence
has
been
one
of
the
big
drivers
of
this,
so
we
do
have
these
deep
soils.
K
One
of
the
problems
is
obviously
spinning
up
the
model
to
get
the
the
deep
peats
that
allow
you
to
get
the
big
carbon
rich
soils
in
the
in
the
arctic,
but
we
are
looking
at
how
do
we
resolve
some
of
those
issues?
But
this
is
definitely
an
active
area
for
the
land
working
group
and
dave
lawrence,
especially
but
the
other.
K
The
other
side
of
it
is
understanding
what
happens
as
we
see
perma,
permafrost,
thawing
and
then
large
carbon
releases
out
of
the
peat
that
is
currently
frozen
in
the
permafrost,
but
dave
lawrence
is,
I
guess,
the
big
driver
for
that.
I
you
know.
I
know
that
peripherally.
It's
not
my
research,
but
I
think
it's
definitely
a
key
area
for
the
glam
model
working
group
and
also
for
ctsm.
K
F
Regarding
the
spin
up
time
of
the
deep
soil,
that
is
also
an
issue
for
ocean
biology,
chemistry,
ice
sheet
coupling
and
so
on,
all
these
processes
that
have
long
time
scale.
So,
for
that
reason
I
would
you
know,
advocate
for
having
a
low
resolution
version
of
the
atmospheric
model
that
you
can
actually
do
these
long
time,
simulations
easier.
K
So
the
one
thing
that
jean
didn't
comment
on
is
the
version
he's
using
fv2
by
one.
He
can
now
get
50
years
per
day
on
cheyenne,
so
that's
getting
there.
Of
course,
we'd
like
to
have
100
200
but
yeah.
If
somebody
wants
to
work
with
the
t31
by
three
marcus.
H
A
A
If
there's
no
other
comments,
we
could
continue
the
chat
conversation
on
the
discussion,
but
also,
I
hope,
to
see
you
all
back
after
our
break
we're
starting
back
again
at
10
30,
and
we
have
a
great
set
of
talks,
a
lot
of
work
with
modeling
different
time
periods
and
comparing
to
the
proxies
and
then,
after
the
end
of
our
second
discussion,
we
will
have
a
a
little
lunch
meet-up
for
those
who
want
to
stick
around
and
chat
on
some
of
these
other.
A
These
issues
that
we've
been
talking
about
maybe
build
some
and
anyway,
let's
have
a
break.
Thank
you.
A
A
A
A
J
A
Okay-
let's
see
so,
we
are
now
at
10
31,
so
I
think
we
should
just
carry
on
here
and
get
started
with
our
second
half
of
talks.
Our
first
student,
our
first
talk
is
jiwoo
han
from
who's
a
graduate
student
at
the
university
of
ucla.
A
Sorry,
she
will
be
talking
on
diatom
inferred,
water
depth
transfer
function
from
a
single
lake
in
northern
california
and
g
woo.
Do
you
want
me
to
play
your
your
recording.
A
E
Okay,
okay,
hello,
my
name
is
jiu
han.
The
title
of
today's
presentation
is
diatom:
infrared
water
depth
transfer
function
from
a
single
lake
in
northern
california
entering
the
21st
century.
There
have
been
many
changes
in
weather
and
climate
conditions
around
the
world,
and
we
are,
we
are
quite
used
to
extreme
weather.
In
particular,
california
has
experienced
water
related
disasters
such
as
severe
drought
and
fires,
frequently
in
the
21st
century,
and
its
scale
and
frequency
have
become
larger,
more
destructive
and
prolonged
to
reduce
damages
and
costs.
E
It
is
important
to
predict
and
prepare
for
the
climate
changes
to
this
end,
climate
modeling
can
be
a
useful
tool.
Climate
modelling
can
predict
future
climate
change,
but
it
can
also
simulate
the
climate
of
the
past
by
modeling
the
past
climate.
We
will
be
able
to
get
a
hint
of
what
will
happen
in
the
future.
E
Therefore,
to
simulate
the
past
hydroclimate
changes
in
california
is
my
research
goal
to
do
so.
Diatoms
reaches
the
phytoplankton
are
chosen
as
a
proxy
indicator
in
this
study
to
investigate
lake
depth
changes.
The
reasons
why
diatoms
are
a
good
indicator
are
that
they
respond
to
lactose
in
environmental
changes
sensitively.
E
They
are
common
and
primary
producers
in
lake
environment
and
they
are
well
preserved
in
lake
sediments.
Due
to
their
silica
shells
using
diatom
records.
I
will
especially
establish
a
water
depth,
transfer
function
that
can
estimate
paleo,
hydroclimate
changes
quantitatively
and
the
and
the
transfer
function
will
be
helpful
to
paleontological
studies
and
projecting
and
preparing
for
future
hydroclimate
changes
eventually.
E
So
the
research
side
of
this
study
is
kelly
lake,
which
is
in
klamath
mountains,
northern
california.
The
location
is
about
10,
kilometers,
away
of
linear
distance
from
the
border
between
california
and
oregon.
The
lake
is
around
at
1
346
meter
site
in
california.
Coast
ranges,
klamath
mountains
have
low
precipitation
during
summer
and
high
rainfalls
during
winters,
with
moderate
seasonality
in
temperature
based
on,
but
based
on
the
backgrounds.
Two
research
questions
can
be
asked
in
this
research.
One
is
the:
is
there
a
correlation
between
lake
dubs
and
diatom
assemblies
in
kelly
lake?
E
Two
is
a
single
lake
with
a
small
size
and
mid
water,
that's
applicable
to
establish
a
diatomic
water
depth
transfer
function.
To
accomplish
this
research.
The
surface
sediment
samples
were
obtained
from
kelly
lake.
The
methanol,
the
lab
members
of
ucla
and
kirby
lab
members
of
cso
fullerton
went
to
the
lake
in
july
2019
overall
40
samples
for
the
top
5
centimeters
of
lake
sediments
were
taken
using
the
mini
glue.
Coral
while
taking
the
samples,
the
observed
depth
and
the
second
depth
were
measured
too.
E
The
lake
was
deep
enough
for
this
study,
which
was
5.2
meters
deep
for
the
deepest
portion.
The
second
depth
were
measured
by
aseki
disk.
It
is
to
see
how
deep
sunlight
can
penetrate
in
the
lake,
because
diatoms
are
a
phytoplankton
and
do
the
photosynthesis,
knowing
that
the
depth
of
forex
zone
is
also
important
to
understand
the
ecology
of
the
diatoms.
E
Also,
the
water
chemistry
was
measured
to
understand
the
lacrosse
environment
better
from
the
measurement.
We
knew
that
cali
lake
was
a
fresh
water
according
to
its
low
salinity
and
is
illegal
to
mesotrophy.
Also,
the
lake
seemed
pretty
warm
during
the
summer.
The
the
map
below
shows
points
where
the
samples
were
taken.
The
red
lines
are
the
three
transacts
that
we
made,
except
the
six
samples
that
were
taken
at
the
margin
or
near
the
lake.
We
collected
all
the
samples
within
the
lake
while
trying
to
hold
various
steps
at
a
regular
interval
along
the
transects.
E
E
Most
species
are
authentic,
which
lives
on
substrates
such
as
plants
or
sediments,
and
there
are
also
a
few
of
pengtonics
diatoms
either
the
diatom
identification
followed
the
taxonomy
after
creamer
and
lynch
bertalot
in
general,
but
some
species
were
classified
according
to
the
updated
information
there
are
about
150
different
species
identified
from
the
site
and
the
dominant
species,
which
exist
more
than
one
or
five
percent.
At
least
one
sample
is
chosen
for
the
further
statistical
analysis.
E
This
diatom
diagram
is
thrown
with
the
21
dominant
species,
which
exist
more
than
five
percent.
At
least
one
sample
using
a
depth
constrained
cluster
analysis.
The
dia
the
diagram
is
divided
based
on
the
water
depth
zones.
The
three
depth
zones
are
named
as
shallow
mid
depth
and
deep
zones.
This
donation
is
assessed
by
one-way
analysis
of
similarity
to
make
sure
if
the
donation
was
correct
due
to
the
time
constraint,
I'm
not
going
over
the
details
of
the
diagram,
but
you
can
read
each
species
fluctuations
in
percentage
along
the
dot
in
kelly
lake.
E
According
to
the
graph,
it
seems
there
is
a
good
correlation
between
diatom
flora
and
water
depth
in
the
league
for
the
remained
statistical
analysis.
The
diatom
species,
which
has
more
than
one
percent
abundance
at
least
one
sample,
was
analyzed
first
of
all,
one-way
analysis.
One
way,
another
analysis
of
similarity
was
conducted
to
test
how
similar
the
diatom
community
is
between
the
data
sums,
and
it
showed
that
there
exists
the
difference
between
the
dobsons.
E
In
the
result
of
the
analysis,
it
offers
p
values
and
r
values.
P
value
indicates
how
significant
the
differences
between
the
groups
and
the
r
values
indicates.
The
strength
of
the
factors
on
the
samples
and
according
to
the
values
we
were
able
to
read
those
wizards
as
there
are
different
diatom
communities
between
the
three
depth
zones
and
lake
water.
That
is
influential
factor
to
the
diatom
composition.
E
With
this
diatom
data
set,
we
established
two
transfer
functions
by
using
two
methods,
such
as
weighted
average
pressure,
listed
squares,
wapls
and
modern
analog
technique.
Mit
after
establishing
transfer
functions,
you
can
compare
them
by
reading
r
square
and
rmsep
values.
Our
r
square
is
the
coefficient.
Determination
and
rmsdp
shows
the
error
between
the
observed
and
the
expected
values,
so
the
lower
the
value,
the
more
significant
it
is.
E
However,
the
reconstructed
depth
in
the
mid
depth
show
quite
over
representation
trend
in
mat
and
wa
pls
showed
better
performance
for
the
meter
depth.
Two
minutes.
Okay,
we
also
demonstrated
residual
scatter
plots
for
the
assessment
of
the
transfer
functions
and
it
showed
that
the
residuals
in
both
models
do
not
have
tendency.
E
To
sum
up,
there
was
a
clear
correlation
between
lake
depth
and
diatom
assemblage
in
kelly
lake.
The
performance
for
both
transfer
functions
was
good,
but
mat
transfer
model
performed
better
than
wa
pls
and
the
diatom
infrared
lake
depth
inference
model
from
the
coast.
Ranges
of
california
was
successfully
established
and
showed
the
relationship
between
diatoms
and
water
depth.
Quantitatively,
there
are
some
future
works
of
this
study
by
adding
more
modern,
modern
diatom
data
to
make
it
more
robust
and
applying
the
down
core
diatom
data
to
reconstruct
past
lake
water
adopts
changes
quantitatively,
which
is
in
progress.
A
You
jihoo,
let's
see
any,
don't
know
if
we
have
really
enough
time
to
take
a
live
question,
so
maybe
we'll
take
some
in
the
chat.
A
Okay,
yeah,
I
would
say
your
your
research
site
looked
awesome
to
go
and
gather
samples
at
yeah
all
right.
Our
next
speaker
is
from
the
university
of
texas
at
austin.
I
believe
ji
jung
is
a
graduate
student,
currently
you're
muted,
by
the
way,
and
you
can
start
sharing.
A
P
P
Awesome
thanks
thanks
for
introduction
and
hello
everyone
thanks
for
having
me
today,
I
will
show
you
our
study
about
path,
changes
in
the
great
plains,
storm
intensity
and
as
well
as
its
relationship
to
land,
surface
temperatures.
P
So
here
is
an
illustration
of
the
dynamics
of
mcs
during
the
warm
season.
Southerly
winds
called
the
great
plains.
