►
From YouTube: 8th PAWS Webinar
Description
The eighth webinar from the Paleoclimate Advances Webinar Series (PAWS) which took place on May 26th 2023.
Kevin Anchukaitis discussed "Progress and uncertainties in global & hemispheric temperature reconstructions of the Common Era" and Karen King discussed "Progress of visual light-based techniques for improving the spatiotemporal coverage of tree ring temperature proxy records for North America"
For more information and to signup for the PAWS Google Group visit:
https://www.cesm.ucar.edu/events/webinars/paws/
A
All
right
well
welcome
to
our
pause
seminar
after
a
long
delay.
We're
back
and
we're
really
excited
about
today's
presentations.
Just
to
remind
you
about
the
format
we'll
have
20
minutes
per
talk
and
then
five
minutes
of
q,
a
and
then
10
minutes
at
the
end
for
General
discussion
and
we'll
be
giving
the
speakers
a
two-minute
reminder
visually
on
screen
when
two
minutes
are
left
in
your
time.
Slot.
A
So
we're
really
excited
today
to
feature
tree
ring
signs
broadly,
so
we
have
two
speakers.
The
first
is
Dr
Kevin
educaitis
who's.
Currently,
a
professor
of
Earth
Systems,
geography
at
the
University
of
Arizona
and
Kevin
has
really
pioneered
using
tree
rings
to
understand
temperature
variability
over
the
Common
Era,
as
well
as
human
environment
relationships,
and
our
second
speaker
is
Dr
Karen
King,
who
is
currently
a
postdoctoral
research
scientist
at
Lamont,
Doherty
Earth
Observatory,
but
will
soon
be
an
assistant
professor
at
the
University
of
Tennessee
Knoxville.
A
A
So
pause
is
led
by
the
CSM
paleoclimate
working
group
and
our
goal
is
to
provide
a
welcoming
space
for
the
open
exchange
of
ideas
and
feature
new
developments
in
paleoclimate
scientists.
So
please
keep
your
comments
respectful.
A
Let
this
be
a
curiosity,
driven
discussion
and
again
just
a
reminder
on
the
format
and
the
recordings
will
be
shared
on
YouTube
after
the
seminar,
so
you
can
return
to
them
anytime.
You
want
so
I'm
going
to
stop
by
share
and
let
Heaven
begin
the
discussion.
C
Thanks
to
the
organizers
for
having
me
and
for
everybody
for
being
here
today
and
I'm
excited
about
the
the
way
this
is
organized,
because
basically,
what
I
get
to
do
is
is
tell
you
about
all
the
problems
and
the
challenges
and
Karen's
going
to
take
over
and
tell
you
about
the
future
and
how
we're
going
to
solve
those,
and
that
seems
about
right.
So
I'm
I'm,
really
appreciating
this
format
and
the
chance
to
talk
about
these
things.
C
C
And
it
was,
this
paper
was
partially
at
least
motivated
by
this
figure
from
a
summary
for
policy
makers
in
2021
towards
the
end
of
that
year
of
the
ipcc-
and
you
know,
sort
of
on
the
right
and
panel
B
is-
is
a
figure
we've
come
to
expect
that
if
you
run
a
climate
model
with
both
human
forcings
and
natural
forcings,
you
can
do
a
pretty
good
job,
reproducing
the
climate
history
of
the
last
150
years
or
so,
and
the
Natural
Factors
can
account
for
the
rise
in
temperatures,
particularly
since
sort
of
the
middle
of
the
20th
century,
but
over
on
the
left
in
the
a
panel
I
guess,
I
was
a
little
surprised
when
I
first
saw
the
SPM,
because
it
presents
a
single
reconstruction.
C
Purportedly
of
global
annual
temperatures,
in
this
case
smooth
to
show
decadal
variability
all
the
way
back
to
the
year.
One
and,
what's
perhaps
most
surprising,
was
the
fact
that
the
error
bars
don't
grow
as
you
go
back
in
time,
as
you
might
expect,
and
I'll
talk
about
why
that
is.
In
fact,
the
error
bars
are
widest
in
a
period
where
we
have
a
substantial
amount
of
proxy
data
there
and
sort
of
the
the
little
ice
age
from
about
1400
in
into
the
the
late
1700s.
C
C
It
builds
on
a
database
that
was
led
by
Julian
milger,
with
a
host
I
think
about
90
other
authors
that
was
put
together
of
temperature
sensitive
proxy
data
for
the
last
2000
years,
and
it
it
uses
this
data
set
to
do
reconstructions
of
temperature
over
the
Common
Era
and
they
applied
a
bunch
of
different
techniques.
C
So
here
I'm
showing
a
figure
from
that
paper
in
the
top
panel,
you
see
the
sort
of
full
spectrum
of
these
reconstructions
and
all
those
different
lines
with
different,
colors
and
and
acronyms
are
different
approaches.
C
You
can
use
to
assembling
a
network
of
proxy
data
into
a
reconstruction,
and
so
you
can
see
in
some
places
they
agree
well,
in
other
places,
they
agree
less
well
and
there's
one
particular
approach
of
Bayesian
hierarchical
model
that
gets
particularly
cold
sort
of
during
the
little
lice
age,
which
is
responsible
for
that
odd
feature
of
the
uncertainty
bounds
being
wider.
Now,
the
focus
of
this
paper
was
really
hey.
If
you
actually
remove
the
sort
of
lowest
frequency.
This
reconstruction
agrees
pretty
well
on
sort
of
this
decadal
multi-decatal
scale.
C
Some
of
that
is
is
almost
certainly
because
of
the
role
of
the
same
climate
or
sorry
same
paleoclinic,
proxy
data
going
into
it,
but
this
was
a
focus
paper
and
they
say
in
this
paper
actually
that
the
uncertainties
for
all
the
reconstructions,
if
you
look
at
them
the
different
ones-
the
CPS
reconstruction,
the
PCR,
that
Bayesian
hierarchical
model,
these
uncertainties
increase
back
in
time
and
are
particularly
large
before
sort
of
a
thousand
CE
or
before
1200
the
year
1200,
as
they
say,
owing
to
the
decrease
in
the
amount
of
input,
proxy
data.
C
And
yet
what
makes
it
into
this
summary?
For
policy
makers
is
the
median
of
those
different
reconstructions
really
sort
of
reflecting
if
we
go
back
the
spread
of
the
central
estimate
of
those
different
proxies,
but
not
the
uncertainty
on
each
one
of
those
methods
themselves
and
that's
where
those
uncertainties
went
and
what
we're
looking
at
is
the
uncertainties
due
to
the
choice
of
methods
and
the
median
estimate,
as
opposed
to
the
full
uncertainty.
