►
From YouTube: CESM Workshop: Biogeochemistry Working Group
Description
The 26th Annual CESM Workshop will be a virtual workshop with a modified schedule on its already scheduled date. Specifically, the virtual Workshop will begin with a full-day schedule on 14 June 2021 with presentations on the state of the CESM; by the award recipients; and three invited speakers in the morning, followed by order 15-minute highlight and progress presentations from each of the CESM Working Groups (WG) in the afternoon.
On 15-17 June 2021, working groups and cross working groups have half-day sessions, some with presentations and some that are discussion only.
B
Ready
thanks,
ryan
thanks
todd,
welcome,
folks
to
the
by
geochemistry
working
group
session
of
the
cesm
workshop
as
you're,
probably
seeing
in
the
pop-ups.
This
meeting
is
being
recorded,
and
that
also
includes
the
chat
so
that
questions
in
the
chat
will
be
able
to
refer
back
to
them
later
after.
B
The
meeting
is
over
we're
going
to,
while
I'm
telling
you
some
logistics
here,
I'm
going
to
show
another
slide
here
about
some
of
the
principles
that
we
try
to
adhere
to,
and
we
want
all
of
our
participants
to
follow
regarding
having
a
respectful
dialogue
amongst
participants
in
our
in
our
meeting
to.
B
I
don't
need
to
read
this
to
you.
You
can
read
it
yourself,
but
these
are
principles
that
we
that
everyone
should
be
adhering
to
after
we
have
a
handful
of
presentations
covering
a
wide
range
of
topics,
we're
going
to
have
a
discussion
session
where
we're
going
to
break
out
into
zoom
meeting
rooms
and
that's
going
to
be
done
randomly
and
we're
going
to
have
some
questions
prompted
to
prompt
discussion
and
I'll
get
to
those.
B
At
the
end
of
my
presentation,
which
is
going
to
come
up
in
just
a
moment
before,
we
that
went
a
lot
faster
than
I
thought
it
would.
I
do
want
to
introduce
the
other
co-chairs
of
the
working
group,
gretchen
keppel,
alex
from
the
university
of
michigan
and
abby
swann
from
the
university
of
washington.
B
Thank
you
to
them
for
your
help.
With
setting
up
this
meeting.
B
B
B
Okay,
I'm
going
to
go
ahead
and
give
an
update
on
some
activities
of
the
working
group,
and
I
guess
one
thing
that
I
did
forget
to
mention
is
that
we
have
20-minute
slots
for
all
of
the
presentations
and
we
would
like
everyone
to
leave
at
least
three
minutes
for
questions
at
the
end
of
their
talk.
If
you're
able
to
give
more,
then
that
would
be
great
and
gretchen
is
going
to
remind
you
as
you
approach
those
time
limits.
B
A
B
So
I'm
going
to
go
into
some
updates.
A
lot
of
this
material
is
repeat
from
what
was
presented
on
monday,
so
apologies
to
those
of
you
that
if
you
get
bored
by
it,
but
it's
better
to
see
the
important
stuff
more
than
once-
and
there
is
some
additional
stuff
in
here
as
well.
B
The
dashed
line
is
showing
this
simulated
atmospheric
co2
in
cesm1
bgc,
and
you
can
see
the
the
growth
of
co2
in
cesm1
was
too
high
growth
compared
to
up
to
the
observed
growth
rate
in
the
atmosphere
over
the
latter
part
of
the
20th
century,
particularly
in
the
bottom
panel.
Where
there's
the
model
minus
observations
you
can
see,
the
slope
of
that
dashed
line
is
positive,
indicating
this
bias
in
growth
rate
in
the
atmospheric
co2
and
that
continues
on
into
the
21st
century
compared
to
the
iam
models
that
are
used
to
construct
future
scenarios.
B
There
are
a
couple
of
colored
solid
lines
also
in
these
plots,
which
are
two
ensemble
members
from
cesm2
with
this
coupled
carbon
cycle
model.
In
this
esm
hist,
when
you
look
at
the
absolute
values
themselves,
the
curves
are
pretty
much
on
top
of
themselves.
It's
hard
to
distinguish
them.
You
do
see
some
variation
in
the
ensemble
members
in
when
we
look
at
the
bias
plot
in
the
bottom
panel
and
there's
a
few
things
that
come
out
of
this
bias
plot
that
are
worthy
of
note.
B
One
is
that
in
the
19th
century,
around
1850,
our
coupled
carbon
cycle
model
produced
atmospheric
co2
concentrations
that
were
slightly
elevated
compared
to
observations,
and
this
was
a
consequence
of
remaining
drift.
After
doing
some
spin
up
in
the
component
models,
the
land
and
the
ocean
model
both
had
drift
of
the
same
sign
in
cesm2,
and
they
both
were
emitting
carbon
into
the
atmosphere.
B
The
drifts
in
the
land
in
the
ocean
after
doing
spin-ups,
were
of
the
opposite
sign
and
we
ended
up
having
a
continual
transfer
of
carbon
from
the
land
to
the
ocean
while
leaving
the
atmospheric
co2
relatively
stable.
B
So
the
fact
that
the
atmospheric
co2
was
slightly
elevated
in
the
1850
is
pretty
much
or
that
it
was
stable
in
cesm1
and
elevated
in
cesm2
is
mostly
just
a
byproduct
of
the
remaining
drift
from
the
spin-ups.
It's
not
something
that
we
were
able
to
ease.
You
would
have
easily
been
able
to
control
from
doing
those
spin-ups.
B
The
next
feature
that
comes
out
in
this
bias
plot
is
the
is
the
up.
Is
the
slope
in
the
mid
20th
century,
in
the
bias
showing
that
the
models
are
accumulating
atmospheric
co2
compared
to
observations?
B
This
is
pretty
much
unchanged
from
cesm1
to
cesm2,
and
this
is
a
feature
that
is
present
in
many
different
cmip
class
models
and
it's
an
ongoing
research
to
understand
why
the
model
by
most
coupled
carbon
cycle
models
are
not
able
to
reproduce
the
atmospheric
growth
rate
of
co2
in
the
mid
20th
century.
B
B
Another
diagnostic
of
the
coupled
carbon
cycle
is
looking
at
the
seasonal
cycle
of
atmospheric
co2
and
I'm
going
to
show
some
figures
of
that
from
barrow
alaska
and
mauna
loa
hawaii
the
dashed
line,
just
like
the
previous
plot.
The
solid
line
is
the
observed
seasonal
cycle
and
barrow
alaska.
There's
a
substantial
around
15
ppm
peak
to
trough
difference
in
co2.
In
the
surface
in
observations
and
in
cesm1
we
were
that
the
amplitude
of
that
seasonal
cycle
was
very
was
low
compared
to
the
observed
seasonal
cycle,
and
that
has
improved
considerably
in
cesm2.
B
B
Amplitude
is
a
notice
is
a
notable
improvement
that
we're
happy
about,
and
then
the
bottom
plot
is
showing
the
amplitude
of
that
seasonal
cycle
as
a
function
of
time,
showing
that
if
we
just
had
looked
at
the
early
part
of
the
20th
century,
around
1920,
cesm1
and
cesm2,
don't
look
that
different,
but
as
the
as
we
progress
through
the
20th
century,
cesm2
grows
considerably
more
than
cesm1
in
where
we
have
the
observational
record
at
barrow.
Cesm2
is
doing
a
decent
job
of
capture
capturing
the
growth
rate
of
this
seasonal
amplitude.
B
B
Possible
the
next
slide,
I'm
showing
some
atmospheric
air
cco2
fluxes
compared
to
a
neural
network,
filling
a
global
map
of
air
co2
flux
and
the
model
does
reasonably
well
compared
to
this
observationally
based
product
for
the
annual
mean
plot
in
the
maps,
and
then
the
zonal
mean
with
the
seasonal
cycle.
The
model
is
doing
reasonably
well
and
the
zonal
average
is
also
looks
pretty
reasonable
compared
to
this
observationally
based
product.
This
is
the
product
of
landchutzer.
I
realize
I'm
running
a
little
short
on
time,
so
I'm
going
to
pick
up
the
pace
here.
B
One
aspect
of
the
ocean
simulation
that
is
a
sore
spot
in
the
model
is
the
deep
ventilation,
the
ventilation
of
the
deep
pacific.
So
if
you
look
in
these
bottom
row,
this
is
a
radiocarbon
simulated
by
the
model
in
the
first
column,
is
a
delta
14c
in
the
model,
and
we
are
actually
off
the
color
scale
shown
here
on
the
middle.
B
The
middle
column
is
an
observational
product
in
the
right
column
is
the
bias,
and
the
model
is
ventilation
of
the
deep
pacific
is
too
weak
compared
to
observations
considerably,
and
so
this
leads
to
excessive
decay
of
radiocarbon,
which
indicates
that
this
sluggish
ventilation-
and
this
has
implications
for
global
nutrient
cycles,
because
we
tend
to
accumulate
nutrients
in
the
deep
pacific
and
we
had
to
put
in
some
clue
just
to
avoid
oxygen
from
really
going
totally
off
the
rails
in
our
cmip
simulations.
B
So
this
is
a
substantial
deficiency
in
cesm2
compared
to
observations.
Moving
on
to
different
flavors
of
activities
of
the
biogeochemistry
group
is
looking
at
analysis
of
simulations
of
the
carbon
cycle,
and
this
is
looking
at
simulations
where.
B
We
typically
do
in
the
climate
model
a
doubling
of
co2,
where
the
radiation
sees
this
doubled,
co2
and
the
surface
components
see
this.
The
bgc
surface
components
also
see
this
increase
in
co2
and
there's
some
experiments
here
where
just
the
surface
components
in
bgc
see
this
increased
co2,
and
it
turns
out
that
the
that
the
terrestrial
system,
seeing
this
additional
co2
leads
to
a
warming
that
explains
as
a
non-trivial
fraction
of
the
warming
that
you
see
in
the
full-blown
doubled
co2
experiments.
B
So
what
you
see
in
this
top
panel
is
the
phys.
So
this
is
the
physiological
contributions
to
warming
from
the
the
terrestrial
system.
Seeing
this
enhanced
co2
and
the
red
here
is
showing
where
you're
getting
warming
just
even
though
the
radiation
in
the
atmosphere
has
seen
the
pre-industrial
co2
we're
getting
warming
just
from
the
terrestrial
system
of
the
physiological
aspect
of
it
and
in
the
bottom
piano
the
fraction
of
the
transient
climate
response.
B
We
see
that
we're
up
a
non-trivial
fraction
of
the
model
warming
is
just
from
this
physiological
response,
and
the
take-home
message
from
this
is
that
the
first
part
of
the
text
says
what
what
what
happens
in
the
carbon
cycle,
particularly
the
terrestrial
carbon
cycle,
doesn't
stay
there.
It
couples
to
the
rest
of
the
model,
I'm
running
short
on
time,
and
so
I'm
going
to
actually
skip
this
slide
here.
B
B
This
is
some
analysis
of
community
projects
of
the
dple,
where
nikki
lovandusky
is
looking
at
the
carbon
cycle
response
the
carbon
cycle
behavior
in
this
community
project
of
the
decadal
prediction,
large
ensemble.
So
this
is
an
experiment
that
was
based
off
of
the
the
lens
experiments
and
then
decadal
prediction
experiments
were.
B
The
black
line
in
these
plots
is
a
reconstruction
of
air
co2,
fluxes
globally
averaged,
and
this
is
from
a
heincast
experiment
forced
with
re-analyses
with
the
ocean
ice
system.
So
we
we
don't
have
a
great
historical
record
of
global
air.
Co2
observed
air
cco2
fluxes
going
back
into
the
mid
20th
century,
and
so
we're
comparing
the
decadal
prediction.
B
Experiments
to
this
reconstruction
force
with
re-analyses
and
the
one-year
predictions
whether
they
are
using
the
the
raw
atmospheric
co2
or
the
detrended,
shows
a
really
good
match
between
the
reconstruction
and
the
initialized
forecast,
and
it's
doing
better
than
we
do
with
the
uninitialized
forecast,
which
doesn't
get
the
wiggles
that
are
present
in
the
in
the
reconstruction
or
the
predictions
that
arise
from
inter-annual
variability.
B
When
we
do
compare
to
the
observational
product,
this
land
shoots
our
neural
net
product.
We
do
see
that
we're
not
capturing
the
full
magnitude
that
is
present
in
that
in
that
product.
For
estimates
of
the
variability
of
the
co2
flux
and
the
understanding,
the
mismatch
between
the
initialized
forecast
and
the
land,
shoots
or
observational
product
is
a
work
in
progress.
B
This
is
really
just
trying
to
sample
some
different
aspects
of
looking
at
the
carbon
cycle.
That's
going
on
in
the
working
group,
some
additional
activities
that
are
going
on
some
aspects
of
these
activities
impact
everyone
who's
doing
carbon
cycle
science
in
the
coupled
model
in
this,
in
particular,
as
you've
probably
heard
in
other
parts
of
this
workshop.
B
The
ocean
model
working
group
is
transitioning
from
pop
to
mom
6
for
cesm3,
and
so
the
carbon
cycle
ocean
carbon
cycle
model
is
going
is
being
ported
from
pop
to
mom,
and
so
what's
really?
What
is
happening
is
that
our
by
geochemistry
library,
marble
is
being
coupled
to
mom,
taking
the
coupling
that
we
currently
have
to
pop
mike
levy.
B
We
expect
we'll
be
using
a
something
close
to
an
isopycno
vertical
coordinate
in
this,
in
the
way
that
mom
implements
this.
This
ends
up,
leading
to
potentially
having
very
thin
layers
at
the
sea
floor
where
you
have
densities
in
your
coordinate
system
that
are
not
present
at
a
particular
location
on
the
globe.
A
B
Okay,
thank
you
and
so
there's
some
additional
details.
B
There
being
worked
out,
there's
a
number
of
activities
that
are
being
worked
on,
including
them
more
functional
groups
in
the
ocean
model,
bgc
improved
sinking,
particle
dynamics
with
some
new
observational
products,
coupling
to
higher
trophic
levels,
there's
a
fishman
project
that
was
mentioned
in
the
paleo
working
group
earlier
today,
and
there's
some
additional
projects
that
I
don't
need
to
read
the
text
to
you
when
we
get
to
the
end
of
the
session
here,
we're
going
to
have
some
discussion
on
how
to
coordinate
bgc
activities
and
I'm
just
going
to
leave
you
with
sort
of
the
motivation
for
this
discussion
section.
B
We
have
a
goal
of
reinvigorating
activity
within
the
biogeochemistry
working
group.
The
number
of
presentations
that
have
been
submitted
to
the
bgc
working
group
has
been
down
in
recent
years
and
there's
a
number
of
reasons
for
that.
Not
all
of
them
are
necessarily
bad.
Some
of
that
is
migration
of
carbon
cycle
talks
to
other
working
groups
where
perhaps
they
have
a
better
fit
for
other
aspects
of
that.
B
But
we
do
want
to
make
sure
that
we
have
a
vigorous
activity
going
on
the
bgc
group,
and
so
we
want
to
have
some
discussion
about
how
to
have
a
more
lively
working
group,
and
so
in
terms
of
thinking
about
that.
B
Well,
a
natural
question
is
well
what
is
the
purpose
or
goal
of
the
bgc
working
group
and
there's
a
few
different
ones
that
the
working
group
co-chairs
have
mentioned,
whether
that's
the
support
of
of
community
projects
beyond
the
scope
of
other
working
groups
like
the
coupled
carbon
modeling
effort,
that
originated
with
the
flying
leap
and
as
coined
many
years
ago,
running
experiments
that
are
animal
community
experiments
that
are
analyzed
by
different
members
of
the
working
group,
such
as
the
c4,
mip
experiments
or
taking
part
in
the
experiments
being
run
out
beyond
the
typical
2100
out
to
2300
or
in
or
analyzing
bgc
aspects
of
other
community
projects,
whether
it's
a
lens
dple
or
clm,
ppe,
and
so
some
questions.
B
We're
going
to
try
to
have
people
have
some
answers
when
we
have
the
breakout
groups
are
what
parts
of
these
aspects
of
vgc
are
you
most
excited
about
which
out?
What's
your
level
of
interest
in
these
activities
or
other
things
that
you're
interested
in
seeing
the
bgc
working
group
and
what
can
the
bgc
working
group
do
to
support
these
activities?
B
B
So
I'll
close
there
and
we'll
look
at
questions.
Take
questions
now.
C
Great-
and
we
already
have
some
questions
in
the
chat-
I
don't
know
if
nikki
wants
to
speak
up
or
I
can
read
the
question.
Can
you
hear
me.
D
B
B
When
we
look
at
those
radiocarbon
ages,
they
do
seem
like
we've
degraded,
compared
to
what
we
previously
in
previous
versions
of
the
model,
but
we
didn't
catch
it
until
too
late
in
the
development
cycle.