Low-Level
jet
develops
and
brings
moisture
from
the
gulf
of
mexico
into
the
central
u.s
and
when
the
low-level
jet
meets
with
the
upper
level,
jet
stream
and
interacts
with
its
embedded
frontal
systems,
mesoscope
convective
systems
would
occur
and
they
could
last
for
from
several
hours
from
several
hours
to
days,
because
the
low
level
jet
is
a
geostrophic
wind
that
is
strongly
affected
by
the
pressure
gradient
between
the
land
and
the
ocean.
P
Here
we
report
our
new
paleoclimate
reconstruction
using
sediments
from
horse
cave
a
horse
cave
is
located
in
central
texas.
Mcs
accounts
for
the
majority
of
its
warm
season.
Rainfall
in
this
region
and
spring
is
the
wettest
season
of
the
year,
because
this
is
when
mcs
occurs
most
frequently,
and
the
pictures
on
the
lower
left
shows
what
the
cave
and
its
segments
look
like.
These
sediments
are
transported
to
the
cave
when,
when
rainfall
occurs,
and
here
we
use
a
multi-proxy
approach.
P
P
On
the
right
hand,
side
we
show
that
the
daily
rainwater
isotopic
composition
are
positively
correlated
with
the
outgoing
lung
wave
radiation,
suggesting
that
the
isotope
values
are
lower.
When
there
is
a
bigger
storm,
we
then
use
the
carbon
isotopic
composition
of
bulk
organic
matter
to
to
reconstruct
changes
in
vegetation
in
this
region.
P
P
The
leaf
wax
time
series
is
shown
in
blue
and
superimposed
on.
It
is
the
percent
titanium
record
in
red
and
at
the
bottom,
is
our
carbon
isotopes
in
green.
Our
multi-proxy
records
from
paws
cave
show
remarkably
coherent
changes
over
the
past
20
000
years,
and
notably
during
the
late
glacial
period.
Our
data
showed
that
the
mcs
activity
was
weak.
P
P
P
The
simulated
low-level
jet
shown
at
the
bottom
of
this
figure
is
consistent
with
our
proxy
records.
The
low-level
jet
shows
a
sudden
jump
near
14
000
years
ago,
and
a
gradual
weakening
from
mid
to
from
early
to
mid
housing,
followed
by
a
gradual
return
towards
the
present,
and
this
suggests
that
changes
in
the
intensity
of
the
low
level
jet
exerts
the
dominant
control
on
past
variations
in
storm
activity,
and
next
we
will
use
trace
to
understand
what
drove
those
changes
in
both
low-level
jet
and
storm
activity
during
the
deglaciation
and
housing
respectively.
P
P
At
as
the
accelerated
retreat
of
the
north
american
ice
sheet,
the
sudden
retreat
of
the
ice
sheet
causes
the
land
to
warm
dramatically
as
a
result
of
changing
albedo,
and
this
warming
drives
a
low
pressure
anomaly
over
the
land
and
a
high
pressure
anomaly
over
the
ocean.
Together,
the
zonal
pressure
gradient
would
be
deepened,
which
would
ultimately
intensify
the
geostrophic
low-level
jet
and
intensify
the
mesoscale
convective
systems
and
then,
during
the
holocene,
the
gradual
nature
of
the
mcs
evolution
suggests
that
it
is
associated
with
the
orbital
insulation
changes.
P
Our
record
shows
that
it
is
in
place
with
the
springtime
insulation
change
rather
than
summer
time
changes.
This
makes
sense
because
of
the
dominant
role
of
springtime
mcs
activity
in
this
region,
and
now,
let's
look
at
the
dynamics
of
the
of
mcs.
On
the
right
hand,
side
this
map
is
showing
the
climate
anomalies
of
pre-industrial
minus
the
mid
holocene,
as
we
can
see
here.
P
Due
to
the
larger
temperature
contrast
and
this
results
in
a
steeper
zonal
pressure
gradient
between
the
land
and
ocean,
which
makes
the
low-level
jets
stronger
today
than
the
main
holocene,
which
also
explains
the
increasing
storm
activity
from
mid-housing
to
the
present,
and
in
conclusion,
we
show
that
past
climate
changes
over
the
central
u.s
reveal
high
sensitivity
of
hydroclimate
and
extreme
weather
to
springtime
land.
Surface
warming.
P
A
Thank
you
teacher
very
interesting.
Let's
see,
I
think
we
should
move
on
and
ask
our
questions
and
have
discussions
in
the
chat.
Our
next
speaker
is
a
graduate
student
at
ohio
state.
D
University
hi
there
thanks
for
the
introduction
and
hopes.
Yes,
that
means
crazy.
D
We
believe
a
robust
method
to
recon
extract
the
acc
bioclean
transport.
It's
helpful
to
access
how
the
southern
ocean
responded
to
the
paleoclimate
conditions,
great
so
physically,
the
acc
is
dominated
by
its
viral
clinic
transport,
as
it
says
in
the
thermal
wind
balance.
This
means
that,
ideally,
the
acc
bio,
clean
transport
can
be
estimated
from
two
or
even
one
vertical
density
profiles
on
ocean
margins
across
the
acc,
which
is
verified
at
the
australia
section.
D
This
is
because
the
horizontal
density
gradient
can
be
approximated
from
two
vertical
density
profiles
and
then
the
single
vertical
density
gradient
reflects
the
horizontal
density
gradient
because
of
oceanic
subduction
process.
So
the
thousand,
so
the
density
around
the
southern
side
is
quite
uniform
and
can
be
approximated
from
the
density
and
the
bottom
depth
on
the
northern
side.
So,
however,
in
reality
we
don't
have
observations
for
even
one
single
density
profile
is
that
we
could
have
estimations
for
three
n
members,
water
masses
and
one
vertical
profile
of
the
calcite
data
18.
D
So
the
question
here
is
how
to
reconstruct
the
density
profile
at
the
lgm,
based
on
these
three
n
members
and
one
single
calculating
profile.
So
bench
stick
lists
are
a
2016
proposed
method.
That's
combined
in
member
water,
masses
and
calcite
profile,
calculating
profile
because
most
of
the
column
north
of
the
acc
can
be
well
approximated
as
a
linear
mixture
of
surface
water
and
intermediate
water
and
bottom
water
and
intermediate
water
in
the
modern
temperature
salinity
and
that
same
water
data
eating
relationship.
D
So,
therefore,
we
access
the
methods
and
australia
section
in
isotope
enable
climate
model.
I
see
example.
In
the
meantime,
we
also
hope
to
find
some
possible
improvements
in
the
reconstruction
method.
We
start
the
most
ideal
case
here.
We
assume
that
we
all-
we
have
temperature,
salinity
and
density
observations
for
these
three
n
members
and
the
vertical
calcite
data
we
can
profile
at
australia
section.
The
density
profile
actually
can
be,
will
reconstructed
reasonably
very
well
so
here
it
shows
the
model
and
the
reconstructed
temperature
salinity
and
the
density
profiles
and
the
button.
D
So
the
dashlights
here
are
from
the
reconstruction
and
the
solid
lights
are
from
the
model.
The
temperature
can
be
reconstructed
very
well,
but
we
can
see
a
large
discrepancy
between
the
model
and
reconstructed
salinity
profile
between
the
intermediate
water
and
the
bottom
water,
mainly
because
of
the
reconstruction
methodology.
D
Fortunately,
the
result
can
be
greatly
improved.
It
will
have
one
additional
poor,
fluid
observation
at
one
kilometer.
D
Yeah,
the
pink
marker
here
refers
to
this
additional
poor,
fluid
observation.
So,
as
shown
in
this
figure,
the
reconstructed
salinity
profile
are
merged
much
closer
to
the
model
profiles
below
the
intermediate
water
depths,
especially
at
the
lgm.
D
So
consequently,
the
reconstructed
density,
profile
and
transport
are
relatively
closer
to
the
model
values
in
this
case.
So
here
is
it's
just
that
adding
more
direct
information
about
temperature
and
salinity
from
four
fluid
could
reconstruct
the
lgm
acc
transport
very
well,
then,
the
observational
condition
is
more
constrained.
D
Now
we
only
have
a
density
observation
for
bottom
water
and
the
salinity
of
surface
water
and
intermediate
water
have
to
be
reconstructed
independently
so
because
the
relationship
between
the
seawater,
dentality
and
salinity
in
the
surface
and
intermediate
waters
is
almost
linear
and
this
linear
relationship
is
verified
to
be
controlled
by
similar
linear
process
as
lgm
as
today
in
the
icem.
D
So
once
we
get
numbers
of
sea
water
data
team
for
surface
and
intermediate
waters,
we
get
estimates
of
salinity.
Unfortunately,
we
cannot
pin
down
the
relationship
this
relationship
at
the
australia
section.
We
verify
that
the
relationship
between
the
surface
between
the
surface
and
intermediate
waters
can
be
reproduced
in
the
south
india
ocean,
as
a
result
of
mixing
between
the
subtropical
dry
waters
and
high
latitude
fresh
water
and
members.
D
So
the
results
are
quite
interesting.
When
we
look
at
the
reconstructed
salinity
profiles
facility
can
be
reconstructed
relatively
okay
at
the
pre-industrial
period,
but
change
dramatically
and
lgm
as
the
reconstructed
scientists
are
quite
overestimated.
D
So
therefore,
the
density
gradient
is
hugely
underestimated.
Additionally,
since
paleo
density
profiles
and
paleo
water
masses,
values
are
reconstructed
from
proxy
data,
instead
of
simply
extracting
them
from
the
model
they
are
subjected
to
a
number
of
synthetics.
Therefore,
we
also
perform
multi-color
estimation
in
this
case
to
access
sensitivity
of
the
reconstructed,
paleocentes
and
acc
transport.
D
So
the
rest,
the
test
shows
that
this
case
would
yield
a
very
large
error
of
the
transport
calculation
and
the
uncertainties
are
in
estimating
surface
and
intermediate
water
temperature
and
salinity
account
for
much
of
the
total
area
on
the
transport
calculation.
D
So
but
but
overall,
we
believe,
that's
a
modest
direct
measurement
of
poor
fluid
of
temperature
and
oscillating
interpret
based
on
the
calcite
that
I
team
profile
can
is
a
is
available
approach.
So
that's
all
about
my
this
work
and
I
welcome
any
comments
and
questions.
A
Oh,
I
think
you
do
you
have
another
slide.
No,
no!
Okay!
Do
we
have
any
questions
we
have
about
a
minute
or
so
to
take.
Otherwise
we
could
don't
see
hands
or
do
I?
No.
A
I
guess
we'll
take
questions
and
comments
in
the
chat.
Then.
Thank
you
ling.
Thank
you.
Thank
you.
Lingway.
Let's
move
on
to
clay
tapers
talk.
A
A
I
All
right
great,
so
thanks
for
having
me
yeah,
so
I
just
wanted
to
talk
on
a
little
update.
We've
made
on
looking
at
western
u.s
ice,
isotopic
change
and
the
last
glacial
maximum,
and
this
is
this
work.
I've
done
in
collaboration
with
quite
a
few
people
shown
here
and
others
that
are
not
on
here
as
well.
I
All
right,
so
why
are
we
interested
in
the
western
u.s
at
the
last
glacial
maximum?
Well,
it's
kind
of
a
unique
hydroclimate
response
at
the
lgm,
and
so
here
in
this
first
panel.
Here
we
have
changes
in
overall
wetter
and
dry
air
conditions
with
drier
conditions
shown
in
red
and
wetter
conditions
shown
in
blue
for
the
last
glacial
maximum.
I
So,
in
order
to
better
understand
the
western
u.s
hydroclimate
response
at
the
last
glacial
maximum,
we
use
the
community
earth
system
model
and
in
this
version
we
have
water,
isotope,
log,
tracers
and
a
version
of
the
model
known
as
icasm
1.3,
and
the
model
generally
does
a
fairly
good
job
at
simulating
the
present-day
isotopic
distribution
of
precipitation.
If
you
want
more
information
about
the
strengths
and
weaknesses
of
the
model
check
out,
some
of
these
references
shown
here.
