►
From YouTube: Biogeochemistry Working Group - 2022 CESM Workshop Day 3
Description
The 27th Annual CESM Workshop will be a virtual event. Specifically, the Workshop will begin with a full-day schedule on 13 June 2022 with presentations on the state of the CESM; by the award recipients; and two presentations from our invited speakers in the morning, followed by order 15-minute highlight and progress presentations from each of the CESM Working Groups (WG) in the afternoon.
To learn more:
https://www.cesm.ucar.edu/events/workshops/2022/
A
Sort
of
logistics
that
we
can
go
ahead
and
discuss
while
we
wait
for
jinji
to
sign
back
in
so
I'm
gretchen,
keppel
alex
I'm
at
the
university
of
michigan
and
I'm
one
of
the
external
co-chairs
of
the
by
geochemistry
working
group
and
I'm
joined
by
abby.
Who
can
introduce
yourself.
B
Hi,
I'm
abby
swann,
I'm
the
other
external
co-chair
of
the
badger
chemistry
working
group,
I'm
at
the
university
of
washington
and
our
end
car
counterpart.
Matt
long
is
the
new
end
car
co-chair
and
is
unable
to
join
us
today.
But
we
do
have
a
recorded
presentation
from
him
and
he
gave
the
update
to
the
large
group
on
monday.
If
you
want
to
go
back
and
watch
that
if
you
didn't
catch
it.
A
Yeah,
so
the
way
we're
structuring
this
session
today
is
we're
starting
off
with
a
keynote
from
jinji
lou,
who
is
leading
carbon
cycle.
A
Modeling
efforts
at
jpl
focused
on
assimilating
observations
of
the
earth
system,
generally
space-based
observations
of
the
earth
system
to
develop
a
process,
level,
understanding
of
of
carbon
cycling
and
biology
cycling,
and
so
we
thought
that
would
be
a
nice
kickoff
to
the
first
section
of
this,
where
we
will
be
presenting
a
few
strategy
talks
to
try
to
generate
ideas
and
and
community
feedback
about
where
we
can
go
next
with
the
biogeochemistry
working
group.
A
So
we'll
have
a
series
of
five-minute
talks
designed
to
be
sort
of
thought,
provoking
about
future
directions
for
bgc
after
jinji's
talk
and
then
we'll
have
about
20
minutes
for
breakout
groups.
So
we
can
sort
of
gauge
interest
and
and
hopefully
refine
some
of
those
ideas
based
on
community
interest.
A
Following
that,
we'll
have
a
break
and
then
four
talks
that
look
at
ocean
biogeochemistry,
another
short
break
and
then
four
talks
that
are
more
focused
on
land
by
geochemistry
and
during
those
breaks.
We'll
have
breakout
rooms
open
so
that
if
you
want
to
use
that
time
for
sort
of
informal
networking,
you'll
have
the
option
to
do
so.
B
B
Great,
we
have
hopefully
not
too
many
technical
difficulties,
but
we'll
see
how
it
goes.
I'm
really
excited
to
have
jinji
joining
us
today
to
hear
about
what
you're
working
on
in
the
nasa
sphere.
Thank.
D
You
so
much
for
the
invitation,
so
I
will
try
to
share
my
screen
to
see
whether
it
works
now.
B
B
To
email
to
me
and
gretchen,
and
hopefully
one
of
us
can
share
it.
Okay,
thank
you.
B
Okay,
so
we're
gonna
start
then
with
matt
long's
talk.
Does
that
make
sense?
Gretchen
oh
wait!
Yeah.
He
was
first
on
the
agenda.
Okay,
so
I
will
share
matt's
talk.
A
E
All
right,
hello,
I
am
matt
long,
I'm
a
scientist
at
ncar
and
I'm
a
new
co-chair
of
the
biogeochemistry
working
group.
I'm
sorry,
I
can't
be
with
you
in
person
today
or
virtually
in
person,
but
I'd
like
to
share
a
few
words
about
some
ideas.
I
have
for
strategic
directions
for
the
budget
chemistry
working
group.
I've
chosen
to
encapsulate
my
ideas
in
the
context
of
this
pyramid
here,
which
entail
the
bottom
levels
of
this
pyramid,
entail
things
that
I
think
we're
already
doing.
E
The
top
three
layers
of
the
pyramid
are
things
that
I
think
we
could
do
more
effectively
and
in
particular,
I
think
they
would
enable
us
to
engage
in
actionable
research
in
the
context
of
climate
mitigation.
I
come
to
this
question
with
a
sense
of
urgency.
We're
we're
familiar,
I
think,
with
seeing
images
like
this
flooding
in
british
columbia
or
droughts
in
brazil,
catastrophic
wildfires
in
california,
or
the
intersection
of
poverty
and
climate
extremes
in
south
asia.
E
This
leaves
me
with
a
sense
of
urgency
to
engage
in
climate
mitigation
net.
Zero
is
a
framework
for
that
has
been
proposed
for
climate
mitigation
that
entails
both
reductions
and
emissions,
reductions
in
emissions
and
the
direct
removal
from
co2
from
the
atmosphere,
either
via
direct
air
capture,
for
instance,
or
the
stimulation
of
natural
snakes.
E
E
What
might
we
use
data
assimilation
coupled
to
an
earth
system
model
for
well?
One
thing
is
to
is
to
be
able
to
do
detection
and
attribution
of
variations
in
the
carbon
cycle.
Here's
a
picture
from
the
global
carbon
project
showing
the
time
evolution
of
the
ocean
carbon
sink
from
both
pco2
based
observations,
as
well
as
a
collection
of
models,
and
you
can
see
that
the
spread
across
these
various
estimates
at
the
end
of
the
record,
for
example,
is
quite
large
sort
of
a
pedogram.
E
E
The
critical
point
here
is
that
this
uncertainty
is
large
relative
to
the
policy
framework
as
enshrined
in
the
paris
accord,
and
you
can
see
on
the
figure
on
the
right
that
brett
has
illustrated
that
uncertainty
in
the
context
of
the
20
30
intended
commitments.
So
the
implication
here
is
that
we
will
not
be
able
to
tell
you
whether
the
paris
accord
has
had
an
effect
on
the
basis
of
bottom-up
estimates
of
the
carbon
cycle.
E
That
brit
is
that
for
which
we're
aiming
to
to
get
funding
and
entails
repeat
transects
using
the
end
card
g5
over
the
north
pacific
sector.
The
idea
here
is
to
generate
to
use
use
these
aircraft
observations
in
concert
with
earth's
the
earth
system,
models
esm
and
data
assimilation
to
improve
estimates
of
the
state
of
the
carbon
cycle,
its
temporal
evolution
and
our
ability
to
effectively
represent
it.
E
Okay.
Finally,
a
few
words
on
engaging
in
proactive
carbon
management.
Many
of
us
are
probably
already
aware
of
the
the
various
technologies
that
have
been
proposed
to
stimulate
negative
emissions
in
here
both
on
land.
But
here
I'm
illustrating
those
in
the.
C
E
A
cree,
a
critical
problem
in
this
space
is
that
you
know
right
now.
The
private
sector
is
spinning
up
with
technologies
and
real
projects
trying
to
are
purporting
to
sequester
co2,
but
the
research
community
really
needs
to
engage
to
ensure
that
these
technologies
are
safe
and
effective.
So
here's
an
illustration,
a
schematic
diagram
of
how
we
might
leverage
a
modeling
framework
to
help
provide
a
backbone
system
for
cdr
verification
and
finally,
just
a
few
words
on
some
thoughts
that
were
not
encapsulated
in
that
pyramid.
E
F
E
A
collective
effort
on
the
methane
cycle
in
csm,
but
it
seems
like
it
could
be
really
important
and
then,
over
the
past
several
years,
I've
been
impressed
at
the
degree
to
which
doing
simple
things
with
the
big
data.
Outputs
that
we
generate
with
your
system
model
is
complicated,
and
I
just
like
to
point
to
this
collection
of
efforts
with
which
I
personally
am
involved
that
are
aiming
to
develop
progress
on
this.
E
I
think
developing,
more
collaborative,
more,
more
collaboration,
centered
on
sort
of
the
analysis
piece
of
our
research
pipeline
is
critical
to
enable
us
to
more
effectively
address
questions
in
the
actionable
science.
E
E
A
All
right
great,
so
I
think
we'll
return
to
jinji's
time
and
won't
have
questions
about
these
speed
strategy
talks
because
we'll
be
doing
that
in
the
breakout
afterwards.
So
jinji.
I
can
go
ahead
and
share
my
screen
and
you
can.
Hopefully
this
will
work.
A
Okay,
great
so
just
let
me
know
when
to
when
to
switch
and,
as
I
mentioned,
we're
we're
very
happy
to
welcome
jinji
liu
for
our
keynote
from
jpl
to
talk
about
understanding,
terrestrial
biogeochemical
processes
with
remote
sensing
observations
and
modeling.
So
I'll,
let
you
know
also
when
you're
at
12
minutes.
Okay,
thank
you.
D
So
much
I
apologize
for
the
technical
difficulties.
I
also
first
first
want
to
acknowledge
the
contribution
from
my
collaborators,
including
paul
bloomberg,
from
caltech
sean
and
anthony
from
gpl.
So
in
this
talk
we
will.
I
will
just
highlight
two
studies
that
use
cell
observation
and
modeling
to
try
to
understand
the
treasure
virtual
chemical
processes,
and
I
will
end
with
some
slots
of
the
future
directions.
D
So
the
terrestrial
bounce
here
next
slide,
please
so
sorry,
but
the
the
top
left
panel
shows
the
trash
about.
Biosphere
really
plays
a
very
important
role
in
the
global
carbon
cycle.
I
mean
it
offsets
more
than
a
quarter
of
the
phosphate
emission.
So
far
so
at
the
future
atmosphere,
co2
growth
is
a
net
effect
between
human
emissions
and
natural
carbon
source
and
sinks.
So
the
remaining
carbon
budget,
the
amount
of
carbon
that
could
still
be
released
into
atmosphere
critically
depends
on
the
changing
of
the
land
and
ocean
carbon
zinc.
D
H
D
Tropical
region
next
slide,
please
so
the
large
model
uncertainties
really
could
be
primarily
contribute
to
two
factors:
the
complex
processes
that
control
carbon
cells
and
sinks
and
also
limited
observations
that
have
been
used
broadly
in
evaluating
and
improving
the
models.
The
process
that
control
the
carbon
source
and
sink
really
spend
several
orders
of
magnitude
in
time,
pearl
duration
and
also
special
skills.
D
Some
processes
occur
over
decades,
such
as
the
permafrost
thought,
and
we
just
could
don't
have
the
such
a
long-term
observation
to
understand
these
processes
and
the
flux
and
observation
that
could
constrain
the
current
balance.
But
it
has
really
limited
observation
over
the
region.
D
We
have
a
very
large
uncertainty
and
the
greenness
observation
that
has
been
available
for
several
decades
really
played
a
very
important
role
to
improve
the
carbon
cycle,
but
it's
not
sufficient
to
constrain
the
carbon
balance
across
the
globe
next
slide,
please,
the
the
increasing
cell
observation
provide
opportunity
for
us
to
constrain
the
global
carbon
balance
and
improve
the
understanding
of
the
coupling
between
the
terrestrial
biochemical
processes
and
climate.
So
the
left
panel
show
is
a
plot
from
a
paper
led
by
dave
schimel.
D
So
he
dubbed
the
brussels
observation
that
mario
carbon
flux
is
processed
as
black
tower
in
the
sky.
So,
with
the
broad
cell
observation,
we
can
constrain
the
net
ecosystem.
Current
balance,
the
column
sealed
observation
from
oco2
provide
global
coverage
in
combination
with
surface
observation.
D
On
the
inversion
of
framework,
we
can
estimate
source
and
sinks
across
the
globe,
so
this
was
this
plot
was
generated
by
the
bottom
right
panel
was
generated
by
osteoscience
team,
so
it
shows
where
the
carbon
has
been
absorbed
from
the
atmosphere
where
the
carbon
has
been
released.
It
also
estimates
uncertainties
as
well.
The
top-down
fluxes
has
a
footprint
about
100
kilometers
is
instead
of
a
few
kilometers
flow
square
kilometers
from
the
flux,
knight
observation,
next
slice.
Please
cell
observation
also
provide
the
constraint
on
the
component
fluxes
to
close
carbon
budget.
D
So
with
cell
observation,
we
could
estimate
gpp
using
the
new,
the
not
very
new
now
so,
let's
use
chlorophyll,
for
instance,
observation.
The
the
the
safe
observation
is
a
bad
product
of
photosynthesis,
so
it
has
been
showing
it's
linearly
related
to
gpp
on
monthly
aggregate
scale,
but
it's
really
active
research.
How
to
best
use
this
observation
and
with
column
c
observation
we
could
constrain
about
mass
burning
and,
at
the
same
time,
with
active
ladder
and
read
observation.
We
could
estimate
above
and
below
ground
bound
mass.
D
So,
with
all
these
observations
from
different
sources,
we
could
constrain
the
global
carbon
balance
and
also
could
calculate
some
really
critical
critical
quantities
such
as
carbon
use,
efficiency,
water
use,
efficiency
and
light
use
efficiency.
D
So,
in
the
rest
of
the
talk,
I
will
just
focus
on
the
two
studies
we
have
done
in
gpl
to
highlight
the
method
we
are
using
to
use
these
observations
to
improve
the
our
understanding
of
virtual
chemical
processes.
So
the
first
in
the
first
study
we
tried
to
address
how
much
enhanced
growth
and
raining
trend
over
northern
highlighted
forests
could
be
attributed
to
warming.
D
So
in
this
study
we
developed
an
emergent
relationship
between
observed
spatial
sensitivity
of
gpp
and
the
greatness
and
with
the
simulated
temporal
trend
due
to
temperature.
So
we
observed
that
the
growing
season
of
safe
constraint
gpp
shares
a
similar
spatial
pattern
with
temperature,
as
shown
in
this
plus
in
every
month
during
the
growing
season.
The
area
that
has
higher
temperature
also
has
higher
gdp
values.
D
The
next
slide
please.
So
then
they
fit
exponential
function
between
the
spatial
distribution
of
the
growing
season,
temperature
and
the
gpp.
So
each
point
in
this
plot
is
a
growing
season,
mean
gpp
and
the
corresponding
temperature
and
model
grade.
So
similarly,
we
also
fitted
exponential
function
with
fluxcom
gp
product
fluxcom
is
a
upscaled
product
based
on
the
fluxline
observation
and
the
spatial
gradient
of
temperature
explains
more
than
six
percent
of
gpu
special
variability
during
the
growing
season,
and
the
exponential
phasing
coefficient
is
about
0.2
next
slide,
please
so.
D
Similarly,
we
fit
exponential
function
between
the
spatial
distribution
of
leafy
leaf
area
index
and
temperature,
and
we
also
see
the
similar
strong
control
of
temperature
on
the
spatial
distribution
of
the
greenness,
while
the
switching
coefficient
is
smaller
than
the
fitting
coefficient
than
gpp.
So
each
panel
in
this
in
this
slide
shows
the
three
products
we
used
like
three
olivia
area
index
products.
We
use
in
this
study
next
step,
please
so
trending
models.
So
then
we
compare
the
the
spatial
sensitivity
with
20
models.
D
The
training
models
really
significantly
underestimate
the
spatial
sensitivity
of
gbp
and
on
the
living
air
index
to
temperature.
So
the
gray
bars
from
a
to
h
shows
the
spatial
sensitivity
of
the
20
models,
either
leaf
air
leaf
area
index
or
gpp
to
temperature.
So
it
shows
it
has
large
range
of
the
spatial
sensitivity.
D
Spatial
sensitivity
and
some
models
shows
very
weak
sensitivity,
and
I
highlighted
here
is
the
clm
4.5
compared
to
the
observation
shown
in
the
blue
bar
also
has
a
smaller
sensitivity,
so
the
the
the
error
bars
in
the
grid
in
from
a
to
h,
represent
this.
The
variability
of
this
special
sensitivity
across
100
year
simulation.
D
It
shows
the
even
though
the
models
has
very
different
special
sensitivity
to
temperature
if
it
doesn't
change
with
time.
So
we
expect
because
this,
the
spatial
sensitivity
doesn't
change
with
time.