B
That's
a
great
question:
I
think
it's
all
about
time,
skills
and
so
for
the
the
bgc.
I
think
this
is
really
hit
us
when
we
do
the
spin-up
simulation,
where
we
run
for
thousands
of
years
to
spin
up
the
bg
the
carbon
cycle,
and
then
that
really
becomes
evident
for
something
like
a
centennial
scale,
transient
experiment.
C
Yeah,
I
can
summarize
from
the
chat
I
don't
will
asked
about
the
changes
in
temperature
that
keith
showed
and
whether
those
were
caused
by
increases
in
water
use,
efficiency
and
lower
land,
late
heat
flux
and
claire
responded
and
said.
Yes,
the
plant
responses
to
co2
are
leading
to
lower
latent
heat
flux
in
those
coupled
simulations
with
drives
warming,
and
then
danica
also
asked
if
that
increases
as
a
reaccounting
for
changes
in
leaf
area
index
and
I'll
just
add
that
that
was
a
analysis
of
about
12
cmp6
models.
C
C
Is
the
due
to
still
model
closure,
which
is
even
though
the
leaf
area
is
increasing?
The
bottle
closure
still
wins.
G
Okay,
so
can
everyone
see
that
yes,
great
and
it
seems
like
I
must
sound?
Okay,
if
you
give
me
a
thumbs
up
great,
so
I'm
gonna
talk
today
a
little
bit
about
seasonal
amplitude
of
co2
exchange
in
the
cesm2
and
rule
of
agriculture
management.
So
this
is
just
a
continuation
of
work
that
I've
been
working
on
for
for
quite
some
time,
so
some
of
you
may
have
seen
parts
of
this
presentation
before,
but
hopefully
these
new
updates
are
interesting
and
thanks
to
everyone.
G
So
I
just
want
to
start
by
highlighting
that,
as
you
all
know,
carbon
dioxide
is
really
important,
plays
a
really
important
role
in
our
climate
and
has
been
increasing
through
time.
But
what
I
want
to
highlight
even
more
is
this
annual
cycle
in
in
carbon
dioxide,
and
the
reason
that
I
want
to
highlight
that
is
because
it's
every
year,
the
carbon
dioxide
concentration,
typically
peaks
in
the
northern
hemisphere
early
spring
and
then,
as
the
northern
hemisphere,
growing
season
progresses.
G
And
so
I
guess
the
difference
between
the
high
and
the
low
values
each
year
is
what
I'm
calling
the
seasonal
cycle
amplitude.
And
this
is
because
this
is
driven
by
plant
processes.
It's
largely
a
measure
of
terrestrial
carbon
dioxide
uptake,
so
the
changes
in
the
amplitude
can
tell
us
something
about
terrestrial
carbon
fluxes
and
in
fact,
there
are
observed
increases
in
the
amplitude
of
the
seasonal
carbon
dioxide
cycle.
So
that's
illustrated
here
for
monoloa
and
furbero,
and
this
is
work
by
raven
and
colleagues.
G
And
so
what
you
see
is
that
there
are
strong
increases
in
seasonal
amplitude,
particularly
at
high
latitudes,
like
barrow,
but
still
even
at
lower
latitudes,
like
model
low.
We
are
seeing
increases
in
carbon
dioxide
amplitude,
so
what
in
fact
contributes
to
the
changes
in
carbon
dioxide
amplitude?
There
are
a
variety
of
factors
that
might
contribute,
but
there's
still
a
lack
of
consensus
in
the
scientific
community
despite
this
continual
investigation
and
so
there's
some
contribution
from
increased
productivity
in
high
latitudes.
G
But
if
you
look
at
the
cement,
5
suite
of
models
of
our
system
models
most
of
these
models,
none
of
these
models,
in
fact
capture
the
trends
and
amplitude,
and
so
this
is
illustrated
here
for
point
barrow
and
alaska,
where
you
can
see
that
the
observations
are
have
a
larger
amplitude
than
any
of
the
cmep5
models.
And
this
is
similar
to
the
results
in
grievan
by
graven
and
colleagues
who
showed
that
models
also
don't
capture
the
amplitude
north
of
45
degrees,
so
very
similar
results.
G
But
if
we
look
into
the
newer
version
of
most
of
these,
or
of
some
of
these
are
system
models
that
were
submitted
to
the
cmf6
suite
of
simulations
you'll
see
that
models
have
improved,
and
so
now
the
the
theme
of
six
models
bracket.
The
observed
changes
in
carbon
dioxide
amplitude.
G
So
I'm
going
to
start
by
looking
at
the
co2
seasonal
cycle
in
each
version,
and
so
this
is
similar
to
what
keith
showed
and
I've
shown
these
plots
in
the
past
as
well.
But
basically,
what
I
want
to
highlight
here
is
this:
is
the
co2
amplitude,
so
the
change
in
co2
amplitude
on
the
x-axis
over
the
over
the
year,
so
starting
in
january
and
ending
in
december?
G
And
what
I
want
to
highlight
here
is
that
cesm2,
which
is
the
purple
line,
has
a
larger
amplitude
in
c
than
cesm1,
which
is
the
orange
line
both
at
monoload
and
a
barrow
and
cesm2
more
closely
matches
observations
which
are
shown
in
the
black
line.
There's
still
there's
it's
still
not
perfect.
We
still
have
some
work
to
do
as
keith
said,
but
the
amplitude
is
improved
in
cesm2.
G
You
can
also
see
that
the
trends
in
the
ces
co2
amplitude
are
also
improved
in
cesm2,
and
so
this
you
can
see
this
furbero
is
on
the
top
again.
Csm2
is
in
the
purple
line,
and
the
observations
are
in
the
black
line
and
you
can
see
that
the
change
in
co2
amplitude
for
cesm2
tracks
the
observations
relatively
well,
whereas
there
was
not
a
strong
change
in
co2,
amplitude
and
ces
m1.
G
Csm2
is
an
improvement,
but
not
not
a
real,
strong
improvement.
I
should
say
for
observations,
but
but
there
is
some
improvement,
and
so
we
are
moving
in
the
right
directions,
and
so,
as
keith
said,
many
of
these
improvements
in
in
co2
in
the
co2
seasonal
cycle
are
coming
from
the
community
land
model
or
the
land
portion
of
the
cesm.
G
G
So
I'm
going
to
switch
from
looking
specifically
at
cesm2
simulations
and
switch
focus
to
the
community
earth
system
models,
land
component,
ceo,
the
the
community,
land
model
or
crm,
and
so
using
land.
Only
simulations
means
that
the
land
is
running
using
observationally,
derived
data
atmosphere
and
we're
not
accounting
for
changes
in
circulation.
G
So
I'm
going
to
isolate
the
impact
of
both
agricultural
management,
which
you
see
here
on
the
bottom,
including
fertilization
and
irrigation
management,
but
I'm
also
going
to
look
at
the
impact
of
forcings
because
other
work.
Previous
work
has
shown
that
these
forcings
are
also
important
contributions
to
the
the
net
carbon
fluxes.
G
Specifically
at
carbon
dioxide
increases
in
carbon
dioxide
and
also
changes
in
climate-
and
I
I
also
want
to
highlight
here
that
I'm
looking
because
these
are
land
only
simulations,
I'm
looking
at
the
net
flux
of
carbon
from
the
land
surface
back
to
the
atmosphere,
the
net
flux
between
those
two
and
this
is
called
net
biome
production.
So
I'm
using
this
as
a
proxy
for
changes
in
co2
concentrations
from
the
land.
G
So
I'm
going
to
start
by
showing
you
or
highlighting
the
forcings
and
then
I'll
move
into
the
management
after
that,
in
a
series
of
next
figures.
So
on
the
on
the
y-axis
here
is
the
net
biome
productivity
amplitude
and
on
the
x
or
yeah.
G
X-Axis
is
the
change
through
time
from
1850
through
2010,
and
I
also
want
to
highlight
that
the
results
that
I'm
showing
you
are
for
the
northern
hemisphere,
so
everything
north
of
30
degrees,
north
and
I'll
start
here
by
highlighting
the
clm
5
results,
which
is
here
so
this
ancient
biome
production
amplitude
over
the
course
of
the
growing
season,
with
seal
all
the
bells
and
whistles
and
clm.
So
it
includes
carbon
dioxide,
climate
change,
irrigation
and
fertilization.
G
So
next
I
want
to
highlight
the
impact
of
climate
change,
so
you
see
climate
change
in
this
orangish
colored
line
here,
and
I
just
wanted
to
highlight
that
we
do
see
an
impact
from
climate
change,
although
it's
not
a
large
impact,
but
we
do
see
it
and
it
starts
to
emerge
later
in
the
20th
century
and
when
we
look
at
spatial
distributions
of
this
change,
you'll
notice
that
most
of
the
impact
of
climate
change
happens
in
the
higher
northern
latitudes.
G
So
so
most
of
the
changes
are
isolated
to
these
northern
latitude
regions,
where
the
temperatures
are
getting
warmer.
So
there
is
increased
growth
and
productivity
due
to
the
warmer
temperatures-
and
I
should
highlight
here
that
all
of
the
colors
in
this
blue
range
mean
that
there
are
increases
in
net
biome
production
due
to
climate
change.
G
I
want
to
contrast
that
by
looking
at
carbon
dioxide
fertilization,
you
can
what
you'll
notice
from
this
figure
is
that
the
magnitude
for
of
changes
in
carbon
dioxide
fertilizat
due
to
carbon
dioxide
fertilization
are
very
similar
to
that
of
climate
change,
but
you'll
notice
that
the
changes
are
more
spatially
widespread,
so
they're
distributed
over
larger
spatial
areas
than
the
changes
in
climate,
and
so
you
see
that
co2
fertilization
is
important
throughout
most
of
temperate
and
high
latitude
regions,
and
I'm
just
going
back
to
this
time
series
plot.
G
You
can
see
that
in
fact
integrated
over
this
larger
spatial
area
globally,
carbon
dioxide
has
a
larger
impact
than
what
climate
had,
and
that
impact
is
really
starts
to
become
evident
in
the
1980s
moving
forward.
G
So
next
I
want
to
dive
into
looking
at
the
impacts
of
fertilization
and
irrigation
management.
So
first
I'll
start
with
irrigation.
The
yellow
or
gold
colored
line
here
is
the
change
in
mvp
amplitude
or
carbon
flux
amplitude.
G
G
But
if
you
look
at
this
regionally
you'll
notice
that
it's
quite
important,
especially
especially
in
south
asia,
where
there
are
large
impacts,
it's
increasing
the
net
biome
production
amplitude
with
a
larger
magnitude
than
we
saw
in
either
co2
or
climate
change,
fluxes
and
so,
but
to
me
this
suggests
that
in
the
future,
as
we
increase
irrigated
area,
irrigation
will
become
globally
more
important,
but
it
is
regionally
very
important
right
now.
G
I
want
to
contrast
this
again
with
agricultural
nitrogen
fertilization
and
just
to
show
that
the
magnitude
of
impact
for
agricultural
nitrogen
fertilization
is
similar
to
that
of
irrigation,
but
it
is
especially
much
more
widespread,
so
these
changes
are
more
widespread
and
also
larger
magnitude
than
that
of
irrigation,
but
also
of
fertilization,
co2
fertilization
and
climate
change.
G
So
agricultural,
nitrogen
fertilization
increases
co2
amplitude
and
if
you
look
back
at
this
time,
series
you'll
notice
that
globally.
This
really
fertilizer
really
does
have
the
largest
impact
on
carbon
carbon
fluxes
between
the
land
surface
and
the
atmosphere,
and
these
changes
are
so.
The
fertilizer
is
a
larger
driver
of
carbon
dioxide
seasonality
than
any
of
the
other
drivers,
and
it's
evident
earlier
on
in
the
time
series
starting
in
the
1950s,
for
example,
when
industrial,
nitrogen
fertilizer
application
became
more
commonplace.
G
So
we
just
want
to
end
by
highlighting
this
is
the
figure
that
I
just
showed
you
on
the
top
with
clm
the
change
in
net
biome
production
amplitude
in
clm
in
the
land
only
simulations,
and
when
you
turn
off
agriculture
and
nitrogen
fertilizer
in
the
in
cesm2.
So
in
a
coupled
model,
you
can
look
at
the
change
in
the
carbon
dioxide
amplitude
and
see
that
the
patterns
of
the
spatial
patterns
of
change
are
actually
similar
to
that
of
mvp.
G
But
you
do
see
changes
a
little
bit
more
distributed
due
to
the
fact
that
this
stuff
cesm2
does
include
atmospheric
circulation,
and
these
changes
in
carbon
dioxide
amplitude
are
actually
quite
large.
If
you
look
at
these
agricultural
areas,
you're
getting
changes
of
up
to
nine
parts
per
million
in
the
amplitude
in
some
of
these
regions.
G
When
you
look
at
the
difference
between
csm2
and
ces
or
cesm1
and
cesm2
you'll
notice
that
the
co2
amplitude
is
larger
in
csm2
than
it
is
in
csm1
and
agriculture
contributes,
is
a
large
contributor
to
those
changes
in
a
carbon
dioxide
amplitude,
and
this
is
primarily
due
to
the
impact
of
industrial,
nitrogen,
fertilizer,
so
fertilizing
our
crops
is
a
really
important
component
of
the
terrestrial
carbon
cycle
and
changing
those
seasonal
flux
patterns.
G
C
A
A
D
Question
is
regarding
agriculture:
there's.
G
Yeah
there
there
is
agricultural
burning,
including
included
in
clm,
so
so
that
is
a
part
of
these
simulations.
I
didn't
look
specifically
at
about
how
agricultural
burning
affects
these
carbon
dioxide
fluxes,
but
I
imagine
that
it
does.
I
think
that
we
don't
fully.
G
We
need
a
better
representation
of
agricultural
fires
actually
because
we
don't
account
for,
for
example,
shifting
cultivation
or
slash
and
burn
fires.
We
just
account
for
general
agricultural
fires,
so
it
likely
contributes,
but
I'm
guessing
that
the
impact
is
under
a
fire
is
underestimated.
In
these
simulations.
G
There's
there's
just
a
lot
of
irrigated
area
in
that
region.
I
think
I
think
a
large
part
of
that
region
is
agriculture,
land
and
a
large
part
of
the
agriculture
land
in
that
region
is
irrigated,
and
so
that's
that's
part
of
the
reason
why
I
think
about.
G
I
think
it's
about
25,
of
total
land
area
is
cropland
and
then
or
no
sorry,
it's
10
of
total
land
area
is
cropland
and
in
that
10
only
25
of
cropland
is
irrigated,
and
so
irrigated
fraction
is
actually
really
small,
which
is
why
you
don't
see
a
lot
of
irrigation
impacts
elsewhere
in
the
world,
but
that
is
one
region
where
irrigation
is
nearly
100
of
the
cropland
area
and
there's
a
really
high
fraction
of
cropland
in
that
region.
G
G
I
think
yeah,
I
think
likely-
does
have
an
impact
on
co2
growth
three,
but
I
you
know
it's
not
something
that
I
probed
for
sure
to
know
for
sure,
and
you
know
I
would
expect
that
it
does
impact
annual
mean
mvp,
but
maybe
not
as
largely
as
as
changing
the
amplitude,
because
we
also
are
harvesting
the
crops
every
year
so
that
biomass
is
removed.
C
I
guess
I'll
ask
a
question
too,
which
is
that
you
showed
them
that
last
plot
a
cesm
spatial
map
of
the
impact
of
fertilization,
and
I
was
wondering
if
you've
gone
and
calculated
the
amplitude
specifically
at
some
of
the
observational
locations
like
monologue
and
barrow,
to
get
from
whatever
simulations
you
have
the
attribution
to
these
specific
things.
You're
looking
at
be
really
cool
to
see
the
attribution
to
all
of
them.
But
if
you
don't.
A
C
G
Yeah,
so
I've
looked
at
it
for
fertilization,
specifically
because
that's
the
only
one
of
these
perturbation
experiments
that
I
have
in
the
fully
coupled
model
and
in
looking
at
fertilizer,
it
does
contribute
it's
not
it's
not
all
of
the
difference,
but
but
it
does
have
an
impact
that
is
actually
measurable.
So.
C
Yeah
specifically
wondering
if
the
attribution
would
be
very
different
at
barrow
and
mauna
loa,
since
you
showed
the
different
spatial
patterns
with
the
climate
effects
being
more
prominent
at
higher
latitudes.
G
H
All
right
good,
not
the
first
one,
though
that's
too
bad,
it's
good.