C
So
Jason
and
I
want
to
talk
a
little
bit
about
this
full
uncertainty
and
in
fact
the
ipcc
had
done
a
pretty
good
job,
particularly
in
the
last
sort
of
previous
two
reports
in
trying
to
convey
this
uncertainty.
So
in
2001,
the
the
famous
hockey
stick
was
featured
again
in
the
SPM,
a
single
reconstruction
of
the
past
thousand
years,
and
there
you
can
see
the
the
uncertainties
really
do
blow
up
about
before
about
sort
of
1600,
reflecting
the
challenges
and
doing
a
global
scale.
In
this
case,
even
just
Northern
Hemisphere
scale
reconstruction.
C
Given
the
relative
paucity
of
high
resolution
proxy
data
in
2007
and
2013,
there
was
sort
of
more
of
a
reconstruction
democracy.
If
you
will
and
the
authors
of
those
two
reports
or
the
Paleo
climate
chapters
in
those
two
reports
used
this
way,
this
sort
of
relative
agreement
between
the
reconstructions
to
convey
that
you
can
actually
see
there's
a
fair
amount
of
change
between
2007
and
2013.,
there's
a
deeper
little
Ice
Age,
perhaps
a
warmer
definitely
a
warmer
medieval
period.
C
C
So
sorry,
in
this
paper,
Jason
and
I
sort
of
reviewed,
you
know
what
we
thought
we
or
how
well
we
knew
the
global
mean
annual
temperatures
over
the
Common,
Era
and
I
I
emphasize
those
words
Global
and
mean
because,
on
top
of
the
challenges
that
are
faced
by
a
declining
number
of
proxy
data
going
back
in
time,
there's
also
the
characteristics
of
those
data
in
resolving
Global
annual
mean
temperatures.
C
One
of
the
biggest
issues
we
face
when
trying
to
do
client
temperature
reconstructions
of
the
Common
Era
is
the
spatial
temporal
distribution
of
proxies.
So
this
is
a
figure
showing
how
those
proxies
change
through
time.
You
can
really
see
the
dominance
of
tree
ring
based
proxies,
certainly
in
the
modern
period,
but
even
going
back
into
the
first
Millennium.
C
C
Key
Rings,
not
surprisingly
reduce
Glacier
Ice
is
more
relatively
stable
through
time,
and
so
the
representation
of
proxies
through
time
is
actually
changing,
and
we
have
relatively
few
temperature
sensitive
proxies
in
this
database
prior
to
about
sort
of
a
thousand
CE,
and
you
can
even
see
the
rapid
decline
after
about
six
or
before
1600
CE,
and
you
can
look
at
this
in
space
too.
So
in
our
best
replicated
time
again,
you
see
that
there's
a
lot
of
triggering
chronologies.
C
They
are
heavily
weighted
towards
the
mid
and
high
latitudes
of
North
America
and
Eurasia.
We
have
some
wonderful
choral
records,
although
most
of
them
are
are
not
long
enough
to
extend
very
back
far
in
time,
and
then
we
have
glacial
ice
records
as
well.
By
the
time
you
get
back
to
500
CE.
This
is
what's
left
a
few
long
chronologies
most
of
these
Bristlecone
Pines
in
the
western
U.S.
You
have
some
ice
cores,
both
in
Greenland
and
Antarctica
tropical
ice
cores,
but
you
really
see
that
the
corals
are
gone.
C
So
what
about
the
seasonality
and
spectral
properties
of
these
proxies?
Well,
Jason
and
I
went
in
and
we
sort
of
looked
at
the
proxies
through
time
and
and
calculated
some
metrics.
So
the
panel
at
the
top
labeled
B
shows
you
the
median
latitude
of
those
proxies
and
you
can
see,
except
for
the
sort
of
most
recent
time
when
corals
sort
of
pull
the
median
latitude
towards
the
tropics,
we're
talking
about
somewhere
between
about
30
and
50
degrees,
north
latitude.
C
There
is
a
widening
of
the
interquartile
range
you
see
at
before
about
800,
as
the
number
of
proxies
drops
and
the
presence
of
Antarctic
ice
cores
to
Ministry
thought.
I
had
20
minutes.
C
Similarly,
a
number
of
proxy
records
recording
different
seasons
is
actually
relatively
stable
through
time,
but
heavily
represented
by
summer
proxies
so
anywhere
from
you
know,
somewhere
in
the
the
50
to
60
summer,
proxies
very
few,
representing
their
local
winter
and
relatively
few,
so
around
25
percent
being
judged
to
be
truly
annual
proxies
and
there's
a
number
of
consequences
to
this.
C
But
one
of
them
is
that,
as
you
look
at
the
sort
of
mean
spectral
properties
of
the
proxies
that
are
available,
what
you
see
is
a
sort
of
a
decline
in
the
higher
frequency
variability
and
increasing
dominance
of
the
lower
frequency
variability
so
that
panel
d,
the
increasing
dark
red
and
the
the
sort
of
change
or
the
weakening
and
the
increase
in
the
light
pink.
As
you
go
back
in
time.
So
changing
a
number
of
proxies.
Changing
spectral
properties
of
these
proxies.
C
If
we
sort
of
separate
these
proxies
as
Emilio,
did
you
also
see
some
interesting
behavior
in
the
very
lowest
frequencies,
so
glacial
glacial
ice,
Lake,
sediments
and
Marine
sentiments
seem
to
have
this
sort
of
remain
from
the
sort
of
early
part
of
the
Common
Era
into
the
little
ice
age,
and
then
the
sort
of
rebound
that
comes
with
anthropogenic
warming
and
and
the
exit
out
of
the
little
ice
age
and
into
the
period
of
of
modern,
anthropogenic
warming,
but
tree
rings,
don't
have
as
clear
a
signal.
C
Decline
from
the
earliest
part
of
the
Common
Era,
so
different
proxies
have
these
different
spectral
and
long-term
Behavior.
C
Also
showed
this
and
tried
to
sort
of
come
up
with
some
explanations
for
why
this
might
be
based
on
seasonality
and
the
host
of
other
things,
but
still
these
these
differences
persisted
the
marine
and
lake
sediment
proxies
the
glacial
ice.
All
had
these
sort
of
declines
in
a
sort
of
millennial,
scale
declines.
Well,
the
tree
rings
were
relatively
flat.
C
C
But
in
most
of
these
you
saw
evidence
that
there
was
particularly
in
the
low
resolution
records,
so
the
lake
sediments
and
the
ocean
sediments
this
kind
of
long-term
decline,
but
that
the
high
resolution
records
didn't
really
show
this,
and
so,
if
you
think
about,
then
this
sort
of
figure
that
was
shown
in
the
ipcc,
the
sort
of
low
frequency
figure
from
the
pages
reconstruction
in
2019.