I
For
for
the
talk
today,
I'll
just
be
looking
at
two
time
periods:
one
is
the
last
glacial
maximum
and
one
is
the
pre-industrial
and
we're
using
one
degree
configuration
of
the
model.
So
we
initialize
these
runs
from
itrace
simulations
and
then
we
increased
the
atmosphere
and
land
resolution
to
one
degree
in
order
to
better
simulate
the
topography
of
the
western
u.s,
and
we
also
added
high
frequency
outputs
to
look
at
storm
activity
and
water
tagging
to
look
at
where
moisture
source
is
originated
and
we'll
be
comparing
these
with
speed.
I
So
all
these
plots
I'll
be
showing
for
the
next
few
slides
are
anomalies
between
the
lgm
minus
the
pre-industrial.
And
here
I'm
just
looking
at
the
isotopic
response
on
the
left
panel.
Here.
This
dashed
area
is
where
we
have
the
western
u.s
as
we've
defined
it,
and
what
you
can
see
is
that
at
the
lgm
there's
a
significant
depletion
in
the
delta
18-0
precipitation
throughout
most
of
the
western
u.s,
and
we
also
see
in
the
right
panel
here
the
precipitation
response
where
we
get
this
drying
in
the
northwest.
I
So
here
I'm
just
showing
the
same
plots,
but
for
winter
delta,
180,
a
precipitation
change
on
the
left
and
winter
precipitation
change
on
the
right,
and
they
generally
reflect
the
annual
mean
signal
primarily
due
to
the
fact
that
they're
largely
driving
the
annual
mean
response
and
so
for
the
next
few
slides
I'll,
be
looking
primarily
at
the
winter
responses
and
in
this
case
I've
defined
winter
as
november
through
march.
I
And
so
this
is
the
delta
html
precipitation
change
in
the
winter
months
from
this
particular
region,
so
moisture
coming
from
this
region
in
the
top
left
panel
and
then
the
delta
180
of
the
water
vapor
associated
with
that
precipitation.
So
what
you
can
see
is
there's
a
clear
correlation
between
the
isotopic
response
of
the
precipitation
and
the
isotopic
response
of
the
water
vapor.
I
We
also
see
that
the
lgm
is
bottom
bottom,
two
panels
that
the
total
amount
of
water
vapor
coming
from
the
central
east
north
pacific
region
and
the
total
amount
of
precipitation
coming
from
the
central
east
north
pacific
region
has
increased
relatively.
So
it's
contributing
more
to
the
overall
signal
at
the
lgm.
I
I
But
we
don't
see
a
significant
increase
in
atmospheric
river
activity,
and
so
this
could
be
related
to
how
the
atmospheric
rivers
are
identified
and
because
we
have
such
cooler
temperatures,
we
have
a
lot
less
water
vapor
in
the
atmosphere
overall.
But
we
we
see
actual
decrease
in
atmospheric
rivers
in
the
northern
part
of
the
the
study
region
and
then
a
kind
of
a
maintenance,
atmospheric
river
activity
a
little
bit
further
south,
and
so,
even
though
we're
not
necessarily
getting
an
increase
in
atmospheric
river
activity.
I
We're
getting
more
or
less
maintenance
of
integrated
vapor
transport,
as
shown
here
in
the
bottom
right
in
the
in
the
southwest
region,
and
that's
somewhat
substantial,
because
we
have
significantly
less
water
vapor
in
the
atmosphere
overall.
So
there's
actually
a
significant
increase
in
the
low-level
jet,
showing
the
bottom
right
here
that
helps
to
maintain
that
vapor
transport,
despite
this
reduction
in
total
water
vapor
of
the
column
at
the
lgm.
I
And
so
one
of
these
figures
just
kind
of
highlight
this-
is
this
different
and
constant
density
lines?
The
blue
is
the
lgm
and
the
red
is
the
pre-industrial
for
a
cross-section
around
40
degrees
north
across
the
western
u.s,
and
what
we
see
is
much
steeper
lines
of
constant
density
at
the
lgm
and
so
what's
going
on.
Is
you
have
this
increase
isentropic
upload
of
these
density
gradients,
leading
to
more
rain
out
and
more
precipitation
efficiency,
as
shown
in
this
bottom
panel
here,
and
what
that
does?
I
And
finally,
this
depletion
signal
is
also
partly
due
to
changes
in
seasonality
of
the
precipitation,
so
winter
precipitation
overall,
is
more
depleted
than
summer
precipitation
for
both
the
pre-industrial
and
the
lgm.
It's
just
shown
in
this
example
on
the
left
here
and
we
also
in
the
lgm,
get
more
of
our
total
annual
precipitation
from
the
winter
months,
so
that
more
depleted
signal
gets
translated
and
in
into
the
annual
signal,
because
you
have
more
of
this
winter
precipitation
at
the
lgm,
as
shown
in
the
percent
differences
in
precip
on
the
right
all
right.
I
So
we
can
compare
with
the
spelathin
records,
and
my
collaborators
have
collected
several
different
speedo
thermal
records
from
our
study
region,
and
these
records
are
transient
records
through
most
of
the
deglaciation,
but
unfortunately
none
of
them
really
extend
all
the
way
or
a
few
of
them
extend
to
present
day.
So
we
can't
really
work
with
anomaly
space
here,
comparing
pre-industrial
lgm
we're
really
just
comparing
the
absolute
values
of
the
lgm
simulations
with
these
speliothm
lgm
measurements.
I
I
You
know
the
precipitation
precipitation
amount,
a
temperature
and
evaporation
and
put
those
values
into
these
forward
proxy
mark
models,
cave
calc
and
car
solution,
and
so
then
we
just
compared
the
speleothem
measured
delta
18
at
the
lgm,
with
our
model,
simulated,
lgm
values,
and
so
the
actual
measurements
are
in
black
and
then
our
comparisons
for
these
different
cave
sites
with
our
different
model,
ford,
proxy
model,
outputs,
are
in
these
different
colors.
I
What
we
find
is
that
overall,
we're
doing
a
fairly
good
job
at
capturing
this
isotopic
signal
at
the
lgm,
and
we
do
an
even
better
job
for
some
of
these
caves,
where
we
have
monitoring
ongoing,
where
we
can
refine
some
of
the
different
knobs
in
the
ford
proxy
model,
as
shown
in
this
bottle
bottom
plot.
Here
we
don't
get
great
comparison
and
cave
without
a
name,
but
that's
a
little
bit
outside
of
the
study
area
and
there's
a
bias
in
the
pre-industrial
configuration
of
icem
for
this
as
well.
I
So
we
think
this
is
kind
of
a
systematic
model
bias
here,
all
right,
so
just
to
summarize
the
icsm
simulates
depletion
and
the
adopted
adrenal
precipitation
in
the
western
u.s
at
the
lgm,
and
this
doesn't
completely
align
with
the
precipitation
itself,
but
the
delta
180
precipitation
response.
I
You
know
measurements,
so
I
think
I'm
probably
out
of
time,
but
what
we're
doing
from
here
on
out
is
trying
to
use
higher
resolution
simulations
in
order
to
better
understand
the
small
scale
variability
in
trying
to
resolve
some
of
these
topographic
features,
which
might
lead
to
more
or
less
change
at
the
lg
and
with
that
I'd
be
happy
to
take
any
questions.
A
Thank
you,
clay.
Nice
talk.
I
think
we
should
move
on
and
take
clay's
questions
or
any
discussions
in
the
chat.
Our
next
speaker
is
chung
feihei,
a
research
assistant
at
ohio
state
university,
and
he
will
be
speaking
on
abrupt,
heinrich
stadial.
One
cooling
missing
in
greenland
oxygen,
isotopes.
J
J
However,
local
to
an
atlantic,
the
greenland
data
rating
seems
logically
unchanged
at
the
onset
of
hs1.
So
even
though
it
shows
a
clear
above
change
at
the
onset
of
jungle
drought,
so
this
is
actually
very
weird,
hopefully
the
possible
that
into
similar
events,
one
had
this
clearable
change,
but
the
other
one
just
doesn't
so.
Our
question
here
is:
is
the
onset
of
hs1,
ready,
muted
in
the
green
and
high
score
record?
J
J
So
four
things
of
ideas
is
quite
similar
to
the
previous
just
20k
simulation,
but
in
iris
we
have
four
runs
with
factorized
forcings,
so
first
run
is
the
glacial
retreating
ice
sheet,
and
on
top
of
this
we
add
orbital
forcing
then
the
greenhouse
gas
forcing
and
finally,
the
mild
water,
forcing
all
these
four
ones
were
integrated
for
9000
years
from
20k
to
11k.
J
That
covers
the
whole
division
period,
and
the
most
important
thing
here
is
that
we
have
stable
water
isotope
incorporating
the
model,
so
we
are
able
to
compare
the
model
data
routine
with
observation
directly
now,
let's
move
back
to
the
green
line.
Let's
see
how
the
climate
change
in
ihs
performs
against
the
always
observation,
so
on
left
axis,
it
is
model
variable
and
on
the
right.
It
is
corresponding
observation
in
each
panel
we
are
doing
our
apple
triple
comparison.
J
J
J
So,
even
though
this
cooling
is
not
seen
in
the
observations,
this
cooling
in
the
model
is
actually
quite
consistent
with
other
proxies
like
the
sea
sub
temperature
in
the
noslantic
and
the
sperm
data
rating
in
kulukev
in
east
asia.
They
are
all
consistent
with
a
slowdown
in
mark
so
here
that
actually
suggests
that
the
data
eating
is
muted,
but
the
temperature
is
not,
then
why?
Because
model
is
wrong
or
something
missing
is
observation.
J
Previous
previous
studies
suggest
that
the
change
of
the
ischool
data
routine
is
composed
by
the
change
of
the
present
sectionality,
as
well
as
the
isotopic
composition.
So
a
increase
in
analytics
would
enrich
the
ischool
diary
routine
because,
for
example,
or
increase
in
the
sum
of
precipitation
would
lead
to
the
isolated
routine
accumulates
more
the
heavy
summer
isotope,
and
for
this
isotopic
composition,
it
is
affected
by
several
factors
like
the
surface
temperature,
the
source
of
the
moisture,
as
well
as
the
strength
of
the
inversion
layer.
J
So,
based
on
this
idea,
we
analyzed
these
two
effects
over
greenland
during
the
last
degradation.
What
I
show
here,
the
radar
curve
is
the
isotopic
composition
effect
and
the
blue
and
orange
curve
are
the
presidency
than
added
effect.
We
see
they're
actually
canceling
each
other
during
the
most
of
the
time,
in
the
last
degree
station
in
particular
in
hs1
and
a
younger
trial,
and
the
net
effect
is
the
black
curve.
It
is
the
change
of
the
ischool
data
routine
and
then
why
these
two
effects
are
canceling
each
other
in
these
two
periods.
J
So
for
this
isotopic
composition
effect,
I
further
analyzes
seasonal
evolution
of
the
surface
temperature.
What
we
see
here,
the
red
curve
in
the
summer
summer,
on
the
surface
temperature
during
last
degradation,
it
is
almost
like
a
constant
and
so
does
that
oc
in
precipitation,
but
for
winter
temperature
it
reduces
a
lot
in
hs1
and
a
younger
drought
and
due
to
temperature
effect,
the
data
reaching
precipitation
also
depletes,
so
a
decrease
in
winter
precipitation
lead
to
this
depletion
in
the
isotopic
composition
and
for
the
precipitation
effect.
J
We
also
checked
the
precipitation
evolution
we
see
quite
similar
to
the
winter
surface
temperature.
The
winter
precipitation
also
reduces
a
lot
in
hs1
and
a
younger
drought,
even
though
the
sum
of
precipitation
increases
in
a
linear
way.
So
as
a
result,
a
blood
change
of
the
winter
precipitation
leads
to
enhanced
precipitationality.