H
H
So
thanks
for
the
invitation
to
present
this
is
building
off
the
same
set
of
simulations
that
keith
talked
about
and
that
danica
talked
about,
but
specifically
I'm
looking
at
inner
annual
and
seasonal
drivers
of
carbon
cycle
variability
represented
by
cesm2
esm,
and
this
is
work
that
I've
been
doing
with
keith
and
donica,
as
well
as
zach,
butterfield
and
gretchen
couple
alex,
who
are
both
sacks,
a
former
graduate
student
at
university
of
michigan
and
it's
a
paper
that
is
just
about
accepted
in
global
biology,
global
biogeochemical
cycles
and
will
be
on
the
ces
m2
special
issue
and
was
made
possible
by
a
nasa
project
that
gretchen
that
I
had
so
before.
H
I
kind
of
dive
into
the
problems
that
we
diagnosed
with
the
model.
I
want
to
highlight
that
clm
and
cesm
have
an
improved
carbon
cycle
representation,
so
these
are
just
two
islam
plots
of
ecosystem
and
carbon
cycle
metrics
for
the
clm
5
gswp3
runs
that
dave
published
on
in
his
2019
james
paper
and
then
for
cesm2
compared
to
the
lens.
You
know.
H
That's
not
what
I'm
going
to
talk
about
here
and
specifically,
I
want
to
focus
on
this
carbon
dioxide
metric
a
little
bit,
which
is
I
like
to
call
this,
the
the
gretchen
metric,
and
this
is
kind
of
what
it
looks
like
in
islam
space
and
so
the
if
you're,
not
familiar
looking
at
eye
lamp
plots.
That
kind
of
you
can
drill
down
from
these.
H
You
know
kind
of
stop
light
kind
of
color
diagrams
to
look
at
individual
metrics,
and
so
this
is
what
co2
looks
like
and
for
clm4,
clm45
and
clm5
shown
here.
You
can
see
that
clm5
scores
better
for
all
of
these
different
metrics,
except
for
inner
annual
variability,
so
this
is
kind
of
a
known
deficiency
in
clm5,
and
I
wanted
to
try
to
investigate
that
a
little
bit
more.
This
is
the
kind
of
cool
diagnostic
that
gretchen
came
up
with
and
what
it
looks
like
you
know.
This
is
what
noaa
observation
suggests.
H
The
inner
annual
variability
of
co2
in
the
atmosphere
is
roughly
0.6
to
1
ppm,
depending
on
where
you
are
latitudinally
and
both
clm4
and
clm5
underestimate
that
underestimate
that
by
quite
a
bit,
so
the
two
kind
of
things
I
want
to
explore
here
are
first
the
what's
causing
this
low
carbon
cycle
variability
in
cesm2
and
then
also
I
wanted
to
look
at
seasonal
drivers
of
carbon
cycle
variability.
H
So
it's
a
little
bit
of
a
whirlwind
to
try
to
squish
these
two
things
into
a
relatively
short
talk
and
just
kind
of
to
ground
everyone
with
what
I'm
talking
about.
So
I'm
using
one
ensemble
member
of
the
cesm2
historical
simulations.
H
So
there's
a
lot
here
and
I'll
kind
of
walk
through
all
these
panels.
I
realized-
I
don't
know.
If
I
have
a,
can
you
guys
see
this?
If
I
just
give
me
a
marker
that
you
can
see.
H
Cursor
a
little
bit
yeah.
I
wonder:
that's
all
right!
That's
good
enough
for
now!
So
in
the
top
left
panel.
H
What's
shown
in
green
is
the
gcp
atmospheric
growth
rate
detrended
and
it's
been
reversed
for
convention,
so
that
positive
flexes
our
net
land
uptake
and
negative
fluxes
are
our
land
losses
and
then
the
the
inner
annual
variability
of
cesm2
is
shown
in
black,
and
you
wouldn't
expect
csm2
to
match
that
you
know
the
exact
wiggles
of
the
observed
because
we
don't
have
el
ninos
and
la
ninas
at
the
same
time,
although
we
do
have
volcanic
eruptions,
so
effects
of
kind
of
tubo
on
the
carbon
cycle
and
other
volcanic
eruptions
should
be
evident,
although
they're
pretty
weak
in
these
simulations.
H
But
the
take
home
is
that
the
magnitude
of
the
wiggles,
the
standard
deviation
of
the
flux
anomalies
is
roughly
half
what
it
should
be.
So
you
know
it's
roughly
0.5
pentagrams
carbon
per
year
is
the
standard
deviation
of
the
nep
anomalies
and
from
the
global
carbon
cycle
project
it
looks
like
they
should
be
closer
to
one
so
because
of
that
low
carbon
cycle
variability,
we
end
up
seeing
relatively
low
carbon
cycle
sensitivity
here,
I'm
just
plotting
nep
anomalies
against
terrestrial
water
storage
and
tropical
temperature
anomalies.
H
So
this
is
supposed
to
be
comparison,
comparison
to
the
humphreys
at
all
paper
for
2018
and
then
for
temperature,
the
cox
at
all
2013
paper,
there's
a
few
that
have
looked
at
temperature
variability
and
you
can
see
for
both
of
these
metrics,
because
the
magnitude
of
the
y-axis
is
compressed.
The
slope
of
the
line
is
too
shallow.
H
You
know
the
the
other
interesting
part
is
that,
even
though
this
is
with
a
fully
coupled
model-
and
so
we
have-
you
know-
modeled
terrestrial
water,
storage
and
temp
temperature
anomalies.
The
magnitude
of
the
x-axis
here
are
kind
of
spot-on,
with
the
observations,
so
it
looks
like
cesm
is
doing
a
relatively
good
job,
representing
the
climate
variability,
at
least
over
the
end
of
the
20th
century
and
start
of
the
21st
century,
but
we're
not
doing
a
good
job
capturing
the
vertical
component.
H
So
what
could
be
causing
that
again?
Nep
is
just
the
difference
between
gpp
and
ecosystem
respiration,
so
a
low
net
flux,
low
variability
in
the
netflix
could
be
caused
by
low
component
variability,
so
low
variability
in
gpp,
or
it
could
also
be
caused
by
this
high
component
correlation.
So
if
gpp
and
ecosystem
respiration
co-vary
at
kind
of
the
same
time
scales,
then
the
net
flux
will
necessarily
have
a
low
inter-annual
variability
and
so
there's
some
preliminary
results
from
matt
wozniak.
H
But
the
take
home-
or
at
least
for
my
purposes
here,
is
that
you
know
clm
doesn't
capture
the
full
range
of
variability
and
the
observations
even
for
this
flex
tower
simulation.
Where
we're
giving
it
observed
meteorology,
so
we
don't
capture
these
kind
of
hot
moments
of
high
productivity
and
we
don't
have
the
same
magnitude
of
range
of
gpp.
H
So
at
least
at
this
one
flex
tower
in
northern
michigan,
there's
some
indication
that
the
the
inputs
here
gpp
are
potentially
too
low
compared
to
the
observations
you
can
look
at
that
spatially
now,
so
these
are
from
the
cpsm
results,
and
these
are
the
standard
deviation
of
detrended
anomalies
and
to
my
eye,
what
kind
of
jumps
out
is
that
there
is
relatively
high
standard
deviation
of
the
anomalies
in
these
arid
savannah
regions,
which
is
in
good
agreement
with
what
a
lot
of
folks
have
seen
that
these
arid
regions
are
potentially
drivers
of
carbon.
H
You
know
have
high
importance
in
terms
of
carbon
cycle
variability,
but
if
you
look
in
the
in
the
core
of
the
tropics,
it's
really
really
low
and
if
you
normalize
for
the
the
fluxes
that
the
normalized
standard
deviation,
the
coefficient
of
variation
is
incredibly
low
and
that's
a
feature.
That's
common
in
in
many
cement
models.
So
this
osullivan
vapor
really
called
it
out
for
the
trendy
models.
H
H
So
the
other
potential,
this
low
variable
gpp
variability-
is
one
potential
driver.
The
other
is
a
high
correlation
between
gpp
and
ecosystem
respiration
anomalies.
So
that's.
This
is
a
plot
of
what
that
looks
like,
and
this
correlation
is
really
really
high.
Some
work
from
what
that
dennis
beldocky
did
at
flexnet
that
it's
kind
of
a
collection
of
flex
sites
suggests
that
the
slope
of
this
line
and
the
correlation
should
be
much
lower
than
what
we're
getting
in
clm
or
cesm.
H
In
this
case,
and
so
I
wanted
to
dive
a
little
bit
into
trying
to
diagnose
this,
and
so
this
is
now
scatter
plots
looking
at
ecosystem
respiration,
anomalies
and
ecosystem
respiration
has
two
components.
So
there's
an
autotrophic
respiration
component
shown
in
green,
a
heterotrophic
respiration
component
shown
in
blue,
and
you
can
see
that
the
anomalies
in
the
system
respiration
are
largely
driven
by
the
autotrophic
component.
H
H
That's
in
clm5
that
makes
plants
pay
the
carbon
cost
of
nitrogen
uptake,
but
the
way
that
it's
applied,
it's
a
relatively
large
flux-
and
you
can
see
here
that
it's
tightly
correlated
with
the
autotrophic
respiration
anomalies,
and
so
you
know
most
of
the
most
of
the
variability
in
autotrophic
respiration
is
instantaneous
and
it's
related
to
this
one
flux,
and
so,
although
it's
not,
we
can't
really
turn
fun
off
in
clm5
to
say
that
it's
fun
that's
causing
this.
H
But
my
guess
is
that,
because
of
because
of
the
way
that
we've
kind
of
put
clm5
together,
that
this
high
correlation
between
fun,
flexes
and
gpp
are
what's
driving
this
high
correlation
between
gpp
and
ecosystem
respiration,
okay,
so
switching
gears
stop
thinking
about
variability.
To
me,
this
is
the
this
is
that's
like
the
useful
part
of
the
talk.
This
is
the
fun
part
of
the
talk,
looking
thinking
about
seasonal
drivers
and
carbon
cycle
variability.
H
So
now
I'm
gonna
start
looking
at
monthly
anomalies
instead
of
these
annual
anomalies,
and
what
I
wanted
to
do
was
use
some
of
the
work
that
gretchen's
lab
had
done.
Looking
at
singular
value
decomposition.
This
was
some
new
math
to
me.
It
basically
works
similarly
to
an
eof
if
you're
more
familiar
with
that
and
you're.
What
happens?
H
I'm
sorry,
and
so
this
is
what
it
looks
like
in
kind
of
more
intuitive
space,
and
so
what's
shown
in
this
figure
in
gray,
is
just
the
climatology
of
gpp.
This
is
for
the
kind
of
high
latitude
ecosystems
north
of
50
degrees
and
the
what
the
units
on
the
y-axis
are
kind
of
meaningless
for
this
type
of
activity,
and
then
time
so
month
of
the
year
shown
on
the
on
the
x-axis,
so
the
climatology
of
gpp
in
northern
latitudes.
H
Has
this
really,
you
know
kind
of
predictable
green
up
in
the
summer
and
then
senescence
in
the
fall,
and
so
what's
what
I'm
calling
the
amplification
vector
is
strongly
correlated
with
that
climatology
of
gpp
so
identified.
I
identified
the
amplification
vector
as
the
vector
there's
of
the
first
two
that
most
strongly
correlates
with
the
climatology
of
gpp.
H
The
other
feature
that
comes
out
of
the
svd
analysis
is
this
theta
estimation
and
so
for
positive
values.
Basically,
you
can
think
of
that
means
that
the
anomalies
are
above
the
zero
line,
and
so
a
data
value
closer
to
one
means
that
you've
got
an
increase
in
the
amplification
of
gpp
and
for
these
grid
cells.
It's
50,
56
of
the
variation
in
gpp
can
be
attributed
to
you
know,
looks
like
this
amplification
vector
the
other.
H
The
second
most
common
mode
of
variability
is
what
I'm
calling
the
redistribution
vector,
and
so
the
redistribution
vector
is
characterized
by
this
kind
of
early
spring
green
up
and
then
subsequently
a
summer
and
fall
decline
in
gpp.
So
there's
kind
of
positive
anomaly
in
the
spring,
followed
by
a
negative
anomaly
in
the
in
the
late
summer
and
fall
as
a
result
of
having
kind
of
balanced.
H
You
know
positive
and
negative
phases.
The
theta
value
is
closer
to
zero
and
for
these
grid
cells,
roughly
26
percent
of
the
variation
can
be
attributed
to
years
that
have
this
kind
of
characteristic
pattern
in
them.
So
moving
forward
I'll
kind
of
go,
go
through
characteristics
of
the
amplification
vector
redistribution
vector,
but
I'll
kind
of
keep
referring
to
this
figure,
because
I
find
it
useful
to
to
understand.
What's
going
on.
H
Perfect,
so
here
is
what
this
amplification
and
redistribution
vectors
look
like
you
know,
roughly
45
of
the
variants
globally
can
be
attributed
to
this
amplification,
vector
and
29
of
the
variants
can
be
attributed
to
the
redistribution
vector,
but
there
are
some
areas,
especially
in
kind
of
the
eastern
united
states,
much
of
central
europe
that
are
really
dominated
by
this
redistribution
vector,
which
is
something
I'll
come
back
to.
H
You
can
look
at
it
in
latitudinal
bands.
So
again,
we've
seen
this
one
before
it's.
It's
kind
of
characteristic,
it's
easier
to
see
at
mid-latitudes,
high
and
mid-latitudes,
but
even
the
tropics
shown
here
for
the
northern
hemisphere
in
the
southern
hemisphere
show
these
kind
of
characteristic
amplification
redistribution
vectors.
So
these
two
modes
of
variability
seem
pretty
pretty
pervasive
and
what
I
wanted
to
do
then
was
then
correlate
the
weights
that
come
out
of
the
sbd
analysis
with
seasonal
anomaly.
H
So
that's
what's
shown
here
is
a
high
correlation
between
the
svd
weights
and
the
jja
seasonal
anomalies
of
gpp.
So
this
is
just
kind
of
a
sniff
test
to
make
sure
things
are
working
right
and
then
I
wanted
to
look
at
anomalies
how
that
correlates
with
terrestrial
water
storage
and
temperature
anomalies,
and
so
there's
a
lot
to
look
at
here.
H
I'll
just
draw
your
eye
to
this
kind
of
the
summer
panel
and
what
you
see
is
that
the
amplification
vector
is
largely
driven
by
water,
so
correlation
positive
correlations
with
water
anomalies
and
mid
and
low
latitudes,
and
then
also
positive
correlations
with
temperature
in
the
high
latitudes.
So
not
surprising
if
you
warm
up
the
arctic,
you
do
see
these
positive
implication
vectors.
H
It
gets
more
interesting
for
redistribution,
and
so
redistribution
is
characterized
again
by
the
sound
of
early
summer
green
up
which
in
the
northern
hemisphere
is
most
strongly
correlated
with
warmer
springs.
So
that's
interesting,
you
get
you
know.
A
warm-up
in
the
spring
leaves
come
on
and
you
get
this
positive
anomaly,
but
then
the
negative
anomaly
is
shown
here
and
you
get
kind
of
negative
anomalies.
H
A
Well,
you've
got
about
two
minutes
left.
H
H
It
also
points
to
some
opportunities
of
doing
ecosystem
predictability
work
and
I
think
smile
would
be
an
opportunity
to
do
that
with
and
then
looking
at
other
models
and
observations
would
be
useful.
This
is
not
work
that
I
have
a
lot
of
time
to
do
so,
if
anyone's
interested
in
doing
it,
I
would
be
more
than
happy
to
share
the
code.
That's
already
on
github
and
talk
more
about
it.
H
Oh
sure
smile
is
the.
I
don't
remember
what
the
acronym
stands
for,
but
it's
this.
It's
part
of
the
eco
earth
system,
prediction
working
group
and
there's
these
series
of
two
year
long
ensemble,
initialized,
ensembles,
that
steve
yeager
is
running
seasonal.
C
Tamil
you're,
a
large
ensemble
thanks
nikki
gingy,
had
a
question.
Are
those
variables
water,
storage
and
temperature
at
the
bottom,
from
observations
or
from
cesm2.
H
C
C
H
I
think
the
fun
flexes
are
pretty
large,
especially
in
future
scenarios,
and
so
I
think
it
kind
of
I'm
hoping
that
with
the
parameter,
perturbation
experiment
that
dave
and
others
are
doing,
that
we
get
to
kind
of
explore
ways
that
we
can
redistribute
the
autotrophic
respiration
fluxes,
maybe
away
from
fun
into
some
of
the
other
components.
H
A
C
A
E
E
I
work
on
legs,
including
all
the
lakes,
towns
and
reservoirs
in
this
case,
because
they
are
very
important
machine
sources,
actually,
the
second
largest
natural
source
after
violence.
The
emissions
are
quite
disprove
disproportionate
compared
to
the
cup
the
surface
area.
Carbon
here
is
a
schematic
diagram
showing
liquid
methane
processes.
E
So
that's
why
evolution
can
be
the
dominant
pathway
in
many
lakes,
diffusion
and
evolution
are
the
most
studied
pathways
by
the
way,
also
another
import.
Another
important
reason
is
that
lakes
are
especially
abundant
in
the
northern
helites.