You
do
notice
that
there's
this
flattening
out
before
a
thousand
CE.
So
where
do
those
Millennial
scale
trends
go?
Were
they
real?
C
C
There's
a
very
nice
paper
by
Lucy
lucay
Luce
in
2021
kind
of
tackling
this
question
and
I'll
call
your
attention
here
to
these
simulations
that
were
done
with
orbital,
forcing
only
using
the
cesm
model
and
the
top
topmost
panel
shows
you
sort
of
global
temperature.
Trends
but
I
want
to
focus
on
the
sort
of
red
and
blue
plots
there,
land
and
sea
in
the
middle
and
then
land
only
in
the
bottom.
C
And
what
you
see
is
the
expected
temperature
Trend
over
the
sort
of
last
millennium
them
depending
on
the
season
and
there's
two
boxes
in
there
and
so
the
black
boxes.
The
heavy
black
line
boxes,
show
the
sort
of
locations
and
seasonal
window
of
the
tree
ring
proxies
that
are
in
Rob,
Wilson
and
I's
end
Trend,
compilation
of
temperature,
sensitive
proxies
and
then
the
dashed
one
which
might
be
harder
to
see
is
from
the
Arctic
2K
compilation
and
a
couple
things
to
draw
your
attention
to,
which
is
that
the
end.
C
Trend
Network
really
samples
a
range
of
expected
Trends
right
anything
from
that
sort
of
early
season,
sort
of
May
June,
even
into
July,
of
either
warming
or
or
neutral
Trends,
and
then
that
late
season
and
high
and
low
latitude
trends
that
come
for
sort
of
Late,
July
August
into
September.
So,
depending
on
the
sensitivity
of
the
proxies
and
their
latitudes
you'd,
have
a
very
different
expectation
of
what
that
Millennial
scale.
C
Trend
might
be,
and
so
mixing
these
over
their
different
seasonal
Windows
could
potentially
obscure
the
millennial
Trend
that
we
might
expect
there
depending.
C
We
also
know
and
there's
some
earlier
papers
by
Nara,
Abram
and
Helen
McGregor,
that
show
this
quite
nicely
that
we
would
actually
expect
a
cooling
Trend,
even
without
the
orbital,
forcing
in
fact
it's
stronger
with
the
other
four
things
because
of
volcanic
eruptions.
Volcanic
eruptions
and
their
distribution
through
the
Lost
millennium
are
in
large
part
responsible
for
the
expectation
that
we
would
have
in
all
season,
particularly
High
latitude,
but
all
latitude
cooling,
at
least
over
the
last
millennium.
C
That's
the
low
frequency,
what
about
the
highest
frequencies?
Well,
what
we
know
is
that
different
proxies
do
a
different
job
of
accurately
representing
volcanic
eruptions
and
we've
actually
known
this
for
a
long
time
we
kind
of
had
to
relearn
it.
So
if
you
go
back
to
the
late
90s,
we've
got
two
temperature
reconstructions,
one
from
Keith
griffa,
using
what
we
call
maximum
late
wood
density
and
Karen's,
going
to
talk
about
a
similar
measurement
to
this
maximum
Lakewood
wood
density.
C
In
a
second
and
then
we
have
the
man
multi-proxy
Reconstruction
from
1999,
and
what
this
is
showing
is
the
average
response
to
very
large
volcanic
eruptions.
What
you
see
is
that
there's
sort
of
a
muted
to
you
know
weak
signal
in
the
multi-proxy
man
reconstruction,
but
quite
a
nice
strong
response
in
the
brifa
wood
density.
We
see
the
same
thing
in
more
recent
reconstructions.
The
earlier
Dirigo
at
all,
2006
reconstruction
has
sort
of
a
muted,
delayed
and
and
weaker
response,
Leah
Schneider's,
mxd
density.
C
Only
reconstruction
has
a
nice
sharp
response
and
a
quick
recovery
and
then
Rob
Wilson.
In
my
end,
Trend
reconstruction
sort
of
has
a
little
bit
of
both.
It
has
a
clear
volcanic
response
because
of
the
mxd,
but
probably
a
longer
tail
of
recovery
due
to
the
presence
of
tree
ring
with
data,
so
even
within
the
tree
ring
proxy,
whether
you
use
something
like
ring
with
or
whether
you
combine
multiple
proxies
or
whether
you
focus
on
something
like
the
maximum
light.
Wood
density
gives
you
a
different
result.
C
I
also
spoke
again
here
about
this
dominance
of
Summer,
sensitive
or
growing
season,
in
many
cases
for
trees
proxies
and
yet
the
pages
2K
reconstruction,
as
reviewers,
have
reminded
us
on
multiple
occasions.
These
are
Global
reconstruction.
Well,
can
it
be
a
global
reconstruction?
This
is
actually
from
the
Emilia
2017
paper
and
what
it's
showing
is
the
local
correlation
between
mean
annual
temperature
and
summer
temperature,
and
while
it's
pretty
good
in
some
places
like
the
tropics,
it's
not
so
good.
C
C
So,
as
Jason
and
I
say
the
results
that
we
have
since
sort
of
the
the
hockey
stick
in
in
the
ipcc
in
20
2001
in
the
SPM
thanks
drifty,
the
results
continue
to
show
that
by
the
late
20th
century,
temperatures
likely
exceeded
those
of
any
time
in
at
least
the
last
Millennium.
C
That
said,
there
are
pretty
big
differences
between
earlier
reconstructions,
like
the
man
at
all
hockey
stick
in
1999
recent
reconstructions
by
sebguya,
Marcus
stuffel
and
their
group,
or
the
n-tran
group
with
Rob
Wilson,
myself
and
others,
the
host
of
ipcc
ar5
reconstructions,
which
you
can
see
in
light
gray
and
that
Pages
2K
2019
that
was
featured
in
the
ipcc.
C
Most
recently,
you
see
it
has
lower
variants,
some
very
different
sort
of
early
medieval
Behavior,
different
responses
to
volcanic
eruptions,
and
so
there
are
real
differences
out
there
arising
from
some
of
these
actors.
That
I've
been
talking
about.
One
of
the
biggest
challenges.
Again,
though,
is
the
proxy
data
themselves.
They
tend
to
be
summer,
biased,
High,
latitude
and
dominated
by
tree
rings.