We
expect
that
there
could
be
some
relationship
between
model
simulated
spatial
sensitivity
with
the
temporal
trend
due
to
temperature,
so
it
turned
out
to
be
the
case.
D
So
apparently,
if
you
go
to
the
next
slide,
please
so
the
next
in
this
slide
in
the
top
top
panels
in
the
top
two
panels,
it
shows
the
spatial,
the
relationship
between
the
spatial
sensitivity
in
the
x-axis
x-axis
accesses
and
the
percentage
change
of
gbp
and
the
leaf
air
leaf
air
index
due
to
temperature.
D
The
models
that
that
have
weaker
spatial
sensitivity
temperature
also
predicts
much
smaller
percentage
change
in
time
due
to
increase
of
temperature.
So
the
linear
relationship
helps
us
form,
an
emerging
relationship
between
the
spatial
sensitivity
and
the
predicted
percentage
change
of
gbp
and
the
leaf
air
index
due
to
temperature.
D
So,
with
this
emerging
relationship,
we
found
that,
if
air,
if
air
index
increased
about
18
percent
per
degree
celsius
and
gpp
increased
about
27
percent
per
degree
celsius,
which
are
more
than
50
or
100
higher
than
the
trendy
model
mean
so
at
the
same
time,
the
emerging
relationship
also
helped
us
reduce
the
uncertainty
by
70
and
24
4,
respectively
for
gdp
and
another
air
index.
D
The
next
step,
please
so
so
the
first
study
we
use
the
emerging
relationship
to
understand
the
high
latitude,
the
warming
effect
on
the
gp
and
the
leaf
air
index.
In
the
second
study,
we
are
trying
to
use
the
data
model
fusion
approach
to
to
to
address
how
climatic
factors
that
drive
the
carbon
flux
interability
vary
across
the
tropical
continents.
D
So,
in
many
studies
in
at
the
study,
how
I
cited
here
treated
the
tropical
region
as
as
a
whole,
but
with
the
with
rich
special
information
provided
by
satellites,
we
could
and
we
could
really
quantify
how
the
how
the
entire
ability,
how
the
dominant
climate
drivers
change
across
the
tribal
continents
next
slide.
Please!
D
Oh,
oh
okay!
So
we
use
sorry,
no,
no,
the
previous
slice
previous
slice,
please.
D
So
we
use
we
use
a
simplified
belgian
chemical
model
cardamom,
so
it
simulates
the
carbon
fluxes
carbon
allocation
and
the
carbon
water
and
energy
coupling.
There
are
four
live
carbon
pools
and
two
dead
carbon
pools,
but
you
need
to
aspect.
D
The
cardamom
is
the
data
fusion
framework,
so
you
optimize
both
the
model
parameters
and
the
initial
conditions
simultaneously,
with
with
a
non-linear
data,
fusion
approach,
it
optimized
this
parameters
in
the
initial
condition
at
each
model
grade,
so
it
ingests
the
top
down
and
b
flux,
estimates,
gpp,
biomass,
burning
and
the
background
biomass
and
below
ground
biomass.
So
all
these
products
help
constrain
the
parameters
that
control
the
carbon
balance.
It
also
simulates
the
total
water
storage
from
race,
so
it
also
can
constrain
the
water
balance
next
slide.
I
D
So
so
this
shows
the
fading
to
the
observations.
So
the
fading
to
observation
is
really
not
an
easy
task
from
this
we
show
because
the
karma
is
a
simplified
model.
Well,
the
observation
includes
all
the
process
like
that.
So
it's
fitting
observation
to
it's
really
not
an
easy
task,
but
this
shows
the
current
model
models.
D
Phase
observation
really
well
for
all
the
variables
next
slide.
Please
next
slide.
So
this
the
shows
highlight
some
parameters.
Initial
conditions,
as
previous
previous
slide.
D
D
And
then
we
use
the
optimized
model
to
sequentially,
replace
the
climate
drivers
with
the
climatology
of
precipitation,
temperature
and
radiation
and
then
quantify
the
relative
contribution
of
the
each
of
the
climate
factors
to
the
entire
ability
of
the
carbon
fluxes.
So
the
top
panel
shows
the
dominant
drivers
for
the
deliverability
of
gpp.
It
shows
over
the
tropical
forest,
amazon
and
congo.
D
Forest
radiation
is
much
is
a
dominant
factor
well
over
the
mountain
region
temperature,
but
over
around
the
shrubland
is
the
precipitation
and
compared
to
the
study
published
like
more
than
two
days
about
two
decades
ago,
constrained
by
the
modis,
they
show
has
much
smaller
region
that
control
the
variation
and
also
the
temperature
pleasure
a
large
important
role
over
the
mountaineers
region.
Next
slide,
please
so
the
for
the
for
the
ecosystem,
restoration
and
the
night
belts
net
biosphere
exchange,
the
precipitation
is,
plays
a
much.
H
D
Little
more
important
role
than
than
the
radiation
and
the
even
over
the
tropical
south
american,
the
congo
region,
so
it
shows
like
really
at
a
great
scale.
The
participation
is
a
more
important
driver
for
the
night
bowser
exchange.
This
is
really
consistent
with
the
study
published
in
2017
using
the
flexcom
observation
by
that,
using
the
statistical
approach
where
you
with
the
the
conclusion
we
draw
from
this
study,
really
is
an
emerging
property
from
data
constraint
model.
D
So
this
really
implies
some
problems
of
using
cell
observation
to
improve
the
model
prediction
at
a
grade
level.
Next
slide,
please.
So
this
is
really
come
to
the
summary.
So
I
really
just
talked
about
three
three
points.
One
is
about
the.
We
could
use
the
cellular
observation
to
constrain
the
carbon
budget
across
the
globe,
but
use
the
two
studies
to
highlight
some
of
the
some
of
the
research
work.
D
We
have
that
we
are
doing
in
a
gpl
using
either
statistical
emerging
constraint,
approach
or
data
model,
fusion
approach
to
improve
understanding
of
the
belgian
chemical
processes.
Next
slide,
please.
So
this
is
my
final
slide
next
slide,
so
the
final
slice.
So
I
think
just
once
one
point
I
want
to
emphasize.
The
cell
observation
really
provide
rich
special
information.
D
We
could
better
use
to
constrain
not
only
the
global
carbon
balance,
but
really
special
spatial
information,
spatial
distribution
of
this
carbon
balance.
This
is
really
critical
to
improve
prediction
and
also
to
increase
the
effort
to
like
using
the
observation
to
constrain
the
model
parameters
and
the
initial
condition-
and
the
last
point
is-
is
really
to
increase
the
dialogue
between
the
modeling
community
and
observation
community
to
better
use
the
current
and
future
cell
observation.
Thank
you.
So
much
for
your
attention.
A
Thanks
gingy,
I
think
we
don't
have
time
for
questions,
but
if
folks
want
to
put
them
in
the
chat
and
jinji
can
try
to
address
those
in
the
chat.
B
A
Yep,
I
think
I'm
up
next.
You
know
five
minutes.
Okay,
that
sounds
good.
Let
me
go
ahead
and
share
my
screen
again.
So
I'm
presenting
just
one
of
these
quick
strategy
talks.
A
I
wanted
to
focus
on
how
we
can
potentially
use
bgc
capabilities
within
cesm
to
guide
adaptation
and
mitigation
science,
so
some
overlaps
with
what
matt
has
discussed,
but
I
guess
I'm
coming
at
it
from
the
perspective
that
if
we
want
to
avoid
catastrophic
climate
change,
carbon
is
ultimately
going
to
have
a
price,
and
I
think
that
provides
some
opportunities
that
we
can
use
csm,
bgc
capabilities
to
track
stocks
and
flows
of
carbon
that
we
can
use
this
model,
which
uniquely
captures
natural
variability
in
the
earth
system
to
try
to
improve
our
capability
to
use
observations
of
co2.
A
That
jinji
discussed
and
matt
mentioned
in
his
talk
to
to
to
be
able
to
sort
of
see
the
signal
that's
emerging
due
to
trends
and
human
activities
and
natural
feedbacks,
rather
than
the
the
variability
in
the
system
and
sort
of
most
importantly,
we
should
be
identifying
partners
or
stakeholders
with
whom
we
can
share
information
and
with
whom
we
can
think
about
how
we
can
further
develop
model
capacity
and
application
of
the
model,
and
I
think
that
ties
into
some
of
the
talks
that
occurred
yesterday
in
the
justice
session
that
you
know,
we've
got
this
amazing
tool
and
we
probably
need
to
start
thinking
about
different
ways
of
doing
science,
moving
away
from
this
cmap
perspective
and
towards
a
user
driven
and
applications
driven
focus
how
to
use
these
capabilities.
A
So
one
example
of
how
we
can
use
our
capabilities
for
mitigation
science.
We
know
that
there
is
a
relationship
between
temperature
change
and
cumulative
carbon
emissions.
This
is
a
figure
from
matthews
paper
in
2009
and
more
recently,
we
can
see
on
that
figure
that
there
are
a
lot
of
processes
that
can
nudge
that
relationship
in
different
directions,
and
many
of
these
are
related
to
biogeochemistry.
A
So,
for
instance,
carbon
cycle
feedbacks
on
land
and
ocean
long-term
permafrost
feedbacks
and
things
like
non-co2
forcing
so
methane,
for
instance,
is
an
important
biogeochemical
compound
that,
in
the
short
term,
can
really
nudge
us
away
from
this
single
transient
climate
response
to
cumulative
emissions
curve
and
and
towards
something
higher.
A
But
it's
also
on
an
actionable
basis,
an
area
that
we
can
have
control
over
and
so
there's
a
lot
of
ways
that
we
can
use
bgc
capabilities
to
better
understand
opportunities
for
nation
in
terms
of
what
we
can
provide
for
actual
carbon
accounting.
One
of
the
key
things
that
we
can
provide
in
terms
of
what
matt
mentioned,
that
inverse
models
haven't
sort
of
converged.
Even
over
the
last
30
years.
A
Csm
bgc
is
sort
of
uniquely
capable
of
describing
at
a
process
level
what
drives
interannual
variability
in
carbon
cycling
and
if
we
can
take
advantage
of
that,
potentially
in
the
context
of
data
assimilation
with
some
of
the
rich
observations
that
jinji
described,
we
would
have
a
much
more
potent
way
of
being
able
to
say
to
policymakers
and
decision
makers.
This
is
what
we
think
the
state
of
the
carbon
cycle
is,
and
this
is
what
remaining
emissions
are.
A
Similarly,
on
the
work
that's
been
going
on
within
bgc
to
include
riverine
transport.
If
we
can
put
carbon
into
that
in
a
truly
coupled
way,
we
have
an
opportunity
with
this
model,
to
keep
track
of
some
of
these
invisible
fluxes
of
of
carbon
that
can
make
carbon
accounting
difficult
and
then
the
uncertainties
can,
at
a
decision-making
perspective
on
sort
of
overwhelm
the
the
values
that
that
policymakers
want.
So
those
are
some
opportunities,
I
think,
there's
also
opportunities
with
bgc
to
do
adaptation.
A
Science,
the
land
model,
is
making
huge
strides
and
including
sort
of
actionable
components,
including
agriculture,
and
being
able
to
model
foreign
vulnerability
and
health
within
fates.
I
think
we
should
tie
some
of
these
impacts
back
to
carbon
storage
and
local
climate
impacts
in
the
fully
coupled
sense.
So
we're
not
just
thinking
about
what's
happening
on
land
and
vegetation,
but
what
does
that
mean
for
carbon
stocks?
What
does
that
mean
for
feedbacks
with
local
weather
due
to
what's
happening
on
the
land?
A
And
for
this
to
be
useful,
we
need
to
form
better
partnerships
with
communities
and
decision
makers.
So
what
information
do
they
deem
actionable,
and
can
we
figure
out
ways
that
we
can
use
uncertainty
in
our
model
rather
than
as
a
hindrance
as
a
way
that
we
can
think
with
stakeholders
about
how
to
do
scenario
planning?
How
can
we
take
our
global
model
and
move
to
a
framework
where
we
have
multiple
stakeholder
groups
that
can
use
the
different
worlds
envisioned
by
ces
and
bgc
to
think
about
what
the
future
might
hold?
A
So
those
are
just
a
few
ideas
about
where
bgc
can
go
next
and
now
we
will
hear
from
abby.
B
I
am
going
to
give
another
perspective
on
things
that
we
can
do
as
part
of
the
bgc
working
group,
and
I
want
to
start
by
saying
this:
is
they
very
much
support
the
ideas
that
have
already
been
presented,
but
this
is
to
present
some
additional
perspective
of
things
that
I
think
bgc
is
uniquely
suited
to
work
on.
So
my
main
point
is
that
the
coupled
system
works
differently
than
the
individual
components
in
isolation
and
that
the
bgc
working
group
is
core
to
investigating
and
building
understanding
of
this
coupled
system.
B
So
maybe
a
bit
one
step
upstream
from
the
application
side
of
things,
but
that
the
there's
really
no
other
place
within
the
cesm
where
people
are
focused
on
why
the
coupling
matters
and
how
the
coupling
changes
the
answer.
So
I'm
going
to
give
just
a
couple
of
examples
of
this
to
motivate
it,
and
then
I'm
excited
to
hear
what
other
folks
have
to
say
in
their
in
the
discussion.
B
So
first
is
just
to
point
out
that
when
we
couple
land
and
atmosphere,
we
get
a
very
different
climate
response
to
changes
on
the
land
surface.
So
this
is
a
study
by
maurice
lagoo,
looking
at
when
you
change
land
surface
properties,
how
much
temperature
change?
B
Do
you
get
for
making
the
surface
darker
when
you
have
only
the
land
model
on
the
left
and
when
you
couple
the
land
model
in
the
atmosphere
model,
you
get
a
lot
more
response
to
the
climate
system,
and
so
this
is
just
an
example
of
where
we
need
to
have
be
thinking
about
changes
in
the
climate
system,
where
we're
considering
all
of
the
components
and
how
they
feed
back
on
one
another.
Not
only
is
the
magnitude
different,
but
the
spatial
pattern
is
different
as
well.
B
B
So
a
single
emissions
driven
trajectory
leads
to
a
range
of
co2
in
the
atmosphere,
outcomes
of
about
300
parts
per
million
across
models,
but
the
parameter
choices
within
a
model
also
matter.
This
is
a
study
by
ben
booth
at
all
from
the
hadley
center
model.
This
is
actually
going
back
to
cmap3,
but
this
is
an
example
where
they
took
a
single
model
and
perturbed
a
whole
bunch
of
parameters
with
the
fully
coupled
carbon
cycle
model.
B
B
Here's
an
example
showing
that
the
uncertainty
of
the
carbon
cycle
doesn't
stay
in
the
carbon
cycle,
so
this
is
the
amount
of
warming
you
get
from
increasing
co2
to
the
radiative
effects
on
the
left
due
to
the
physiological
effects
on
the
right.
It
looks
small
due
to
the
plant's
responses
to
co2,
but
it
is
up
to
20
percent
locally
of
the
warming
and
it's
even
six
percent
in
the
global
average
just
due
to
how
the
plants
are
responding
and
again
that
is
going
to
vary
across
models
and
across
parameter
choices.
B
These
physiological
effects
show
up
in
additionally
in
things
like
rainfall
as
cape
cooperman
showed
a
few
years
ago,
so
I
just
wanted
to
show
you
really
quickly.
One
plot
we've
done.
This
is
the
very
first
look
at
a
perturbed
parameter
ensemble
within
not
the
fully
coupled
carbon
cycle,
but
with
land
atmosphere
and
a
slab
ocean.
B
To
tell
you
that
this
is
being
driven
at
the
boundaries
by
things
like
the
dry
surface,
layer,
parameterization
and
biomass,
heat
storage
and
stomatal
conductance
parameter
choices
with
a
lot
more
work
to
do
to
understand
that
in
a
spatial
pattern.
But
I
wanted
to
point
out
that
this
is
quite
a
large
range,
even
when
we
include
the
changes
over
ocean.
So
in
fact
it's
1.23
degrees
celsius
range
across
these
parameter
choices
and
global
mean
temperature,
which
is
you
could
compare
to
something
like
three
degrees
across
all
cmip5
models.
B
So
this
is
quite
a
large
range
just
from
these
parameter
choices
and
again.