E
About
40
of
the
world's
lake
area.
Is
north
of
50
degree
north,
meaning
that
they'll
probably
undergo
great
impacts
by
climate
change
and
speaking
of
climate
impacts?
There
are
many
ways
that
this
can
work.
Most
directly.
Warming
will
enhance
microbial
activities,
while
methanogens
are
found
to
be
more
responsive
to
warming
than
methanol
troughs,
meaning
that
methane
production
will
increase
faster
than
oxidation
and
for
lakes
that
freeze
in
winter
they'll
have
longer
as
free
days,
meaning
longer
emission
periods.
E
Also,
there
are
impacts
related
to
permafrost,
thawing
and
except
from
that
climate.
Climate
warming
can
also
work
with
other
factors
such
as
lake
eutrophication,
meaning
that
they
are
getting
more
nutrient
rich
liqueutrification
will
exacerbate
the
warming
impacts.
I'm
just
making
several
examples
here
and
this
list
can
go
on
and
on
and
they're
all
potentially
very
important
topics.
E
So,
given
the
necessity
to
study
like
methane,
our
estimations
so
far
still
have
large
uncertainties,
actually
the
largest
among
all
natural
sources.
So
these
two
data
synthesis
studies
show
the
gap
between
top
down
and
bottom-up
estimations.
Top-Down
here
means
satellite
data,
inversion
and
bottom-up
means
modeling
or
statistical
extrapolation.
E
E
E
Well,
most
data
are
for
diffusive
fluxes
only
because
they're
easier
to
measure,
but,
as
I
mentioned,
evolution
can
be
the
dominant
pathway
and
also
most
observations
only
lasted
for
less
than
three
days
because
of
the
measurement
difficulties
which
is
far
from
enough
to
capture
the
lake's
annual
emission
potential,
and
there
are
many
measurements
for
very
small
lakes,
which
are,
of
course,
highly
productive
and
very
and
the
results
can
be
sensational,
but
when
it
comes
to
regional
or
global
estimations,
the
mapping
of
these
very
small
lakes
can
be
a
large
uncertainty
source.
E
E
An
intrinsic
issue
with
statistical
extrapolation
is
that
it's
not
spatially
or
temporally
explicit
and,
for
example,
for
northern
legs.
It
uses
an
estimated
at-free
length
for
all
legs
or
ballasts
to
either
way
causes
large
uncertainties
and
for
modeling.
Since
there
are
more
than
one
million
lakes
over
the
world,
the
process
can
be
very
time
consuming.
E
So
with
this
in
mind,
the
goal
of
this
study
is
using
a
modeling
approach
to
reduce
the
uncertainties
in
detonation
of
global
electromagnetic
emissions
and
to
make
projections
of
future
missions
under
different
climate
scenarios
now
so
to
investigate
the
spatial
distribution,
which
is
a
wonderful
thing
of
model
I'll
briefly
introduce
the
model
I'm
using
specifically
how
it
compares
to
cell
m
the
lake
model
in
clm
and
then
go
through
the
simulation
design
quickly,
because
the
modeling
details
are
not
what
we
want
to
get
into
here.
E
E
It's
a
participant
model
in
the
esme
ismi
project,
the
intersectoral
model,
intercomparison
project,
so
that
it
has
been
applied
in
several
studies
and
also
model
intercomparison
work.
The
two
models
have
share
many
similarities.
They
are
both
1d
process
based
and
they
are.
The
thermal
modules
are
quite
similar,
they're,
both
both
based
on
the
hostile
mode
model,
meaning
that
they
simulate
key
processes
like
turbulent
heat,
fluxes,
iv,
diffusivity
and
thermal
conductivity,
and
relief,
forcing
in
very
similar
ways.
E
The
paper
has
just
been
accepted
and
other
from
that,
in
terms
of
the
methane
model
cell
m
only
has
well.
The
the
mathematical
in
cell
m
is
mainly
aimed
for
valence,
which
are
in
some
way
similar
to
lakes,
but
also
have
many
different
mechanisms
from
lakes,
for
example,
the
gas
transportation
are
quite
different
and
also
prescribed
bulk
oxygen
concentrations
to
the
grades,
instead
of
simulating
them
explicitly
so
for
model
calibration
and
validation.
We
carefully
filtered
the
data
and
selected
those
like
for
northern
lakes.
E
E
There
was
a
lot
of
data
organization
work
and
finally,
we
simul.
We
ran
simulation
for
the
first
decade
and
the
last
decade
of
the
century
and
considered
two
climate
scenarios,
the
rcp,
4.5
and
8.5,
and
here
are
the
results
for
the
present.
We
simulate
a
mean
annual
emission
of
25.8
plus
minus
9.4
kilogram.
E
The
temperate
zone
will
become
the
largest
source
under
rcp
4.5
and
under
speed.
8.5,
the
arctic
and
subactive
region
will
also
exceed
the
tropics.
E
Comparing
to
the
previous
studies.
Our
resolution
of
present
emission
is
on
the
lower
end
of
these
ranges.
So
for
this
rows
and
twitter
paper
published
very
recently,
they
did
include
lakes
smaller
than
even
0.001
square
kilometers,
but
I
only
like
truncated
their
results
here.
Just
for
comparison,
our
result
is
in
some
way
more
reasonable
to
reconcile
with
the
top-down
admission,
while
there
are
definitely
limitations
there
and
I'll
go
to
that
shortly.
E
So
it
shows
that
the
subarctic
and
arctic
region
will
experience.
Thus
various
increase
in
methane
emission
around
60
higher
than
the
temperate
and
almost
more
than
triple
the
tropics,
and
this
region
is
also
the
most
sensitive
to
the
climate
intuitively
and
and
from
the
simulations
with
more
than
ten
percent
increase
with
one
degree
warming,
so
that
more
than
double
the
tropics.
E
Well,
despite
the
different
climate
impacts
on
different
regions,
the
spatial
distribution,
on
the
contrary,
is
not
changing
much.
So
here
are
the
global
mapping
and
the
latitudinal
distribution
of
methane
fluxes,
and
there
are
scenarios
the
mean
annu,
any
methane
flux
peaks
at
around
10
degree
north
and
decreases
towards
the
post
and
the
house.
Balls
on
the
maps
are
more
or
less
the
same.
E
So
here
is
a
summary
of
our
current
results.
We
estimate
and
predict
global
lag
methane
missions
for
the
first
and
last
decade
of
the
century.
E
We
found
that
the
tropics
is
the
largest
source
at
present,
but
the
temporal
arctic
and
starbucks
regions
will
become
more
important
in
the
future,
especially
the
arctic
and
subarctic
is
the
fastest
growing,
as
well
as
most
climate
sensitive
and
therefore
should
raise
more
attention
in
the
future
projections,
and
probably
more
studies
should
be
done
on
the
mechanisms
of
optical
lakes,
responding
to
climate
and
finally,
the
spatial
patterns
within
this
century
do
not
seem
to
be
largely
affected,
so
our
studies
have
some
has
many
limitations,
and
this
small,
like
issue
kind
of,
comes
to
the
floor
throughout
my
talk.
E
So
here's
the
thing
with
it,
the
smallest
lake
in
the
dataset,
we're
using,
is
0.1
square
kilometer,
there's
one
other
spatially,
explicit
dataset
called
global
water
bodies
or
glowable,
including
lakes
as
small
as
point
zero,
zero
two
square
kilometer,
but
to
resolve
these
leaks
on
satellite
images
can
lead
to
pretty
large
uncertainties
and
also
that
data
does
not
include
like
depth
information.
E
So
we
still
want
to
hedge
your
legs
for
our
purpose
and
in
terms
of
the
statistical
extrapolations
studies
mentioned
previously,
they
included
even
smaller
lakes
and
the
total
surface
area,
and
abundance
of
these
leaks
obviously
cannot
be
resolved
on
satellite
images.
So
the
relationship
is
obtained
using
a
parietal
distribution.
That's
shown
here.
E
E
For
example,
like
all
lakes,
without
traffic
status,
data
are
considered,
oligotrophic,
meaning
that
they
are
neutron,
poor
and
thus
less
productive.
That
means
that
our
estimation
is
potentially
conservative
and
well.
There
are,
of
course,
other
limitations
and
we're.
I
was
trying
to
improve
our
study
and
experiment
design,
so
this
is
my
last
slide
so
for
future
work.
We
want
to
further
investigate
like
methane
mission
under
climate
change.
E
We
plan
to
like
other
impact
factors
due
to
climate
change.
For
example,
the
promote
force
thaline
has
been
shown
to
be
significant.
They
will
increase
the
sediment
carbon
content
due
to
the
input
of
all
the
organic
carbon
and
also
permafrost
affected
lakes.
Will
experience
drainage
or
expansion
relatively
rapidly,
so
that
will
affect
the
total
emission
largely
also,
and
also
as
mentioned,
electrification
can
be
another
potential
topic,
so
these
all
require
more
data
collection
and
modeling
and
yeah.
That's
the
last
slide
of
my
talk
and
thank
you
all
for
listening.
C
C
C
And
I
actually
have
a
question:
you
said
that
your
prediction
for
the
future
or
projection
for
the
future
is
that
the
spatial
pattern
of
emissions
wouldn't
change
very
much.
But
you
also
mentioned
that
you
expected
the
tropics
to
become
relatively
less
important
or
maybe
the
emissions
to
become
relatively
higher
at
high
latitudes.
And
so
I
was
just
wondering
if
you
could
reconcile
those
two
statements,
because
it
seems
like
if
you
have
a
shift
from
one
region
to
another
you'd
get
a
shift
in
the
spatial
pattern.
E
Let's
go
to
this
one,
so
I
want
to
mention
that
the
optics
will
exceed
will
become
more
important.
I
meant
a
total
emission
like
adding
up
over
the
whole
region,
but
in
terms
of
the
spatial
distribution.
E
If
you
look
at
latitudinal
distribution
curves
the
flux
which
is
in
the
unit
of
metamorph
per
square
meter,
the
pattern
of
that
still
stays
more
or
less
unchanged,
so
meaning
that
for
tropical
lakes,
they
will
still
have
larger,
larger
emission
per
square
meter.
But
if
you
are
adding
up
adding
up
emission
for
the
whole
region,
they're
becoming
less
important
than
the
arctic,
does
it
make
sense
to
you.
C
A
Yep,
my
question
is:
is
trying
to
understand
how
you
initialize
the
model
like
where,
where
does
the
organic
matter
come
from
to
generate
the
methane
emissions?
Are
you
is
that
a
prognostic
quantity
or
is
it
a
prescribed
quantity.
E
Yeah,
that's
that's
pretty
scrapped.
We
collected
data
from
previous
studies
like
in
like
measurements
on
lake
sediment,
organic
carbon
content
and
like
we
assigned
different
organic
carbon
content
to
lakes
in
different
regions
like
for
yadona
lakes,
there
was
a
value
and
for
thermal
cost
legs
there
was
a
value
and
in
that,
in
that
fashion,
yeah
we
use
a
prescribed
value
for
each
leak.
E
A
E
Yeah,
that's
a
very
good
question
because
the
model
can
do
that.
There
is
there.
Are
those
processes
like
a
cabin
boreal
and
like
yeah
the
carbon
dynamics
in
the
model,
but
well
this
comes
to
the
parameter
issue
because
I'm
just
not
sure
we
have
enough
data
to
bracket
these
parameter
value
values
involved
in
the
in
those
processes.
E
So
I
just
decided
that
probably
that's
a
problem
for
future,
we'll
just
use
the
prescribed
values
for
now
yeah
makes
sense.
C
C
I
Okay,
guys
really
glad
to
present
some
results
from
the
model
we
really
developed.
It's
not
really
developed.
It's
more
like
we
improve
a
little
bit
based
on
crm
model,
so
this
model
we
call
single
micro.
You
know
the
name
is
really
emphasize
the
microbial
roles
on
the
biochemical
processes,
so
particularly
in
the
current
version
we
emphasize
on
the
soil
carbon
table
and
also
on
the
methane
dynamic.
So
that's
why
you
see
the
title
here
is
c
milliliter
microbial
mechanism
for
soil,
carbon
and
methane
dynamic
mix
with
a
cm
mac
model.
I
I
So
this
is
the
one
of
the
report
came
out
about
two
years
about
10
years
ago
is
2011.
So
this
is
really
a
report
by
the
american
academy
of
microbiology,
so
emphasize
the
incorporating
microbial
process
into
the
clan
model.
So
then,
after
that
number
model
has
been
developed,
one
of
these
you
see
here
is
alison,
steve,
addison,
mimic
and
bellwether
and
corbus
manned
resolves
many
different
models
and
also
one
of
the
models
about
really
published
by
josh.
It's
about
2003,
so
josh
always
mentioned
that
it's
a
toy
version
model,
but
it's
more
like
the
first.
I
I
think
the
very
explain
explicitly
represent
the
microbial
processes.
That
is
one.
So
what
I
present
here
is
is
we
call
stem
michael,
so
you
see
the
name.
So
that
is
really
so
you
see
this
is
a
structure.
So
all
those
what
I
present
here
is
the
biochemical
part,
because
this
model
is
really
built
on
the
default
serum
4.5.
I
So
all
the
hydrology
thermal
plants
are
exact
same
so
with
the
cm
4.5.
So
what
we
did
is
really
based
on
the
biochemical
part.
We
added
two
parts:
one
is
the
biochemical
selection,
so
we
add
bacterial
fungi
and
dissolve
organic
matter
and
also
for
the.
We
also
add
another
called
the
microbial
function
group
based
metamodel.
I
So
so
this
is
the
really
summarize
the
component.
We
have
been
revised
or
some
revisions
we
have
done
on
the
4.5,
so
the
bacterial
part
bacteria
fungi.
So
we
separate
two
fungal
groups
and
also
these
are
organic
matter.
This
is
for
the
better
chemical
and
also
the
methane
model
I
want
to
emphasize
is.
I
We
did
have
a
very
kind
of
like
explicit
variable
here,
so
it
includes
dissolve
organic
matter,
so
acid
acid,
hydrogen
seed,
classic
methyl
genesis,
hydrotropic,
methyl
genesis,
anaerobic
methane
and
oxidation
and
anaerobic
methane
oxidation
and
for
each
layers
we
have
all
those
processes
and
also
for
the
three
paths
with
for
the
methane
transport
include
the
plant
diffusion
plant,
media
transport,
diffusion
evolution
and
also
another
things
I
revision
we
have
done
is
so
because
current
so
column
based
is
something
it
sometimes
cannot
works.
I
Well,
so
we
reorganize
the
chemical
a
little
bit,
so
we
use
the
predominant
pft
type
to
parameter
some
microbial
soil
processes
in
that
way,
for
example,
the
defaulting
factor.
So
this
is
we
really
parameterize
to
make
sure
that
is
each
plant
function.
Tab
is
individual
value,
so
in
that
way
we
can
much
better
to
simulate
the
sort
of
carbon
distributions
and
also
we
bring
some
soil
parameters
and
all
the
microbial
parameters
out
into
the
parameter
files
and
the
second
to
events.
I
I
So
in
the
next
two
slides,
I
don't
want
it's
a
lot.
It
really
shows.
I
showed
little
bit
the
what
we
do,
the
approach.
We
call
the
microbial
micro
economic
approach
to
improve
the
model.
So
for
these
two
slides,
I
don't
want
to
see
a
lot
release.
This
is
just
show
the
coach
we
use.
So
one
thing
I
don't
emphasize
we
in
this
approach
and
also
in
a
later
approach.
We
divided
microbial
mechanism.
We
tried
to
put
that
maximum
in
the
serum
we
included
disposal.
I
Environmental
filtering
like
could
be
response,
diversification
and
low
extinction.
So
those
is
the
mechanism
we're
trying
to
further
improve
the
cm
microbus
models.
So
then,
in
the
next
few
slides
I
try
to
introduce
from
two
parts.
So
one
is
the
better
chemistry
the
more
like
the
carbon,
so
carbon
dynamic
and
second
one
is
the
methane
dynamics.
So
this
is
the
first
part.
So
first
part
here
is
that
we
really
emphasize
how
the
bacteria
and
fungi
really
regulate
to
the
carbon
flow.
I
So
you
you
guys
know
that
in
this
deforestation
version,
so
we
have
all
those
cascade
right
from
each
pose
we
have
upstream.
So
I
think
this
is
the
actually
covering
the
cloud.
This
figure
worked
very
well.
It's
really
show
that
how
this
flow
from
upstream
to
the
downstream
so
for
each
individual
flow
from
up
stream
to
downstream
we
added
bacteria
control
and
the
fungal
control
element,
and
for
each
of
those
we
have
different
scene
ratio,
and
then
they
have
different
physiology
properties
right
because
they
have
different
physiology.
I
So
then,
when
this
carbon
flow
through
the
bacteria
and
fungi,
so
you
have
the
co2
release,
then
you
can
simulate
how
this
microbial
function.