C
Until
you
get
back
into
the
first
Millennium,
where
the
first
Millennium
presents
a
real
challenge,
so
Jason
and
I
made
a
few
recommendations
and
we'd
be
happy
to
talk
about
this
during
questions,
we
ask
the
community
to
use
and
evaluate
multiple
reconstructions,
so
one
reconstruction,
even
if
it's
from
a
Consortium,
even
if
it's
2
000
years
and
purports
to
be
Global,
mean
and
annual,
has
positive,
positives
and
negatives
as
do
any
reconstruction,
and
so
using
and
evaluating.
Multiple
reconstruction
is
important
depending
on
what
your
purpose
of
using
these
reconstructions
is.
C
You
have
to
clearly
Define
the
intended
uses
using
a
multi-proxy,
smoothed
reconstruction,
or
one
that's
dominated
by
tree
ring
with
data
is
probably
not
as
good
for
studying
volcanic
eruptions
as
a
reconstruction
that
is
dominated
by
the
wood
density
proxies
like
the
one
Karen's
about
to
talk
about
we're,
encouraging
the
whole
Community
to
continue
to
do
spatial
reconstructions.
The
spatial
data
that
we
can
get
is
often
a
clue
as
to
atmospheric
circulation
and
where
the
strengths
and
weaknesses
are
in
our
reconstructions,
just
to
name
a
few
and
then
always
always.
C
C
D
E
Thank
you,
Kevin
for
this
nice
talk
well,
since
two-thirds
of
the
planet
are
covered
by
the
ocean.
I
am,
of
course,
wondering
about
ocean
data
which
seem
to
be
I
mean
you
show
these
two
maps
like
one
in
the
later
period,
one
in
the
earlier
period,
the
earlier
period
had
almost
no
ocean
data.
E
Although
then,
in
a
later
figure
you
showed
like
50
records
from
Ocean
sediments.
Can
you
talk
a
little
bit
about
about
that?
What.
C
Yeah
so
there's
obviously
a
tremendous
amount
of
ocean
data
out
there
covering
different
time
scales
for
the
pages
2K
purpose
when
we
were
trying
to
cover
the
last
2000
years.
Things
like
you
know
whether
the
sediment
core
comes
all
the
way
to
the
present,
whether
it
has
the
sedimentation
rate
to
resolve
parts
of
the
Common
Era,
whether
it
has
the
age
model
Precision
to
be
able
to
use
to
all
sort
of
come
into
play.
C
A
lot
of
the
methods
that
were
used
in
the
2019
paper.
That
I
showed
you
and
I'm
not
on
that
I
should
say
so.
I
can't
speak
to
everything
they
did,
but
a
lot
of
those
methods
require
a
pretty
precise
or
exact
annual
age
model
or
the
ability
to
reasonably
annualize.
So
one
of
the
limitations
in
using
the
Abundant
ocean
data
that
we
do
have
comes
down
to
resolution
age,
model
control,
and
things
like
that.
C
So
there
are
lower
resolution
Marine
records
from
the
common
air
era
in
the
database,
but
a
lot
of
those
were
not
able
to
be
applied
with
the
different
methods
that
were
used.
Some
of
them
were
some
of
the
more
so
yeah
and
I
think
you
know
Finding
ways
where
we
can
use
those
data
on
these
time.
Scales
effectively
is
also
a
really
important
thing.
You
know
we
really.
When
it
comes
to
data,
we
need
all
hands
on
deck,
so
to
speak,
every
bit
of
information
that
we
can.
C
E
F
Talk
yeah
well,
I
have
a
question
like
it's
maybe
offer
the
topic,
but
so,
in
addition
to
temperature,
precipitation
is
also
a
very
important
variable.
People
are
trying
to
represent
it,
but
we
know
that
temperature
is
so
far
I
think
it
is
the
motor
easiest
variable
to
reconstruct,
given
what
we
have
right
now.
Given
those
limitations
and
presentation
has
a
much
weaker
spatial
covariance
and
is
more
local
I'm
just
curious.
What
do
you
think
of
precipitation
field
reconstruction
with
children's?
C
Yeah,
that's
a
great
question
I,
so
my
senses
and
Jason
and
I
sort
of
wrestled
with
whether
to
tackle
this
and
we
decided
we
already
had
enough.
So
we
didn't
sort
of
go
into
this
and
confine
ourselves
to
temperature.
But
it's
a
really
good
point
and
your
everything
you
said
is
exactly
spot
on.
It
makes
reconstructions
of
moisture,
particularly
on
large
scales,
even
harder.
So
my
sense
is
that
most
sort
of
field
reconstructions
have
thus
far
sort
of
focused
on
the
regional
to
Continental
scale
and
so
said.
C
Okay,
here
we
we
feel
like
we've
got
enough
proxies
we've
got
an
extensive
network.
The
network
is
sensitive
in
a
way
to
local
Hydra
climate.
That
we
think
is
reasonable.
We've
selected
a
metric
but
I
think
when
you
start
to
move
to
Global,
hydroclimate
and
regions
where
you
don't
have
a
lot
of
hydroclimate
proxies
or
we
Face
the
same
issues.
I
just
mentioned
to
to
Andreas
that
we
might
have
great
lake
sediment
records,
but
maybe
they're
discrete
they
sort
of
record
low
stands
and
high
stands
or
something
question
becomes.
C
How
do
you
incorporate
those
data?
How
do
you
deal
with
age
model?
Change
and-
and
we've
tried
to
do
some
of
this
in
in
the
Horn
of
Africa
and
sort
of
the
Great
Lakes
region
of
Africa,
with
some
of
the
higher
resolution
records
and
one
of
the
things
you
really
have
to
deal
with.
Is
this
time
uncertainty?
So
any
you
know-
and
this
goes
for
temperature
too.
C
One
big
Advance
would
be
methods
to
really
robustly
and
interactively
and
quickly
use,
say,
ensembles
or
or
directly
incorporate
age
model
uncertainty,
and
you
know
that
may
come
with
a
trade-off.
It
may
mean
that
okay,
we
only
really
trust
our
ability
to
resolve
decadal
or
multi-decatal
variability
and
we're
going
to
have
to
leave
the
sort
of
you
know
single
year
and
Flash
droughts
behind.
So
that's
probably
going
to
represent
a
trade-off,
but
I
think
Regional
and
Continental
scale
is
still
a
a
reasonable
Target
for
hydroclimate
reconstructions.
D
G
All
right
great,
thank
you
guys.
So,
let's
get
started,
can
you
guys
see?
Okay,
yeah,
okay,
excellent?
Well,
thanks
again,
I'm
super
happy
to
get
to
talk
to
you
guys
about
what
I've
been
working
with
over
the
last
couple
years.