This
is
an
illustration
just
of
this
idea
that
the
coupled
system
works
differently
and
I
think
the
biology
chemistry
working
group
is
uniquely
poised
to
be
thinking
carefully
about
the
implications
of
coupling
and
how
the
individual
components
may
be
more
important
or
dif
important
in
different
ways
when
they're
coupled
together.
A
Great
thanks
so
now
we
are
we're
reserving
about,
I
think,
maybe
we'll
we
might
end
up
eating
into
the
break
a
little
bit.
If
that's,
okay,
we
had
a
20
minute
break
schedule,
but
maybe
we
can
move
that
to
a
15-minute
break
scheduled
and
then
we
can
have
like
10
minutes
in
breakout
groups
and
10
minutes.
To
recap,
in
a
larger
group,
to
sort
of
discuss
what
areas
you
all
find
most
compelling
in
terms
of
the
directions
that
we
could
take
the
bj
bgwc.
A
What
are
you
interested
in
in
doing
what
capabilities
would
you
like
to
be
added?
What
connections
would
you
like
to
be
made
and
hopefully
sort
of
a
look
at
what
the
modeling
efforts
at
jpl
are
and
a
few
visions
for
what
we
could
be
doing
will
help
stimulate
that
discussion.
A
So
I
think
we've
got
a
google
slide
deck
that
folks
can
put
their
ideas
in
I'll
paste
that
into
the
chat
and
then
we're
going
to
break
into
like
oh-
and
we
already
did
great
thanks,
we'll
begin
to
like
six
different
groups
and
you
can
have
10
minutes
to
brainstorm
and
then
we'll
come
back
as
a
group
and
we
can
share
a
vision
before
we
have
a
break
so
yeah.
B
C
Yes,
you
can
so
my
name
is
jackie,
schumann,
I'm
based
here
at
end
car
and
someone
else
in
my
group
nominated
me.
So
I
will
report
on
our
group.
We
feel
that
we
should
continue
to
leverage
the
the
work
with
the
ppe
to
for
assimilation
of
remote
sensing
observations
with
that.
The
connections
between
observations
and
models,
focusing
on
how
we
reduce
uncertainty.
C
We've
already
seen
examples
today
with
the
ppe
and
zarakis's
work
that
there
is
a
lot
of
sensitivity
to
parameter
choice,
and
so
there's
a
feeling
that
we
should
embrace
uncertainty.
You
know
we
can
tune
up
the
model,
but
that
uncertainty
is
inherent
to
the
process,
and
so
something
like
fates,
which
adds
complexity,
has
the
capacity.
C
It
may
not
reduce
uncertainty,
but
it's
going
to
give
us
more
information,
and
so
that
information
can
be
more
directly
connected
to
observations
you
know
outside
of
remote
sensing,
but
also
inherent
to
remote
sensing
ocean-based
work
specifically
ocean-based
carbon
dioxide
removal.
Looking
into
where
does
this
matter
for
policy
cesm
is
a
key
tool.
It's
not
necessarily
being
used
in
back
of
the
envelope
calculations
and
so
looking
at
the
downstream
effects
and
global
impacts,
as
with
the
land
or
the
ocean.
C
These
are
not
limited
to
specific
regions,
and
so
in
the
coupled
framework,
you
know
we
can
see
that
things
extend
across
the
ocean
and
land
respectively,
and
that
was
our
quick
jam
session.
B
Awesome
thanks
jackie
and
then,
since
we
do
want
to
give
people
a
time
to
have
some
break.
Can
we
just
hear
one
comment
from
a
couple
of
other
groups?
I'm
gonna
nominate
group
one
to
give
us
one
thing.
They
talked
about
that
they'd
like
to
share
hello.
C
This
is
holly
olivarez.
Let's
see
how
do
I
pick
one
thing:
I'm
gonna
choose
well.
What
we
were
talking
about
was
not
particularly
abandoning
the
cmip
process,
but
also
not
continuing
to
put
so
much
weight
into
it
as
the
thing
and
so
and
all,
but
at
the
same
time
that
we
can't
have
just
one
model
that
is
shifting
our
our
priorities.
So
a
thousand
percent
on
board
that
we
shift
our
we
shift
how
we're
using
our
tools,
but
that
we,
it
would
be
great
if
we
have
more
than
just
cesm.
C
That
is
doing
this
in
this
time
frame.
Thanks.
C
Keith,
do
you
want
to
do
it
or
should
I?
I
don't
think
we
ever
go
for
it?
Kristin?
Okay,
so
we
just
had
a
our
discussion
more
blended
into
doing
actionable
science
like
basically
reevaluate,
which
community
runs
could
be
done
with
for
be
useful
to
stakeholders
actually
like
in
terms
of
conservation
or
fisheries,
because
it's
just
become
clear
that
you
know
a
lot
of
the
potential
collaborators.
Don't
even
know
the
model
ones
exist
and
what
output
is
actually
available
from
them.
C
Just
in
terms
of
being
useful-
and
we
know
this
is
like
a
big.
This
actual
science
has
been
a
big
focus
here
at
incar
and
yeah,
making
sure
that
we
communicate
these
results
more
effectively,
so
I'll
leave
it
there
since
we're
being
quick.
B
C
J
J
You
can
chime
in,
but
one
thing
that
came
out
of
our.
J
B
B
K
To
say
methane
was
something
that
came
up
in
our
group
as
being
really
important,
especially
after
monday's
discussion
on
the
importance
of
methane,
and
we
don't
really
have
a
methane
cycle
in
the
model
right
now
which
we
can
run
interactively.
So
that
could
be
an
effort
that
we
could
put
into
for
the
bgc.
B
Great
okay,
thanks
everyone.
I
wanted
to
give
everyone
a
few
minutes
to
take
a
break
before
we
reconvene
and
15
minutes.
I
think
there's
gonna
be
breakout
rooms
if
you'd
like
to
stay
and
chat
with
other
folks,
or
we
will
see
you
back
here
in
15
minutes
for
talks.
A
And
if
you
are
a
presenter,
make
sure.
D
A
Ahead
and
try
sharing
your
screen
during
the
break
and
if
you
can't
share
your
screen
like
do
I've
gotten
slides
from
several
of
you,
but
make
sure
you
send
to
either
abbey
or
me
your
slides.
If
you
can't
share
your
screen.
Thank
you.
B
Great
and
then
could
we
get
some
optional
breakout
room
started.
That
would
be
great.
B
I
J
C
I
C
C
A
A
A
We
can
do
it
half
and
half.
That
sounds
good,
so
I
could
keep
time
for
the
first
session
and
you
can
do
introductions
sure
all
right
sounds
good.
B
And
we'll
give
you
a
warning
nicola
at
10
minutes
or
yeah.
Okay,.
B
We
are
going
to
get
started
with
a
series
of
four
talks
and
then
we'll
have
a
short
break,
just
a
quick
break
and
then
four
more
talks.
After
that,
the
first
group
is
loosely
centered
around
ocean
topics
and
the
second
group
loosely
centered
on
land
topics,
but
these
are
all
things
that
projected
onto
the
coupled
system
and
so
I
hope,
are
relevant
to
everyone
who's
gathered
here.
First
up
we
have
nicola
wiseman.
You
want
to
go
ahead
and
share
your
screen
and
you're
also
muted,
as
well.
L
All
right,
hi
everyone,
my
name
is
nicola
wiseman,
I'm
a
graduate
student
at
the
university
of
california,
irvine
and
today
I'll
be
talking
about
phytoplankton,
variable
elemental,
stoichiometry
and
how
that
modifies
marine
carbon
fluxes.
L
So
let
me
just
think:
okay,
there
we
go
so,
as
you
know,
the
bgc
working
group
most
of
us
know
that
the
carbon
export
of
the
ocean
is
linked
to
nutrients
via
a
process
called
stoichiometry
where
in
the
surface
ocean,
we
have
phytoplankton
undergoing
photosynthesis,
where
they
take
up
carbon
dioxide,
nitrogen
phosphorus
iron
and
other
micronutrients
assimilate
into
phytoplankton
biomass
at
these
ratios
and
expo
where
it
is
therefore
exported
to
the
deeper
ocean,
and
so
I'm
going
to
be
focusing
on
this
sort
of
last
part.
L
This
portion
of
the
marine
biological
pump
right
here,
where
we
have
this
phytoplankton
biomass,
where
these
elements
are
assimilated
together
at
this
ratio.
So
recently,
what
we've
been
doing
with
the
cesm
bec
is
looking
at
these
carbon
to
nitrogen
ratios.
So
phytoplankton,
carbon
and
nitrogen
ratios
play
a
key
role
in
the
coupling
of
the
carbon
and
nitrogen
cycles,
where
there
are
significant
variations
in
different
ocean
environments.
L
Experiments
have
shown
that
including
variable
c
to
n
is
necessary
in
order
to
accurately
account
for
total
c
to
end
of
p
variability,
as
well
as
nutrient
cycling.
So
what
we
want
to
know
is
what
are
the
downstream
effects
of
implementing
variable
c
to
n
and
what
impact
does
it
have
on
the
global
biogeochemical
cycles?
L
So
these
are.
This
is
just
some
more
motivation
of
why
we
need
variable
carbon
denatured
to
models
so
most
earth
system
models
have
used
fixed
red
field
c
to
n
ratios
previously,
and
this
has
been
found
to
underestimate
the
ocean
dic
inventory
and
that
as
well,
that
the
marine
nitrogen
cycle
is
more
sensitive
to
biological
processes.
L
When
variable
c
to
n
is
included,
so
the
marine
nitrogen
cycle
we
care
about,
because
it's
key
to
global
net
primary
productivity
as
most
of
the
oceans
are
nitrogen,
limited
and
perturbations
to
nitrogen
export
and
nitrogen
fixation
affect
the
global
nitrogen
supply
and
denitrification
in
particular,
which
serves
as
a
loss
of
nitrogen
from
the
marine
ecosystem
is
very
sensitive
to
export
and
oxygen
availability,
which
are
also
highly
variable
within
earth
system
models,
and
particularly
the
ocean
biogeochemical
models.
L
So
I'm
presenting
some
results
here
using
the
cesmbc
version
1.98,
where
we
have
three
phytoplankton
classes,
our
small
phytoplankton,
diazotrophs
and
diatoms,
where
we
now
have
fully
variable
c
to
end
p,
f
e
to
s
I,
this
has
recently
been
updated
to
have
some
updated
iron
sources
which
include
hydrothermal,
vents
and
bottom
scavenging,
in
addition
to
the
standard,
atmospheric
deposition,
sediments
and
rivers.
L
Sort
of
metric
for
phytoplankton
stoichiometry,
so
I'm
just
going
to
go
through
this
a
little
bit
so
now,
each
for
each
nutrient,
n,
p,
fe
or
si
it
is
its
ratio
to
carbon
is
calculated
from
this
sort
of
schematic
where
at
high
ambient
concentrations
in
the
surface
waters
of
the
nutrient
x,
the
x,
the
nutrient
carbon
ratio
is
at
its
maximum,
which
we
prescribe.
L
We
also
prescribe
this
optimal
x
concentration
and
when
the
nutrient
concentrations
fall
below
that
value,
the
phytoplankton
decrease
their
uptake
of
the
nutrient
accordingly
and
reduce
their
quotas
until
the
minimum
is
reached.
So
this
allows
for
fully
variable
c
to
end
p,
to
f
e
to
s.
I
we
also
have
some
additional
constraints
on
our
p
quotas
and
our
s.
I
quotas
for
p
quotas
when
nitrate
is
particularly
low.
L
Both
the
n
and
p
uptake
are
not
so
not
just
the
p,
but
the
n
and
p
uptake
are
reduced
accordingly,
in
order
to
maintain
appropriate
end
to
p
values
and
for
s
I,
which
are,
is
important
for
our
diatoms,
when
the
concentration
of
si
decreases
as
cytocarbon
decreases,
just
as
we
would
expect
from
the
schematic.
But
we
also
have
an
iron
constraint
on
them
where,
when
si
is
replete
but
but
iron
is
low,
the
side
to
carbon
increases
accordingly.
L
So
here
are
our
new
dynamic
ranges
that
we've
implemented
for
each
phytoplankton
group.
So
the
these
ranges
for
c
to
end
to
p
were
constrained
from
the
go
ship
pom
observations
from
five
different
cruises
around
the
globe.
L
L
So
we
tuned
these
parameters
accordingly,
based
on
observations,
as
well
as
sort
of
trying
to
match
to
observed
surface
concentrations
of
nutrients
from
world
ocean
atlas
and,
as
you
can
see,
they're
they're
highly
variable,
they're
group
specific,
and
so
what
I'm
going
to
show
here
is
the
results
of
comparing
a
300-year
pre-constant
pre-industrial
co2
run,
whereas
we're
averaging
over
the
last
20
years
versus
one
where
fully
fixed
stoichiometry
is
included.
L
So
I
see
a
c
to
ndp
of
96
to
16
to
1
and
ironed
a
carbon
of
7
and
an
s
I
to
n
of
1
compared
to
these
values
here
and
so
I'll
get
into
those
results.
Here's
just
that
plm
database
again,
really
quick.
L
So
we
have
observations
from
the
atlantic
ocean,
the
pacific,
as
well
as
the
indian
ocean
basins,
and
I'm
going
to
be
comparing
them
to
the
cesm
export
at
100
meters,
and
one
thing
to
note
is
just
that:
the
export
at
100
meters,
the
pom
observations
include
not
just
phytoplankton
zooplankton,
but
also
heterotrophic
bacteria
and
detritus
that
are
not
present
in
cesm.
L
So
here
we
just
have
the
nutrient
of
the
carbon
to
nutrient
concentration
or
the
phytoplankton,
slash
export
c
to
n
c
to
p
on
the
y
axis
versus
dissolved
ambient
nutrients
on
the
x
axis,
and
this
is
just
showing
you
know.
When
we
have
lower
concentrations
of
nutrients.
We
have
increased
carbon
to
nutrient
value,
so
our
blue
is
our
observations.
Our
red
is
our
export
from
the
phyto
from
csm
and
then
our
black
is
the
cesm
phytoplankton
alone.
So
it's
just.
L
This
is
the
main
takeaway
here
is
just
that
our
export
is
in
relative
agreement
with
our
pom
observations,
which
is
good.
So
then
we
went
on
forward
with
some
of
the
more
science
questions.
Actually
so
here's
just
comparing
those
surface
pom
observations.
So
in
the
background
here
we
have
the
cesm
output
again.
This
is
export
at
100
meters
over
laid
on
the
top,
and
these
squares
are
the
pom
observations
for
these
transects
and
overall
there's
sort
of
general
agreement
between
the
observations
and
the
model
output.
L
Here's
carbon
to
phosphorus.
Again,
we
sort
of
one
of
the
really
key
things.
Is
we
see
these
elevated
values
in
the
oligotrophic
gyres
where
nutrients
are
limiting,
and
then
we
see
lower
c
to
p
values
in
our
upwelling
regions
where
nutrients
are
in
in
reasonable
supply
and
then
here's
our
endopee
as
well?
So
one
of
the
main
takeaways
from
this
is
that
getting
our
surface
nutrient
fields
correct,
is
really
key
to
calculating
or
correctly
simulating
our
c
to
end
of
p
in
the
surface
ocean
and
gradients.
L
So
here
I
have
some
of
those
results
from
the
comparison
simulations
where
again,
we
use
fixed
stoichiometry
on
one
side
and
variable
stoichiometry
in
the
other,
and
these
are
maps
of
nutrient
limitation
where
purple's
nitrogen
blue
is
iron.
Green
phosphorus,
yellow
silicon,
great
temperature
and
orange
represents
a
pleat
ocean.
L
So
obviously
we
see
very
strong
differences
here
between
the
variable
stoichiometry
and
the
fixed
stoichiometry
versions
where,
in
the
fixed
stoichiometry,
we
are
seeing
a
transition
from
the
unlimited
ocean
that
we
expect,
and
that
has
been
observed
in
bottle
incubations
transitions
primarily
to
a
si
and
p
limited
ocean.
We
also
see
large
increases
in
si
limitation,
where
n
or
iron
and
or
fe
were
previously
limiting.