J
Therefore,
it
leads
to
enrichment
in
the
ice
core.ot.
So
here
we
find
that
actually
it
is
a
winter
clamp
that
determines
these
two
effects.
A
decrease
in
winter
temperature
leads
to
a
depletion
in
the
isotopic
composition
and
a
decrease
in
winter
precipitation
lead
to
this
enhanced
precipitancy
than
that
effect.
J
So
the
question
is
what
control
this
winter
climate
change
in
the
greenland?
Actually,
we
find
it
is
because
the
sea
ice
winter
time
sea
ice
in
the
north
landing.
What
I
show
here
is
the
leading
leading
mode
of
the
maximum
covariance
analysis
between
the
winter
time
sea
ice
and
the
climate
change
in
the
north
atlantic.
J
The
leading
mode
accounts
for
more
than
99
of
the
total
variance,
so
they
suggest
that
the
winter
sea
ice
and
the
climate
change
in
the
non-atlantic
and
greenland
area
coherent,
and
this
coherence
is
because
the
sea
is
controlled,
the
water
source
and
energy
source
for
the
green
land.
First
for
the
water
source,
for
example,
an
expansion
of
the
ice
from
lgm21
will
shut
off
the
evaporation
in
the
north
landing,
so
the
precipitation
over
greenland
reduces
in
winter.
Therefore,
it
leads
to
enhanced
presence
of
another
effect
well
for
this
energy
source.
J
So
expansion
of
the
sea
ice
would
also
shut
off
the
heat
flux
from
ocean
to
the
atmosphere.
So
there
is
a
widespread
cooling
in
the
north
atlantic
during
heritage
events
due
to
temperature
effect.
The
data
eating
water
vapor
also
depletes,
and
therefore
the
data
routine
precipitation
over
over
greenland
also
depletes.
So,
basically,
is
linked
to
the
depletion
in
the
isotopic
composition
effect.
J
Well,
at
the
onset
well
from
lgm
to
hs1
the
cs.
Expansion
is
much
smaller
compared
to
from
boiling
error
to
younger
gel,
because
the
lgm
is
much
colder
than
the
boiling
area,
so
the
sea
ice
energy
is
much
softer.
Therefore,
a
smaller
cs
expansion
lead
to
these
two
effects
actually
lightly
cancels
each
other
due
to
a
largely
mutated
dioding
response
at
the
onset
of
hs1,
but
later
at
the
end
of
the
younger
trial
that
it
is
automated
by
this
isotopic
depletion.
J
Therefore,
we
see
a
clear
above
change
at
the
onset
of
the
younger
gel
so
further.
Our
studies
suggest
that
there
is
a
blood
cooling
at
the
onset
of
hs1,
because
expansion
on
the
sea
ice.
Therefore,
based
on
this
analysis,
we
suggest
that
if
we
rely
on
the
data
waiting
to
infer
the
cooling
in
past
hemorrhage
events,
so
these
coolings
are
likely
to
be
underestimated.
J
J
A
Thank
you,
chung
fei
nice
talk
and
we
have
a
minute,
maybe
for
a
question.
If
anyone
has
a
question
for
chung
faye.
K
A
Sometimes
people
are
catching
up
in
the
chat
with
previous
talks.
It's
kind
of
hard
to
juggle
all
of
these
windows
and
try
to
listen
to
a
talk
and
also
comment
on
previous
talks,
so
bear
with
everyone.
A
Yeah,
let's
move
on
next,
we
have
hannah
zanowski
who's,
a
post-doc
at
the
university
of
colorado.
Here
in
boulder.
She
will
be
speaking
on
pacific
ocean
14c,
radiocarbon
evolution
during
the
deglaciation.
In
the
c
I
trace
simulation.
Q
Awesome,
thank
you
esther.
I
should
also
let
everyone
know
that
this
fall.
I
will
be
starting
as
an
assistant
professor
at
the
university
of
wisconsin-madison,
so
pretty
big
deal
so
just
to
set
the
stage
for
everyone
here.
Q
The
glacial
circulation
of
the
pacific
ocean
was
actually
a
lot
different
at
the
lgm
than
it
was
today,
and
you
can
kind
of
see
this
from
this
sort
of
schematic
from
matsumoto
at
all
2002
on
the
right
in
this
picture,
which
basically
shows
sort
of
the
deep
inflow
of
the
bottom
waters
outflow
of
north
pacific,
deep
water,
and
also
some
inflow
of
north
pacific
invading
water
and
such
and
so
the
major
differences
between
the
lgm
and
present
day.
Q
And
in
these
two
time
periods-
and
you
can
kind
of
see
that,
from
these
two
sections
from
radal
2020
of
lgm
and
modern
day,
carbon
14
age,
where
you
see
really
really
old,
carbon
14
ages
in
the
deep
pacific
and
relatively
younger
ages
in
the
upper
ocean-
and
so
one
of
the
questions
that
we
have,
though,
is
how
do
we
go
from
this
lgm
circulation
to
the
circulation
in
the
present
day,
and
so
to
be
able
to
answer
that
we're
going
to
use
the
simulation
called
c?
I
trace.
Q
It
was
run
from
22
000
years
ago
to
present
day
it's
forced
with
a
bunch
of
different
fluxes
from
the
trace,
21
simulation,
and
it
also
has
isotope
tracers
enabled
in
it.
So
we
can
kind
of
get
at
these
sort
of
circulation
changes,
particularly
by
looking
at
carbon
14,
and
that's
what
I'm
going
to
look
in
the
pacific.
So
this
figure
on
the
right
is
just
a
sort
of
summary.
Q
If
you
will
of
the
sort
of
major
forcings,
we've
got
the
prescribed
atmospheric
carbon
14
in
the
black
line,
the
increasing
atmospheric
scene
2,
which
is
also
prescribed
in
this
model,
as
well
as
some
of
the
major
sort
of
circulation
metrics
and
the
fresh
water
horsing
in
the
northern
and
the
southern
hemisphere.
So
this
model
is
capable
of
sort
of
modeling
the
periods
of
abrupt
change
throughout
this
time
period,
so
yeah
so
diving
right
into
the
carbon
14,
though.
So.
Q
What
I
have
here
are
zonal
mean
sections
of
carbon
to
14
at
various
time
periods.
Throughout
this
model
simulation,
the
top
is
the
actual
carbon
14
values.
The
bottom
is
the
carbon
14
age
and
so
largely
kind
of
what
you're
seeing
from
this
is,
even
though
we've
kind
of
subtracted
the
atmosphere
out,
there's
this
large
scale
sort
of
decline
in
the
carbon-14
values
in
the
ocean
and
sort
of
this.
Q
You
know
basically
overall
making
the
waters
younger
over
time
as
well.
Now,
if
you
kind
of
compare
this
to
present
day,
though
this
model
is
still
fairly
biased,
so
we
take
the
present
day,
values
of
the
c
trace
c.
I
trace
stimulation.
Q
We
subtract
the
loadout
values
from
that
and
this
kind
of
gives
you
a
sense
of
what
the
biases
are
in
this
model,
and
so
that
does
then
of
course,
lead
to
relatively
older
ages
within
the
model,
and
that
again
is
tied
to
the
fact
that
the
circulation
in
general
is
far
too
sluggish
within
here.
So
nevertheless,
though,
we
can
go
ahead
and
then
compare
our
isotope
the
isotopes
within
the
model,
though,
to
core
records
around
the
pacific
ocean.
So
that's
what
I'm
doing
here
we're
looking
at
carbon
14
ages.
Q
These
are
benthic
minus
planktonic
ages.
For
these
cores,
which
are
in
purple,
the
model
is
in
black
and
then
the
gray
line
over
time
is
basically
sort
of
a
corrected
version
of
this
model
or
an
adjusted
version
where
basically,
I'm
just
shifting
the
entire
record
up
or
down,
based
on
the
bias
relative
to
the
modern
day,
bias
relative
to
galota
and
that's
just
to
kind
of
get
a
sense
of
of
how
much
this
bias
can
actually
sort
of
affect
the
ages
that
we
get
so
again.
Q
There
are
a
lot
of
differences
compared
to
these
cores,
but
at
the
same
time,
if
you
kind
of
look
at
these
things
closely,
you
do
see
that
this
model
is
actually
capable
of
seeing
or
at
least
simulating
some
of
the
carbon
14-h
variations
that
we
do
see
in
these
core
records.
And
so
I
think
that
is
actually
really
interesting.
Q
And
so
these
are
sort
of
point
wise
comparisons
between
like
one
grade
cell
in
the
model
and
then
the
course
so
to
understand
something
about
carbon
14
circulation
and
a
particular
circulation
within
the
pacific
ocean.
Then
what
I'm
doing
here
now
is
I'm
just
showing
you
averages
in
sort
of
larger
boxes
within
those
regions
and
the
same
regions
that
contain
those
cores,
and
so
this
is
now
bottom
again.
Benthic
minus
platonic.
Q
If
you
will
carbon
14
over
time
within
these
same
regions
and
in
the
blue
and
then
what
I
have
are
two
other
ocean
tracers
that
we
use
in
this
model
ideal
age
which
basically
is
set
to
zero
at
the
surface
and
it
just
ages
in
the
interior.
Q
That's
a
measure
of
the
circulation
age
and
then
ventilation
age,
which
is
set
to
zero
at
the
surface,
but
only
if
the
water
is
actually
come
into
contact
with
the
atmosphere,
and
so
this
is
to
kind
of
give
you
effect
the
effect
of
like
what
happens
underneath
ice
in
particular
when
you
don't
actually
have
direct
exchange
with
the
atmosphere
and
some
of
that
sort
of
disequilibrium
that
we
get
so
overall.
What
you
see
is
this
sort
of
similar
behavior
between
all
three
three
tracers
generally
ventilation.
Q
A's
is
closer
to
carbon-14
and
then,
but
there
are
major
differences
between
the
14
and
both
ideal
age
and
ventilation,
age,
and
that,
of
course,
is
because
carbon
14
has
this
full-blown
surface
reservoir
age
effect,
which,
because
it
can't
takes
a
really
long
time
to
equilibrate
with
the
atmosphere.
So
if
we
think
about
carbon-14
as
a
circulation
age
plus
a
reservoir
age
and
plus
some
sort
of
residual
that's
associated
with
mixing,
we
can
actually
get
some
sense,
then
of
what
those
sort
of
other
two
processes.
Q
So
this
figure
on
the
on
the
right
is
just
an
a
ratio
of
ideal
age
to
carbon
14
h,
and
so
these
are
depth
versus
time
profiles
now
of
those
same
regions
and
what
you
get
is
that
yeah,
in
some
time
periods,
you're
seeing
this
relative
match,
where
you
have
these
greener
colors,
where
ideal
age
and
carbon
14
age
track
each
other
well,
but
there's
plenty
of
other
times
and
and
different
depths
as
well,
where
you
know
this
reservoir
age
effect
and
potentially
this
residual
as
well
kind
of
strongly
impact
the
carbon
14
ages
that
you
see
so
this
kind
of
helps,
you
figure
out
what
you're
actually
measuring
when
you're
comparing
to
these
proxy
records.
Q
So
nevertheless,
though,
if
you
actually
do
want
to
use
carbon-14
age
purely
as
a
circulation
age,
tracer,
a
really
simple
way
of
doing
that
is
basically
just
and
you
can
do
this
in
a
model.
That's
the
glory
of
having
a
model.
Is
that
you
have.
You
know
time
points
everywhere.
You
can
do
this
by
basically
subtracting
the
ages
of
the
source
waters
from
some
other
region.
Basically,
and
so
example
of
this
is
in
the
southwest
pacific
and
the
deep
western
boundary
current,
we
know
that
the
waters
generally
come
from
the
deep
ross
sea.
Q
So
if
you
subtract
the
ages
of
those
two
things,
you
can
get
a
sense
of
what
the
actual
changes
are
over
time,
and
so
now,
what
you
see
is
that
all
three
of
these
tracers
actually
track
each
other
pretty
well.