G
So
just
to
add
it's
not
just
me
it's,
you
know,
I'm
a
piece
of
the
puzzle,
I
think
in
a
large
working
group
of
people,
not
just
in
North
America,
but
across
the
Northern
Hemisphere
working
with
this
new
tech
technique
called
Blue
intensity,
as
Kevin
mentioned,
it's
kind
of
beef
up
kind
of
our
knowledge
and
our
current
understanding
of
how
tree
rings
are
able
to
capture
temperature,
and
if
we
can
do
you
know
a
better
job
in
the
future
at
incorporating
these
new
techniques
as
they
come
up
to
improve
our
regional
and
local
scale.
G
Paleo
temperature
estimates
Through
Time,
so
thinking
about
you
know,
climate
proxies
from
tree
rings
in
North
America,
where
I
primarily
work.
We're
really
lucky
in
the
way
that
we're
really
well
replicated
in
terms
of
you,
know,
time
and
space
in
terms
of
trees
being
able
to
capture
hydroclimate
variability,
and
so
this
is
evident
with
you
know:
Ed
Cook's,
living
Blended,
drought
Atlas,
as
well
as
this
new
seasonal
precipitation,
Atlas
and,
of
course
you
know
tree
rings
are
valuable
in
this.
G
This
way
that
we
have
these
annual
resolution
proxies,
and
so,
however,
we
do
have
this
relative
paucity
of
temperature
proxies
from
tree
rings
in
North
America,
especially
as
Kevin
mentioned.
This
is
not
just
you
know,
unique
to
North
America,
but
at
the
lower
latitudes.
We
have
kind
of
these
problems.
G
Getting
these
really
robust
records
of
temperature
from
tree
rings,
but
where
this
has
been
able
to
be
done
in
the
past
is
limited
to
high
latitude
and
high
elevation
locations,
primarily
from
conifers,
and
then
the
metrics,
which
all
really
spend
a
lot
more
time.
Talking
talking
about
kind
of
some
of
the
reasons
why
we
have
this,
this
pattern
is
because
the
metrics
we're
using
so
predominantly
ringwith
has
been
used
in
the
past
until
you
know,
semi
recently,
last
couple
decades,
this
emphasis
on
a
new
technique
rather
than
measuring
radial
growth.
G
We're
now
measuring
densy
metric
growth
from
year
to
year.
This
technique
called
maximum
ring
density
or
mxd
is
I'll
refer
to
the
rest
of
the
talk,
and
then
this
newer
technique,
which
is
largely
related
to
ring
density
and,
in
fact,
gives
kind
of
this
representative
measure
of
annual
ring
density,
a
technique,
that's
a
visible
light
technique,
called
Blue
intensity
or
bi.
G
So
the
emphasis
on
mxd
in
the
1980s
and
90s
really
for
North
America,
comes
from
the
work
from
Brooklyn
schweingruber.
This
hemispheric
scale
development
of
this
network
of
tree
ring
chronologies
that
are
sensitive
to
summer
temperatures,
and
so
why
this
emphasis
on
mxd
well,
unlike
measuring
radial
growth,
how
wide
or
how
narrow
a
ring
is
each
year
these
guys
are
looking
at.
You
know
how
dense
or
not
dense
a
ring
is,
and
this
is
related
to
the
amount
of
cell
wall
material.
G
That's
deposited
each
year
after
the
cessation
of
radial
growth,
and
so
this
image
here
is
just
showing
you
kind
of
what
that
looks
like
for
those
of
you
guys
that
aren't
familiar
with
treatment:
growth
right,
the
the
dark,
dense
Parts
here.
That's
that
secondary
cell
wall.
The
idea
that
the
warmer
the
temperatures
it's
more
favorable
for
the
metabolic
pathways
that
make
these
secondary
wall
cell
wall
material
and
so
also
kind
of
thinking
about.
Well.
G
G
So
what
I've
done
is
these
are
total
ring
with
chronologies
I'm,
looking
at
the
climate
response
based
on
crew,
T-Max
and
team
bean-
and
it's
just
appears
since
correlation
for
current
season
growing
season,
so
starting
January
through
September,
and
what
kind
of
the
the
issue
is.
Is
you
know
radial
growth?
G
G
Trend
data
set
that
they
put
together
and,
as
you
can
see
with
this
histogram
here
on
the
bottom,
you
still
see
that
the
ring,
with
only
chronologies
kind
of
dominate
the
lower
end
of
the
explained
variance
over
their
calibration
period,
whereas
the
Composites
between
mxd
and
then
mxd
and
bi
and
ringwith
tend
to
have
a
stronger
signal.
So
that
said,
you
know
we
do
have
kind
of
differences
in
tree
ring
growth
chronologies
in
their
response
to
temperature,
but
also
we
have
these
spatial
gaps
as
well.
G
So
mxds
worked
really
well,
but
it
does
have
its
limits,
and
this
is
something
that
even
you
know,
schwein
grouper
wrote
about
frequently
it's
pretty
expensive
and
it's
pretty
time
consuming,
and
so,
if
we
want
to
have
these
large-scale
efforts
to
not
just
push
our
paleo
temperature
records
back
through
time,
but
also
update
them,
it
might
not
be
the
exact
I
guess
or
the
the
ideal
approach
for
doing
this,
and
so
that
kind
of
gets
on
to
this
idea
of
switching
from
these
x-ray
based
techniques
that
look
at
density
to
these
visible
light
techniques
and
so
kind
of
to
demonstrate
what
that
kind
of
looks
like
it
might
not
be
such
a
linear
jump
to
some
people.
G
G
Well,
if
we
break
down
the
parts,
the
two
Basics
geometric
parts
of
the
cell
right
you've
got
the
cell
Lumen
the
space
and
you've
got
that
cell
wall
material,
and
this
is
you
know,
consistently
documented
in
the
wood
technology,
literature
in
the
wood
physiological
literature
and
it's
it's
kind
of
interesting
being
able
to
incorporate
those
two
bodies.
Work
now
into
tree
rings
to
kind
of
better
understand
what
we're
looking
at
the
differences
between
these
two
metrics.
G
And
so
then
we
can
see
that
cell
wall
area
is
positively
correlated
with
the
density,
as
I
mentioned
in
previous
slide.
But
then
it's
not
until
the
work
of
Shepherd
at
all
in
North
America
he's
working
in
Maine
with
red
Spruce.
Where
he
starts
to
look
at
you
know,
can
you
say,
can
you
make
a
relationship
between
actual
x-ray,
densitometry
invisible
light
reflectance
across
a
ring,
and
so
what
he's
able
to
show
is
that
late,
wood
density
and
late
wood,
brightness
or
the
amount
of
reflected
light
in
that
light,
wood
portion?