Communication
change
has
impact
on
these
flows
and
also
for
the
disorder
matter,
so
here
is
really
sure
that
how
the
actual
fungi
can
use
this
organ
matter
and
then
when
it's
respirated
reduce
the
co2
back.
So
this
is
the
general
structure
for
that.
I
So
with
that,
so
I
think
start
two
or
three
years
ago
the
u.s
has
started
trying
to
use
this
model
and
we
collect
a
lot
of
data
from
different
sites
across
global.
We
choose
non-natural
bounds
for
each
bomb.
We
select
more
than
two
sides
more
than
two
sets.
Those
is
the
sign
name.
So
every
little
information
you
can
find
these
papers.
So
all
those
different
side.
We
have
one
side
for
the
calibration
and
one
side
for
validation,
so
all
those
represent
individual
set
and
on
the
left.
I
So
this
column
left
side
is
for
the
the
model.
Simulate
fungal
biomass
on
the
right
side
is
model
simulate
bacterial
bowel
mass
so
for
each
set.
I
do
want
to
allow
you
guys
more
like
know
that
is
you
see
here
so
the
different
color
of
that
represents
your
so
that
we
change
the
skill
a
little
bit
to
match
the
value,
but
the
value
that
the
skill
of
the
x
axis
is
different,
because
the
key
reason
for
that
is
because
the
microbial
diameter
is
known
is
largely
larger
compared
with
the
existing
variable.
I
So
we
try
to
keep
this
to
make
a
comparison,
but
you
see
the
value
are
different,
so
this
is.
I
think
this
is
a
urgent
issue
for
the
community
if
you
are
able
to
better
simulate
the
recovery
rivals.
So
this
is
the
distribution.
Then
this
is
the
scatter
plot
to
show
that
if
we
put
all
those
for
each
bomb,
so
you
see
roughly
so
it's
a
comparison
here
so
for
the
the
model
can
explain
about
70
percent
of
the
biomass
and
about
26
of
the
bacteria
biomass
on
average.
I
So
this
is
not
very
good,
but
roughly
we
are
able
to
reach.
This
is
the
current
accuracy
for
the
model.
So
after
that
we
did
after
we
have
used
the
model
to
simulate
the
bacteria
and
fungal
dynamic.
Then
we
further
look
like.
So
how
can
we
use
this
model
to
simulate
the
carbon
processes
so
in
the
next
kind
of
a
few
slides?
So
we
will
emphasize
model
simulator
the
macro
respiration,
so
all
those
different
sets
out.
So
you
see
so
each
panel
here
represents
as
one
side.
I
So
the
the
red
star
represents
the
observation
data.
So
the
black
dot
is
the
model
simulate
so
for
many
sites,
so
the
name
of
the
site
represents
different
site
and
the
bounds
you
see
on
the
bottom.
So
this
is
micro
respiration.
So
then
this
is
that
we
also
further
evaluate
how
the
microbial
seasonal
variations
can
have
impact
on
the
different
carbon
fluxes.
So
this
is
the
respiration,
so
bacteria,
this
is
the
biomass
of
the
lecture.
So
I
want
to
let
you
guys
know
that
is
here
so
the
different
color
of
that
is.
I
We
use
two
version
models,
so
one
version
models.
We
use
the
default
cmq
model
and
this
red
color
represented
model.
We
remove
the
seasonal
variation
of
the
microbial
control,
particularly
on
the
temperature
and
moisture,
so
we
use
any
average
of
the
temperature
moisture
and
to
control
the
microbial
biomass,
scroll
and
dice,
and
everything
else
is
same,
so
this
is
in
this
way.
We
compare
these
two
virtual
model.
Then
we
can
further
evaluate
how
the
microbial
seasonal
variations
has
impact
on
the
carbon
emission.
So
so
this
is
biomass.
I
You
see
so
most
of
them
most
of
the
balance,
so
the
seasonal
variation
did
have
a
smaller
see,
a
smaller
biomass
of
the
bacteria
and
fungal.
So
this
is
respiration,
and
this
is
the
carbon
loss
from
bacterial
fungi.
So
these
laws
eventually
represent
the
microbial
vibration
right
because,
in
the
current
version,
all
carbon
release
from
soil
to
atmosphere.
It
must
be
of
the
bacteria
and
found
it.
So
this
is
the
cell,
so
the
valley
is
the
results.
Innovation,
black
is
the
width
synthetic.
I
So
you
see
the
different
this
carbon
loss
and
this
is
coming
again
from
the
fungal
and
vector
right.
So
you
see
the
difference
due
to
the
time
limitation.
I
don't
want
to
go
to
detail,
but
those
results
we
show
here.
Then
we
further
to
estimate.
So
if
we
add
everything
together,
so
how
about
the
microbial
respiration
and
then
what
we
got
is
this
one
microbial
respiration
and
carbon
storage?
I
So
then,
really,
if
you
compare
this
so
the
original
model
and
model
without
season
variation,
so
we
did
find
that
when
we
remove
the
seas90,
the
respiration
is
declined
and
then
so
organic
carbon
increase.
In
this
we
compare,
we
found
that
the
microbial
signality
overall
is
stimulated
microwave
activity
and
then
lead
to
higher
respiration
and
the
lower
little
lower
spoken
tumor.
So
this
is
the
founder
of
this
study.
I
So
this
is
some
prelim
results.
We
are
currently
working
on
try
to
use
this
model
to
simulate
global
dispute
of
the
different
variables.
So
do
you
see
so
organic
microbial
respiration?
So
this
is
the
along
the
liquid
distribution
and
the
right
panels
show
the
spatial
distribution
of
different
variables
so
bacterial
fungi
and
dissolve
organic
matters
distributions
okay.
So
I
have
a
little
bit
time
to.
I
want
to
introduce
the
method
model
and
some
application
results.
I
So
this
is
the
model
structure
so
similar
with
the
previous
version,
but
we,
we
add
a
little
more
content
here.
How
can
we
use
the
current
version
to
link
with
different
field
measurements
so
profile
that
these
different
variables
do
you
see
as
the
s8
and
the
co2
methane
oxygen,
and
also?
How
can
we
use
this
to
connect
with
the
genetic
information?
So
someone
has
been
done.
Some
are
still
ongoing
work,
okay,
so
this
is
the
some
of
the
work
done
by
another
student
yui,
so
she
work
on
this
model
and
simulate
the
use.
I
This
model
simulate
the
methane
fluxes
and
net
excess
exchange,
and
also
excellent
respiration
at
arctic
for
the
different
length
types.
So
this
is
the
trough.
This
is
for
the
low
center
polygon
center.
This
is
the
lowest
program
rims,
so
for
the
different
landscape.
We
did
have
this
model
the
simulation
of
the
methane
fluxes.
I
So
this
is
for
the
either
we
have
the
same
tabs.
So
this
is
compared
with
operation.
So
then
we
further
use
this
to
simulate
different
types
right.
You
see
so
we
do
have
different
values,
different
patterns,
since
the
pattern
of
the
different
landscape
types.
This
has
been
published
two
years
ago
after
we
done
this
one.
So
we
further
committed
on
the
side
level.
Can
we
do
the
upscaling
to
see
how
this
plot
level
measurement
by
stand?
I
How
can
we
upskill
to
the
ecommerce
tower
domains?
So
this
work
is
just
done
about
one
month
ago.
So
then
this
is
the
result
for
this
result,
so
we
use
the
model
by
using
the
cm
micro
model
and
also
we
come
this
model
with
the
three
type
of
the
flip
flop
model.
The
three
for
the
first
marble
is
this
three,
so,
first
one
the
homogeneous
footprint
model
and
gradient
for
the
model
and
the
dynamic
different
model.
So
the.
I
Okay
thanks,
so
the
difference
between
three
food
models,
the
homogeneous
model
we
assume
so
in
a
circle.
Distance
from
the
tower
every
grid
is
have
exact
same
confusion.
So
that's
why
we
call
homogeneous
the
green
infinitive
models.
We
have
the
side
the
tower
in
the
center,
but
the
distance,
the
contribution
to
the
term
element
is
inverse
to
the
distance.
So
if
we
move
away
from
the
tower,
the
contribution
is
slower
is
lower.
So
this
is
the
contribution
for
the
the
different
grid
to
the
e-commerce
method,
fluxes
and
then
dynamic.
I
I
So
then
this
is
for
one
side,
but
then
we
have
five
sides
so
in
the
arctic,
the
five
side
so
and
also
for
the
different
years
different
bonds,
because
some
months
we
don't
have
data.
So
that's
why
you
only
see
few
months
and
particularly
of
those
months,
is
in
the
summer
or
in
the
full
season,
but
in
the
winter
season,
many
of
those
data
available.
So
we
use
those
to
compare
with
our
visions.
So
roughly
so
we
can
see
that
the
homogeneous
and
gluten
are
dynamic.
I
Most
of
those
months,
the
dynamic
model,
combined
with
the
cm
micro
model.
We
have
much
better
permanent
performance
to
simulate
the
methane
dynamic.
Even
some
months
we
don't
have.
The
dynamic
model
is
not
best,
but
still
it
promotes,
promotes
performance
much
better
right.
So
you
see
the
r
square
here
is
still
very
high,
so
those
and
many
other
side,
for
example
like
the
may
of
2011,
so
the
dynamic
model
is
much
better
compared
with
other
tools
so
anyway.
So
this
is
the
the
key
conclusion
for
this
space.
I
We
are
dynamic,
photography,
model
plus
the
cmake
model.
We
are
able
to
much
better
to
simulate
the
methane
fluxes,
the
electronics
domain
top
domains.
I
Okay,
so
I
think
that's
all
the
content,
some
ongoing
workers.
We
are
still
working
on
this
trying
to
get
more
data
to
further
improve
model
and
also
test
model
for
different
microbial
mechanism
and
also
further
to
evaluate
all
the
microbial
contribute
to
the
land
asthma
feedback,
particularly
on
the
co2
and
methane
and
the
network
that
is,
we
are
still
working
on
it,
but
countries
particularly
on
the
method
and
the
co2
practices.
I
So
I
also
want
to
acknowledge
the
contribution
from
the
people
over
the
past
few
years,
so
many
or
student
has
contributed
to
the
data.
Convenient
model
also
contribute
your
client
data
to
use
data
to
validate
them.
Okay.
So
that's
all
thanks.
A
Yep
there's
about
two
and
a
half
minutes
left.
I
Thanks
will
so
this
in
this
current
version,
so
with
the
bacteria
and
fungals,
the
contribution
for
the
dynamic
is
particularly
on
the
environment
factor
so
because
each
of
those
we
simulate
the
carbon
assimilation
and
also
carbon
respiration
right
then,
and
also
this
is
for
the
substrate
environment
factors,
so
temperature
volatile
control
of
the
growth
size,
increase
rate,
that's
rate
and
also
substitute
how
much
carbon?
How
much
do
you
see
as
available
in
the
environment
so
all
those
together
to
really
control
the
carbon
valid
abundance
of
the
bacteria
and
fungal
yeah.
H
I
I
think
yes,
so
this
is
the
fungal
biomass,
bacterial
biomass.
You
see
here
right
so
the
general
pattern
we
do
have
that,
but
exactly
value
we're
still
looking
at
the
details,
matters
yeah,
so
I
do
want
to
emphasize
once
this.
It
represents
the
model
simulator
and
also
observation
data.
So
this
observation
data
is,
we
collect
a
global
kind
of
meta-analysis.
J
C
A
F
C
J
J
Go
ahead
all
right!
Thank
you.
So
just
sharing
some
work
about
a
data
assimilation
work
that
we're
performing
with
clm
across
the
western
united
states.
This
is
a
collaboration.
There
has
been
a
collaboration
with
the
university
of
utah
and
was
funded
from
nasa's
carbon
monitoring
system
grant,
and
the
timing
is
good.
This
was
just
accepted
in
the
journal
james
this
week.
Next
slide.
J
So
yes,
so
why
monitor
carbon
in
the
western
united
states-
and
I
this
has
to
do
with
the
substantial
carbon
resources
in
the
western
united
states,
but
also
the
vulnerability
of
our
forests
susceptible
to
forest
fire
insect
infestation
in
drought
mortality,
and
so
I
just
put
up
the
us
drought
monitor
for
this
for
last
week.
J
It's
probably
actually
worse
than
this
now
because
of
the
extreme
heat,
but
this
is
a
particular
early
season
for
exceptional
drought
to
occur
in
the
western
united
states,
and
a
little
background
of
2012
through
2015
was
a
particularly
bad
drought
period
that
caused
kind
of
large
areas
of
mortality,
and
so
this
is
a
bit
ominous
to
get
this
sort
of
drought
this
early
on
in
the
season.
So
there's
the
motivation
for
moderating
the
carbon
and
the
carbon
flux.
J
The
challenge
is
that,
given
the
the
mountainous
terrain,
this
is
challenging
to
monitor
with
traditional
carbon,
carbon
monitoring
techniques
like
flux,
towers
or
atmospheric
versions,
and
advanced
that
advance.
The
slide.
J
There
we
go,
and
so
so
it's
difficult
to
do
the
traditional
flux
monitoring
through
top
down
atmospheric
inversions,
and
so
what
we're
doing
here
is
we're
using
the
bottom-up
modeling
approach,
using
a
clm
as
a
starting
point,
but
then
bringing
in
remotely
sense
remotely
sense
observations
to
help
improve
it.
Next
slide.
J
It's
a
quick
overview
on
the
left.
The
cartoon
is
essentially
showing
how
the
assimilation
setup
works
with
with
dart,
and
so
we
drive
clm,
create
80
different
instances
of
clm
through
80
member
ensemble
of
cam
4,
dart
reanalysis,
so
80
equally,
like
likely
realizations
of
the
atmosphere,
and
we
use
that
to
create
the
model
spread.
J
Specifically,
we
don't
update
all
variables,
but
all
variables
are
influenced
due
to
the
internal
mechanisms
within
clm,
so
so
carbon,
so
water
and
the
fluxes
are
changing
because
we're
updating
the
state
and
so
a
more
specific
way
to
look
at
this.
On
the
right
hand,
side
of
the
slide
going
from
the
bottom
up,
so
we're
taking
in
we're
estimating
the
biomass
observed
prior
state
in
the
green
text
there
or
we're
getting
that
through
a
forward
operator
that
assembles
the
estimated
biomass
by
looking
at
the
fundamental
variables
like
life
stem
carbon.
J
J
And
then
we
take
that
increment
and
regress
that,
upon
the
other
unobserved
variables
within
the
clm
model
state,
and
so
we
take
that
and
then
we
apply
it
to
and
I'm
showing
a
grid
cell
up
there.
So
we
apply
all
those
updates
to
all
the
individual
pfts
and
columns
in
the
restart
file
of
clm.
So
some
pretty
heavy
lifting
considering
how
complex
clm
is
next
slide,
so
just
a
little
bit
more
about
the
methods.
J
So
we
we
create
a
single
instance
spin
up
kind
of
similar
to
what
a
lot
of
initial
initialization
to
get
the
soil
carbon
and
to
get
the
initial
biomass
correct.
We
use
a
one
degree
grid
for
this.
One
of
the
unique
things
we
probably
do
is
to
use
the
grid
met,
met,
forcing
which
is
specific
for
the
western
united
states.
That
removes
a
lot
of
biases
in
warm
temperatures
and
low
precipitation
that
you
get
for
a
lot
of
data
atmospheres
for
the
western
united
states,
and
then
we
transition
into
the
assimilation
run.
J
The
time
window
is
from
1998
to
2011,
and
we
do
something
that
we
call
looping,
which
means
we
cycle
over
the
assimilation
window
three
times
in
sec
in
succession,
so
there's
three
kind
of
the
the
red
lines
right
there
and
that
better.
It
gets
us
from
the
free
simulation,
using
no
observations
and
closer
to
the
observations
in
a
more
gradual
pattern.
J
We
also
use
adaptive
inflation,
which
essentially
is
just
a
technique
to
make
sure
that
we
create
or
maintain
the
correct
ensemble
spread
between
all
our
different
80
realizations
during
the
assimilation
process
to
make
sure
that
it
doesn't
doesn't
collapse
on
us
next
slide,
and
so,
as
I
mentioned
before,
we're
using
monthly
above-ground
biomass
observations
and
leaf
area.
I
use
the
observation
term
loosely.
These
are
really
data
products,
so
these
are
really
like
spectral
reflectances
that
are
then
ground
truth
to
get
the
land
surface
property
that
we
care
about
like
leaf
area.
J
We
use
an
observation,
rejection
threshold
of
three
sigma
such
that
we
can
remove
observations
that
are
too
far
removed
from
what
clm
is
telling
us,
and
this
is
to
protect
against
systematic
biases
that
may
be
unaccounted
for
in
the
uncertainty
and
then,
finally,
we
can
localize
both
in
space.