L
L
The
spatial
distribution
doesn't
change
much,
except
that
our
p
limitation
gets
much
stronger,
and
so
these
observations
of
nutrient
limitation
have
been
sort
of
in
general
agreement
with
bottle
experiments
from
the
past,
as
well
as
modeling
studies
that
have
looked
at
these
surface
maps
of
nutrient
limitation.
L
So
obviously
you
know
when
we
use
fixed
stoichiometry
versus
variable,
we're
having
very
direct
impacts
on
the
surface
nutrient
cycling,
which
leads
me
to
some
key
global
fluxes
that
I
have
here
so
on
carbon
and
nitrogen
fluxes.
Again,
we
have
our
variable
simulation
versus
our
fixed
npp,
poc
export
nitrogen
fixation
and
water
column
denaturation.
L
So
when
the
fixed
stoichiometry
is
used
instead
of
this
variable,
we
have
a
decrease
in
all
of
these
parameters,
all
of
these
values
and
fluxes.
So
we
get
a
decrease
in
that
primary
productivity
of
about
1.4
percent.
We
have
a
decrease
in
poc
export
of
about
16,
a
decrease
in
nitrogen
fixation
of
up
to
40
percent
and
a
decrease
in
water
column.
Denaturation
of
75-
and
one
thing
to
note
here,
is
that
our
variable
case
is
in
good
agreement
with
previous
estimates,
from
other
studies
based
on
observations,
inverse
models,
etc.
L
So
it's
just
really
key
to
sort
of
point
out
here
that
our
fixed
version
is
really
underestimating
a
lot
of
these
carbon
and
nitrogen
pools
and
fluxes.
L
So,
in
summary,
our
dynamic
stoichiometry
is
necessary
for
understanding
ocean
biogeochemistry
and
our
global
nutrient
cycles.
L
A
Maybe
questions
in
the
chat
as
or
one
quick
question
as
we
get
the
next
slides
up.
B
Yeah,
so
why
don't
we
go
ahead
and
have
our
there?
We
go
great
job
amanda,
we'll
get
our
if
somebody
I'm
not
seeing
any
immediate
hands.
So
why?
If
you
have
questions
for
nicola,
please
put
them
in
the
chat
or
contact
her,
but
go
ahead
and
have
a
discussion
in
the
chat
there
and
we'll
move
on
to
amanda.
So
amanda
faye
is
up
next.
I
So
just
a
quick
overview.
Large
volcanic
eruptions,
such
as
pinatubo,
which
erupted
in
june
of
1991
act
to
cool
the
climate
due
to
the
injection
of
aerosols
and
other
particles
into
the
stratosphere,
obviously
much
more
complicated
than
that,
but
that's
just
a
basic
overview.
So
our
experimental
setup
involved
creating
a
new
large
ensemble
with
the
sole
change
being
the
manipulation
of
the
volcanic
aerosol
mass
mixing
ratio
forcing
so
we
replaced
the
five
years
1991
through
1995,
with
five
previous
years,
1986
through
1990
to
simulate
a
non-eruption
condition.
I
I
I
want
to
touch
on
another
detail
of
the
experiment,
though,
because
we
actually
re-ran
the
original
cesm1
large
ensemble
on
the
cheyenne
machine.
We
found
that
the
small
changes
caused
by
the
compiler
computer
switch
impacted
our
ability
to
do
a
true
one-to-one
comparison,
because
we
were
running
the
no
pinatubo
on
cheyenne.
So
the
preliminary
analysis
I
did
looking
at
the
impact
of
the
computer
change
showed
that
the
changes
were
on
the
scale
of
what
we
would
see
is
variability
within
members
of
a
large
ensemble,
depending
on
what
variable
you
want
to
look
at.
I
I
I
So
we
subtract
no
pinatubo
from
the
original
lens
and
we'll
go
through
some
interpretation
of
these
plots
just
to
jump
right
into
the
results.
We
find
that
the
two
ensembles
show
that
the
ocean's
physical
state
and
the
surface
fluxes
respond
very
quickly.
Post-Eruption
and
changes
in
the
interior
are
long-lasting.
Oxygen
is
immediately
injected
into
the
upper
ocean
post
eruption
and
then
it
transits
to
depth
where
it
permanently
changes.
The
permanent
increases,
the
interior
ocean
inventory
and
the
global
ocean.
Carbon
sink
increases
by
nearly
0.3
petagrams
of
carbon
per
year
in
1992.
I
So,
first,
looking
at
the
physical
response
just
to
understand
a
little
bit
of
that
on
the
top
plot,
I'm
showing
a
sea
surface
temperature
for
the
29
ensemble
members
of
both
cesm
in
blue
and
the
cesm
no
pinatubo
in
red
and
the
ensemble
means
are
the
thicker
lines
and
the
time
series
below
shows
the
difference
between
that.
So
again,
cesm
minus
the
no
pinituba,
and
so
what
we
see
is
that
this
model
shows
a
forced,
global,
sst
decline
of
nearly
2.2
degrees
celsius
immediately
after
the
eruption.
I
You
can
see
it
right
here
and
then
this
anomaly
here
and
that
the
individual
ensemble
members
range
in
magnitude
of
cooling,
of
course,
but
the
one
sigma
spread
is
about
0.08
degrees
celsius.
So
the
cooling
persists
that
signal
of
cooling
persists
and
the
ensembles
are
statistically
significant
out
to
1996..
That's
that
thicker
line
here
and
I'm
not
including
maps
here,
but
I
wanted
to
mention
that
we
see
that
the
cooling
induced
by
pinatulo
is
highly
structured
in
space
and
time
and
the
indian
and
pacific
ocean
basins
are
playing
the
key
roles.
I
Looking
below
the
surface
now,
this
is
a
cooling
signal
spreading
deeper
into
depths
with
time
from
this
half
molar
of
the
top
one
kilometer
of
the
ocean
and
again
cesm
minus
no
pinatubo.
So
the
blue
is
showing
a
cooling
signal.
I
Stippling
covers
the
areas
where
there
is
no
significant
difference,
so
at
95,
confidence,
bounce
and
blue
indicates
that
cooling
caused
by
the
eruption
of
pinatubo,
and
we
see
it
significantly
in
that
top
100
meters.
But
it
does
persist
for
about
five
years
in
that
surface
layer,
but
also
a
significant
cooling,
albeit
much
smaller
in
magnitude.
Precipitate
persists
at
depth
all
the
way
through
2004
and
beyond.
Actually
again,
we
go
out
to
2025,
but
I'm
limiting
the
time
shown
here.
Just
for
sake
of
highlighting
the
most
interesting
parts.
I
We
see
that
after
1996
the
surface
kind
of
returns
back
to
indistinguishable
from
the
no
penitubal
and
pinatubo.
You
know
they're
oscillating
here,
and
that
signal
is
different.
I
If
we
look
at
depth
so
now
we're
going
to
extend
our
view
down
to
deeper
depths
and
I'm
showing
the
ocean
heat
content
for
the
ensembles
inventory
down
to
1000
meters,
again
cesm
in
blue
and
no
penitubal
and
red,
and
you
can
see
that
the
ocean
heat
content
is
significantly
reduced
due
to
the
eruption
in
cesm,
there's
a
large
drop
in
blind
ensemble
here
and
unlike
sst,
which
we
shot,
we
saw
in
the
last
slide.
The
ocean
heat
content
does
not
rebound
to
no
pinatubo
levels.
I
There
is
a
consistent
difference
between
these
two
lines,
all
the
way
through
2025.
In
fact,
so
the
external
forcing
of
this
pinatubo
eruption,
is
having
a
lasting
impact
on
the
ocean
heat
content
in
this
model.
I
Next
slide,
I
want
to
reiterate
that
there's
so
much
to
look
at
and
we
have
the
full
output
of
this
model
and
we've
just
scratched
the
surface.
So
we
took
a
quick
look
at
amok,
found
significant
anomalies
here,
showing
a
1995
through
1999
mean
amok
anomaly
from
cesm
minus
no
pinatubo.
I
So
this
would
be
a
strengthening
of
amok
circulation
for
this
time
period
in
the
ensemble
with
pinatubo
and
if
you
just
excuse
me,
zoom
in
on
the
45
degrees
north
here
in
these
time
series
we're
showing
both
ensembles
and
then
the
difference
on
the
bottom,
and
this
fourth
signal
is
showing
a
increase
amoc
anomaly
of
about
one
sphere
jump.
But
I
also
plotted
the
three
largest
members
and
three
smallest
members
of
the
29
member
ensemble,
and
you
can
see
that
in
some
members
the
imac
anomaly
is
as
high
as
five
sphere
drops.
I
So
that's
something
to
take
away.
There's
lots
more
to
look
at
in
regard
to
circulation
and
other
physical
variables.
So
I
hope
others
are
able
to
utilize
this
model,
but
jumping
now
to
biogeochemical
variables,
time
series
plots
of
oxygen
and
dic
inventory
on
the
left,
and
then
the
ensemble
mean
differences
for
various
steps
on
the
right,
we'll
go
through
it.
So
the
left
we're
looking
at
the
top
kilometer
reflecting
you
can
see
on
the
top.
I
Roughly
so
looking
at
various
steps
on
the
right,
you'll
see
that
significant
inventory
anomalies
for
both
the
upper
and
deep
ocean
exist
here
for
these
first
few
years,
meaning
an
increased
oxygen
content
with
the
eruption
in
the
upper
ocean
through
1998
and
impacts
to
the
oxygen
in
the
full
depth
inventory.
This
darkest
orangish
red
color
extend
for
the
entire
35
years
of
this
experiment,
so
significant
increase
in
oxygen
in
the
ocean
due
to
the
eruption
of
pinatubo.
I
Looking
at
carbon.
Now
the
gulp
of
carbon
taken
up
by
the
eruption
or
after
the
eruption
amplifies
the
increase
in
carbon
trend.
So
carbon
in
the
ocean
is
increasing
trend
as
opposed
to
oxygen.
I
So,
in
contrast
to
the
oxygen
inventories
which
increase
with
the
depth
of
integration
shown
in
those
orange
lines,
the
carbon
anomalies
are
consistent
across
the
depth,
which
indicates
that
the
largest
anomalies
are
concentrated
in
those
upper
layers
of
and
they
dominate
over
smaller
differences
at
depth.
You're
at
10
minutes.
I
Excellent
just
want
to
touch
on
some
other
work,
that's
being
done
on
this
project,
so
so
this
also,
these
runs
are
being
utilized
by
holly
oliveira's
in
the
levandusky
group
at
cu
boulder,
and
the
question
they're
interested
in
is
what
is
the
impact
of
pinatubo
in
a
transect
framework.
So,
specifically,
the
bulk
of
state
of
the
ocean.
I
Observations
were
collected
in
the
years
following
the
eruption
with
the
will
stragos
effort,
and
so
questions
like
were
these
baseline
observations
impacted
by
the
eruption
that
coincided
with
this
effort
and
how
could
that
potentially
change
the
trends
and
changes
in
ocean
climate
that
were
interpreting
from
these
observations
collected
in
the
early
90s?
I
This
is
a
quick
slide
of
holly's
work,
which
I
won't
jump
into,
but
she's
creating
a
virtual
transect
around
the
ocean
and
can
plot
that
transect
both
for
cesm
and
no
pinatubo,
and
look
at
differences
and
they're
finding
some
really
cool
stuff.
Looking
at
variability
they'll
also
touch
on
cfcs
and
biogeochemistry
trends.
I'm
excited
to
see
how
that
work
develops
so
just
to
wrap
up.
We
have
29
ensemble
members
of
cesm
run
with
a
single
change
to
the
forcing
to
simulate
a
no
pinatubo
world.
I
B
B
I
We
haven't,
I
mean
specifically
separated
them,
but
you
can
see
that
it's
not
all
driven
by
solubility
the
surface
temperature
doesn't
account
for
all
of
that
change,
both
in
time
and
space.
So
definitely
some
combination
of
the
two.
K
C
B
B
F
Thank
you.
So
I
actually
I
changed
a
little
bit
of
my
of
my
topics.
My
name
is
elio
and
I'm
a
phd
candidate
in
university
of
california
irvine
and
today
I'm
going
to
talk
about
modulating
the
ocean
carbon
sink
by
multiple
ocean
processes
under
climate
vermill.
So
this
study
is
a
little
based
on
a
classic
ceism
studies.
We
based
on
c
ism
circulation-
and
we
combine
this
with
an
offline
bgc
model
to
study
the
effect
of
ocean
processes
on
global
carbon
uptake.
F
So
this
figure
is
a
schematic
of
carbon
uptake
and
carbon
transport
in
the
ocean
by
ocean
carbon
value
chemistry
workshop,
and
this
figure
shows
that
ocean
plays
an
important
role
to
take
up
anthropogenic
carbon
from
atmosphere
and
the
ocean.
The
total
ocean
carbon
have
two
components:
one
is
a
natural
carbon
and
one
is
isoprogenic
carbon.
F
So,
as
we
can
see
from
this
from
this
figure
like
the
biological
pump,
which
transfer
carbon
from
surface
ocean
to
deep
ocean
could
affect
air
co2
flux
by
modulating
the
surface,
dic
and
also
the
circulation.
Here
we
marked
as
overturning
circulation.
The
circulation
can
also
distribute
carbon,
distribute
carbon
in
the
ocean
and
also
can
affect
air
cco2
flags
and
also
the
sea.
F
F
So
here
is
our
for
our
methodology.
We
first
based
on
the
first
version
of
cesm
1.0
to
integrate
integrate
from
play
industrial
to
your
200,
the
business
as
usual
warming
scenario.
Rcp
is
ap
8.5
with
the
resolution
is
around
1
degree
horizontally
and
10
to
250
250
meter
vertically,
so
based
on
the
based
on
the
cism
outputs.
Here
we
show
the
cism
global
carbon
annual
global
carbon
uptake,
so
we
can
see
the
ocean
carbon
uptake
peaks
at
around
5
gram,
5
gram
carbon
per
year
at
20
80s
and
then
decreases
afterwards.
F
But
to
answer
this
question,
it
is
hard
to
distinguish
different
processes
only
based
on
the
coupled
climate
model.
So
what
we
are
trying
to
do
is
to
retrieve
the
cesm
circulation
to
build
up
transport
operator
and
then
combined
this
with
an
offline
bgc
model
so
that
we
can
easily
set.
We
can
easily
separate
the
row
of
different
ocean
process
to
the
carbon
uptake.
So
here
is
our
equation.
The
first
c
here
is
the
ocean
ocean.
F
It
sorry,
it's
a
dissolving
organic
carbon
in
the
ocean
and
tr
here
is
like
why
I
just
mentioned
the
transport
operator
from
csm1
outputs,
which
includes
information
from
both
advection
and
diffusion,
because
we
run
we
have
the
model
outputs
for
450
years
from
pre-industrial
to
2300,
and
we
have
the
annual
outputs.
So
we
have
450
annual
transport
operator
to
contribute
to
the
circulation
part
in
the
model,
and
then
we
have
the
biology
contribution
in
the
model.
F
The
biology
contribution
is
solved
from
a
linear,
mineralized
bgc
model,
which
is
not
shown
here
and
also
the
other
term,
is
the
air
co2
flux
which
dominate,
oh,
which
is
the
hop
which
is
how
much
carbon
that
ocean
can
take
up
from
atmosphere,
and
the
second
equation
here
is
the
alkalinity
equation.
The
alkalinity
is
also
needed
to
solve
dynamically,
because
augmented
is
plays
very
important
role
in
controlling
air
co2
flux.
F
So
here
is
our
new
model
and
we
try
to
set
up
a
fuel
sensitivity
tests
to
to
understand
the
contribution
from
each
ocean
process.
So
here
is
our
sensitivity:
experiments
the
first
one
is
the
control
experiment.
That
means
we
fixed
atmospheric
pco2.
F
We
fixed
biology
effects,
circulation
and
sea
surface
temperature
at
pre-industrial
level,
and
I
keep
this
as
a
control
experiment
and
for
the
other
experiments
we
try
to
include
different
terms
which
changes
with
climate.
So
the
second
experiment
I
call
is
circulation,
which
means
I
only
allow
circulation
changing
with
changing
waste
climate.
F
That
means
like
the
difference
between
the
second
experiment,
with
the
first
experiment,
could
show
the
circulation
change
the
impacts
of
changing
circulation
of
natural
carbon
uptake,
because
we
do
not
include
changing
a
riding
atmospheric
pco2
and
the
third
experiment
here
is:
we
have
writing
atmospheric
pco2
at
rcp,
ecp,
8.5
warming
scenarios
and
then
we
keep
increasing.