Similarly,
you
can
do
this
in
the
northeast
pacific,
where
you
just
subtract
the
deep
north
atlantic
age.
Q
This
is
average
between
two
and
four
kilometers,
which
is
more
or
less
where
you'd
expect
to
have
north
atlantic
water
coming
in
eventually
as
north
as
pacific,
deep
water,
and
so
it
doesn't
match,
perhaps
particularly
as
well,
because
there
are
other
water
masses
within
that
region
or
other
type
waters
that
come
from
different
places.
But
again
you
can
kind
of
see
this
similar
behavior
across
the
models.
So
just
to
conclude
so
despite
we
have
these
mean
state
biases.
Q
I
think
it's
pretty
it's
pretty
interesting,
given
that
this
is
a
fairly
low
resolution
model
and
then
this
carbon
14
age
itself,
if
you
can
really
tell
that
it
reflects
the
circulation
age
and
this
reservoir
age
change
now,
and
we
can
use
this
model
to
do
a
lot
of
different
and
interesting
analyses
as
a
result
of
having
all
of
these
sort
of
traces
and
all
of
these
things
actually
wrapped
up
within
within
this
long-term
simulation,
so
really
exciting.
Chris,
the
output
is
actually
now
publicly
available.
Q
There's
a
data
set
has
a
doi,
and
if
you
want
to
learn
more
about
it,
you
can
visit
the
ci
trace
website
here.
So
I
will
leave
it
at
that,
and
I
think
I
blew
through
that
an
incredibly
blistering
pace.
A
Well,
everyone's
it,
I
guess,
if,
with
your
permission,
everyone's
talk
is
gonna
be
posted
eventually,
so
if
there's
details,
people
missed
and
any
talk
and
and
not
just
pointing
at
you
hannah
yeah.
These
talks
go
really
fast
for
me
because
I'm
trying
to
juggle
different
things
but
they're
all
going
to
be
posted
on
the
website.
A
So
we
can
refer
to.
You
know,
look
at
them
later
and.
A
And
so
this
was
an
I
just:
do
we
have
enough
time
for
yeah
question?
This
was
an
ocean,
forced
experiment
and
do
you
have
any
feel
for
how
that
might
have
impacted
any
of
these
results?
I
know
you
used
use
atmospheric,
forcing
from
the
trace
violation,
correct
yeah.
A
Any
I
mean
if
this
would
be
coupled.
Would
there
be
any
because
I
know
this
carbon
tracers
are
forced
anyway
from
the
atmosphere,
but.
Q
Yeah,
I
think
it
does
actually
impact
like
the
eventual
carbon
values
and
the
deep
in
the
in
the
ocean.
Just
because
you
kind
of
have
to
you're,
not
ocean,
isn't
really
allowed
to
out.
You
know
the
circulation
isn't
actually
allowed
to
come
in.
Oh
yeah,
the
atmosphere
right,
so
so
that's
kind
of
the
biggest
thing,
and
even
if
you
force
it
like
that,
you
can
sometimes
actually
then
lead
to
values
that
I
think
are
lower
than
they
would
they
should
be
otherwise.
So
there
isn't
bias
with
that.
Q
I
do
think
a
fully
coupled
simulation
in
all
honesty
would
be
the
best
way
to
go.
So
if
we
ever
get
these
things
in
through
the
atmosphere
and
in
the
land
model,
I
think
that
would
be
most
ideal,
but
this
is
a
good
first
step.
Q
Q
Right,
so
this
is
kind
of
the
interim
step
which
is
like
all
right,
just
force
it
with
that.
I
do
know
that,
for
the
most
part,
the
model
ocean.
Actually
this
the
ocean
in
this
model,
does
track
the.
Q
I
think
the
temperature
and
salinity
pretty
well
compared
to
trace
21,
but
I
know
that
there
are
some
biases
because
we
have
they
have
to
do
a
lot
of,
I
think
surface,
restoring
just
to
kind
of
get
everything
and
pretty
strong
service,
restoring,
I
think,
and
particularly
in
the
nordic
seas
and
stuff,
like
that,
so
that
might
eventually
lead
to
sort
of
differences
between
the
two
simulations.
At
the
end
of
the
day,.
A
I
do
see
andreas
hand
raised,
but
we
probably
need
to
move
on
to
the
the
next
speaker.
So
maybe,
if
you
can
ask
your
question
in
the
chat
andreas,
so
our
next
speaker
is
from
he's
a
researcher
at
santa
barbara.
R
Thanks
for
introduction-
and
this
is
qinghua-
and
so
the
talk
I
will
present
today
is
the
work
led
by
my
student
and
did
with
my
coworker
and
the
colleague
from
a
different
institute
and
the
university
in
united
states
and
china
and
from
the
topic
you
can
see.
The
main
idea
is
that
the
time
scale
we
try
to
focus
on
the
2000
the
past
2000
years,
and
we
focus
on
some
data.
It's
called
a
last
millennium,
real
reconstruction
or
reanalysis.
R
Our
mr
data
version
two
and
the
goal
is
to
understand
circulation
change
over
the
two
pa
to
2000
years,
and
we
try
to
understand
some
role
of
internal
variability
like
the
ipo
interdicator
pacific
oscillation.
How
does
the
tropical
signal
or
tropical
forcing
drive
a
circulation
chain
over
the
two
pole
or
two
hemisphere
over
like
the
multiplication
time
scale,
and
also
to
compare
this
to
this,
like
recurrent
mode
recurring
mode
with
the
recent
40
years
of
our
ability?
R
So
this
is
the
main
goal,
and
another
thing
I
like
to
draw
your
attention
is
that
if
you
look
at
the
this
panel
here
is
this
is
from
three
levels
or
is
the
linear
trend
and
using
annual
mean
data
and
z200
as
a
means
of
200
millibar
height,
a
500
millibar
height
in
the
middle
and
the
surface
temperature,
and
this
is
the
ford
three
data
site
era
5
and
there
is
a
75,
a
large
ensemble
historical
run
and
mainly
reflect
is
the
historical
I
mean
the
co2
forcing
a
response
to
pseudo
faulting.
R
Another
one
is
a
large
example.
Many
many
models.
Each
model
have
a
40
from
40
to
100
members,
so
add
them
together,
so
csm1
and
the
jfdl
model.
Another
several
model.
Add
them
together.
So
the
idea
here
is
that
if
you
look
at
this
map,
you'll
find
some
discrepancy,
but
there's
also
some
similarity.
The
similarities,
the
global
warming
signal
warm
everywhere
and
in
the
model
it
tends
to
give
a
really
uniform
change
everywhere,
like
like
in
the
model
here,
you
select,
the
hat.
R
The
height
change
is
the
has
a
maximum
in
the
topics,
but
in
observation,
if
you
look
the
same
variable
same
pattern:
the
same
trend-
you
don't
find
at
least
a
maximum
in
in
the
topics,
and
if
you
look
this
surface
temperature
you'll
find
that
there's
some.
We
call
a
non-uniform
feature:
that's
like
the
maximum
warming
over
the
arctic
and
a
little
bit
cooling
around
the
southern
ocean
and
no
change
over
india,
eastern
pacific.
But
the
model
give
give
us
like
a
co2
force
in
favor.
R
Is
the
el
nino
like
pattern
and
you
know
warming,
and
so
we
think
there's
some
mismatch
or
like
the
discrepancy
we
try
to
understand.
Why
is
that?
We
think
it's
the
due
to
internal
variability,
because
we
know
the
model
can
get
to
this
arctic
amplification
due
to
this
co2
or
like
a
party
feedback
there.
But
but
the
problem
is
the
maximum
warming
like
the
height
of
chain
is
not
the
maximum.
It
has
no
maximum
over
the
polar
rating.
R
It's
the
maximum
height
of
rise
over
the
tropics,
but
in
observation
there
is
like
a
co-located
between
the
two
two
circulation,
the
two
two
variable
in
a
different
level.
So
this
means
the
the
maximum
high
rise
in
the
arctic,
but
also
maximum
surface
temperature
in
arctic
and
the
cooling
here
in
the
surface,
but
also
at
the
lower
office
of
japanese
high
in
in
southern
hemisphere.
So
this
means
that
the
dynamic
behind
the
two
warming
scenario
it's
a
bit
different,
something
maybe
due
to
its
internal
variability.
R
So
we
think
it's
a
probability
is
the
tropical
forcing
or
something
related
to
the
ipo?
And
this
is
a
very
simple
plot.
As
this
one
from
observation,
we
just
simply
remove
zonomin
component
from
something
like
this:
the
shorter
period,
because
this
internal
mode
and
the
internal
mode
have
some
like
the
arctic
period
like
the
upper
upper
trend
from
1999
to
2013..
So
it
just
simply
remove
the
the
zonon
component
from
this
variable.
R
We
can
find
there's
a
very
clear
tropical,
driven
like
a
cooling
driven
electric
pattern,
propagate
propagating
to
the
two
hemisphere,
so
favor
warming
order,
greenland
and
arctic
and
the
favorite
cooling
over
southern
ocean.
So
there's
a
proper
reason
to
explain
this,
the
symmetry
between
these
the
polar
change
like
the
article
maximum
warming
or
called
a
a
or
something
like
the
no
change
or
southern
ocean
or
close
to
antarctica.
So
this
is
the
main
idea.
So
we
want
this
reason
for
the
years
we
want
to
see.
R
This
is
the
how
how
how
the
similar
mode
behave
over
the
past
2000
years.
So
we
looked
at
some
like
2000
years
data,
but
before
that
we
simply
calculate
uf
like
using
the
recent
40
years.
We
find
that
this
mode
is
a
leading
mode
in
a
different
level,
just
like
to
show
like
article
warming,
corresponding
to
southern
ocean
cooling
and
all
connected
like
tropical
cooling.
R
So
it's
a
more
like
ipo
signal
so
because
we
give
a
name,
it's
called
ipo,
bts
ipo
means
ipo
related,
bipolar
teleconnection,
something
like
that,
and
so
this
is
from
another
paper
just
to
show
a
very
typical
ipo
mode.
So
this
is
just
a
opposite
sign,
so
there's
the
classical
ipo
pattern,
and
so
now
we
look.
R
This
is
the
lmr
data,
so
lmr
data
is
the
reconstruction
data
combined
model
cssm4
solving
year
simulation
with
some
proxy
data
everywhere,
just
like
a
globe
in
globe
and
things
that
they
approached
data
is
a
it's
a
more
it's
available,
it's
a
more
intense
over
like
a
northern
hemisphere.
So
we
looked
up
just
a
simple
uf
for
the
southern
hemisphere.
For
this
this
variable
in
this
domain
and
for
2000
years
we
find
the
same
remote.
So
this
is
the
you
have
like
you
have
two.
R
R
We
can
find
a
similar
feature
like
like
a
cooling
over
the
eastern
pacific
or
ipo
like
a
cooling
phase
or
an
active
phase
corresponding
to
correspond
to
like
the
article
warming
and
the
southern
ocean
cooling,
and
if
you
put
the
polar
protection
you
see
most
of
its
ice
car
data
like
a
process
data
cluster
around
the
screen-
and
these
are
triggering
data-
show
like
the
eurasia,
cooling,
okay
in
just
out
of
phase
with
this
article
warming.
So
people
call
this
like
a
worm.
R
Rt
code
content,
cold
continent
pattern,
so
it
still
seems
like
it's
not
very
new
just
occur.
Recent
30
years
probably
had
occurred
many
many
times
over
the
past
2000
years.
That
is
his
ideas.
Otherwise,
simply
using
this
approximate
data,
the
coral
data
represent
the
tropical
forcing
correlated
with
the
other,
like
other
parts
of
data
everywhere,
so
still
can
find
an
out
of
phase
connection
between
the
tropical
forcing
with
the
green
land
but
the
same
phase
with
eurasia.