So
what
realm
of
influence
the
observations
have
on
the
model
and
also
we
localize
in
state
space?
J
So
we
can't
adjust
all
500
variables
all
columns,
all
pfts
at
once,
so
we
have
to
make
a
choice
of
what
are
what
is
most
related
to
dictating
the
carbon
cycle,
and
so
here
I've
just
given
some
of
the
fundamental
state
variables
that
that
we
adjust
both
related
to
the
carbon
in
the
biomass
and
the
nitrogen
in
the
biomass
next
slide
and
so
finally
to
the
results.
And
so
when
we
include
these
remotely
sense
observations
within
clm.
J
We
show
about
a
30
reduction
in
the
biomass
and
the
leaf
area
with
those
plots
above
and
so
you
can
see,
the
leaf
area
is
getting
closer
from
from
the
black,
from
the
free
simulation
to
the
blue
observations
and
the
dynamic
nature
of
it.
The
biomass
takes
a
little
bit
longer
and
takes
the
three
loop
period
before
it
gets
more
similar
to
the
observations.
J
Even
though
we're
reducing
dramatically
the
leaf
year
in
the
biomass,
the
main
flux
is
offset,
so
the
cumulative
nep
or
the
net
carbon
uptake
over
the
western
u.s
doesn't
really
change
from
the
free
run
to
the
assimilated
run.
And
so
we
show
that
in
the
bottom,
in
the
bottom
figure
right
there,
the
component
fluxes
are
decreasing
but
kind
of
offsetting
so
that
the
net
carbon
uptake
isn't
really
changing
next
slide,
and
so
just
more
diagnostics
to
understand
how
the
assimilation
system
is
working.
J
What
we
typically
do
is
look
at
acceptance
rate,
and
so
those
are
kind
of
the
the
marks
in
the
pink
or
the
salmon
colors.
So
the
top
is
the
observations.
The
total
observations
possible
and
then
also
shown
in
the
observations,
assimilated
and
the
idea
generally,
is
to
to
maximize
the
number
of
observation
observations
to
assimilate,
rejecting
those
that
are
that
are
just
too
far
away
and
looking
at
the
root
mean
square
error.
J
We
definitely
want
it
to
go
next
slide,
so
I've
just
shown
a
kind
of
like
regional
grid
cell
average
behavior.
So
what
are
the
individual
pfts
doing?
So?
I've
just
shown
the
the
major
pfts
that
make
up
the
region
between
the
temper
and
boreal
forest,
and
so
we
can
show,
as
we
would
expect
from
leaf
area
in
the
top
left.
There.
We
see
reductions
in
the
leaf
area
and
also
an
increase
in
amplitude
in
the
red
and
the
assimilated
lines
as
well.
J
One
thing
we
do
notice
is
that,
although
the
natural
vegetation
types
are
fairly
amenable
to
being
adjusted,
the
the
crops
are
not
the
crops
generally
and
I'm
showing
an
example
where
it's
not
really
budging
at
all,
but
generally
they're
more
difficult
to
adjust,
and
this
is
presumably
because
of
the
way
that
crops
are
on
tighter
trajectories
as
far
as
the
timing
of
their
growth
and
senescence
senescence
or
harvest,
and
so
we're
thinking
of
ways
to
get
around
get
around
this
I'll
talk
about
later
next
slide,
and
so
how
does
clm
five
dart
r
simulation
run?
J
J
And
so
you
could
see
the
machine
learning
approach.
It's
basically
saying
that
this
area
is
showing
a
lot
of
carbon
uptake
where
clm5
dart
is
saying:
no,
it's
more.
It's
a
weak
carbon
uptake
or
more
neutral,
and
that's
regardless,
if
it's
the
free
run
of
the
assimilated
run,
and
so
the
reasons
for
this
could
be
the
way
that
disturbance
history
is
accounted
for.
The
machine
learning
approach
doesn't
necessarily
account
for,
like
the
site,
history
or
disturbance
history.
J
J
So
but
it's
important
here
not
to
say
like
which
one
is
correct,
I'm
just
kind
of
showing
these
kind
of
as
bookend
estimates
of
of
the
carbon
uptake
next
slide.
J
J
The
the
assimilation
is
just
is
fairly
neutral
through
the
for
the
entire
period,
and
I
should
say
this
is
1998-2011,
so
the
severe
drought
period
from
2012
to
2015
hasn't
come
on
quite
yet,
and
so
we're
a
bit
skeptical
of
this,
or
at
least
wanted
to
probe
this
a
little
bit
more.
So
we
kind
of
looked
at
a
little
bit
more
diagnostics
next
slide.
J
And
so
this
is,
of
course,
a
water
limited
area,
the
western
united
states.
So
I
looked
at
this
one
of
the
soil,
moisture
limitation
diagnostics
in
sealand,
5,
the
highest
water
limitation
with
the
darkest
red
colors
and,
as
you
might
expect,
it's
mostly
water,
limited
within
the
great
basin
and,
to
some
extent
the
higher
elevations.
J
What
we
know
for
sure
is
that
it's
definitely
influencing
the
spatial
pattern
of
the
gpp
in
terms
of
correlation,
and
so
we
know
that
it's
still
it's
having
an
impact,
and
one
of
the
reasons.
Why
is
that
snow
is
definitely
one
of
the
key
water
cycling
variables
in
the
western
united
states,
so
at
least
the
modeled
snow
that
we
have
and
I'm
showing
that
in
the
bottom
left
the
spring
snow,
water,
equivalent
or
sweet
it's
biased,
low.
J
We
also
adjusted
to
look
at
the
sensitivity
of
the
assimilation
to
the
various
adjusted
variables,
so
I
show
those
kind
of
the
circled
color
boxes
there
from
top
to
bottom,
using
a
minimum
amount
of
adjusted
variables
to
more
of
a
maximum
amount
of
adjusted
variables,
and
I
show
the
adjusted
variable
behavior
on
the
left,
column
and
kind
of
some
of
the
other
variables
on
the
right,
and
you
can
see
the
adjusted
variables
are
pretty
consistent,
regardless
of
kind
of
the
flavor
of
the
simulation
that
I'm
doing
here.
What
really
starts
to
vary.
J
There
are
variables
such
as
the
soil,
carbon
or
root
carbon,
because
they're
they're
not
explicitly
included
in
the
adjusted
state
variables,
and
so
this
has
an
impact
on
the
cumulative
nep
on
the
bottom
of
the
slide,
showing
some
spread
there.
Although
it
still
indicates
it's
still
a
weak
carbon
sink
and
does
not
agree
with
the
with
the
strong
uptake
of
the
fluxcom
estimate.
J
Okay,
cool
next
slide,
so
just
reviewing
some
key
points
so
assimilating
the
observations
of
biomass
and
leaf
area,
although
it
does
reduce
the
biomass
and
gives
us
what
we
think
is
a
more
reliable
state
of
the
system,
it
doesn't
really
change
the
carbon
uptake
and
and
furthermore,
when
we
kind
of
probed
some
of
the
different
assimilation
settings
that
that
kind
of
weak
carbon
uptake
was
was
fairly
robust.
J
Another
estimate
of
carbon
uptake
flexcom
was
much
more,
was
much
stronger
in
this
regional
carbon
uptake,
and
we
think
that
could
be
reasons
due
to
the
disturbance
history.
The
methods
of
the
different
types
of
approaches
and
kind
of
necessitating
the
need
for
including
more
water,
cycling
variables
within
our
seal
and
five
dart
assimilation
next
slide.
J
So
where
can
we
go
from
from
here?
And
so
I
talked
about
the
water,
cycling
variables,
and
so
one
of
the
cool
things
of
the
western
u.s
is.
We
do
have
snotel
sites
that
measure
the
stonewater
equivalent,
so
it
is.
It
is
possible
to
use
those
as
observations
to
directly
update
the
snow
layers
in
clm
and
the
different
layers
of
soil
water
in
the
subsurface.
J
There's
been
an
explosion
in
solar,
induced
fluorescence
measurements
across
the
western
united
states,
both
increasing
temperature
resolution
and
in
time,
and
so
those
have
a
strong
empirical
relationship
with
gpp
and
so
we're,
including
infrastructure
and
dart,
to
be
able
to
ingest
those
observations
right
now.
Another
approach
we
could
take
is
is
really
leveraging
more
high
resolution.
J
Land
cover
maps
that
give
more
metadata
more
specific,
more
specificity
to
what
the
biomass
observations
are
actually
telling
us.
So
it
could
allow
us
to
map
specific
adjustments
to
the
pfts
and
so
getting
back
to
the
to
the
crop
pft.
We
couldn't
get
it
adjusted.
If
we
could
apply
a
smarter
forward
operator,
we
could
get
more
specificity
in
our
adjustment
directly
to
to
the
crop
and
then
next
the
final
spatial
resolution.
I
said
you
might
be
asking
like:
why?
Don't
you
just
go
to
a
super
fine
spatial
resolution?
J
J
But
what
we
found
in
work
that-
and
I
don't
have
time
to
show
now-
is
once
you
get
below
one
degree
resolution.
It
doesn't
really
give
you
much
improvement
in
the
carbon
dynamics,
and
that
could,
because
you
know
we're
missing
fundamental
representation
that
that
could
be
represented
in
in
the
hill
slope
hydrology
model
and
also
both
in
the
hill
slope
hydrology
and
then
also
I'll
forget
what
I
was
going
to
say,
but
oh
and
also
parameter
estimation,
and
so
we've
also
shown
that
it
could
be
more
important
instead
of
spatial
physical
resolution.
J
Ecological
resolution,
meaning
better
plant
functional
type
representation,
is
still
important
in
this
area,
and
so
that
brings
to
the
last
point
on
the
bottom
right
that
I've
talked
about
the
data
simulation
research,
the
ensembl
common
filter
in
terms
of
state
adjustments.
So
things
like
biomass
and
leaf
area.
J
But
it
is
possible
to
use
it
in
kind
of
a
parameter
estimation
mode,
where
it
kind
of
has
a
more
like
a
stronger
impression
on
the
model
where,
if
you
update
the
state,
if
it's
fine
as
long
as
you
still
have
data
to
keep
on
updating
it,
but
eventually
the
model
could
lose
memory
of
it
and
then,
and
so
a
more
kind
of
fixed
improvement
could
be
through
a
parameter.
Estimation
approach
next
slide.
A
J
A
J
Got
like
30
seconds
left
sure
perfect
yeah.
This
slide
got
a
little
messed
up,
but
anyway,
this
is
just
to
show
all
the
different
models
that
we
support,
not
just
clm
five
in
here
the
links
to
the
documentation,
we've
just
kind
of
refurbished,
our
documentation
for
for
dart,
and
so
please
take
a
look.
If
you
have
any
questions
and
thank
you.
C
Great
thanks
so
much.
I
think
we
should
move
on
to
the
next
talk,
but
I'll
note
that
there
were
a
couple
of
questions
in
the
chat.
So
if
you
want
to
have
a
look
at
those
about
crops
and
snotel
and
one
more
I
haven't
had
a
chance
to
read
yet
so
go
ahead
and
continue
that
conversation
next
up,
we
have
nikola
and
I
hope
I
pronounced
that
correctly.
C
K
Hi
everyone,
my
name,
is
it's
nicola
wiseman,
I'm
a
a
phd
candidate
at
uc
irvine,
and
so
today
I
we're
gonna
take
a
little
bit
of
a
a
deviation
from
some
of
the
some
of
the
land
and
lake
talks.
We've
had
so
far,
and
I'm
gonna
talk
about
our
ocean
bgc
in
cesm,
and
so
what
I'm
going
to
focus
on
here
is
the
biol
biogeochemical
impacts
of
plankton
stoichiometry
in
the
cesm.
K
So
this
is
some
work
that
I've
been
doing
over
the
past
couple
years
with
keith,
moore
and
adam
martini,
at
uc
irvine,
as
well
as
bench
mining
at
the
bigelow
marine
lab.
K
So
I
mean
I
feel
like
this
is
exactly
what
this
group
is
really
focused
on,
but
one
of
the
the
ocean
plays.
You
know
a
really
key
role
in
in
carbon
storage,
as
well
as
our
the
ocean
and
land
both
play.
K
Key
storage
isn't
key
roles
in
carbon
storage,
with
about
a
half
of
carbon
emissions
being
left
in
the
atmosphere,
25
going
up
in
a
land
25
going
up
in
the
ocean,
and
the
uptake
by
the
ocean
is
really
driven
by
both
our
solubility
and
our
biological
pumps
and
so
for
our
biological
pump.
It's
what
it's
really
driven
by
are
our
phytoplankton
and
so
phytoplankton.
K
So,
with
the
background
here
being
dissolved
nitrate
and
then
in
the
panel
b,
being
dissolved
phosphate,
and
so
what
we
really
see
here
is
we
see
the
elevate.
Our
h
and
lc
regions
are
high
nutrient
chlorophyll
regions
where
this
iron
is
limiting
in
our
north
pacific
equatorial
pacific
southern
ocean.
K
We
see
our
phosphorus
and
nitrogen
coal
imitation
in
the
atlantic,
sub-tropical
dryers,
and
then
we
also-
I
don't
have
it
listed
in
the
key
here,
but
we
also
have
si
limitation
as
well
and
some
of
our
upper
ocean
or
our
high
latitude
regions.
It
was
really
driven
by
sort
of
the
blue
cycle
of
our
phytoplankton.
K
We
know
that
variable
stoichiometry
within
our
phytoplankton
modifies
the
carbon
export
surface
nutrients
and
the
spatial
patterns
of
our
nutrient
limitation,
so
we
have
our
primary
productivity
by
our
photo
phytoplankton
photosynthesis.
Taking
up
I'm
specifically
going
to
be
focusing
on
nitrogen
phosphorus
and
iron.
Here
they
take
up
these
nutrients
as
well
as
carbon,
to
build
their
biomass,
and
then
the
ratio
of
those
nutrients
is
then
can
be
exported
into
the
deep
ocean
which
is.
This
is
what
we
really
care
about
when
we're
talking
about
oceans
and
biogeochemistry.
K
Is
this
carbon
export
from
the
surface
layer
and
that
long-term
storage
and
sort
of
getting
an
idea
of
the
role
of
how
that
might
may
change
sort
of
in
a
in
a
future
climate
scenario?
So
within
the
cesm
model,
I'm
working
with
the
out
focusing
on
that
region
here
I'll
get
to
that
one
second
yeah!
So
here
due
to
some
recent
expansion
of
methods
and
determining
iron
to
carbon
ratios
of
our
phytoplankton
we've.
Actually,
the
database
on
these
specific
cellular
measurements
has
really
been
expanding.
K
This
is
thanks
to
work
by
entwining
and
collaborators
where
we're
able
to
measure
cell
specific
iron
to
carpet
ratios
instead
of
just
bulk
pom
iron
to
carbon,
which
has
been
done
in
the
past,
and
so
this
is
a
collection
of
measurements
that
have
made
in
the
past
a
couple
decades
of
iron
to
carbon,
and
one
thing
that
we
really
see
here
is
that
these
ratios
can
vary
more
than
tenfold.
K
So
you
know
our
lowest.
Measurements
are
down
in
our
single
digits,
but
we
also
have
you
know
in
our
in
our
high
iron
regions,
like
our
the
saharan
dust
plume
and
our
coastal
regions,
you
can
have
very
high
highly
elevated
intercarbon
ratios
upwards
of
you
know,
60
to
90
micro,
mole
per
mole
here,
even
and
even
farther
above
that,
depending
on
the
organism
and
depending
on
the
conditions.
So
we
get
our
lowest
values
in
our
iron
limited
regions.
K
Our
h
lcs
that
high
that
high
latitude
north
pacific,
equatorial
pacific
southern
ocean
and
then
again
we
see
the
more
elevated
values
in
those
dust
plumes,
as
well
as
in
our
coastal
regions.
So
this
is,
you
know
a
data
set.
That
is,
that
is
being
expanded
slowly,
but
surely,
but
one
of
the
things
that
we've
been
able
to
look
at
with
this-
you
know
you
know:
diet
set
of
iron
to
carbon
is
the
relationship
between
cellular
iron
to
carbon
and
the
local
dissolved
iron.
So
this
is
a
plot.
K
This
is
just
from
the
observations
that
have
been
gathered.
We've
got
this
linear
relationship
that
really
becomes
pretty
pretty
clear,
with
with
iron
to
carbon
versus
dissolved
iron
at
those
locations.
So
there
there's
this
this
correlation
between
the
two
and
it's
consistent
with
this
idea
of
frugality.
Where
you
know
in
our
low
iron
conditions,
organisms
are
able
to
reduce
their
cellular
iron
quote
as
their
cellular
iron
uptake
and
then
when
iron
is
in
surplus,
we
have
this
luxury
uptake
that
is
able
to
occur.