We
keep
include
a
biology
term,
allow
biology
primary
production
and
export
production
changing
with
changing
waste
time,
and
then
we
have
another
experiment.
F
So
the
difference
between
the
third
one
and
the
first
one
can
give
us
like
the
contribution
from
atmospheric
co2
increase,
and
then
we,
the
difference
between
the
or
the
fourth
one
and
the
third
one
gives
us
the
biology
contribution
and
also
we
can
separate
the
contribution
of
circulation
on
both
natural
carbon
and
anthropogenic
carbon,
and
also
we
can
know
the
sea
surface
temperature
changes
on
the
global
climatic.
F
So
before
I
show
the
contribution
of
each
of
each
ocean
process,
I
would
like
to
show
a
global
pattern
of
air
co2
flags
at
pre-natural
level
or
at
the
end
of
21st
century
and
at
the
end
of
23rd
century.
So
we
can
see
the
positive.
The
positive
here
means
the
ocean
takes
up
co2
from
atmosphere
and
active
means
ocean
releases
co2
from
atmosph
to
atmosphere.
F
So
at
pre-natural
level
we
can
see
the
natural
carbon
out.
Geysing
high
plants
are
hypers
either
out
hypothesize
the
up
fighting
region
and
the
yin
guy
thing
happens
either
down
by
the
region
and
the
air
co2
flux
keeps
increasing
to
the
end
of
21st
century
and
but,
as
time
goes
by,
we
can
see
the
air
co2
fluxes
decreases
keeps
decreasing
after
2100.
F
F
And
now
I
will
show
you
like
the
air
co2
flux
in
different
in
different
sensitivity,
experiments.
So
the
right
line
here
is
the
ocean
co2
uptake,
as
we
suppose,
the
circulation
and
biology
agreement
isoprene
natural
levels.
That
means
we
own
the
right
land
means
the
air
co2
flux,
which
will
only
allow
atmospheric
pco2
increases
with
time.
But
we
keep
all
the
other
ocean
processes
constant
at
pre-natural
level,
and
the
blue
line
here
is
the
is
we
still
allow
the
atmospheric
pco2
increasing,
but
we
also
allow
the
changing
circulation
into
the
model.
F
That
means
like,
after
the
atmospheric
co2,
doesn't
keep
increasing,
that
fast
right,
10
minutes,
okay
and
and
then
the
pink
one
show.
The
pink
white
includes
the
changing
of
biology
on
the
base
based
on
the
based
on
the
blue
line.
So
we
can
see
that
the
decrease,
export
production
or
export
primary
production
will
a
little
bit
reduce
the
carbon
uptake
from
the
atmosphere
and
the
black
line
is
a
full
time
material.
So
the
difference
between
the
pink
one
and
the
black
line,
showing
the
increased
sea
surface
temperature
reduce
the
carbon
uptake.
F
Then
I
try
to
separate
the
contribution
from
changing
circulation
to
ocean
carbon
uptake,
so
the
blightline,
the
black
line
here,
shows
the
shows
that
reduce
carbon
uptake
by
the
changing
circulation,
and
I
separate
it
into
the
right
line,
which
is
only
the
carbon
of
the
nitro
carbon
cycle
and
the
blue
line,
which
is
the
anthropogenic
carbon
uptake.
So
we
can
see
that
the
right
line
actually
is
a
positive
number.
That
means,
even
though
we
have
slowing
over
time
circulation
under
climate
forming,
it
promotes
ocean
to
sequester
more
carbon
in
the
deep
ocean.
F
So
we
can
see
like
from
the
from
the
panel
here
there.
The
slowing,
overtaking
circulation
supplies
turn
more
nitro
carbon
in
the
d-position
which
increase
the
deposition
eic,
but
the
slowing
overton
circulation
most
prevent
anthropogenic
carbon
transporting
into
deep
ocean.
That
means
that
the
carbon
the
surface
ocean
could
be
more
easily
to
get
saturated
and
prevent
to
or
prevent
from
absorbing
more
carbon
from
atmosphere.
F
Yeah,
so
that's
almost
everything
I
have
and
there
are
some
take-home
messages
and
yeah,
so
I
just
want
to.
I
just
want
to
mention
that
the
overtime,
the
slowing
of
over
10
circulation
is
very
important
in
modulating
carbon
carbon
uptake,
which
is
non-negligible
yo
yeah.
That's
everything.
Thank
you.
B
Great,
thank
you.
Let's
see
if
you
wanted,
yasir
wants
to
go
ahead
and
pull
up
slides.
We
could
take
one
quick
question
in
the
interim.
Otherwise
we
can
have
questions
in
the
chat.
B
Not
seeing
any
hands
up
so
let's
go
ahead
and
move
on
to
our
next
talk
and
if
you
have
questions
for
you,
please
put
them
in
the
chat
and
we
can
have
a
discussion
there.
So
go
ahead.
K
Hi
everyone
yeah,
so
this
is
essentially
work
done
with
the
new
eddy,
resolving
capabilities
of
csm
with
ocean
by
geochemistry
turned
on
and
the
main
question
that
I'll
be
exploring
today
is
what
controls
the
oxygen
budget
in
the
upper
control,
pacific
and
two
key
points
is
that
eddies
are
important.
Gerbil
and
mixing
is
also
important
in
this
region.
K
So
this
this
work
is
motivated
by
the
fact
that
the
global
oxygen
inventory
has
been
declining
in
the
last
few
decades.
We
don't
fully
understand
why
we
know
that
ocean
warming
is
expected
to
reduce
the
oxygen
content
due
to
gas,
solubility
dependence
on
temperature
and
the
dependence
of
stratification
and
ventilation
on
temperature.
K
But
in
the
equatorial
pacific,
where
most
of
the
oxygen
or
substantial
portion
of
the
oxygen
loss
is
happening,
we
don't
fully
understand
the
dynamics
there
there
well,
especially
since
the
models,
don't
really
reproduce
that
oxygen
loss
in
the
equatorial
pacific,
and
we
know
that
the
models,
including
csm,
do
a
pretty
bad
job
with
representing
oxygen
in
this
region,
as
shown
by
the
lack
of
agreements
or
a
lack
of
stippling
here
in
the
models
to
the
right
that
show
that
different
models
do
different
things
depending
on
their
biases
and
the
processes
they
might
be
resolving.
K
So
in
in
this
work,
we
are
considering
something
more
fundamental,
which
is
really
what
physical
processes
balance
oxygen
consumption
in
the
upper
control
pacific,
because
we
don't
have
enough
observations
to
constrain
the
oxygen
budget.
K
So
we
use
the
the
csm
at
one
tenth
of
a
degree
coupled
to
pec
for
biogeochemistry,
and
we
started
with
a
five
year
annual
cycle
core
four
stimulation.
This
is
spun
up
for
15
years
for
physics
and
only
a
year
for
bgc.
So
it's
not
ideal
for
understanding
processes
like
what's
driving
the
biases,
but
the
physics
is
operating
on
bgc
that
resembles
observations.
K
So
the
the
main,
the
main
region
that
we're
looking
at
is
the
eastern
and
central
equatorial
pacific.
So
the
the
top
left
panel
shows
the
mean
oxygen
distributions
of
the
26.2
iso
pigments,
and
you
could
see
the
omz
extending
all
the
way
to
the
western
pacific
in
csm
and
a
ton
of
oxygenated
waters
going
all
the
way
to
the
eastern
pacific,
as
as
far
as
we
we
know,
is
mostly
driven
by
the
transport.
K
Effective
transport
of
oxygen
to
the
eastern
pacific
by
the
equatorial
current
system
and
the
panel
to
the
bottom
left
shows
a
section
along
125
west.
That
shows
the
signature
of
some
of
these
attractive
pathways.
K
However,
we
we're
looking
here
at
the
potential
contribution
of
eddie's
and
vertical
mixing,
given
the
fact
that
these
two
processes
have
been
pretty
well
documented
in
models
and
observations
that
are
having
big
impacts
on
the
heat
budget,
but
the
mixed
layer
and
the
upper
thermocline,
and
essentially,
how
do
these
different
processes
balance
the
consumption
of
oxygen
in
the
upper
thermocline
in
this
region.
K
So
here's
some
results,
the
well.
What
I'm
showing
here
is
essentially
a
section
between
100
west
and
160
west,
from
0
to
500
depth
and
in
the
x-axis.
I
show
latitude
so
essentially
averaging
over
the
eastern
and
central
equatorial
pacific
and
looking
at
the
contribution
of
these
different
processes
mean
adduction
by
the
currents
and
upwelling
eddy
advection,
vertical
mixing
or
the
oxygen
flux
due
to
vertical
mixing
and
sources
and
sinks
of
oxygen,
essentially
microbial
consumption
of
depth
and
production
by
plankton
near
the
surface.
K
And
what
you'll
note
is
that
the
consumption
of
oxygen
in
the
upper
thermocline,
essentially
in
the
50
to
about
300
meter
depth,
the
the
decrease
in
oxygen
due
to
these
microbial
respiration,
is
essentially
balanced
by
a
combination
of
mean
advection
and
induction
vertical
mixing.
So
this
is
quite
different
from
the
idea
that
the
mean
of
action
by
the
current
system
is
balancing
consumption
across
the
upper
ocean,
so
mixing
and
eddy
effection
will
likely
play
some
key
roles,
at
least
based
on
these
simulation
and
a
resolving
scales.
K
So
I'm
gonna
go
through
each
one
of
these
and
get
an
idea
of
some
of
the
processes
behind
them.
We,
this
is
essentially
a
a
section
along
125
west
in
the
bottom
that
shows
these
transport.
It's
fair
drop
based
on
adcp
data
from
johnson
and
stroma,
and
others
that
shows
that
the
euc,
the
north
subsurface
counter
current
the
control
under
currents.
K
K
This
is
great,
but
what's
interesting
is
that
eddy
infection
plays
a
really
key
role
here,
especially
north
of
the
equator
and
all
the
way
down
to
200,
meter
depth
and
that's
associated
with
the
propagation
of
tropical
and
stability
vertices
in
the
equatorial
pacific,
especially
north
of
zero
degrees
along
the
equator,
and
it
reflects
two
processes
based
on
some
lagrangian
particle
simulations
that
we've
done
through
this
work,
eddie
steering,
which
essentially
leads
to
a
northward
or
polar
infection
of
particles,
as
these
vortices
are
passing
as
well
as
subduction
subduction
down
to
depth
of
about
50
to
150
meter
depth
along
the
leading
edge
of
these
vortices.
K
So
here
to
the
panels,
to
the
left,
show
lagrangian
particles,
particles
being
fed
to
a
passing
vortex,
and
we
see
after
15
days
30
days.
Some
of
these
particles
gain
depth
of
about
50
to
100
meters.
As
these
as
these
vortices
are
propagating
towards
the
west
and
essentially
subduct
oxygen
along
sloping,
iso
pigments
along
that
leading
edge,
the
eddy
steering
component
is
essentially.
K
Through
the
anti-cyclonic
flow
of
the
eddies
redistributing
particles
from
from
about
zero
to
two
degrees
north,
where
the
euc
passes
through
all
the
way
up
to
about
five
to
six
degrees
north,
so
we
could
see
that
oxygenated
waters
from
the
equatorial
under
currents
can
be
taken
word
and
oxygenates.
The
upper
thermocline
along
the
path
and
then
finally,
vertical
mixing
is
a
critical
component
of
oxygenating.
The
upper
thermocline
here-
and
this
is
also
tightly
associated
with
tropical
stability.
Vortices.
K
We
see
that
most
of
the
oxygen
flux
due
to
vertical
mixing
is
is
associated
with
these
cold
phases
of
the
tropical
stability
waves
as
they're
passing,
and
that's
that
leads
to
an
air
c
flux
of
oxygen,
a
fairly
intense
parenthesis,
flux
of
oxygen,
essentially,
a
local
ventilation
of
oxygen,
as
these
eddies
are
passing
and
we
can
see
their
signature
also
in
surface
apparent
oxygen
utilization.
So
we
essentially
see
that
the
thermocline
is
is
being
ex
thermocline.
A
You've
got
two
more
minutes,
including
your
q,
a.
K
Okay,
I'll
wrap
it
up
then
so.
Finally,
I
just
want
to
touch
on
how
these
different
processes
might
be
impacting
the
seasonal
variability
and
maybe
international
variability,
and
we
find
that
vertical
mixing
is
is
quite
an
important
component
of
the
variability,
and
we
see
that
as
steady,
kinetic
energy
increases
from
spring
to
fall.
Essentially,
the
the
flux
of
oxygen
do
vertical.
Mixing
is
also
intensified
and
that's
more
or
less
in
line
with
what
we
see
with
heat
budget
in
the
equatorial
pacific.
K
So
I've
got
some
key
points
here
if
you
want
to
take
a
screenshot
for
a
summary,
but
I'd
like
to
leave
it
with
some
questions
for
future
work,
and
this
is
tightly
related.
The
the
dynamics
associated
with
tropical
stability
vertices
include
a
lot
of
sub
mix,
submissive
scale
processes,
so
it'd
be
interesting
to
see
how
submissive
scale
features
like
fronts
along
the
leading
edge,
facilitate
the
mixing,
the
subduction
of
oxygen
and
other
transport
processes.
K
So
with
that,
I've
got
some
papers
there
and
and
thank
the
co-workers.
B
Thank
you.
We
can
take
a
question
and
just
to
give
everyone
a
heads
up.
B
The
next
thing
coming
on
our
agenda
is
a
break
and
we're
going
to
reconvene
for
those
talks
at
the
half
hour
of
whatever
time
zone
you're
in
so
if
you
need
to
take
that
break
now,
but
we
can
definitely
take
a
question-
and
I
just
wanted
to
thank
all
of
the
speakers
from
this
session
for
giving
us
a
really
nice
broad,
look
at
all
of
the
many
different
aspects
of
or
some
survey
of,
the
aspects
important
in
the
ocean
from
the
physical
mixing
to
the
assumptions
about
biology
to
the
interactions
with
events
in
the
atmosphere.
B
That's
really
great
to
see
the
the
breadth
of
what's
included
in
this
in
this
working
group.
So
if
we
have
questions
for
yester,
we
can
take
those
now
you
can
type
in
the
chat
or
raise
a
hand.
C
A
How
is
your
mean
operator
just
defined?
Does
the
eddy
term
include
the
seasonal.
K
Yeah
so
let's
see.
G
K
See
this
okay,
frank
and
everyone?
Sorry,
yes,.
M
K
A
little
bit
of
math,
the
eddy
component
is
essentially
calculated
from
the
climatological
monthly
mean,
so
I
have
not
been
able
to
remove
the
rectified
effects
with
the
seasonal
cycle,
especially
that
these
tropical
stability
waves
are
strongly
seasonal,
yeah
and
especially
after
talking
to
analina
and
many
others.
We've
tried
this
in
the
past,
it's
much
more
complicated
than
that.
K
So
so
far,
we've
just
I've
been
following
the
grief
ease
at
all
method,
where
we
just
take
the
anomaly
from
the
climatological
monthly
mean,
I
think
I
it
would
be
great
to
chat
some
more,
especially
with
you
and
anna
on
on
how
we
can
develop
a
method
for
taking
the
seasonal
seasonality
of
the
eddy
component
into
into
account
yeah.
Hopefully,
that
answered
your
your
question.
K
B
Sure
great
and
I
think
with
that
we
should
give
everyone
a
chance
to
have
a
break
we're
going
to
have
some
breakout
rooms
right
now.
So
if
you
want
to
try
to
catch
one
of
the
speakers,
you
can
look
for
which
breakout
room
they
might
be
in
and
you
can
go
ahead
and
chat
with
folks.
There,
we'll
reconvene
for
our
next
set
of
talks
at
the
half
hour,
so
see
everyone
in
a
few
minutes.
A
C
A
We'll
do
the
the
time
similar
where
abby
will
let
you
know
when
you
are
at
ten
minutes
and
then
you've
got
two
minutes
left
for
either
q
a
or
wrapping
up.
G
A
All
right
so
we'll
be
getting
started
in
a
minute.