R
Like
the
trailing
data
so
last
question
we
would
like
to
ask
how
important
is
a
cssm4
to
to
generate
the
signal?
Probably
this
that
everything
we
see
here
is
simply
due
to
the
model
itself
rather
than
just
the
proxy
data,
so
we
just
are
using
the
original
cssm4
data
like
this.
That's
a
thousand
years
of
free
running
without
a
simulation
simply
calculate
uf,
but
we
find
that
you
have
no
uf
very
similar
to
this
observed
just
a
tiny
connection,
but
we
try
something
mix
some
together
to
see.
R
If
we
fall
system,
it's
a
model
to
capture
this
mode,
we
still
can
get
some
of
them,
so
we
find
a
half
from
you.
Have
one
and
a
half
from
you
have
three
add
them
together.
We
can
get
this
similar
mode
like
teleconnection,
just
like
the
cooling
here
and
the
article
warming.
So
this
last
slide.
If
we
put
everything
together,
it
seems
like
the
so.
This
is
observation.
Another
observation
like
a
reconstruction
data,
so
it
seems
like
warming,
cooling
and
the
cooling
and
eurasia
cooling,
but
a
model.
R
If
we're
using
this
reconstruction,
we
still
can
get
some
remote
like
a
cooling
warming,
cooling,
cooling,
but
it's
pretty
weak
so
like
and
just
a
time
series
with
this
mode.
It
seems
like
a
lot
of
indicator
variability
and
the,
but
the
model
itself
give
really
a
noise
and
pretty
random
randomness
feature
so
the
so.
I
think
the
the
main
conclusion
is
this
mode
is
probably
real,
and
but
it's
the
main
sources
for
this
mode
is
come
from
two
okay.
R
They
come
from
two
one
is
where's
the
proxy
data,
but
another
one
is
it's
a
six
sm4,
so
you
can
think
like
the
gc
select
like
make
it
something
like
the
tissue,
and
this
approximately
just
like
a
seasoning
and
included
ingredients
mainly
come
from
system
four,
so
that
just
to
prepare
set
up
this
table
set
up
this
a
gradient
ingredient,
but
the
product
data
just
changes
the
flavor.
Something
like
that.
So
so
I
want
to
stop
here
and
take
any
question.
Attendee
may
have.
A
That's
because
I
was
muted,
thank
you.
A
I
think
we
should
keep
moving
and
looks
like
the
chat.
Has
I
don't
see
any
questions
meaning
to
you,
but
let's
put
them
in
the
chat
and
move
on
to
the
last
speaker,
olivia
truax
from
the
university
of
otago
who's,
a
phd
student
there
in
new
zealand
and.
L
L
A
A
B
I'm
I'm
back.
Can
you
hear
me.
O
Good
is
that
is
that
working?
Yes,
yes,
it
works.
Okay,
yeah,
my
camera
decided
to
anyway.
I
will
try
to
share
my
screen.
O
Yes,
okay,
success;
yes,
good
great!
Thank
you!
Okay,
thanks
for
bearing
with
me
everyone
hi,
my
name
is
olivia
truax
and
I'm
a
phd
student
at
the
university
of
otago
in
new
zealand,
as
esther
mentioned,
and
I'm
here
today
to
talk
to
you
about
my
research.
O
My
research
reconstructing
past
interpreted
climates
using
paleoclimate
data
assimilation
and
it's
actually
really
exciting
to
be
following
ching
wadding,
because
I'm
talking
about
a
lot
of
the
same
set
of
observations
but
from
the
perspective
of
a
geologist,
so
high-resolution
proxy
records
from
the
past
1000
years
are
an
essential
source
of
information
to
augment
the
instrumental
record
of
particular
interests
are
the
little
ice
age
or
lia
thought
to
have
had
relatively
cold
global
temperatures
and
the
medieval
climate
anomaly
or
mca
when
north
atlantic
temperatures
were
warmer.
O
Several
last
millennium,
paleoclimate
reconstructions,
indicate
colder
tropical
pacific
conditions,
and
a
persistent
la
nina-like
mean
state
during
the
mca
and
a
shift
to
a
more
el
nino-like
mean
state
and
an
equator
word
position
of
the
southern
hemisphere.
Westerly
winds
during
the
lia
previous
paleoclimate
studies,
including
crosstalk
at
all
2021,
which
I've
highlighted
here,
have
hypothesized
that
changes
in
antarctic,
surface
temperatures
and
cx
dynamics
across
the
mca.
Liaa
transition
are
primarily
driven
by
mean
state
changes
in
enzo
and
the
southern
angular
mode,
which
is
the
position
of
the
southern
hemisphere
westerly
winds.
O
However,
the
limited
spatial
extent
of
paleoclimate
reconstructions,
particularly
in
the
southern
hemisphere,
as
you
can
see
in
this
plot,
really
is
a
challenge
to
to
reconstructing
dynamical
making
dynamical
inferences
using
kind
of
these
disparate
reconstructions.
O
And
so
here
I
examine
the
impact
of
large
scale
modes
of
atmospheric
variability
on
common
era.
Antarctic
climate,
using
the
last
millennium
re-analysis,
which
you
were
introduced
to
in
the
last
talk.
So
the
lmr
is
a
paleoclimate
data
assimilation
which
fuses
the
information
contained
in
the
proxy
records.
In
this
case
the
page's
2k
proxy
network
within
the
dynamical
constraints.
The
climate
model.
O
But
so
the
results
of
a
data
simulation
can
be
productively
compared
to
the
last
millennium
climate
simulations
to
assess
the
information
added
by
the
proxy
network,
because
all
the
trends
and
the
temporal
structure
in
the
reconstruction
is
driven
by
the
proxies,
and
so
this
framework
allows
dynamical
hypotheses
regarding
the
impact
of
changes
in
large-scale
modes
of
atmospheric
climate.
Variability
such
as
this
one
to
be
rigorously
tested
in
a
dynamically
consistent
context.
O
So
my
question
that
I'm
trying
to
answer
does
the
lmr
support
a
link
between
last
millennium
climate
in
antarctica
and
multi-centennial
trends
in
sam
and
enso,
and
so
just
by
starting
off
and
looking
at
lia
mca
anomaly
plots.
The
answer
is
kind
of
yes.
So
if
we
look
at
temperature
and
pressure
anomalies
and
the
regression
between
surface
air
temperature
and
mean
sea
level
pressure
and
the
amino
3.4
index,
we
can
see
that
last
millennium
changes
in
southern
hemisphere,
climate
in
the
lmr
are
characteristic
of
these
large
scale
mean
state
changes
in
el
nino.
O
Southern
oscillation
mean
annual
temperatures
in
both
the
central
and
eastern
equatorial
pacific
are
warmer
during
the
lia
compared
to
the
mca,
which
is
consistent
with
previous
work.
Suggesting
a
radiatively
forced
la
nina-like
mean
state
during
the
mca,
so
annually
resolved
climate
fields
actually
allow
us
to
identify
the
relative
influence
of
changes
in
enso
and
sam
on
last
millennium
climate
in
antarctica,
because
they
have
distinct
signatures.
O
So
looking
at
mean
sea
level,
pressure
mean
sea
level,
pressure
and
temperature
anomalies
centered
in
the
pacific
sector
of
west
antarctica
are
consistent
with
tropical
influence
associated
with
a
la
nina-like
pattern
prior
to
1450
ce.
So
that's
a
up
here
on
the
left.
However,
this
trend
is
pretty
small
note.
The
change
in
color
bar-
and
this
makes
sense,
due
to
the
lower
much
lower
number
of
proxy
records
which
are
assimilated
in
the
first
part
of
the
last
millennium.
O
So
after
1450
ce
sea
level,
pressure
is
higher
in
the
pacific
sector
of
west
antarctica.
So
through
here,
and
though
this
atmospheric
pattern
resembles
the
impact
of
el
nino,
it's
not
strictly
pacific
centered.
The
annular
component
is
consistent
with
the
pronounced
negative
trend
in
the
southern
annual
mode
during
this
time
period
and
the
combined
impact
of
el
nino
and
sam
acting
on
surface
in
antarctica.
O
So
I
was
interested
in
checking
the
lmr
based
estimates
of
southern
hemisphere,
last
millennium
climate,
against
a
complementary
data
set.
So
I
further
compared
temperature
anomalies
in
the
lmr
simulation
in
a
in
an
lmr
simulation
which
excludes
all
antarctic
proxies
to
an
out-of-sample
database
of
qualitatively
assessed.
Paleo
temperature
records
compiled
in
mooning
at
all,
2019
and
so
warm
anomalies
are
denoted
by
the
red.
Dots
and
cool
anomalies
are
denoted
by
the
blue
dots
and
both
the
lmr
and
these
out-of-sample
antarctic
records
show
a
characteristic
dipole
pattern.
O
After
1450
ce,
most
regions
of
antarctica
cool,
while
the
eastern
ross
sea,
which
is
just
in
here
and
parts
of
the
amundsen
amundsen
sea
sector
of
the
southern
ocean.
Warm
and
good
agreement
between
paleoclimate
reconstructions,
using
different
proxy
databases
and
really
different
methodologies,
provide
support
for
the
inference
that
these
mid
and
low
latitude
teleconnections
are
primary
driver
of
last
millennium
climate
in
the
lmr.
O
So
I
went
ahead
and
compared
the
results
of
the
lmr
to
the
ccsm4
to
have
a
look
at
what
the
model
prior
was
using
or
looked
like,
and
the
pattern
of
surface
temperature
and
pressure.
Differences
between
the
lia
and
mca
in
the
ccsm
board
is
notably
different
from
that
in
paleoclimate
records
and
paleoclimate
records
and
data
assimilation.
O
The
regionally
heterogeneous
dynamically
driven
pattern
in
the
lmr
and
other
paleoclimate
records
is
really
different
from
the
global
pattern
of
cooling
in
the
gcm,
a
decrease
in
sea
level,
pressure
over
the
southern
high
latitudes
or
a
southward
shift
in
the
southern
hemisphere.
Westerly
winds
in
the
ccsm
4
during
the
lia
is
consistent
with
an
increased
temperature
gradient
and
polar
amplification
of
cooling,
but
it's
at
odds
with
the
precisely
opposite
interpretation
from
paleoclimate
records,
which
suggests
that
the
westerlies
actually
moved
north
across
this
time
period.
O
So
the
results
of
my
analysis
underscores
that
the
ccsm
board
doesn't
simulate
the
low
frequency
dynamic
climate
variability
that
the
geologic
community
infer
from
proxy
records
in
the
lmr
assembly.
Assimilated
proxies
from
the
mid
and
low
latitudes
are
adding
new
information
about
these
low
frequency,
dynamical
relationships
compared
to
unconstrained
model
simulations
and
if
we
look
at
time-saving
analysis,
mtn
analysis
of
the
enso
index
from
the
lmr
down
here,
as
well
as
the
cssm4
and
the
cesm
last
millennium.
Ensemble
simulations
show
shared
significant
periodicities
in
these
kind
of
hot,
more
high
frequency
and
so
bands.
O
If
we
have
a
look
at
the
nino
3.4
index
in
the
ccsm
bore,
which
is
just
down
here,
we
can
see
that
it
basically
just
tracks
global
mean
surface
temperature,
so
it's
responding
to
the
forcing
by
volcanic
solar,
solar
and
volcanic,
forcing
rather
than
the
type
of
multi-centennial
dynamic
climate
variability
that
we
get
in
the
lmr.
A
Thank
you,
olivia
yeah
and
I
know
you've
looked
at
our
your
our
last
millennium
ensemble.
Did
you
do
any
comparisons
to
to
that?
Those
simulations?
You
have
a
lot
more.
You
can
maybe
yeah
effective
internal
variability
a
little
bit.