K
This
sort
of
this
follows
along
with
some
of
the
work
that
has
been
done
by
the
martini
group
at
uc
irvine,
looking
at
phytoplankton,
stoichiometry,
specifically
c
to
ndp
and
pom
stoichiometry
as
well,
but
so
this
is
from
a
2015
paper
where
these
dots
here
are
showing,
or
these
sort
of
blocks
are
showing
field
observations
variability
in
you
know
we
have
p2c
versus
po4
concentration,
endoc
versus
no3
concentration,
and
so
there
is
a
lot
more
variability
in
our
p
to
c
ratios
versus
our
end
to
c
ratios.
K
But
one
thing
that
is
really
key
is
that
in
the
sort
of
low
nitrate
regions
or
in
when
we,
when
nitrate
is
very
low
and
nitrates
limiting
there
is
a
lot
more
variation
that
you
sort
of
don't
see
along
the
line,
and
nitrate
is
one
of
the
most
limiting
nutrients
in
the
ocean.
So
this
is
sort
of
a
region
that
we
really
do
care
about.
Is
these
low
nitrate
values,
where
end
of
c
may
be
more
variable.
K
K
K
and
then,
in
our
variable
run,
we
have
c
to
p
varying
about
you
know
100
to
200,
and
the
p
is
also
variable,
but
these
two
are
linked
by
a
constant
c
to
n
ratio
of
120
to
16..
K
So
it
is
the
it's
the
p,
that's
really
varying
here,
even
though
you
know
we
have
variable
end
of
p
here,
it's
the
p,
that's
changing
and
n
is
fixed,
so
for
our
iron
to
carbon.
We
have
a
range
of
for
our
three
phytoplankton
groups.
We
have
our
small
phytoplankton,
our
diatoms
and
our
diazotropes
for
our
small
phytoplankton
and
our
diatoms.
We
have
a
range
of
three
to
ninety
micromole
and
then
for
our
diazotrophs.
We
have
a
range
of
six
to
eighty.
K
As
diazotropes
have
higher
iron
requirements
due
to
the
metalloproteins
that
are
found
within
them
in
order
to
undergo
nitrogen
fixation,
and
then
we
also
have
our
variable
si
to
c
for
our
diatom
group,
and
so
these
are
300
year
runs.
These
are
two
300
year
simulations
where
we
analyzed
the
last
20
years.
K
K
We
have
our
maximum
nutrient
to
carbon.
This
is
that
we
prescribe,
and
then
we
have
this
optimal
nutrient
concentration
for
growth,
and
so
when
the
ambient
concentration
falls
below
this,
the
phytoplankton
will
reduce
their
quotas
accordingly,
in
order
to
sort
of
account
account
for
those
lower
lower
nutrient
conditions
until
they
reach
a
prescribed
minimum.
And
so
this
applies
for
both
the
p
phosphorus
iron
and
silica.
K
And
so
overall,
we
end
up
with
the
ambient
concentration
determining
the
ratio
for
nucro
and,
as
I
said
before,
the
the
nitrogen
to
carbon
ratio
is
fixed
at
16
to
120,
where
the
uptake
is
dependent
on
the
available
nitrate
and
ammonium
so
with
including
these
three
variable
nutrients.
I'm
gonna
focus
on
iron,
really
quick
for
a
second.
So
with
this
new
formulation
of
this
wider
iron
range
of
roughly
three
to
three
to
ninety
csm
is
able
to
better
capture
the
iron-to-carbon
ratios
that
we
measure
in
observations.
K
K
So
here
we
have
the
this
is
so
this
is
going
now
going
into
that.
Those
two
runs
that
I
was
just
talking
about,
so
we
have
the
variable,
stoichiometry
run,
fixed
stoichiometry,
run
and
then
the
difference
between
the
two
versus
our
observations.
K
So
as
a
result
of
increasing
the
variable
or
including
variable
stoichiometry
in
the
model,
it
modifies
the
efficiency
of
export
by
the
biota
and
therefore
the
surface
nutrient
distribution.
So
here
we
have
the
iron,
and
the
main
thing
that
we
see
here
is
that
in
the
variable
model
you
have
a
lot
more
iron
uptake
in
our
coastal
regions
and
so
therefore
lower
surface
iron
compared
to
our
fixed,
stoichiometry,
braun.
And
then
here.
This
is
a.
K
To
look
at,
but
we
have
our
nitrate
phosphate
silica,
and
so
again
we
have
variable
the
variable
run
at
the
top
fixed
in
the
middle
and
then
the
difference,
and
then
I
just
I've
included
world
lotion
analysis
here
at
the
bottom,
for
the
nitrate
and
the
phosphate,
and
the
main
takeaway
here
is
in
these
difference.
Plots
which
is
just
shows
that
the
h
and
lcs
expand
significantly
in
our
fixed,
runs
due
to
widespread
iron
limitation.
K
We
also
get
a
reduction
of
surface
phosphate
in
the
subtropical
gyres
here
in
the
fixed
run,
and
this
is
due
to
the
c
to
p
ratio
being
too
low
or
p
to
c
being
too
high,
so
again
we're
having
two
efficient
export
of
phosphorus
in
regions.
That
would
other
where
phytoplankton
would,
in
the
variable
run,
be
reducing
their
quotas.
Accordingly,
therefore
increasing
their
c2p
and
rsi
as
well
sort
of
follows
that
same
pattern
as
our
phosphate.
K
So
these
are
our:
these
are
maps
of
newt
growth
limitation
by
nutrients,
so
purple
is
nitrogen,
blue
is
iron,
green
is
phosphorus,
yellow
is
silicon,
so
in
our
again
our
variable
versus
our
fixed
run
and
if
I
just
sort
of
focus
on
you
know
our
diatoms
here
in
general,
we
do
see
a
major
expansion
of
these
iron
limited
regions
and
the
hlc's
in
particular,
there's
also
a
major
expansion
of
p
limitation,
like
I
said
before,
due
to
the
increased
export
of
nutrients,
and
we
also
see
expansion
of
si
limitation
in
the
high
northern
altitudes.
K
E
K
This
is
our
surface
chlorophyll
in
our
two
runs
and
then
again
our
difference
overall,
lowering
our
the
variable
nutrient
quotas,
where
they
are
able
to
sort
of
lower
their
nutrient
quarters
nutrient
quotas
accordingly,
in
in
regions
where
they're
limited
you
know
by
phosphorus
or
iron,
is
able
to
maximize
their
efficiency,
and
so
you
get.
K
You
know
these
greens
and
reds
here
in
this
bottom
plot,
we're
getting
more
chlorophyll
in
our
variable
run,
because
they're
able
to
sort
of
utilize
and
recycle
these
nutrients,
and
then
we
have
our
particulate
fluxes
again
variable
fixed
and
then
the
difference
and
in
general
overall
we
see
that
export
is
a
pretty
much
yeah,
except
for
in
this
sort
of
western
region.
Variable
stoichiometry,
just
maximizing
this
use
of
nutrients,
leading
to
increases
in
productivity
and
carbon
export
relative
to
the
fixed
run,
and
so
even
after
300
years.
You
know
these.
K
The
poc
export
is
converging
a
tiny
bit,
but
it
would
take,
I
think,
like
a
couple
thousands
of
years
in
order
to
get
these
to
actually
converge.
So
as
soon
as
you
sort
of
introduce
this
variable
stoichiometry,
it
really
has
a
first
order
impact
on
the
carbon
cycle.
K
And
I'm
actually
going
to
skip
sort
of
to
our
summary,
really
quick.
So
again,
we
have
these
first
order
impacts
on
carbon
cycle
by
including
variable
stoichiometry,
that
is
modifying
the
biological
carbon
pump
and
our
c
co2
fluxes.
Our
variable
stoichiometries
strongly
impact
surface
nutrient
and
distributions
of
patterns
of
limitation
and
accounting
for
this
variable.
Stoichiometry
is
really
it's
critical
to
projecting
the
ocean
bioqueue
chemical
response
to
ongoing
global
warming.
K
Our
next
step
isn't
to
include
variable
embassy,
which
is
currently
fixed,
but
there
is
some
variation
as
evident
from
observations,
especially
in
our
oligotrophic
gyres
and
so
it'll
be
very
interesting
to
see
what
accounts
for
what
happens
with
this
as
we
go
to
increasing,
including
this
variability
as
well.
Thank
you.
C
Great
thanks.
We
have
a
couple
of
questions
already.
I
saw
one
hand
that
went
up
and
then
went
down.
So
if
you
still
want
to
ask
your
question
and
also
some
clapping,
I'm
going
to
go
with
nikki's
question
in
the
chat
first,
which
chlorophyll
map
looks
more
like
the
satellite
observations
that.
K
Is
a
great
question
I
I
should
have
included
that
one
in
here.
It's
it's
definitely
the
variable.
The
variable
map
is
is
much
closer
to
matching
our
to
matching
satellite
observations
because
the
especially
with
the
fixed
run,
you
really
get
a
major
underestimate
in
the
equatorial
pacific,
upwelling
region
of
of
chlorophyll,
as
well
as
there's
some
dampening
of
the
bloom
cycle
for
in
the
high
latitudes
as
well.
B
Sure,
thanks
nicole,
for
that
great
presentation,
a
lot
of
the
story
you
were
explaining
in
context
of
the
iron
limitation
say
in
the
equatorial
pacific
in
terms
of
the
impact
on
the
hnlc.
Have
you
done
or
considered
doing
experiments
where
you
don't
do
all
of
these
variable
stoichiometries,
but
do
them
one
at
a
time.
K
K
I
don't
know
if
we
have
one
with
just
turning
off
the
p
or
the
s
I
but
would
be,
would
be
easy
to
do
so,
and
that
might
be
something
that
could
make
you
know
as
I,
I
will
be
doing
sort
of
a
full
look
with
the
variable
versus
fixed
and
sort
of
a
long-term
climate
simulation
down
the
road,
and
so
that
would
definitely
be
something
that
could
potentially
tell
an
interesting
story
as
part
of
that
project.
C
Daryl
you
had,
I
think
you
had
your
hand
up
earlier
and
then
it
disappeared.
I'm
sorry!
I
missed
whose
hand
that
was
but
go
ahead.
Well,
my.
G
I
have
a
bunch
of
questions.
One
of
them
got
answered,
so
my
understanding
from
this
is
that
the
export
ratio
doesn't
change,
but
the
amount
of
productivity
changes
because
better
utilized
is
that
correct.
G
Is
there
more
ballasting
if
you
have
more
iron,
like
I
understand,
if,
like
the
plankton,
has
more
iron
and
it's
exported,
then
it's
taking
more
iron
out
of
the
surface
into
the
deep
ocean,
but
like
does
it
so
so
does?
Did
you
did
you
change
the
export
module
and
no
okay?
That
was
my
question.
Did
yeah
that
was
yeah.
K
As
far
as
like
not
changed,
yeah
did
not
change
export
at
all.
It's
fully
just
being
driven
by
the
by
the
phytoplankton
changes
in
this
okay.
G
So
that
was
my
basic
question
and
then
my
other,
like
big
question,
and
I've
often
thought
this
about
this
model.
You
know
just
we
have
these
like
super
plankton
like
plankton,
functional
types
right,
and
so
you
know
in
reality,
there's
a
great
diversity
of
diatom
species
that
do
different
things
with
their
stoichiometry,
and
so
I
mean
you
showed
some
really
nice
plots
of
matching
the
observations.
But
like
is
there
a
limit
to
plankton
functional
types
for
this?
G
K
That
is
something
that
actually
is
being
worked
on
by
one
of
the
other
students
in
my
group
we
are
working
on
a
oh.
I
think
it's
like
a
seven
phytoplankton
model.
Seven,
I
would
I
could
count
them,
but
we
are.
That
is
something
that's
being
worked
on
in
conjunction
with
some
of
the
scientists
at
ncar
in
our
group.
So
with
with
marble
the
sort
of
new
biogeochemical
model,
they
will
be.
We
one
of
the
great
things
about
marble.
K
Is
it's
modular
and
so
you're
able
to
add
and
remove
phytoplankton
groups
as
as
you
see
fit
so
we
are,
there
is
a,
I
think.
It's
eight,
I
think
it's
eight
phytoplankton
there's
an
eight
phytoplankton
model
that
is
currently
being
worked
on
and
they
will
have
you
know
these
individual
sort
of
variable
stoichiometries.
K
So
I
think
it'll
be
interesting
to
see
you
know
the
one
one
of
the
issues
with
doing.
That,
though,
is
then
it's
it's
more
knobs
to
turn,
and
so
it
can
get
very
difficult
to
sort
of
narrow
down.
What
we
really
think
is
correct,
and
a
lot
of
that
is
limited
by
observations
and
whether
or
not
we
have
sort
of
you
know
specific
species,
stoichiometry
observations,
and
so
I'm
always
a
proponent,
more
more
studies,
more
fieldwork
and
then
we'll
we'll
throw
that
into
the
model
as
we
get
it.
B
It
was
a
brief
follow-up
to
to
the
first
question
from
cheryl,
so
I
don't
think
you
showed
community
composition
nicola,
but
I.
B
K
Yeah,
I
do
have
them.
If
you
send
me
an
email,
I
can
absolutely
find
them
and
show
them
to
you,
but
I
just
don't
have
them
in
this
in
this
specific
presentation,
but
I
yeah
I'm
pretty
sure
there
are
definite
shifts.
It
would
be
really
interesting.
Small
dominated.
C
This
is
great
discussion,
but
I
think
we
really
have
to
get
going
move
on
to
the
if
we
want
to
have
the
full
group
discussion.
A
About
the
priorities
for
the
the
working
group,
I
think
we
need
to
move
on
to
that
sorry
that.
C
Okay,
next,
we
were
as
keith
mentioned
earlier.
We
were
hoping
to
have
a
group
discussion
and
I
don't
know
if
keith
wants
to
try
to
pull
that
slide
up
right
now,
but
we
are
going
to
send
people
into
breakout
groups,
so
we
could
hopefully
have
some
conversations
in
slightly
smaller
groups
about
what
the
what
the
working
group
might
do
as
I'm
trying
to
paste
the
link
to
our
slides
in
there.
C
What
the
working
group
might
want
to
work
on
in
the
future
and
prioritize
and
how
we
could
how
we
could
coordinate
those
activities
to
bring
people
together
around
this
topic.
So
this
is
a
google
slide
deck
where
we
can
take
some
notes.
We
can
share
this
link
also
within
the
breakout
groups,
and
then
I
think
we're
planning
to
head
into
probably
three
breakout
groups,
and
I
don't
know
if
todd
was
going
to
set
those
up
or
if
I
should
set
them
up.
C
E
A
A
A
A
A
C
C
Okay.
Well,
I
hope
that
people
were
able
to
have
some
conversations
in
their
groups.
I
don't
this.
I
thought
we
could
share
out
some
of
what
we
talked
about
in
the
different
groups,
but
I'd
be
happy
for
it
and
not
to
be
me
talking
so
does
anybody
want
to
volunteer
something
that
came
up
in
their
group
that
they
wanted
to
share
and
I'll
keep
taking
some
notes.
H
H
I'll
go
for
it.
The
dave
asked
the
provocative
question
as
his
memo
during
the
during
the
meeting,
which
was
like.
Does
the
biogeochemistry
working
group
need
to
exist
and
we
can
kind
of
came
up
with
some?
Well,
it's
not.
H
It's
not
really
doing
any
model
development,
and
so,
if
it
does
need
to
exist,
it
needs
to
exist
kind
of
more
for
scientific
questions,
maybe
more
similar
to
what
the
paleoclimate
working
group
does,
and
it
also
needs
to
exist
in
some
form,
because
it
has
a
really
nice
computing
allocation,
which
is
useful
for
the
development
of
biogeochemistry.
So,
even
though
the
working
group
itself
isn't
doing
the
development,
its
computing
allocation
is
very
useful.
H
Yeah
there
was
more,
there
was
more
to
it,
that's
on
our
on
our
sheet,
but
that
was
the
the
most
interesting.
I
think
part
of
the
conversation
was
like
a
bit
of
soul-searching
about
whether
it's
necessary.
A
I
would
say
that
an
answer
to
you
know
sort
of
the
need
for
science
questions.
That
was,
I
thought
where
our
group
spent
most
of
our
time
was
thinking
about
whether
there
were
questions
that
we
could
answer
about,
how
the
coupled
climate
system
would
respond
to
negative
emissions
and
whether
we
can
make
a
constrained
model
prediction
to
understand.
A
You
know
10
years
from
now,
given
a
certain
level
of
atmospheric
co2.
To
what
extent
is
it
because
folks
did
not
meet
their
emissions
production
contributions
or
the
extent
to
which
natural
sinks
might
have
changed?
A
C
A
C
A
Of
related,
but
to
what
gretchen
was
just
talking
about,
but
I
think
you
know
it's
important
to
recognize
that
to
try
and
understand
the
carbon
cycle,
you
have
to
actually
be
able
to
measure
the
fluxes
of
carbon
and
we
can't
actually
do
that
right
now
at
large.