We've
got
four
talks
that
are
more
focused
on
land
biogeochemistry,
but
obviously
ties
to
the
coupled
system
after
that,
after
those
four
talks,
we'll
have
some
time
at
the
end
for
sort
of
a
wrap-up
discussion.
So
it's
another
opportunity
for
folks
to
ask
questions
to
any
of
the
speakers
or
to
follow
up
on
any
of
the
discussion
that
you
had
in
your
small
groups
or
in
the
breakouts
so
we'll
have.
A
The
ideal
is
like
ten
and
two
talks,
ten
minutes
of
presentation,
two
minutes
of
question,
but
everyone
please
feel
free
to
use
the
chat
during
and
after
the
presentations,
so
we'll
kick
off
with
songweng,
we'll
be
talking
about
re-parameterization
as
models
move
from
carbon
only
to
carbon
nitrogen
models.
So
thanks
son.
G
Hello,
thank
you,
our
president,
in
place
of
professor
law.
Today,
my
name
is
seongwang
a
phd
student
in
chinese
academy
of
sciences.
G
Thanks
for
giving
me
the
opportunity
to
share
a
study
with
you,
a
research
about
reprimanization
required
after
model
structure
change
from
capone
to
carbon,
nitrogen
company
terrestrial
exist
ecosystems,
construct,
broadcast
measuring
thirty
percent
of
human
emitted
carbon
and
play
a
critical
role
in
mitigating
climate
change,
and
the
terrestrial
carbon
cycle
is
strongly
regulated
by
nitrogen
availability
in
ecosystem.
Nitrogen
can
regulate
the
carbon
size
remaining
by
these
three
weeks.
The
formation
or
gain
matter
requires
a
certain
amount
of
latitude
coming
with
processes
of
plant.
G
Hence,
there
is
a
growing
awareness
that
the
terrestrial
natural
cycle
is
crucial
to
accurately
predict
the
carbon
cycle
by
resistance
models
and
the
different
climate
change
scenarios
and,
more
and
more
and
say
some
models
account
for
nitrogen
dynamic.
Here
is
the
models
in
siem
6.
We
can
see
that
more
than
half
of
the
system
models
have
incorporated
the
nitrogen
cycle.
G
As
a
result,
in
almost
all
ssr
models,
considering
latching
limitation
will
reduce
the
gpp
and
its
carbon
storage
without
latching
company
you
could
exceed
some
carbon
carbonyl
model
would
overestimate
the
existing
carbon
impulse
in
the
sequential,
because
the
potential
latching
limitation
is
not
considered.
It
sounds
reasonable
right,
however,
the
the
common
stories
in
the
real-world
existence.
They
are
not
changed
according
to
whether
or
not
a
model
considers
nitrogen
processes.
The
the
common
electric
interactions
always
exist
in
the
real
world.
G
G
We
can
observe
some
characteristics
with
many
methods
like
a
failed
experiment,
incubation
effects,
observations
and
acetone
logging,
and
then
we
can
transfer
this
information
to
model
for
model
construction,
then,
in
in
return,
the
model
is
used
to
describe
the
function
of
the
system.
If
the
model
obtains
the
information
of
the
existing
by
parameter
after
data
simulation,
the
model
can
will
simulate
and
predict
the
function
of
the
corresponding
system
and
the
parameter
in
the
model
will
contain
the
information
from
of
the
x-system.
G
B
G
B
G
Contained
in
the
observations
is
emphatically
represented
in
estimated
parameters.
However,
when
a
carbon
and
larger
recovery
model
is
calibrated
using
the
same
observations,
the
nitrogen
related
information
is
is
no
longer
emphatically
represented
in
the
carbon
relative
parameter.
It's
what
actually,
it
was
reflected
on
the
carbon
and
larger
interactions
in
the
model.
G
In
order
to
answer
these
questions,
we
conduct
a
research
using
data
simulation
to
estimate,
parameters
of
a
carbon
only
model
and
a
carbon
lateral
company
model
based
on
a
failed
experiment.
In
qinghai
tibet
plateau
the
estimated
parameters
were
used
in
their
respective
model
to
simulate
carbon
and
nitrogen
dynamics
during
the
experimental
period.
G
Here
are
the
main
results
of
the
primarization
by
data
simulations
when
found
that
adding
a
natural
module
increase,
the
carbon
exists
rate
or
most
carbon
poor.
It's
a
little
different
from
what
we
know
before.
Usually,
we
think
that
natural
limitation
will
decrease
exactly
so.
Why
why
we
get
the
upside
results?
G
I
want
to
use
a
configuration
to
explain
this
result
in
the
traditional
accessor
models.
We
think
that
the
national
limitation
information
degree
in
the
carbon
and
large
economy
model
is
larger
than
the
carbon
only
model.
I
show
this
relationship
by
the
balance
tilted
towards
the
carbon
and
electric
recovery
model.
G
So
what?
What
is
the
real
result?
In
the
experimental
period
to
model
both
wheels?
We
can
see
that
in
the
experimental
period,
two
models
both
will
estimate
the
most
exist
and
poor
in
comparison
with
observations
in
ambient
and
vomit
treatment.
The
gene
line
is
a
pro
size
simulated
by
the
cover
only
model
and
a
green
line
is
the
process
simulated
by
the
company
model.
G
G
The
constraint
parameters
are
different,
but
if
we
use
the
corresponding
parameters
models,
the
simulation
results
are
similar,
but
if
we
use
the
parameter
without
retuned
by
digital
simulation,
when
we
incorporate
the
nitrogen
cycle
to
a
model,
the
ecosystem,
carbon
storage-
will
be
underestimated,
and
here
are
some
painful
messages.
Based
on
this
result,
we
get
a
concept
figures.
The
x,
I
said,
is
the
latching
limitation
degree
of
the
observations
and
the
website
is
you're.
G
Okay,
the
the
blue
line
is
the
simulation
of
the
carbon
only
and
carbology
coupling
after
data
simulation,
and
the
orange
line
is
the
simulation
of
company
model
without
reprimanization.
If
there
is
no
national
limitation,
okay
or
stimulation
are
same,
but
if
there
is
nitrogen
limitation.
G
In
the
observations
they
exist
from
carbon
storage
simulated
by
the
company
model
without
reprimandation,
we
are
much
lower
than
the
model
with
data
simulation.
Hence
the
assistant
models
may
underestimate
both
future
terrestrial
carbon
filtration
and
the
potential
carbon
climate
feedback,
because
the
then
overcomes
the
nitrogen
constraint
and
we
highlight
the
necessity
or
to
replace
a
parallelization
after
motor
structure
change
from
the
carbon
only
to
a
carbon
and
latching
coupling.
G
However,
the
carbon
od
model
can
not
capture
the
change
of
carbon
or
nitrogen
limitation
information,
because
the
latching
process
are
not
explicitly
simulated
by
the
camera
only
mode,
but
the
company
model
may
be
better.
Yes,
when
affirmed
the
significance
of
the
carbon
latching
company
model.
B
Great,
thank
you
so
much.
We
could
take
one
quick
question
or
go
ahead
and
type
questions
in
the
chat
as
well.
G
A
J
Hello
thanks
for
having
me
today,
it's
a
presentation
about
these
proposed
changes,
we're
looking
for
feedback
with,
but
but
before
I
get
into
that.
I
just
wanted
to
talk
just
for
a
second
about
what
islam
is.
I
realize
that
a
lot
of
new
people
have
come
to
this
meeting
and
we
used
to
talk
about
this
a
great
deal,
and
I
lam
also
is
one
of
those
acronyms
in
the
doe
system
that
becomes
kind
of
overloaded,
and
you
hear
it
to
mean
a
lot
of
things.
So
you
know
people
could
be
talking
about.
J
It
stands
for
international
land
model
benchmarking
and
they
use
they
could
be
talking
about.
The
community
group
of
scientists
enthusiastic
about
model
benchmarking,
of
which
you
can
be
a
part
if
you
would
like
it
could
be
talking
about
the
results
that
we
catalog.
I
have
a
bunch
listed
here,
there'll
be
a
link
at
the
end
of
my
presentation.
If
you
want
to
grab
the
slides
and
go
explore
the
links
one
of
those
appeared
in
the
ar6
chapter,
it
was
in
one
of
the
slides
earlier
in
this
session.
J
It
could
be
talking
about
our
software
package.
It's
a
hosted
at
github,
you
can
install
it
with
conda
and
a
new
version
has
been
under
development
for
a
while.
Slowly
by
me,
I'm
using
x-ray
we're
hoping
that
will
kind
of
enable
new
kinds
of
comparisons
things
the
community
has
wanted,
or
you
could
be
talking
about.
Our
data
sets.
J
I'm
aware
that
there's
a
lot
of
people
in
this
community
that
use
these
data
sets
because
they're
formatted
in
a
very
convenient
way
and
a
large
collection
of
quantities
that
are
relevant
for
the
modeling
that
we
do.
In
that
vein,
we
have
a
lot
of
new
additions.
J
There's
a
biomass
data
set
from
from
the
european
space
agency.
That's
in
there
that
I
quercened
out
that's
available
and
some
nasa
funded
weekend
data
that's
available
the
class
data
set.
These
are
unique
collection
of
a
blending
of
a
lot
of
products,
but
they
provide
uncertainty,
estimates
and
they
also
close
the
mass
and
energy
budget
which
is
fairly
unique.
I'm
not
aware
of
anything
else
that
does
that
we
got
teas
biological
nitrogen
fixation
in
there,
as
well
as
a
surface
soil,
moisture
product
from
a
publication
by
wang
and
mao.
J
So
you
can
see
those
results
if
you
go
to
the
cmp6
link
here
and
check
it
out,
and
I
will
say
you
know
that
that
if
you
would
like
to
make
contributions,
a
lot
of
those
data
sets
that
we
added
came
from
suggestions
here,
I
think,
will
at
least
suggested
two
of
them,
and
we
like
suggestions.
You
can
email
them
to
me
or
you
can
go
to
this
github
repository
and
and
make
a
and
open
an
issue
with
the
suggestion.
J
But
you
could
also
be
talking
about
when
you
say
I
am
the
methodology
that
we
have
and
that's
what
this
talk
is
really
meant
to
be
is
is
a
to
point
out
some
consequences
of
a
lot
of
the
decisions
that
we've
made
and
suggest
maybe
a
potential
change
to
to
plug
some
holes
in
that
methodology.
J
So
I've
shown
this
slide
a
bunch
of
times
before,
but
this
is
how
we
score
our
bias.
We
make
essentially
kind
of
three
choices
here
and
the
first
is
to
normalize
the
absolute
value
of
the
bias
by
the
standard
deviation
of
the
reference.
J
That'll
come
back
in
just
a
moment,
and
we
then
map
that
to
the
unit
interval
using
an
exponential,
it's
a
fairly
standard
thing
to
do,
and
then
we
integrate
over
the
globe,
but
we
integrate
by
by
a
weight
and
that
weight
is
the
so-called
mass
of
the
the
quantity
there,
which
is
in
this
case,
is
just
shown
as
like
the
mean
of
the
reference
variable.
So
what
why
are
we
doing
this?
J
So
what
you're
seeing
here
is
a
distribution
of
biases
from
the
cmap
five
and
six
set
of
models
with
respect
to
gpp
from
fluxcom
a
lot
of
what
I'm
going
to
show.
You
is
gpp
from
flexcom
and
I'm
showing
you
these
kind
of
violin
plots
and
splitting
it
all
out
in
different
regions
which
correlate
to
whitaker
biomes
they're
regions
that
we
geographical
regions
we
constructed
from
the
parameter
space
and
backing
out
the
latitudes
and
longitudes.
J
So
if
you
kind
of
can
compare
the
distribution
of
scores
in
these
regions,
you'll
see
that
they're
pretty
evenly
distributed
except
tropical
rainforest,
and
that's
because
the
variability
in
the
rainforest
is
relatively
low.
It's
just
gpp
is
always
on
and
therefore
it
doesn't
and
we
get
this
kind
of
different
effect
here
and
the
scores
from
the
tropics
are
just
poor.
J
The
second
effect
is
this
mass
weighting
that
we
use,
and
so
what
I've
done
here
is,
if
you,
if
you
take
kind
of
the
full
contribution,
if
you
try
to
make
a
plot
of
how
each
pixel
contributes
to
the
overall
score.
So
this
this
plot
down
here
on
the
bottom
left,
is
showing
the
area
of
every
cell
and
its
gpp,
divided
through
by
the
total
sum
and
what
it
gives
you.
J
It
is
true
that
a
lot
of
tropics
scores
are
very
similar
to
our
global
scores,
implying
that
this
is
a
problem
beyond
just
gpp.
So
what's
the
proposed
change?
Well,
what
we're
trying
to
do
by
normalizing
these
errors
is
we're
trying
to
make
the
air
somehow
commensurate
globally.
But,
of
course,
like
we're
picking,
you
know,
the
globe
is
a
very
different
place
as
you
as
you
go
across
biome
across
latitude.
J
There's
a
lot
of
changes,
so
we're
going
to
try
suggest
is
to
capture
those
changes
by
instead
looking
to
score
in
the
set
of
biome-like
regions.
Here,
I'm
using
these
whitaker
regions,
you
could
pick
copenhagen
regions.
You
could
pick
anything,
but
the
main
idea
that
I'm
putting
forward
here
is.
We
want
to
pick
regions
inside
which
the
the
errors
that
you
can
see
are
commensurate
in
order
of
magnitude,
and
so
because
these
regions
are
reflect
biomes,
where
the
climate
is
similar
and
the
vegetation
perhaps
is
similar.
J
Then
we
we're
we're
going
to
say:
okay
inside
this
region,
these
errors
can
be
treated
as
commensurate,
and
so
then,
what
I
do
is
for
every
region
and
every
variable
and
across
a
selection
of
models,
in
this
case,
the
five
five
versus
six
subset,
I'm
going
to
compute
the
98th
percentile,
the
absolute
value
of
the
bias,
you
could
say
the
worst
value
of
the
bias
98
percentile
is
just
meant
to
kind
of
make
sure
we
skip
over
any
kind
of
anomaly
or
something
essentially,
what
I'm
asking
is
how
bad
have
the
errors
been?
J
So
how
does
it
look
again
for
gpp
and
fluxcom
the
98th
percentile
of
biases,
I'm
showing
in
the
top
left
panel,
and
it
it
does
what
you
would
expect
large
biases
in
the
tropics
that
taper
off
as
you
go
to
higher
latitudes
and
then
on?
The
top
right
figure
is
csm
twos
by
gpp,
and
you
can
see
our
old
bias
score
map
down
here
on
the
bottom
left,
and
maybe
people
haven't
stared
at
those
very
much.
J
But
you
know
for
like
dave
lawrence
and
many
of
the
clm
group
that
have
you
you'll
notice
that
a
lot
of
times
those
maps
aren't
terribly
helpful
they're.
Just
they
get
very
poor
scores
in
a
lot
of
regions.
J
You
can
see
why
we
were
mass
weighting
like
these
dry
regions,
in
australia
or
even
in
the
in
the
west
here,
where
maybe
there's
not
so
much
gpp
you're
getting
very
bad
scores
and
regions
that
don't
matter
so
much,
whereas
the
the
the
the
new
way
we're
scoring
the
bias
really
takes
care
of
a
lot
of
that
problem
and
leads
to
a
bias
score,
which
has
much
better
correspondence
with
the
bias.
J
When
I
glance
up
at
the
bias
and
look
at
this
score,
I
feel
there's
a
better
correspondence
and
it
also
highlights,
for
example,
where
the
bias
is
it
is
anonymously
different.
So
if
you
look
at
csm2's
high
latitude,
gpp
bias,
you'll
notice,
there's
this
high
bias
and
up
in
these
high
latitudes,
which
is
anomalous
it's
different
than
csm1
and
with
respect
to
a
lot
of
other
models
too
in
the
collection,
and
so
this
way
of
defining
errors
now
kind
of
highlights
that
it
makes
these
maps
much
more
useful
to
kind
of
tease
out.
J
Okay,
thank
you.
So
that's
our
proposal
and
it
is
it's
kind
of
a
rethinking
of
how
we
could
approach
scoring.
It
removes
this
normalization
by
variability
exponential
mapping
mass
weighting,
a
lot
of
things
which
were
kind
of
necessary
but
subjective
decisions
that
we
made,
and
it
leads
to
a
better
bias
to
score
correspondence
and
give
some
context
to
our
scores.