O
Yeah
I
looked
at
across
the
ensemble
in
and
you
never
get
the
kind
of
centennial
variability
that
the
proxy
records
imply
or
have
been
interpreted
to
represent
yeah
okay,
and
we
also
we
had
a
look
as
well
at
the
high
top
version
of
the
last
million
ensemble
to
see
if
there
might
be
something
there
in
the
solar,
because
there's
a
solar
link
between
and
with
the
southern
annual
mode-
and
we
didn't
find
anything
there
either.
A
All
right,
I
think
anything
else
can
go
in
the
chat
and
we
should
probably
move
on.
We
do
have
another
discussion,
we're
kind
of
running
out
of
time
before
noon
here,
but
I'm
going
to
share
my
screen
again
because
we
put
into
the
agenda.
A
Some
discussions
for
us
to
think
about,
or
some
questions
for
us
to
think
about
during
the
discussion,
and
I
just
want
to
open
up
the
floor
to
anyone
who
might
have
something
to
say
about
anything
that
we've
heard
today
or
any
other
contradict
contribution
they'd
like
to
to
make.
I
do
want
to
thank
all
the
speakers
for
your
great
talks
and
for
making
this
a
great
working
group
session
with
so
much
interest
in
submissions
before
we
go
and
then
just
open
it
up.
A
The
questions
that
we
came
up
with
for
this
discussion
are:
how
can
we
better
understand
past
climate
changes
through
combining
model
and
proxy,
and
then
what
do
we
need
to
do
to
improve
the
models?
We
saw
a
lot
of
indications
that
the
models
aren't
getting
everything
right,
but
they
are
getting
some
things
right.
A
Any
of
the
speakers
who
spoke
just
now
on
proxies
proxy
model
comparisons.
I
think
one
thing
that
seems
I
I
know
clay
in
his
talk.
A
He
compared
to
proxy
offline
proxy
models,
which
kind
of
helps
translate
model
output
to
proxies
lets
us
do
a
better
comparison.
I
Oh
yeah,
I
mean
just
on
that.
I
mean
I
think
it
was
interesting
and
you
can
definitely
add
uncertainties
to
the
records
with
some
of
these
for
proxy
modeling.
But
surprisingly,
and
stuff
I
didn't
have
time
to
show.
Is
that
even
with
really
just
simple
assumptions
about
delta
180
precipitation
and
a
temperature
fractionation
effect,
the
the
results
were
fairly
similar.
So
it
was
like.
A
I
think
what
olivia
showed,
how
the
the
models
aren't
necessarily
getting
our
centennial
scale
variability
in
good
agreement
with
proxies
and
I'm
you
know,
I'm
not
sure
what
what
that's
why
that
is.
Is
it
that
we're
not
we're
dampening
out
these
longer
time
scale
variability
modes
of
variability
or
it's
we're
not
getting.
You
know
it's
part
of
the
internal
variability
they're,
not
forced.
A
L
Yeah
I
brought
that
up
in
the
chat
as
well,
but
there's
a
I
think.
The
bigger
point
is
so.
I
think
that
data
simulation
is
emerges.
A
really
cool
way
to
you
know,
merge
the
model
and
proxy
information.
Obviously
that's
something
that
you
know:
we've
been
working
on
too,
but
I
do
want
to
make
the
point
that
you
really
do
need
to
be
aware
of
what's
going
into
those
products,
because
you
know,
for
example,
in
the
kc
lmr,
it's
just
very.
L
They
just
took
everything
and
used
a
linear
regression
and
put
it
in
there,
and
so
it
didn't
account
for
anything
like
proxy
redmine
right.
So
proxy
systems
don't
record
the
climate
system
faithfully
in
the
spectral
domain
right
they
always
red
in
it,
and
that
could
be.
You
know
if
you're
still
iothem
or
drip
water
in
your
cave.
L
If
you're,
a
coral,
there's
low
frequency
changes
that
may
be
biological,
and
so
we
now
have
more
advanced
proxy
system
models
that
try
and
sort
of
get
at
that,
and
they
actually
can
be
used
in
the
assimilation
framework
and
improve
that.
But
we
should
always
be
careful
about
saying
models.
Don't
have
enough
internal
variability
because
that
that
actually
like
once
you
bring
in
the
proxy
symbol,
the
differences
start
to
go
away
right,
suggesting
that
it's
really
about
comparing
apples
to
apples
and
oranges
to
oranges.
L
L
So
it
sort
of
to
me
emphasizes
the
real
importance,
both
with
just
normal
proxy
model
comparison
or
with
data
simulation
of
making
sure
to
see.
People
and
model
people
are
talking
there
and
working
together,
because
if
proxy
people
have
the
domain
expertise,
they
know
these
records
inside
and
out
and
they
can
determine
the
impact
of
that
record
might
be
on
a
data
simulation,
for
example,
or
even
you
know
they
might
be
able
to
say
in
proxy
model
world.
A
Yeah,
I
think,
tighter
collaboration,
it's
definitely
something
that
we
could
do
better
on
and.
F
Can
I
add
to
the
discussion?
Okay,
because
I
think
that's
a
good
point
that
jessica
brings
up
that
there
could
be
proxy
biases.
You
know
that
lead
to
the
running
in
the
proxy
record,
so
that
could
be
one
explanation.
The
other
explanation,
of
course,
could
be
in
the
models
that
they're
missing
some
low
frequency
variability,
and
I
think
we
should
remember
that
the
models
are
still
incomplete
in
a
sense,
for
example,
that
they're
not
coupled
into
interactive
ice
sheets.
F
So
if
there's
ice
sheet
variability
that
could
contribute
some
lower
frequency
variability
or
maybe
dynamic
vegetation-
I
don't
know,
but
you
know
there,
there
may
be
some
issues
in
the
models
as
well.
Okay,.
O
Olivia
yeah,
I
was
just
gonna
say
that
I
totally
agree
with
all
of
jessica's
points.
I
am
actually
a
geologist
by
training
and
came
to
the
lmr
as
a
way
to
try
to
generalize
from
these
antarctic
records,
because
a
lot
of
them
were
kind
of
point-to-point
comparisons
and
initially
I'd
been
really
curious
about
how
I
could
use
models
to
interpret
my
records
and
so
yeah.
O
I
would
just
agree
with
a
lot
of
a
lot
of
what
has
previously
been
said
and
add
that
I
think
that
yeah
working
closely
with
folks
who
know
the
proxy
records
is
good
and
is
also
helpful,
because
I
think
that
the
proxy
community
can
sometimes
make
these.
You
know
point-to-point
comparisons
between
individual
records,
which
yeah
aren't
necessarily
realistic.
Given
the
climate
system.
R
Yeah
so
yeah,
I
really
appreciate
it
coming
from
jesse
and
oh
great,
the
psm
is
mainly
for
emphasize
this
linear
relationship.
So
I'm
just
wondering
anybody
here
or
other
groups-
have
some
interest
to
develop
this
lmr,
using
like
a
machine
learning
method
because
to
try
to
using
some
some
more
advanced
master
to
to
take
into
account
this
non-linear
relationship
so,
and
I
think
that's
maybe
some
promising
way
promising
direction
to
go
like
further
yeah.
That's
it.
Thank
you.
Thank
you.
N
Yeah
yeah
so
yeah
I've
kind
of
powered,
my
niche
in
the
sort
of
data
proxy,
our
model
proxy
comparison
world,
looking
at
hydrologic
changes
and
one
of
the
things
that
I've
noticed
is
like
for
modeling.
These
regional
scale
variability.
N
It's
really
important
to
have
these
high
resolution
models,
but
a
problem
I
run
into
consistently
is
like
the
the
length
of
time
that
these
models
have
been
run
for
because,
like
I,
I
kind
of
question,
sometimes
if
say
like
20
or
even
50
years,
like
these
high
resolution,
data
sets
is,
is
adequate
to
capture
sort
of
what
could
be
like
the
sort
of
tail
end
of
events
that
contribute
to
our
actual
observations
of
what's
being
deposited
in
in
response
to
hydrologic
changes.
N
A
Yeah
good
good
point:
anyone
dan,
let
me
know
when
you're
ready,
ching
hwa.
Do
you
still
have
your
hand
up?
Oh
dan,
are
you
ready?
Oh
sorry,.
P
K
P
Is
really
interesting
conversation?
I
guess
you
know
some
of
the
things
raised
by
the
questions
about
you
know
how
to
interrogate
the
fidelity
of
the
model's
representation
of
lower
frequency
variability,
and
you
know
related
to
andreas's
comment.
I
guess
I
don't
know
what
you
could
change
about
the
model
to
change
how
it
generates
variability
on
decay
level
to
centennial
time
scales,
but
I
think
moving
forward
with
some
of
these
data
simulation
approaches.
P
Like
the
lmr
takes
output
from
a
model
and
then
sort
of
stitches
that
together
to
represent
each
of
the
proxy
years,
but
you're
sampling
there
just
from
sort
of
how
the
model
realizes
internal
variability,
but
we
know
that
that's
there's
more
to
it
right,
there's
more
there's
model
errors.
How
do
we
sample
model
errors?
How
do
we
try
to?
P
You
know
correct
more
of
the
model,
so
I
would
you
know
it'd
be
really
exciting
to
think
about
running
some
ensembles
or
thinking
about
sensitivity,
experiments
where
you
know,
maybe
in
cesm
you're
varying
some
of
these
parameters
in
past.
A
Yeah
it
looks
like
it
went
away:
oh
no,
it's
still
there.
I
don't
know
how
to
do
it
either.
So
I'm
going
to
stop
sharing
my
screen
because
it's
it
hinders
me
with
a
lot
of.
K
G
Yeah,
I
think
important
thing
is
that
the
models
capture
the
variability,
but
what
we
notice
is
there's
even
larger
problems,
and
so
the
moles
have
some
problems
to
capture
the
mean
state
and
for
many
studies
we
are
doing
here.
The
deep
sea
circulation,
at
least
in
the
pacific,
is
not
adequately.
G
I
hope
that
will
be
improved
in
the
future.
I
mean
we
had
in
the
past
ocean
circulation
models
where
the
deep
sea
circulation
were
well
addressed
and
that's
always
a
problem.
If
you
couple
the
atmosphere
with
the
ocean
that
you
get
these
these
ventilation
in
the
inner
quadrant
seas,
but
I
think
that's
actually
a
main
focus
that
we
need
to
get
the
mean
state
of
the
climate
right
and
then
we
can
focus
on
the
variability
of
the
climate.
A
G
A
A
Commentary,
how
are
we
doing
with
our
I
mean
you
know.
A
We
now
have
some
geotracers
in
the
model
they're
expensive,
to
run
in
every
application
and
would
be
great
if
they
were
less
expensive,
but
maybe
the
lower
resolution
configuration
might
help
there
to
run
more
geo
tracers,
but
we're
making
progress
there,
but
I
suppose
we
have
more
progress
yet
to
make
before
we're
fully
there
and
then
also
with
forward
proxy
modeling
data
assimilation.
A
All
of
these
can
really
push
us
forward
in
understanding
all
these
issues,
so
anyone
want
to
mention
anything
about
the
models.
A
And
what
I've
seen
with
mom
is
the
current
coupled
mom
versions
are
a
little
bit
even
more
sluggish
in
the
overturning
circulation
strength,
so
I
think
they
are
working
on
that
to
improve
it.
I
think
it's
getting
better
at
least,
and
the
ocean
model
working
group
is
having
their
session
concurrently
with
ours,
so
yeah,
so
yeah.
We
do
have
plans
for
in
the
next
in
dur.
A
In
this
allocation
we
hope
to
be
able
to
get
some
experti
experience
running
with
mom
to
test
out
that
coming
ccsm
three
model
before
it
turns
into
ccsm3
gives
some
feedback
from
the
paleoclimate
applications
on
that.
So
andreas.
N
I
I
did
have
something
to
say
actually
this
time.
I
did
raise
my
hand
this
time.