G
G
So
then,
a
few
years
later,
we
finally
get
the
work
by
Danny
McCarroll
from
2002,
where
he's
able
to
say
maximum
density
has
a
strong
negative
relationship
with
the
blue
minimum
reflectance.
So
that's
from
the
visible
spectrum,
just
that
blue
wavelet
of
light.
So
if
you
invert
it
in
theory,
you
can
have
this
kind
of
surrogate
parameter
for
this
mxd.
G
That
was
that
was
kind
of
the
thought
at
the
time
and
so
kind
of
how
that
translates
to
today
in
the
way
that
I've
been
using
it.
If
we
look
at
what
blue
intensity
is
well
simply,
it's
the
measure
of
reflected
blue
wavelet
light
across
an
annual
ring.
So
if
we
think
of
the
visible
spectrum
breaking
that
down-
and
so
you
can
do
this
with
a
camera
or
a
scanner,
so
using
a
scanner-
or
you
know
both
when
you
scan
an
image
of
a
core,
you
end
up
getting
something
that
looks
like
this
right.
G
G
If
we
look
at
again
that
similar
example
1954
looking
at
the
late
wood
portion
in
tree
rings,
we
talk
about
marker
years
right.
If
you've
got
an
exceptionally
dry
year,
you
might
have
a
narrow
band.
Well,
in
this
case,
it's
not
always
as
visually
obvious
as
this
example
here,
but
you
might
have
say
later,
lighter
late
wood
bands,
which
is
what
we're
seeing
here
with
1954
compared
to
the
rings
around
it.
The
inverse
1958,
for
example,
is
pretty
dark
and
dense
looking
well.
G
G
Inversely
1958
prior
to
2021
was
the
warmest
on
record,
so
I
hope
this
example
kind
of
clarifies
any
questions
you
you
might
have
about
that
and
again
I'm
happy
to
talk
more
about
the
kind
of
ins
and
outs
of
of
blue
intensity
and
how
it
works
in
the
future
and
so
kind
of
my
introduction
to
how
all
this
really
kind
of
built
upon
itself.
Well,
I
was
really
inspired
by
this
idea
of
having
this
tree
ring
density.
G
Network,
the
Griffin
schwein
grouper
network,
but
kind
of
the
idea
of
being
able
to
build
upon
it
with
mxc
seemed
a
little
bit
unrealistic
for
me
and
then
thinking
about
the
spatial
gaps
in
the
end
Trend
Network
fast
forward
to
Summer
of
2017..
This
is
me
as
a
new
grad
student
learning
from
Rob
Wilson.
This
technique,
blue
intensity,
and
at
this
point
it
had
really
only
been
used.
I
think
once
or
twice
in
North
America
for
a
tree
ring
temperature
reconstruction
and
so
I
was
really
inspired
by
by
his
teachings
there.
G
And
so,
as
I
mentioned,
with
the
Paul
Shepard
paper,
a
few
slides
prior,
he
had
done
some
work
in
the
1990s
looking
at
Light
reflectance
in
Maine
with
red
Spruce.
Well,
you
know
being
inspired
from
that
summer.
At
nadf,
I
went
and
I
tried
blue
intensity
on
a
kind
of
Southern
range
periphery
of
red
Spruce
in
the
Smoky
Mountains,
not
far
from
where
I'll
be
teaching,
which
is
kind
of
cool
full
circle.
G
But
we
found
that
we
had
strong
positive
relationships
between
the
late
wood,
blue
intensity
and
the
late
summer,
growing
temp
yeah
late
summer,
temperatures,
and
so
you
know
this
was
really
exciting.
But
of
course
like
anything
in
science,
you
know
you
need
replication
to
kind
of
say
well.
Is
this
this
a
mistake?
G
That
I
showed
earlier
showing
the
difference
between
latewood
blue
intensity,
which
are
the
chronologies
on
the
top
and
their
total
ring
with
counterparts
on
the
bottom
again,
just
a
simple
Pearson's
correlation.
For
current
year
temperature,
you
can
see
the
late
wood
blue
intensity,
especially
in
the
late
summer
months.
It
has
really
strong
positive
and
temporally
stable
relationships
with
temperature,
whereas
the
ring
with
counterparts
do
not
so
this
was
really
promising
in
terms
of
can
we
use
this
new
metric
in
previously
unexplored
locations
to
get
these
perhaps
refined,
temperature
reconstructions
and
so
through.
G
The
course
of
my
graduate
work.
I
was
able
to
develop
these
Regional
reconstructions
of
Maximum
and
mean
summer
temperature,
so
these
were
for
the
Intermountain
West
and
more
recently
we
got
a
millennial
length,
temperature
record
for
the
Pacific
Northwest
and
so
kind
of
fast
forward
to
present
times
what
I've
been
working
on
at
Lamont,
with
a
bunch
of
folks
from
all
over
just
been
really
exciting.
G
Is
this
synthesis
of
not
just
mine
and
others
work
with
blue
intensity,
but
with
all
the
the
ring
density,
I'm
kind
of
to
see
if
we
can
make
these
spatial
field
reconstructions?
That
are
a
you
know,
a
better
I
guess
tool
to
to
answer
certain
questions
as
Kevin
was
mentioning
right.
It's
like
being
able
to
be
explicit
with
the
type
of
reconstruction
you're
using
for
the
ques
for
really
specific
questions
so
for
in
this
case,
I
was
really
interested
in
this
idea
of
competing
hydroclimate
signals
in
the
tree.
G
This
includes
early
wood
density,
minimum
density
and
the
same
is
true
with
blue
intensity
as
well.
So
so
far,
I've
really
kind
of
just
talked
about
the
use
of
late
wood
blue
intensity.
But
as
you
would
expect
with
you
know,
density
right,
there
is
so
much
more
information,
potentially
that
the
trees
are
telling
you
that
I
think
is
worth
a
lot
more
exploration
and
and
refinement
in
the
future,
and
so
looking
at
just
the
response,
the
density
predictors
versus
the
blue
intensity
predictors.
G
We
can
see
that
you
know
we
do
have
these
seasonal
differences
among
predictors
themselves
and
I.
Think
that's
what
Kevin
was
saying
like
more
replication
and
more
kind
of
chances
to
look
at
these
similar
parameters
in
across
space.
We
kind
of
get
a
better
idea
of
not
just
the
biological
basis
of
these
new
predictors,
but
also
kind
of
a
better
understanding
of
how
these
these
parameters
kind
of
work
in
a
multi-proxy
fashion,
and
so
looking
a
little
bit
at
this
gridded
reconstruction.
G
Some
results
for
the
calibration
verification
statistics,
we're
actually
able
to
explain
quite
a
bit
of
variance
over
our
calibration
verification
period.