You
know
continental
scales
and
we
can't
do
that
because
the
atmospheric
co2
inversion
techniques
essentially
don't
work,
but
if
they
did
work,
they
don't
have
the
spatial
resolution
or
the
process
information
that
would
actually
allow
you
to
really
understand.
A
What's
going
on
the
earth
system
models
are
the
key
to
understanding
the
processes
that
are
driving
things,
yet
they
aren't
generally
used
for
flux
estimation.
So
I
I
see
a
big
you
know
gap
but
possible
opportunity
for
interconnecting
these
two
sort
of
fields
of
study
to
try
and
use
earth
system
models.
In
a
flux
estimation
framework
you
know,
brett's
talk
was
was
starting
down
this
direction.
A
I
thought
that
was
really
exciting
a
lot
of
hurdles,
but
I
think
that
the
process,
information
and
versus
the
models
could
really
be
brought
to
bear
in
ways
that
hasn't
previously.
C
So
I
was
wondering
if
that
gets
to
now,
I'm
forgetting
it
was.
Maybe
well
said
that
it
stated
that
the
budget
chemistry
working
group
doesn't
do
any
model
development,
but
it's
a
question
of.
D
D
One
one
thing
that
I
didn't
mention
in
our
group
because
I
had
to
go
briefly,
but
the
what's
what's
neat
about
biogeochemistry
is
that
we
have
these
new
observational
systems
that
are
coming
online,
like
in
the
physical
realm.
Observational
systems
have
existed
for
quite
some
time
and
they
really
haven't
evolved,
but
we
have
new
observations,
particularly
in
ocean
biogeochemistry.
D
We
have
biogeochemical
argo
measuring
nitrogen
in
the
ocean,
measuring
pco2
and
ph
in
the
ocean,
and
csm
could
be
a
test
bed
for
for
budget
chemical
aussie
type
experiments
in
a
way
that
no
other
model
has
really
really
gone.
No
one
else
has
really
gone
for
that.
Yet
so,
and
probably
there
are,
there
are
similar
things
you
could
think
of
for
the
terrestrial
carbon
cycle
as
well,
where
the
the
bgc
component
of
the
land,
carbon
of
the
land
model,
could
be
used
for
an
aussie
type
test.
C
D
D
So
why
don't
you
put
them
into
your
model
and
let
them
a
vector
ground
with
the
lagrangian
flow
and
sample
the
subsample,
the
model
by
geochemistry
at
you
know
at
certain
locations,
and
then
you
can
get
a
sense
of
whether
you
can
back
out
the
large
scale
patterns
or
the
small
scale
patterns.
Or
what
have
you.
F
F
So
we
sort
of
felt
like
the
development
had
to
be
in
those
working
groups.
But
that
said,
you
know.
Yes,
there
was
a
huge
need
for
the
biogeochemistry
working
group
and
I
think
maybe
the
need
is
sort
of
what
nikki
has
been
talking
about
in
terms
of
community
experiments
and
even
what
brits
talking
about
in
terms
of
sort
of
can
we
actually
get
flux,
estimates
out
of
an
earth
system
model?
Now
we
need
to
think
of
a
community
project
that
would
be
driven
by
the
biogeochemistry
working
group.
F
They
would
involve,
would
evolve
development
but
it'd
be
developing
in
terms
of
creating
like
this
whole
sort
of
carbon
prediction
system,
for
example,
you
know:
can
we
actually
do
something
like
that?
That
would
touch
into
other
aspects,
because
there's
the
earth
system
prediction
working
group,
you
know
there's
the
large
ensemble
activities
going
on
with
other
working
groups.
But
again
you
know
we
can
create
sort
of
like
a
a
big
activity.
F
That's
led
by
the
bio-geochemistry
working
group
that
spawn
spans
across
all
aspects
of
csm
and
that's
that's
the
need
and
that's
the
rationale
for
the
bgc
working
group.
I
kind
of
like
what
brit
says,
because
it
puts
another
value
to
both
the
land
model
and
the
ocean
model,
not
just
providing
sort
of
surface
fluxes
to
the
atmosphere
or
looking
at
impacts
of
climate
change,
but
actually
sort
of
becoming
a
really
important
scientific
tool.
F
I
think
what
brett
was
brett
was
talking
about
today
was
actually
really
kind
of
interesting
of
comparing
his
assimilated
product
to
fluxcom.
That
to
me
seemed
like
kind
of
interesting
right.
There's
a
lot
going
on
right
there
to
explore.
B
I
have
heard
from
some
corners
some
pushback
about
making
sure
that
you
balance
that
with
bottom-up
pi
work,
and
so
we
need
to
make
sure
that
we
still
have
that
we're
welcoming
to
pi
initiated
what
we
used
to
call
entrepreneurial
science
with
the
large-scale
projects,
and
so
I
think,
there's
a
balance
to
be
struck.
There.
B
That
that
can
be,
if
that
can
be
a
good
counter
argument.
G
Yeah,
so
just
to
add
to
that
keith
in
our
group
we
talked
about
how
there
needs
to
be
more
people
who
know
how
to
do
what
you
do
and
and-
and
you
know,
run
these
like
coupled
climate
simulations
and
that
one
way
to
do
that
is
have
pi's
outside
of
in
car,
get
funding
for
post,
docs
and
grad
students
who
could
learn
those
skills.
So
we
can,
you
know,
take
that
out
of
in-car
and
spread
it
around
more.
G
Yeah,
I
think
that,
having
you
know
there's
a
lot
of
interest
from
the
outside
community
and
at
least
in
talking
to
me
for
having
fully
coupled
simulations
and
I
think,
having
projects
that
are.
You
know
if
we
target
a
few
of
these
larger
group
projects
in
this
within
this
group
and
provide
those
simulations
for
analysis
with
the
fully
coupled
simulations
that
could
provide
another
tool
and
then
you
know
you're
right
then
postdocs
and
other
people
have
have
capability
of
analyzing
those.
G
I
realize
that
some
of
those
are
available
through
cmip
and
also
through
the
lens
ii
work,
but
we
could
potentially
provide
additional
simulations.
G
The
other
thing
that
I
wanted
to
mention
is
just
that
you
know
we
focused
a
lot
and
we,
I
think
this
group
still
focuses
a
lot
on
co2,
but
nitrogen
is
also
really
important
for
reactive
chemistry
in
the
atmosphere,
for
example,
if
we're
leaching,
and
so
that's
an
important
component,
and
it
was
really
interesting
to
see
all
the
talks
on
methane
and
thinking
about
including
methane
in
some
of
these
projects,.
A
I
was
just
gonna
say
we,
you
know,
we
talked
about
this
a
little
bit
too,
and
these
are
sort
of
these
intractable
problems
that
we
can.
We
always
talk
about
and
never
make
progress
on,
closing
the
nitrogen
cycle,
closing
closing
the
methane
cycle
or
you
know
doing
prognostic,
and
maybe
the
working
group
can
be
that
can
be
initiate
this.
But
I.
A
B
A
A
A
C
But
even
for
carbon,
we're
still
missing
some
big
pieces
of
the
integration
right,
especially
if
we're
thinking
about
estimating
carbon
fluxes,
it
seems
like
there's
some
major
infrastructure
pieces
that
are
missing
there.
We
talked
in
our
group
about
riverine
inputs
to
the
ocean
from
land,
but,
as
far
as
I
know,
is
something
that's
always
on
the
list
for
this
group,
and
we
never
have
made
progress
on.
C
A
I
think
I
think
you're
right
that
you,
you
know
if
you,
if
you
tried
to
tweak
the
parameters
in
you,
know
clm
and
marble,
you
know
to
try
and
match
atmospheric
co2
exactly
you
would
you
run
into
structural
problems
eventually,
but
you
may
have
so
much
tunability
that
you
can
actually
match
the
real.
You
know
atmospheric
co2
without
you
know
necessarily
getting
it
right
for
the
right
reason,
but
I'm
looking
forward
to
you
know
the
ppe.
A
I
think
that's
the
right
acronym
from
the
from
the
clm
parameter
ensemble
and
just
you
know,
to
see
how
much
you
know.
How
much
can
you
swing
and
say
the
seasonal
cycle
amplitude
or
the
atmospheric
growth
rate,
just
by
selecting
parameters
within
a
reasonable
range,
and
I
talked
to
matt
a
little
bit
about
you
know
the
limitations
of
doing
something
similar
with
the
ocean
model.
A
I
think,
if
I
understand
correctly,
it
has
to
do
with
the
sort
of
the
time
scales
of
of
you
know
in
the
way
that
the
physics
in
the
ocean
sort
of
sets
the
stage
for
the
badger
chemistry,
but
but
he
had
an
idea
for
doing
something
where
you
were
nudging
the
deep
ocean
and
just
doing
your
parameter
ensemble
on
you
know
things
that
mattered
for
the
for
the
near
surface
layer.
So
you
know
we.
It
may
be
that
once
we
have
some
experience.
A
Looking
at
ppe
that
you
know
we
find
out,
there
is
a
way
that
you
could
team
up
with.
You
know
the
folks
in
dart
I
mean
ncar.
Has
these
big
efforts
in
data
simulation
that
haven't
really
been
focused
on
cesm?
Specifically,
it's
my
impression
so
there's
also
an
opportunity
there
yeah
well
with
brett,
moving
to
end
card
that
helps,
at
least
from
the
land
side.
Right
he's,
got
tons
of
land
experience
and
really
interested
in
land
data
simulation,
so
it
could
really
plus
andy
fox
still
being
around.
B
And
we've
got
dan
manrein
in
the
oceanography
also
works
with
the
dark
group.
So
we've
started
some
conversations
about,
but
they've
kind
of
stagnated
in
recent
in
recent
months,
but
we
are
trying
to
move
forward
with
trying
to
have
more
interaction
with
the
dart
group
ocean
bgc
and
I
I
would
love
to
be
able
to
take
lessons
learned
from
clm
ppe
and
apply
them
to
marvel
ppe.
D
Share,
I
have
a
comment
about
keith's,
like
opening
opening
presentation
for
this
subgroup
meeting,
which
was
that
he
said,
we've
noticed
there
have
been
fewer
presentations
in
submitted
and
I
just
want
to
say
that
that's
partially
my
fault
and
I
apologize,
but
I
didn't
have
the
bandwidth
to
have
myself
or
anybody
in
my
group
submit
presentations
either
this
year
or
last
year
because
of
the
pandemic,
and
I
suspect
that
that's
true
in
many
of
the
working
groups.
G
I'm
gingerdo
from
gpl.
K
G
E
These
talks
is
very
interesting,
but
I
think
it
would
be
great
if
the
modeling
community
have
more
interactions
with
the
observation
community
to
incorporate
more
observations
into
the
models.
I
think
bright
stock
trying
to
use
dart
system
to
automatic
optimize.
The.
G
Fluxes
is
also
very
interesting,
but
I
a
long
time
I
want
to
say
that
top-down
flexing
motion
is
not.
I
mean,
still
have
some
usage.
C
I'm
actually
wondering
if
gordon
or
nikki
had
any
thoughts
on
that,
because
I,
whenever
I
think
about
this,
I
think
about
your
paper,
where
you
show
that
it
we
can't
actually
constrain
the
future
carbon
cycle
by
using
the
past
observations.
Do
you
think
that
there
are
any
aspects
of
carbon
cycling
that
would
be
more
useful
for
constraining
the
future
or
is
all
hope
lost.
C
D
D
D
Way,
no,
no,
I
mean,
I
think,
certainly
we
could
we
could.
We
would
do
better
to
improve
our
our
model
observation
match,
based
on
the
few
observations
that
we
have,
but
I
think
we
can
also
use
our
model
to
promote
the
collection
of
additional
observational
based
data,
so
I
think
we
can
use
it
in
reverse
to
argue
for
why
we
needed
more
in
the
past
and
we
didn't
have
it
and
what
we
can
be
doing.
Moving
forward.
A
A
C
A
If,
if
with
with
a
much
richer
set
of
constraints,
there
is
actually
but
just
to
be
a
little
more
optimistic
that
maybe
maybe
there
is
a
chance.
F
Hey
well,
somebody's
got
to
go
negative
here.
So
I'll
do
that,
which
is
you
know?
I
I
do
think
we
have
to.
I
agree,
you
know
more
data
will
help
solve
the
problem,
but
will
it
actually
it
will
help,
but
it
won't.
Actually,
you
know,
eliminate
the
problem
and
I
I
don't
know.
I
still
have
this
view
that
we
have
to
live
in
this
world
of
uncertainty
that
we're
there's.
F
If
we
just
look
at
the
diversity
of
life
in
the
world,
there's
no
way
we
can
ever
truly
capture
the
difference,
all
the
different
life
forms
and
organisms
and
the
diff
even
just
the
spatial
variability
of
ecosystems,
and
so
maybe
we
have
to
start
living
in
this
world
of
uncertainty.
I
don't
know
what
that
means.
You
know.
F
I
know
it's
perceived
as
being
negative
that
we
should
actually
be
converging
on
an
answer,
but
I
have
this
view
that
maybe
there
are
multiple
answers
out
there
and
we
probably
need
to
start
living
in
that
world
of
multiple
answers,
and
so
those
are
so
those
are
some
of
the
things
that
nikki
and
I
were
trying
to
get
at.
You
know
with
what
we
were
doing,
but
what
I
also
want
to
end
on
something
is
optimistic,
which
is,
if
you
look
at
the
other
paper
yeah
that
nikki
led
on
carbon
cycle
prediction.
F
There
is
some
predictability
in
the
carbon
cycle.
So,
despite
all
this
uncertainty
and
all
this
stuff,
we
actually
can
do
some
things
that
are
going
to
be
really,
I
think,
helpful
in
terms
of
looking
at
how
the
carbon
cycle
is
actually
evolving
in
the
future
and
maybe
sort
of
how
we
want
to
why
it's
changing
and
what
we
could
actually
do
to
sort
of
make
predictions
about
how
it
will
change
over
the
next
year
or
so
so
yeah.
F
D
One
of
the
things
that
jerry
meal
has
been
struck
by
in
this
predictability
work
is
that
we
often
find
in
the
biogeochemical
world
higher
or
longer
lasting
predictability
than
you
do
for
physical
variables,
and
he
has
hooked
on
to
that.
So
we
have
sold
a
physical
climate
scientist
on
budget
chemistry
through
these
prediction
simulations.
I
think,
there's
more
that
we
can
do
like
that.
I
think
is
biogeochemist.
C
I
can
think
of
a
land
example
of
that
too,
potentially
anyway,
that
you
might
be.
You
know
some
knowing
something
about
the
vegetation
and
what
you
expected
to
do.
It's
going
to
tell
you
something
about
what
you
expect
in
the
physical
climate
system
and
that
may
have
a
longer
term
memory
associated
with
it
than
what's
influencing
the
physical
climate.
C
B
Thank
you,
everyone
for
participating
in
this
conversation.
This
is
really
useful
to
hear
these
different
perspectives
in
input.
Oh.
C
We
discovered
that
we
have
an
email
list,
but
probably
nobody's
on
it,
like
I'm,
not
sure,
I'm
on
it,
so
we
were
hoping
that
we
could
try
to
get.
We
could
share
this
around
and
try
to
get
people
who
are
interested
to
sign
up
so
that
we
could
at
least
have
a
hope
of
contacting
people
and
coordinating
some
of
these
ideas
going
forward.
At
least
it.
You
know
it's
obviously
like,
as
people
are
interested,
but
we
didn't
otherwise
know
how
to
contact
the
people
who
were
interested.
B
A
Off-Site
might
actually
be
a
lot
easier.
I
think
for
there's.
The
tea
leaves
that
I've
been
reading
are
that
you
know
a
large
actual
workshop
won't
happen
until
2022,
but
there
might
be
an
opening
up
of
having
like
a
10
person
meeting
sometime,
this
fall,
but
yeah
somebody
wants
to
organize
something
up
in
breckenridge.
B
C
G
One
thing
that
I
was
just
thinking
is:
I
wonder
if
this
group
would
be
interested
in
an
aspen
global
change,
institute
meeting
another
meeting
in
the
mountains.
G
A
C
Cycle
yeah:
well
thanks.
Everyone
for
participating
really.
B
Yes,
thank
you.
Thank
you
very
much.
Everyone,
and
we
hope
to
see
you
soon
again
and
yes,
nikki,
says
in
person
would
be
great
too.
C
A
Have
to
chat
yeah
now
feel
free
to
keep
on
going.
I
mean
okay,
thank
you.
Yeah!
I'm
gonna
sign
off
thanks.
Everyone
thanks.
A
Okay,
I
have
to
apologize
that
that
I
had
to
start
driving
home
during
the
breakout.
Okay,.
A
That's
why
I
was
I
like
this
discussion,
some
big
complex.
You
know,
chemistry,
question
that
involves.