Yeah
I
was
gonna,
build
off
of
what
arnie
was
saying
about,
like
the
models,
recording
mean
states
and
and
extremes
and
trying
to
get
the
mean
state
right,
and
I
feel
like
sometimes
like
when
you
do
a
deep
dive
in
the
proxies
there's
a
question
of
like
what
they
are
actually
recording.
N
Not
just
hydrologic
proxies
like
I
was
talking
about
earlier,
but
temperature
proxies
as
well
and
an
example
this,
and
it
would
be
like
text
86
and
asking
like
whether
this
is
recording
actually
a
mean
temperature
or
is
it
more
of
a
seasonal
temperature?
N
And
so
we
should
it's
just
something
that
we
should
take
into
account
when
looking
at
our
models,
is
it's
not
like?
Our
proxies
aren't,
always
recording
mean
states.
F
I
will
you
know
since
well.
I
will
just
state
like
again
that
that
I
think
the
importance
of
a
low
resolution,
fast
version
is
really
important.
I
mean
we've,
we've
seen
lots
of
nice
applications
on
high
resolution
modeling
and
that's
that's
a
nice
thing
too,
but
I
just
want
to
hope
that
people
do
not
forget.
We
need
these.
We
need
to
need
models
and
run
run
long
enough
that
have
include
these
capabilities
of
isotope
and
and
carbon
cycle
and
so
on,
and
so
so.
F
For
that
reason
I
think
it's
really
crucial
to
have
a
working
as
as
low
resolution
as
possible.
I
would
say
a
version
that
you
can
use
for
these
applications
for
equilibration
carbon
cycle
ice
sheet,
coupling
and
so
on,
and
that
would
also
enable
more
of
the
oxygen
isotope
ability
you
know
if
we,
if
we
imagine
that
the
bulk
of
the
paleoclimate
data
from
the
ocean
at
least,
is
probably
oxygen
isotopes
and
we
have
used
only
a
tiny
fraction
of
that
of
that
data.
F
You
know
lgm
and
maybe
trace,
but
you
know
fully
coupled
oxygen,
isotope
or
system
modeling.
That
involves
ice
sheets
right
that
because
their
effects
on
melt
water
and
so
on
and
affects
o-18,
so
it
it
involves
a
fully
coupled
ocean
atmosphere,
ice
sheet
vegetation
carbon
cycle
model
that
you
can
run
for
these
long
time
scales.
So
that's
why
I'm
making
this
keep
repeating
this
plug
for
the
low
resolution
model,
even
though
it
I
may
sound
like
a
broken
record,
but
okay
I'll
stop
here.
A
Yeah,
I
think
it's
been
something
of
a
wishful
hope.
I
guess
I'm
trying
to
find
the
right
words
here
to
have
water
isotopes
in
the
ice
sheet
model,
and
I
don't
know
if
betty
can.
Oh
betty
have
your
hand
raised.
K
Well,
it
wasn't
so
much
that,
but
just
in
general
both
what
andreas
just
said
and
just
put
in
the
chat
and
is,
I
think
you
know
part
of
when
you
think
of
the
water
isotope
version
icsn
1.2-
that
was
a
community
effort
of
very
bright
grad
students
and
post-docs,
so
jessie,
nussbaumer
and
tony
wong
and
jean-zoo
all
contributed
to
getting
that
going
and
jessie's
still
involved
with
m6,
so
that
and
the
low
resolution
model.
K
If
this
group
could
get
some
really
bright,
postdocs
or
students
that
want
to
work
on
this,
if
you
can
engage
with
the
developers
in
land
ice
ocean
land,
I
know
their
water.
Isotopes
is
a
big
push,
but
all
of
this
it
would
be
really
great,
even
in
mom's
sex,
to
make
sure
we
get
a
resolution
that
works
for
us
would
be
great
to
get
the
community
involved.
So
we
can
help
you
coordinate
that
get
in
touch
to
whatever
needs
to
be
done.
K
Bill
I'm
just
joining
and
I
came
in
in
the
middle,
but
I
heard
something
about
isotopes
and
ice
sheets
and
I
was
wondering
if
someone
could
fill
me
in
on
what
the
what
the
desire
is.
A
To
have
a
to
be
able
to
track
water
isotopes
coming
off
fluxes
to
you,
know,
atmosphere
and
ocean
land
coming
from
the
ice
sheet.
So.
K
K
That
that's
that's
that's
when
it
still
belongs
to
clm
right,
but
if
you
want
to
transport
isotopes
through
the
ice,
that's
not
a
that!
That's
a
moderate,
not
a
huge
effort
for
the
ice
sheet
model,
because
the
ice
sheet
model
is
already
set
up
to
handle.
Tracers
the
the
bigger
effort
might
be
getting
things
through
the
coupler
to
the
ice
sheet
model.
K
So
so
that
that's
something
I'd
be
glad
to
talk
about.
If,
if
that's
of
interest,
but
it's
something
it's
something
that
we've
talked
about
a
little
bit
through
the
years.
But
we've
never
quite.
A
A
Yeah
and
it
might
require,
you
know
someone
to
really
embrace
the
project
and.
K
But
but
guntree
and
I
could
certainly
help
on
the
ice
sheet
ice
sheet.
Modeling.
C
I
have
a
quick
question
since
it
seems
like
yeah.
I
was
just
curious
while
we're
talking
about
the
ice
sheet
isotope
and
all
that
seems
like
the
land
model
is
a
roadblock
sort
of
kind
of
here,
and
the
same
was
like
dynamic
vegetation.
So
in
the
meantime,
is
it
possible
to
like
make
a
compost
set
with
the
ice
sheet
model
with
the
rest
of
cesm,
but
maybe
like
a
biome
model
connected,
so
you
can
do
like
dynamic
vegetation.
C
They
can
do.
You
can
do
like
actually
interactions,
and
I
know-
or
I
heard
that
there's
some
effort
on
making
a
like
make
it-
maybe
a
box
model
for
the
water
isotopes
for
the
land.
So
I
was
just
wondering
like
waiting.
Is
there
a
benefit
like
during
those
intermediate
years?
Well,
those
more
sophisticated
things
are
developed.
Maybe
we
could
still
do
science
using
a
maybe
simplified
configuration.
A
I
know
jesse
is
here
in
our
list
here.
I
don't
know
if
he's
on
but
yeah,
there's
a
simplified
land
model
called
slim
and
it's
been
talked
about
as
a
potential
place,
that
a
simple
bucket
model
could
be
implemented
where
you're
just
computing,
you're
sort
of
collecting
the
isotopic
composition
of
the
water
and
land
very
simply
and
then
computing
the
fluxes.
A
K
Yeah
sorry,
I
had
to
contribute
to
this
earlier,
but
I'm
being
one
of
those
terrible.
You
know
virtual
people
who
was
like
half
working.
So
my
my
fault,
anyways
yeah,
yeah,
see
it
clm
in
terms
of
isotopes
and
csm2
is
kind
of
the
sticking
point
to
my
knowledge.
It's
not
anyone's
fault.
It's
just
the
the
people
who
are
assigned
to
it
are
just
really
busy
but
yeah.
So
there
is
a
bucket
model.
There's
a
fancier
bucket
model
that
actually
david
noon
and
rich
fiorella
developed
for
clm5
for
isotopes.
K
And
canopy
interception
which
are
kind
of
the
big
ones,
the
catches-
I
don't
know
how
it's
not
gonna,
do
anything
for
snow.
So,
if
you
want
to
look
at
like
cold
regions,
I'm
not
sure
how
well
it's
going
to
work,
but
you
can
always
add
another
bucket
to
the
bucket
model
so
and
yeah,
and
the
advantage
of
this
bucket
model
is
actually
all
in
cam.
K
It
just
needs
the
flux,
this
regular
physical
fluxes
from
clm,
so
it
should
be
relatively
portable
yeah,
that's
what
I'm
using
for
cam
six,
because
cam
six
has
to
run
with
clm5.
So
I
need
a
bucket
model
to
check
dance
chips
and
cam.
Six.
K
Yeah,
but
if
you're
like
dying
to
use
it,
you
know,
okay,
if
I'm
happy
to
share
in
some
ways
the
more
eyes
the
better,
because
I
haven't
really
done.
Basically,
it
runs
and
it
passes
the
eye
test,
but
I
haven't
done
any
sort
of
like
rigorous
quantitative
validation.
Yet
so
it's
kind
of
used
at
your
own
risk,
but
whoever.
K
I
It
if
they
want
right.
K
C
A
Vegetation
data
sets
are
input,
it's
kind
of
run
in
an
offline
mode,
so
so
climate
model
information-
you
know
climatology-
is
given
to
biome
from
the
climate
model
and
then
biome
runs
and
produces
distribution
of
biomes
now
and
maybe
potentially
pfts.
A
That's
that
would
be
a
slightly
new
development
but
biomes
and
then
the
biomes
are
transferred
to
pfts,
given
some
assumptions
about
what
pfts
are
in
what
biomes
and
how
much
of
whatever
and
then
that's
input
in
a
data
set
surface
data
set
and
then
the
climate
model
is
run
with
that,
so
it's
kind
of
a
periodically
updated.
It's
not
it's
not
like
you
know
it's
periodically
updated,
not
run
in
in
a
coupled
situation,
so
someone
would
have
to
take
on
that
project.
I
think
bill.
K
Yeah,
I
was
just
wondering:
do
you
or
or
betty
see
a
long
term
path
for
either
biome
or,
let's
say
fates
being,
and
maybe
fates
makes
more
sense
in
terms
of
being
part
of
the
runtime
flow?
But
I
was
wondering:
does
fates
do
what
biome
does
is
that
a
possible
long-term,
dynamic
vegetation
model
that
would
would
do
what
we
wanted
to
do
for
paleo
runs.
A
Let's
see,
I
don't
see
sam
levis
was
here
earlier.
He
worked
on
the
old
dynamic
vegetation
model.
I
think
fates
is,
is
the
the
newer
version
that
is
part
of
csm.
Currently,
I
I
think
we
need
to
get
some
projects
going
using
it.
I
I
don't
know
of
anything
where
biome
would
be
brought
in
as
an
as
a
sort
of
an
online
component.
I
don't
know
of
anything
like
that,
but
fates
is
already
in
ctsm,
but
I'm
not
sure
it's
in
production
mode.
A
Yeah
and
rosie
gave
a
talk
to
our
working
group
a
few
summers
ago
about
fates
and
it
looked
really
promising,
especially
for
deep
time
purposes,
because
it
kind
of
represents
niches,
so
yeah.
I
think
I
think,
that's
probably
the
long.
The
long-term
plan
path
for
dynamic
veg
and
in
the
short
term,
we've
been
using
biome
in
an
offline
in
an
offline
way,
and
if
someone
is
really
good
wants
to
try
to,
I
don't
know
about
creating
a
comp
set
ren.
A
I
think
that
might
be
overstepping
it
a
bit,
because
a
lot
of
infrastructure
would
need
to
happen
to
have
that
work
in
a
kind
of
a
seamless
way.
Now
it
take
to
to
get
biome
updated.
It
takes
a
lot
of
human
interaction
with
with
everything
it's
not
automated
at
all.
So.
C
Yeah
I
mean
it
would
be
good
to
even
have
more
than
one
dynamic
vegetation,
given
that
we
know
from
like
a
betty
stock.
How
strong
and
marcus
talk
yesterday
like
how
big
in
the
influences
so
it'd
be
good
to
sort
of
have
like
a
different
like
different
dynamic
vegetation
turned
on,
and
then
you
can
see
like
almost
do
a
mini
ensemble,
to
see
what
the
simulation
results
might
be.
A
A
Anyone
else
are
we
in
our
informal
lunch
discussion
period
here,
it
seems
like
we've
gone
over.
I
think
stephanie's
still
here,
but
I
thought
she
had
to
leave
at
some
point.
I
have
to
leave
soon
so
yeah.