Because
now,
if
you
got
a
poor
score,
it
means
that
your
score
is
low
relative
to
what
models
have
been
able
to
do.
J
You
know
for
the
last,
whatever
10
years
or
so
so
that,
of
course,
there's
there's
a
lot
of
caveats.
Maybe
we
haven't
thought
through
of
everything,
so
we're
welcoming
some
comments
either
now
or
later,
the
slides
are
there
and
thank
you
for
your
attention.
A
All
right
thanks
if
folks,
have
questions,
please
put
them
in
the
chat
gabe.
If
you
want
to
start
putting
your
slides
up
or
if
anybody
has
a
quick
question,
you
can
feel
free
to
ask
nate
directly
while
gabe
gets
his
slides
up.
A
All
right:
well,
I
know
that
the
island
folks
love
to
get
feedback.
So
if
anyone
has
any
thoughts
about
this,
I'm
sure
nate
would
like
to
hear
them
afterwards,
all
right
so
now
we'll
turn
to
gabe
cooperman
who's
going
to
be
talking
about
competing
influences
on
evapotranspiration
and
how
they
modulate
vegetation
climate
impacts
in
the
cesm
c4
mips
simulations.
H
Thanks
yeah,
I'm
gabe
cooperman
from
the
university
of
georgia,
and
I
want
to
share
you
know-
maybe
some
somewhat
preliminary
results
where
we've
been
revisiting
some
of
the
vegetation
climate
impacts
that
were
identified
in
csm1
and
other
cement,
5
models
using
csm2
and
sort
of
unfolding,
some
of
the
competing
influences
of
plants
on
evapotranspiration,
and
I
thought
that
would
be
useful
to
highlight
to
this
group
as
motivating
how
how
important
plants
are
for
for
the
climate
change
signal
that
we're
seeing
in
fully
coupled
simulations.
H
This
is
work
that
is
being
led
by
my
graduate
students,
ashley
cornish
and
a
lot
of
kordak
who
are
supported
by
a
doe
rgma
pro
project,
that's
led
by
abby,
swann
and
others,
and
so
broadly
we're.
You
know
we're
interested
in
how
co2
affects
the
climate
system,
and
so
we
have
rated
warming
that
comes
with
a
host
of
of
climate
impacts
that
have
consequences
for
ecosystems,
but
ecosystems.
H
Don't
sort
of
respond
passively
to
these
climate
changes
they're
integral
to
the
the
changes
themselves
and
there's
been
work
that
has
showed
how
plants
influence
precipitation
patterns,
flooding,
drought
and
heat
waves
in
response
to
rising
co2
and
largely
that's
through
their
influence
on
evapotranspiration
through
a
couple
of
mechanisms,
one
is
reducing
stomatal
conductance,
so
stomata
don't
open
as
much
under
higher
co2
conditions
and
that
reduces
transpiration
rates.
H
Another
mechanism
is
through
increases
in
leaf
area
and
co2
fertilization
that
affect
plant
growth
and
that
can
affect
the
surface
albedo.
It
can
affect
transpiration.
It
can
affect
how
much
rain
water
is
intercepted
by
the
by
the
canopy
and
then
re-evaporated
into
the
atmosphere
and
in
csm1
this
the
model
conductance
effect
was
was
the
largest
and
so
that
led
to
a
broad
decrease
in
total
vapor
transpiration
due
to
plants
responding
in
this
case
to
four
times
co2,
and
so
that
means
less
moisture
going
to
the
atmosphere
which
can
affect
precipitation.
H
H
That
can
warm
the
atmosphere
and
it
can
also
influence
atmospheric
circulation
in
ways
that
affect
moisture
transport
and
subsequently
precipitation
as
well,
and
so
we
we've
been
separating
these
plant
effects
from
from
radiatively
driven
greenhouse
effects
using
these
idealized
co2
only
simulations
as
part
of
c4mep,
where
you,
independently
control
the
co2
concentration
in
the
atmosphere
and
the
surface
components
of
the
model,
in
this
case
we're
looking
at
clm,
and
so
I'm
going
to
show
you
some
results
from
three
simulations:
a
full
simulation
where
co2
was
increased
from
pre-industrial
to
four
times
co2
in
both
the
atmosphere
and
the
land
surface,
a
simulation
called
radiation
where
we
just
increase
it
in
the
or
it's
just
increasing
in
the
atmosphere
and
a
physiology
stimulation
where
it
only
increases
in
the
land
surface
and
again,
using
these
simulations,
we
can
see
contributions
from
both
radiation
and
physiology
to
a
lot
of
large-scale
climate
changes.
H
This
is
showing
the
pattern
of
precipitation
change
at
the
end
of
these
simulations
relative
to
pre-industrial
and
the
four
times
co2
full
simulation
on
the
left.
That
pattern
looks
really
similar
to
the
rcp
8.5
results
and
you
see
large
increases
over
most
of
the
tropics,
except
for
the
amazon,
and
that
has
contributions
from
both
physiology
and
radiation.
H
In
particular,
south
america
and
the
maritime
continent
have
a
strong
physiology
contribution,
and
if
you
take
these
the
runoff
from
these
models
and
you
downscale
it
for
flood
statistics,
this
is
showing
the
100
year
flood
return
period,
changing
as
a
result
of
co2
increases.
H
You
see
strong
plant
effects
due
to
physiology
and
radiation
as
well,
and
so
we
can
see
you
know
these
blue
colors
in
the
in
the
full
simulation
representing
about
25
or
50
year,
return
periods
from
what
used
to
be
a
100
year,
return
period,
big
contributions
from
physiology
and
radiation.
Some
places
one
dominates
in
some
places
they
amplify
one
another.
H
Likewise,
when
you
look
at
drought,
metrics
that
take
the
surface
conditions
into
account,
like
precipitation,
minus
evaporation,
you
see
a
mitigating
effect
of
plant
responses
relative
to
the
radiative
responses,
kind
of
reducing
drop
stress
and
the
full
simulation
that
you
wouldn't
see.
If
you
looked
at
something
like
the
palmer
drop
index
or
something
that
took
more
atmospheric,
focused
metrics,
and
then
we
also
see
plant
influences
on
on
heatwave
statistics,
so
this
is
looking
at
the
heat
wave,
total
number
of
days
and
intensity.
H
This
is
showing
the
the
ratio
of
the
physiological
driven
changes
to
the
rate
of
driven
changes,
and
so
you
can
see
that
the
positive
numbers
of
the
physiology
is
amplifying
the
radiatively
driven
changes
and
in
some
local
regions
as
much
as
20
to
40
percent,
so
significant
contribution
due
to
physiological
changes.
H
H
So
this
is
looking
at
the
total,
evaporation
change
in
csm1
and
csm2,
and
you
can
see
it's
much
smaller
csm2
and
that
means
the
climate
impacts
are
also
smaller.
So
this
is
the
pattern
of
precipitation
change,
which
is
much
weaker
in
csm2
than
it
was
previously.
H
The
runoff
pattern
is
similar,
but
again
it's
much
weaker
in
csm2,
and
this
has
consequences
for
for
drought
and
flooding
conditions
and
changes
in
temperature
increase
in
both
versions
of
the
model
almost
everywhere,
but
those
those
big
hot
spots
in
south
america
and
north
america
and
some
of
the
high
latitudes
are
much
smaller
in
in
csm2
than
they
were
in
csm1.
H
So
these
are.
These
are
just
climatological
changes.
You
know
the
first
20
years
versus
the
last
20
years
of
these
simulations,
but
you
know,
I
think
they
they
capture
a
lot
of
the
signal
you,
you
would
see
if
you
looked
at
some
of
those
higher
level,
statistics
of
flooding
and
heat
waves,
and
so
we
wanted
to
sort
of
as
an
initial
step
kind
of
parse
out.
H
What's
driving
these
changes
in
evaporation,
so
we
looked
at
the
components
of
evaporation
so
starting
with
transpiration
changes
in
the
bottom
here,
pre-industrial
values
in
the
top
and
then
the
global
mean
changes
on
the
right
and
you
can
see
the
csm1
has
higher
transpiration
rates
in
part
due
to
more
precipitation
in
this
version
of
the
model.
When
run,
it
has
a
coupled
model,
but
the
declines
are
actually
pretty
similar
in
both
model
versions.
In
fact,
in
some
regions
and
in
the
global
mean,
the
csm2
declines
are
larger
than
the
csm1
declines.
H
For
the
second
largest
component
to
evapotranspiration
soil,
evaporation
again,
we
see
csm1
had
a
larger
soil,
evaporation
than
csm2
again
related
to
precipitation
differences,
but
both
models
had
very
little
sensitivity
to
co2
changes
where
they
really
differed.
Was
this
third
component,
the
canopy,
evaporation
change
where
there's
a
significant
increase
in
csm2
and
very
little
change
in
csm1
and
and
so
in
csm2.
H
The
canon
evaporation
is
offsetting
the
declines
in
in
transpiration,
and
so
the
overall
et
changes
is
is
much
smaller
because
of
that
and
this
results
in
part
due
to
a
much
stronger
increase
in
leaf
area
in
response
to
to
increasing
co2.
H
So
again,
the
plot
on
the
right
is:
is
the
global
mean,
and
you
can
see
that
blue
line
for
csm2
is
much
steeper
than
the
green
line
for
csm1
there's
higher
lai
in
the
pre-industrial
climate
in
csm1,
but
by
the
end
of
the
simulation
there's
higher
lai
in
in
csm2,
and
so
that
means
more
rainwater
can
be
collected
onto
the
leaves
and
re-evaporated
into
the
atmosphere
more
more
easily,
and
so
that
contributes
in
the
canopy
of
operation
changes.
H
H
That's,
not
necessarily
true
for
other
variables
of
carbon
uptake.
This
is
the
mvp
change
they're
more
similar
early
on
in
the
simulation,
and
they
have
a
much
stronger
sensitivity
in
csm2
that
leads
to
big
changes
going
forward
and
there's
similar
similar
patterns
for
gpp
and
total
carbon
storage,
where
they're
more
similar
kind
of
at
the
beginning
of
the
simulation
and
diverge.
H
But
the
leaf
area
in
particular,
is
not
sort
of
well
constrained
for
for
the
present
day
range
here.
H
So
I'll,
just
kind
of
conclude,
with
some
some
main
points
and
some
questions
you
know,
so
we
looked
at
these
competing
influences
on
evapotranspiration.
We
found
that
in
csm1
it's
really
just
transpiration,
that's
changing
and
not
canopy
evaporation.
So
we
had
strong
climate
impacts,
but
with
this
cancellation
effect
in
csm2,
the
climate
impacts
were
much
smaller.
H
H
Evaporation
is
very
sensitive
to
leaf
area
changes,
but
we
also
know
that
cam
and
most
our
system
models
ring
kind
of
too
weakly
and
too
frequently
compared
to
observations,
and
so
does
that
allow
for
an
un
unrealistically
large
control
of
leaf
area
that
too
much
water
can
collect
on
these
leaves
and
then
re-evaporate,
because
it's
raining
so
so
weakly
and
the
other
thing
with
this
finds
is.
You
know
we
see
differences
in
both
clips
for
pre-industrial
and
present-day
lai
distributions,
as
well
as
the
sensitivity
and
so
there's
a
need
to
sort
of
understand.
H
What's
driving
those
differences
is
it?
Is
it
different
sensitivities
in
terms
of
total
overall
productivity
is
the
initial
density
in
the
pre-industrial
climate
allowing
for
more
plant
and
leaf
area
growth
in
response
to
higher
co2
and
more
carbon
allocation
to
leaves-
and
these
are-
these
are
some
things
that
would
be
good
to
constrain,
to
understand
their
influence
on
the
climate.
A
Hey
thanks
so
much
gabe,
it
looks
like
dave.
Lawrence
has
a
question.
M
So
I
I'm
sure
you've
already
made
this
connection,
but
I
think
if
we
get
to
the
stage
with
the
perturbation
experiment,
where
we
do
run
a
large
ensemble
of
of
cl
on
five
simulations
with
different
valid
parameter
sets.
I
think
that
was
going
to
be
a
really
rich
resource
to
evaluate
this
kind
of
thing,
with
a
lot
less
of
that,
like
everything
changed
kind
of
aspect
to
it,
yeah.
H
I
agree
it's
so
hard
to
isolate.
What's
going
on,
yeah,
yeah
and
sensitivity,
experiments
with
holding
leaf
area
fixed
and
and
and
allowing
everything
else
to
change,
I
think
will
also
help
unfold.
A
Oh
sorry,
yes,
dave
lawrence
is
gonna,
be
our
last
talk
of
the
session
he's
going
to
be
talking
about
bioenergy
expansion.
We
can
see
your
slides,
hopefully
he's
not
muted.
Thanks
for
closing
out
our
session,
we'll
have
some
discussion
following.
M
Great,
so
thanks
yeah,
so
I'm
going
to
present
this
work
about
by
energy
expansion
versus
rear
affordation
mitigation
scenarios.
M
I'm
representing
or
work
with
I've
been
doing
with
yon
yon
chang
who's
at
the
you've,
probably
seen
her
talk
at
the
lawn
mower
working
group
or
budget
chemistry
working
groups
before
she
used
to
be
a
piano
now
now
she's
at
the
national
university
of
singapore
and
part
of
the
reason
why
I'm
presenting
is
because
it's
just
a
very
large
time
change
for
her.
M
So
you
know
the.
M
I
think
this
work
ties
in
well
with
that
initial
discussion
we
had
before
the
before
this
session
got
going.
So
the
idea
here
is,
you
know:
land-based
mitigation
strategies
are
obviously
going
to
be
required,
also
known
as
natural
climate
solutions
to
achieve
1.5
degrees,
c
or
two
degrees,
climate
targets
and
some
sectoral
analysis.
Studies
have
suggested
that
there
is,
you
know,
obviously,
potential
from
land-based
solutions
to
mitigate.
M
You
know:
10
to
15
gigatons
of
co2
per
year,
equivalent
per
year
by
2050,
which
is
about
20
30,
30
percent
of
the
mitigation
that
we
need
to
achieve
at
low
temperature
targets,
and
the
figure
on
the
right
shows
the
some
nice
paper
by
by
stephanie
rowe,
that
kind
of
catalogs
the
different
possible
methods,
and
really
it's
you
know,
in
addition
to
reducing
deforestation,
is
a
good
mitigation
potential.
You
know
reforestation
or
out
forestation
and
bioenergy.
M
Carbon
capture
and
storage
are
the
two
most
promising
and
highest
mitigation
potential,
and
these
are
the
strategies
that
are
also
employed
by
iams
to
develop
mitigation
pathways.
M
M
Obviously,
traditional
crops
can
be
used
as
biofuel
feud
stocks,
but
really
perennial
grasses,
such
as
switchgrass
and
miscanthus,
are
better
and
are
what
you
know
really
be
used
in
the
real
world
in
most
instances
because
they
have
higher
productivity
and
water
use
efficiency
as
well
as
lower
demands
for
not
irrigation
and
fertilization
than
than
traditional
crops,
and
so
yonyan
introduced
these
buyer
energy
crops
into
into
clm5,
and
so
we're
using
that
capability
within
this
study,
so
you
know,
there's
alternative
mitigation
pathways.
M
This
is
the
shared
socioeconomic
pathways
concept
and
within
that
concept
there
are
multiple
pathways
that
can
achieve
each
mitigation
target.
So
if
you're
trying
to
get
to
you
know
global
warming
target
of
less
than
2wc,
which
is
roughly
the
rcp
2.6
category,
there's
different
pathways
that
they
can
get
there.
For
example,
the
ssp1
scenario,
which
is
shown
there
in
the
bottom
left
here,
which
is,
has
you
know,
low
challenges
to
both
mitigation
and
adaptation?
It's
called
that
colloquially
known
as
the
sustainability
pathway.
You
know
that's
mainly
an
affordation
or
reforestation-based
pathway.
M
It
involves
a
decreasing
population,
assumption
and
less
needs
for
food
and
less
meat
consumption
and
then
there's
the
ssp2
pathway,
which
also
can
get
to
rcp
2.6
and
that's
known
as
the
intermediate
challenges,
pathway
or
the
middle
of
the
road,
and
that
pathway
within
gcam
at
least,
is
mainly
a
bioenergy
based
mitigation
pathway.