G
So
we've
got
so
far
pretty
pretty
good
skill
of
this
reconstruction
in
places
where,
before
we
really
haven't,
been
able
to
do
so
and
again,
I
don't
know
if
I
said
it
before,
but
these
these
temperature
estimates
are
based
on
positive
relationships
between
tree
growth
and
in
current
year.
Temperatures
which
again
is,
is
really
exciting
and
kind
of
a
pretty
new
development
and
just
kind
of
looking
at
skill
of
this
reconstruction.
G
So
far,
not
only
do
we
have
good
calibration
verification,
but
the
spatial
patterns
of
the
leading
modes
of
temperature
variability
were
able
to
really
capture
in
these
reconstructions
as
well,
which
was
really
exciting,
so
kind
of
just
showing
that
further,
and
then
you
know
so
these
are
summertime,
reconstructions,
so
being
able
to
to
see.
G
Well,
you
know
what
is
the
signal
of
volcanic
cooling
across
space,
so
we're
looking
pretty
good
so
far
and
in
terms
of
current
year
response
following
major
volcanic
eruptions
for
summertime
Cooling,
and
this
is
across
the
region
of
North
Western
North
America,
it
was
particularly
surprising,
is
even
at
these
lower
latitudes
to
this
fourth
leading
mode
here,
we're
you
know
at
30
degree
30
to
32
degrees
latitude,
we're
getting
reliable
temperature
estimates
where,
before
it
just
has
not
really
been
able
to
to
happen
with
tree
rings.
G
So
this
was
very,
very
encouraging
and
then
also
thinking
back
to
that
original
work
with
mxd
trying
to
use
this
new
and
developing
parameter
to
kind
of
understand
how
mxd
behaves
as
well
being
able
to
compare
the
blue
intensity
based
reconstructions
with
these
pre-existing
reconstructions.
G
Well,
not
only
are
we
able
to
kind
of
view
use
bi
is
kind
of
the
pseudo
proxy
updates
to
mxd,
but
when
we
combine
them,
we're
able
to
also
push
our
reconstructions
back
through
time,
which
is
also
pretty
exciting
and
so
for
the
really
specific
question
that
I'm
kind
of
looking
at
now
with
this
example
is
how
we
can
parse
out
the
relationship
between
hydroclimate
variability
and
temperature
back
through
time
using
tree
rings.
So,
as
I
mentioned
right,
this
reconstruction
is
using
positive
relationships
between
current
year
temperatures
and
tree
rain
growth.
G
As
such,
it's
also
wholly
data
independent
of
the
living
Blended,
drought,
Atlas
and
the
seasonal
precipitation
Atlas.
So
that's
also
kind
of
pretty
exciting
to
say
we
can
do
this.
Independent
evaluation
of
paleo
estimates
through
time
with
tree
rings
and
then,
lastly,
kind
of
the
bigger
picture
of
where
I'm
going
is,
as
I
mentioned
at
the
beginning.
This
is
not
just
me.
This
is
I
get
to
be
the
ringleader.
G
Gridded
reconstruction-
and
this
is
really
involved
just
a
ton
of
people
from
different
I
guess,
subfields
of
dendrochronology,
whether
it's
dendro
climate,
the
archaeological
people
kind
of
helping
us
out
finding
these
historical
samples
as
old
sources
of
paleoclimate
information.
So
there's
a
lot
of
work
to
be
done,
but
I
am
actually
extremely
optimistic
at
our
abilities
so
far
and
kind
of
the
directions
moving
forward.
Third,
blue
intensity
I'll
just
kind
of
wrap
this
up
with
it's,
not
just
you
know,
North
America,
but
over,
especially
the
Northern
Hemisphere.
G
This
technique
has
just
absolutely
exploded
over
the
last
five
or
so
years.
If
we
look
at
the
current
we're
almost
current
Global
state
of
the
blue
intensity
network,
if
we
look
at,
you
know
from
Danny
mccarroll's
initial
work
in
2002
to
2017,
you
know,
starting
in
in
northern
Europe,
these
high
latitude
places,
but
in
the
last
five
years
alone,
just
the
amount
of
work
that's
been
done
is
incredible,
and
this
is
not
just
point-by-point
locations.
This
is
also
experimenting
in
refinement
with
different
species,
especially
Rob
Wilson's.
Recent
work.
G
That
said,
bi
is
a
pretty
new
technique
that,
as
kind
of
this
new
technique,
there's
a
lot
of
refinement
and
metadata
and
tracking
I.
Think
that
really
needs
to
to
continue
as
we
move
forward
with
this
proxy
record,
and
this
starts
with
sample
collection
and
kind
of
making
sure
that
you
know
we
are
moving
forward
and
continuing
to
refine
this.
G
This
method,
in
a
way
that
you
know,
is
somewhat
ubiquitous
across
the
lab
groups
working
together
so
starting
from
sample
collection
to
data
stewardship
at
the
end,
there's
a
lot
to
be
refined
and
to
be
examined
and
tested
and
replicated,
and
my
hope
is
that
you
know
the
relative
accessibility
of
this
method
will
allow
for
that
in
the
future.
D
D
A
A
question
that
was
a
great
series
of
talks
for
Karen
I
noticed
that
when
you
plotted
the
global
map
of
blue
intensity
chronologies
for
a
lot
of
gymnosperms
on
that,
so
I
was
wondering
if
they're
intra-specific
considerations
and
if
there's
some
taxes
that
might
not
be
well
suited
to
this
technique.
G
Totally
yeah
and
so
I
think
you
know,
starting
with
conifers.
It
has
a
lot
to
do
with
you
know.
You've
got
that
light.
Wood
and
you've
got
the
early
wood
differentiation,
but
it's
also
color
of
the
wood.
As
it's
a
visible
light
technique.
You
have
to
deal
with
these
biases,
not
just
from
you
know,
differences
in
growth
in
the
trends
when
we
talk
about
detrending
right
well,
now
you
have
Trends
in
color
and
so
trying
to
get
something.
G
That's
ubiquitous
across
earlywood
latewood
in
terms
of
Heartwood
sapwood
changes,
resin
Ducks
things
like
I've,
seen
a
couple
folks
trying
to
work
on
Oak
blue
intensity
with
Oak
and
I'm,
not
quite
sure
how
they
get
around
all
the
vessels
and
and
everything
but
yeah.
There's.
Definitely
there's
people
working
on
it,
so
I
think
Milo.