But
there's
also
a
growing
population
and
rising
food
needs
that
go
along
with
this.
M
With
this
scenario,
and
so
we've
been
collaborating
with
gcam
group
to
have
these
scenarios
run
both
by
the
same
integrated
assessment
model
and
then
yonya
just
processed
that
to
turn
it
into
then
use
change
files
that
we
can
use
within
within
clm,
and
this
is
what
it
looks
like
from
the
from
the
global
integrated
picture,
so
the
p2
in
red.
There
is
the
you
know.
M
This
is
the
change
of
bioenergy,
so
a
very
massive
increase
in
the
amount
of
bioenergy
crops,
some
deforestation
that
goes
along
with
that
to
accommodate
that
need
for
bioenergy
crops.
Some
of
it
is
conversion
of
cropland
or
grasslands,
but
some
of
it
requires
forests
and
there's
also
an
increase
in
fort
and
cropland
to
feed
that
growing
population
in
ssp2
and
then
their
reforestation
pathway
also
has
some
bioenergy
increase.
M
So
keep
that
in
mind,
but
there's
reforestation,
affordable
stations
will
increase
the
forest
area
and
a
decrease
in
cropland,
because
there's
a
less
population
and
less
food
needs
for
the
crops.
So,
if
you
look
at
the
spatial
pattern,
you
know
it's
pretty
complex,
you
know
it's
in
the
ssp2.
M
There's
buyer
energy
cops
added
everywhere.
There's
you
know
some
deforestation
to
accommodate
that
bioenergy
and
there's
an
increase
in
cropland
in
many
places
around
the
world
to
accommodate
the
the
needs
for
food
in
the
ssp1.
There's
some
increased
bioenergy
crops,
there's
both
reforestation
and
deforestation.
M
You
know
to
to
create
new
cropland
in
in
different
places
to
accommodate
the
some
bioenergy
uses.
So
the
main
point
of
this
figure
is
that
it's
a
complex
you
know
set
of
changes,
regional
changes
in
these
two
different
scenarios,
even
though
we're
calling
them
bioenergy
and
reforestation.
M
So
the
questions
we
want
to
ask
in
the
study:
are
you
know
how
do
these?
How
equivalent
are
these
pathways
from
a
carbon
mitigation
perspective
when
you
evaluate
them
within
within
csm2?
And
secondarily,
are
there
significant
neurosurface
climate
consequences
associated
with
the
different
pathways?
So
we
did
a
fairly
simple
experiment.
M
We
ran
two
sets
of
experiments
with
three
ensemble
members,
each
from
2015
to
2100,
both
of
them
using
the
exact
same
radiative
forcing
and
co2
levels
from
ssp
1-2.6,
but
using
the
ssp
1-2.6
land
use
scenario
in
one
of
them,
which
we
call
the
reforest
and
using
the
ssp2
dash
2.6
line
to
use
the
other
one
which
you
call
bio
crop.
M
You
know
that,
in
addition
to
the
co2
fertilization,
that
those
reforced
or
a
forced
regions
are
gaining
carbon,
as
we
expect,
and
so
that's
showing
that
it's
you
know
working
as
intended
in
a
mitigation
carbon
based
perspective,
the
bio
crop
version
also
scenario
also
gets
there
pretty
close
to
the
same
as
the
reforest
in
terms
of
total
amount
of
carbon.
M
Well,
one
of
the
reasons
it's
so
different
is
that
there
is
a
lot
of
land
use
change
emissions
in
that
bio
crop
simulation
to
create
all
that
bio
energy
area,
land
area,
so
there's
a
big
increase
or
difference
in
the
land
use
change
emissions
and,
furthermore,
there
will
be
a
lot
of
addition
of
this
additional
sync
capacity
with
that
deforestation.
M
So
that's
a
loss
of
a
potential
carbon
gain
and
then
there's
also
more
fire
emissions
that
happen
on
that
bio
crop
land
and
that's
actually,
mostly
due
to
degradation-induced
fires
from
the
tropics
they're.
Coming
from
that
deforestation.
M
So,
even
though
you're
creating
more
forest
area
in
the
reforested
run,
there's
actually
more
fire
emissions
coming
from
the
biocompany,
so
complex
set
of
responses
going
on
for
bex
that
you
know
the
fossil
fuel
substitution,
the
carbon
offset
and
carbon
capture
and
storage
efficiency
are
highly
uncertain.
So
you
know
for
bio
energy
to
work.
You
have
to
convert
that.
M
You
know
those
crops
into
fuel
somehow
use
that
fuel
to
power
cars
to
to
create
electricity,
and
then
you
also
have
to
have
mexicans
to
capture
that
that
carbon
and
so
there's
a
huge
amount
of
uncertainty
that
goes
on
in
in
the
effectiveness
of
these
things,
and
that
you
know
this
looks
at
that
these
these
spots
here
show
you
know
the
amount
of
carbon
offsets
from
you
know
not
having
to
use
fossil
fuels
and
the
associated
uncertainty
with
that
for
these
two
different
scenarios,
and
then
certainty
is
related
to
the
future
biomass
yields,
which
you
know
we're
not
capturing
all
the
technological
advances
we
kind
of
that.
M
You
know
the
energy
conversion
technology
is
also
affected
there,
and
the
bottom
here
shows
that
the
carbon
accumulation
you
get
from
from
the
carbon
capture
and
storage
aspect
and
again,
there's
uncertainty
there.
How
effective
is
that
carbon
caption
storage?
Is
it
80,
effective,
60,
effective,
90,
effective,
and
that
has
different
consequences.
So
if
you
kind
of
add
all
these
things
together
and
say
well
what?
If
you
include
all
those
uncertainties?
M
This
is
what
is
a
more
realistic
or
honest
assessment
of
the
potential
for
the
bioenergy
for
the
differences
between
these
two
scenarios
and
basically,
what
it
says
is
that
you
know
if
there's
rapid
and
large
technological
advances
you
know,
buyer
energy
could
be
a
considerably
larger,
effective
carbon
sink
than
the
reforest.
That's
this
up
here
in
this
lot
a
brand
of
the
of
the
uncertain
estimate,
but
with
slower
technological
advances,
the
bio
crops
could
totally
fail
in
some
tentacles
and
they
could
actually
be
a
carbon
source
due
to
all
that
land
use,
change
and
deforestation.
M
That
happened
in
terms
of
reforestation.
You
know
the
effectiveness
that
there's
a
question
of
whether
environmental
conditions
support
the
growth
of
a
healthy
forest.
So
this
is
the
map
showing
the
change
in
trees
and
the
reforest
run,
and
this
shows
that
gaining
carbon,
what
you
see
is,
you
know,
there's
gain
and
carbon
in
the
in
the
tropics.
M
That's
co2
fertilization
there's
gain
in
carbon
due
to
the
reforestation
effect,
but
in
some
places
you've
planted
forests
like
here
in
the
central
us
and
there's
essentially
no
carbon
gain,
and
basically
what
we
can
find
is
that
you
know
if
you
look
at
that,
there's
there's
some
places
around
the
world.
If
you
focus
here
on
that
on
these
different
probability
density
functions,
some
places
in
the
world.
This
is
the
co2
related
fertilization
effect
where
those
high
and
constant
forest
fraction.
M
Those
places
tend
to
gain
a
lot
of
carbon
in
the
future,
and
this
is
where
there's
basically
grasslands.
They
don't
gain
carbon.
That's
as
you
expect,
and
then
this
gray
curve
here
shows
what
happens
for
the
reforested
regions
and,
basically,
what
it
says
is
in
some
regions,
you're
gaining
quite
a
bit
of
carbon
and
some
reasons,
you're.
Really
not.
M
We've
actually
been
able
to
trace
that
back
to
the
fact
that
in
some
places
the
climate
just
does
not
support
the
growth
of
that
forest
and
here's
a
pdf
showing
you
know
different
regions
around
the
world,
especially
in
the
tropics.
Where
there's
enough
moisture
it
does
gain
enough
carbon.
You
know
in
places
like
the
central
us,
where
actually
in
csm
there's
a
low
bias
of
precipitation,
it
doesn't
gain
any
carbon
also
across
europe.
M
Great
okay
last
point
here
is,
you
know,
there's
disparate
impacts
in
terms
of
the
surface
climate,
so
this
is
the
difference
in
temperature
for
the
bio
crop
versus
reforest
on
the
daily
maximum
temperature
temperature,
see
in
the
tropics
you're
getting
you
know,
warmer
conditions
in
the
bio
energy
expansion,
as
you
might
expect.
So
you
know
converting
to
croplands.
You
know
tends
to
warm
the
climate,
but
in
the
higher
mid,
latitudes,
you're,
actually
seeing
cooling
in
the
buyer
engine,
that's
a
little
bit
counter
intuitive.
M
You
might
have
expected
that
you
know
if
there's
more
force,
there'd
be
cooling.
What's
interesting.
Here
is
in
that
bio
crop
scenario.
You
actually
get
enough
cooling
that
it
totally
offsets
all
of
the
warming
in
so
many
places
around
the
world,
and
so
why
is
that
cooling
happening?
Well?
It's
mainly
because
of
the
fact
that,
like
I
said
before
some
reason,
those
forests
don't
grow.
M
If
the
forests
don't
grow,
you
don't
get
the
et
cooling,
you
don't
get
the
you
know
the
mix,
the
better
mixing
from
the
roughness
effect,
and
you
just
don't
get
that
cooling
so,
where
newly
granted
forests
go
well,
deforestation
results
in
cooler,
temperatures
where
they
don't
go
well,
both
will
be
high
by
energy
food.
Coffee,
tea
drives,
albedos
drive,
cooler
temperatures.
M
So,
to
summarize,
you
know
the
effective
carbon
sink
associated
with
land-based
mitigation
is
really
dependent
on
a
lot
of
assumptions
related
to
that's
technological
advances
and
also
on
what
the
environmental
conditions
are,
that
support
growth.
There's
a
ton
of
caveats.
A
couple
climate
biases
really
come
into
play
here
and
there's
also
a
lot
of
land
model,
structural
limitations
that
are
having
a
big
impact
and,
furthermore,
there's
a
single
model
study.
We
should
do
this
with
multiple
models
to
see
what
the
effects
are.
I
will
stop
there.
A
All
right,
thanks
dave
and
thanks
to
all
of
our
speakers
for
a
really
interesting
session,
we've
got
about
10
minutes
for
open
discussions.
So
if
folks
have
questions
for
dave
or
any
concluding
thoughts,
they
want
to
share
with
the
group,
we
would
welcome
all
of
those.
A
B
I
didn't
think
I
should
call
on
myself
dave
thought
it's
really
interesting.
I'm
curious.
You
said
it
was
due
to
changes
in
e.t.
Did
you
also
check
about
the
influence
of
albedo,
because
I
would
guess
the
bioenergy
crops
would
be
brighter
than
the
trees
as
well.
M
B
C
C
Yeah
there
was
a
real,
interesting
discussion
going
on
in
the
chat
between
ichi
and
ghouling,
wang
about
the
results
from
songs
and
ichi's
presentation,
and
so
I'd
like
to
sort
of
follow
up
on
that,
because,
if
the
gist
of
it
that
I
seem
to
get
from
the
presentation
and
even
then
the
chat
was
that
yeah,
we
might
have
to
start
we're
moving
to
this
world
where
we're
we're
tuning
up,
clm
and
parameter
estimation,
but
that's
very
much
dif
dependent
on
the
model
structure
we're
moving
into
this
world.
C
I
guess
where
maybe
we're
going
to
have
to
retune
the
model
every
time
there's
a
structural
change
to
it.
So
I'd
like
to
hear
ichi's
thoughts
on
that
you
know
is
that
sort
of
really
the
implications
of
what
was
coming
across
from
their
study,
because
we
haven't
really
talked
about
that
too
much.
Yet.
N
Well,
it's
a
great
question
and
you
know
I
recently
I
thought
about
this
issue.
A
lot
and
probably
you
know
most
striking
results
was
from
crm
4.0.
You
know
when
clm
4.0,
you
incorporate
the
nitrogen
imitation
and
then
the
you
know:
global
total
soil,
carbon
storage.
You
only
get
to
like
500
something
and
they
in
the
real
world.
You
know
based
on
data
most
of
the
time
it's
that
about
2
000..
N
N
N
That's
real-world
process
and
the
state
will
not
change
because
of
our
model
structure.
Change,
yeah,
unfortunately,
and
also
also
one
thing-
is
that
you
know
for
modern
community.
Actually
this
is
quite
another
we
have
been.
We
have
used
the
model
for
60
some
years,
but
most
of
the
time
we
tuned
the
parameter
to
make
the
model
simulation
reasonable.
N
But
unfortunately
you
know
I
understand.
Most
of
you
probably
spend
a
lot
of
time
and
and
work
very
hard
to
tune
the
model
parameters.
But
none
of
the
effort
has
been
documented
in
literature
and
we
understand
you
know
we
have
discussed
how
to
modify
model
structure,
but
the
community
hasn't
paid
enough
attention
on
parameterization
issue.
B
I
want
to
just
follow
on
that
and
say
it
seems
like
the
implication
from
what
you're
saying
is
that
you
prefer
a
comparison
more
like.
I
lamb,
like
benchmarking,
the
models
against
the
observations,
whatever
the
structures
are
within
that
model,
rather
than
the
way
that
we
often
do
a
lot
of
experiments
where
we
change
one
one.
B
You
know
structural
element
and
then
look
at
how
the
answer
is
different,
because
those
that's
not
really
a
fair
comparison.
If
the
structure
should
be
tuned
differently
to
the
observations
that
at
all
match
up
with
what
you're
suggesting.
N
Well,
I'm
not
sure
if
I
fully
probably
get,
can
you
say
it
again.
Sorry.
B
Sure
I
guess
I
mean
like
a
really
common
way.
We
might
do
something
is
add
a
new
feature
or
say
like
we're
talking
about
adding
nitrogen
cycling,
and
then
we
compare
it
to
with
or
without
that
component,
but
that
maybe
isn't
the
right
comparison.
The
right
comparison
is
to
compare
the
both
separate
structural
versions
tuned
separately
to
some
third
independent
thing,
like
islam,
might
be
used
to
like
compare
it
against
a
benchmark.
N
Yeah,
probably
let
me
let
me
put
you
know
when
songs
submit
to
the
manuscript
we
get.
We
get
the
two
reviews.
The
reviewer
was
astounded
by
the
point
we
tried
to
make
and
also
they
asked
if
you
always
tune
your
model
and,
what's
you
know,
what's
good
the
mode
about,
so
they
that
the
two
review
really
forced
us
to
think
about
this
issue.
You
know
that
why
we
need
model.
If
you
keep
tuning,
you
know,
whenever
you
change
mode
structure,
your
tune
again,
you
re-parameterize
it
again.
N
So
if
we
take
the
nitrogen
as
the
issue,
you
know
the
conceptually,
we
want
to
use
nitrogen
model.
Cnc
carbon,
nitrogen,
coupled
model
to
to
examine
how
ecosystems
respond
to
climate
change
under
the
nitrogen
limitation
issue.
So
so
we
really
you
know.
N
B
M
M
B
J
Sure
sure
I
mentioned
I'm
really
bad
at
zoom
chat,
so
I'm
not
sure
how
much
of
it
came
through,
but
we've
had
what
we
call
iomb.
Somebody
suggested
a
different
acronym,
but
for
whatever
reason
we
settled
on
iomb,
the
initial
work
was
done
by
olu
oluwashin
and
currently
it's
undergone
by
weiwei.
So
it
does,
it
does
exist
and
it's
out
there.
Weiwei
has
a
paper,
hopefully
to
publish
very
soon
on
it
and
we
just
don't
tend
to
present
as
much
on
it
not
being
ocean
people
ourselves.
J
A
But
right
well
being
mindful
of
the
time
we
should
probably
break,
but
thanks
everybody
for
the
excellent
talks
and
the
great
discussion.
I
know
that
abby
matt
and
I
welcome
continued
thoughts
about
how
we
can
move
the
bgc
working
group
forward,
especially
now
that
matt
has
joined
as
the
new
co-chair
so
feel
free
to
continue
the
discussion
with
us
at
any
point.
If,
if
new
ideas
are
prompted
by
the
presentations
today,.