G
Shreed
ball
is
also
trying
to
get
around
this
idea
of
color
bias
by
basically
making
what
he's
calling
surface
intensity
it's
it's
a
composite
black
and
white
image
where
in
theory,
if
that
works
out
and
he's
able
to
replicate
that
among
different
species,
you
no
longer
have
this
kind
of
color
issue.
So
hopefully
that
opens
the
door
for
new
species
and
such
but.
D
H
Thank
you.
Thank
you,
cara
Calvin
and
the
Karen
for
the
one
of
our
talks.
I
have
one
quick
question
for
Cara
and
you
show
the
eof
analysis
for
the
response
for
the
volcanic
eruption
and
it's
very
interesting
to
see
that
the
cooling
response
is
like
consistent
throughout
the
eof1
to
uf4
and
I
was
wondering.
Is
there
any
like
explanation
for
the
special
pattern
difference
to
the
cooling
response
like
why?
There's
like
a
special
pattern
through
uf1
UF
War,
show
the
different
special
pattern
to
the
volcanic
Cooling
yeah.
G
So
I
think
certainly
part
of
it
is
the
difference
in
the
seasonal
response
and
the
strength
of
the
response
of
the
chronologies
at
each
grid,
Point.
So
so
I'm
thinking
in
terms
of
like
the
southern
kind
of
range
periphery,
chronologies.
It's
you
don't
get
this
as
strong
of
a
response,
as
you
might
in
the
Pacific
Northwest
in
terms
of
their
ability,
the
tree's
ability
to
capture
temperature.
G
So
that's
that's,
probably
part
of
it
and
then
I
think
it's
also
got
to
do
with
probably
the
distribution
of
chronologies
kind
of
factoring
into
that.
Obviously,
there's
some
pretty
big
spatial
gaps,
so
I
showed
the
I
think
it
was
eof4
his
kind
of
into
Mexico
and
into
Eastern
or
sorry
Western
Texas.
You
don't
really
have
a
ton
of
triggering
chronologies
there,
so
you're,
relying
from
on
chronologies
that
are
further
away.
I
think
that's
also
part
of
it
too.
There's
just
increased
uncertainty
there
as
well
Kevin
jump
in.
G
If,
if
you
have
anything
to
add
I,
think
that's
a
little
bit
like
so
your
wheelhouse.
C
C
Some
of
those
regions
are
probably
sharing
data
sharing
spatial
covariance,
that's
larger
than
the
the
region
that
the
rotated
wants
to
find
so
and
I
also
think
you
know,
given
the
given
this
the
magnitude
of
some
of
those
eruptions
and
the
impact
on
North
America,
so
1600
1601
that
there's
some
pretty
strong
signals
that
are
probably
yeah
field
wide
and
that's
really
driving
that,
but
I
think
I
think
Karen
got
them
all
right.
Yeah.
G
G
That's
a
good
point,
so
I
mean
I
think
the
bulk
of
what
I'm
looking
at
with
the
temperature.
It's
these
parameters
that
come
truly
from
the
late
wood
Port
of
of
the
ring,
and
so
when
you
think
about
not
only
when
the
lignification
or
the
the
densification
of
the
ring
is
happening
it's
in
the
late
summer
season,
so
it
kind
of
makes
sense.
G
The
biological
memory
of
blue
intensity
and
density
is
not
the
same
as
say
ring
width
right,
so
it's
going
to
be
more
of
an
I
think
an
immediate
response,
so
that
is
kind
of
a
limitation.
As
you
know,
Kevin
was
mentioning
right:
I
wouldn't
use
blue
intensity,
especially
late
with
blue
intensity,
to
make
a
annual
reconstruction
of
temperature,
it's
definitely
limited
to
to
summer
seasons.
G
That
said,
though,
I
mean,
with
the
examination
of
other
metrics
say
if
you
start
to
look
at
the
early
wood,
and
you
see
an
improved
response
say
into
spring
temperature,
so
yeah
I
would
say
as
a
whole
for
now
with
what
I've
shown
in
North
America,
we
really
are
focused
on
summer
season
as
like,
the
predominant
he's
in
yeah.
D
Maybe
can
I
ask
a
question
about
the
temperature
Trend
over
the
last
Millennials,
so
in
the
past
I've
seen
people
have
deals,
the
paleoclimate,
the
data
simulation
approach
and
there
might
be
a
couple
of
products
on
there.
So
maybe
could
you
please
comment
on
how
reliable
those
are
or
what
are
the
like
perspectives?
We
could
improve
as
those
products.
C
Yeah
so
I
think
we've
we've
continued
to
learn
a
lot
since
you
know
the
late
1990s
about
some
of
the
the
strengths
and
weaknesses
of
our
our
reconstructions,
I
I.
Think
you
know,
especially
when
we
think
about
how
to
do
sort
of
model
data
comparison,
keeping
some
of
the
the
sort
of
signal
biases.
You
know
summer
sensitive
the
potential
that
that,
given
the
latitude
and
the
seasonal
sensitivity
we
we
might
expect
you
know
in
different
chronologies
in
different
locations.
Different
orbital
signals,
whether
it's
triggering
with
or
density
or
bi.
C
We
might
expect
to
have
volcanic
signals,
so
I
think
you
know
the
two
recommendations
we
sort
of
made
about
being
cognizant
of
the
strengths
and
weaknesses
of
the
underlying
proxy
data
is
important,
and
then
you
know
more
generally,
not
just
selecting
one
of
these
reconstructions.
So
even
if
it's
the
latest
one,
even
if
it
says
its
annual
mean
Global
mean
you
know,
all
of
these
reconstructions
have
made
choices.
Some
of
those
choices
are
are
not
well
constrained
right.
C
This
isn't
this
isn't
sort
of
something
where,
where
all
the
the
things
are
sort
of
come
from
come
to
us
from
frequences
statistics,
we're
making
choices
that
we
think
are
reasonable
and
different
groups
can
make
different
reasonable
assumptions.
So
I
think
the
use
of
ensembles,
whether
they're
ensembles
of
opportunity
using
all
the
available
reconstructions
or
what
we're
moving
towards
an
end
trend
is
generating
our
own
Ensemble.
So
you
know
10
or
sorry,
100,
200,
000
possible,
reconstruct
actions
reflecting
different
choices,
different
data
selection
and
things
like
that.
C
You
know
that
provides
a
way.
You
know,
perhaps
particularly
the
modeling
Community,
to
have
ensembles
of
models
and
then
ensembles
of
reconstructions
and
and
that
may
make
the
the
act
of
or
the
process
of
doing,
comparisons
between
paleoclimate
modeling
and
you
know,
however,
paleo
climate
reconstructions
can
best
be
used
at
least
more
robust
or
or
maybe
perhaps
better
said-
that
a
fuller
accounting
for
uncertainties.