►
From YouTube: CESM Land Model Working Group - 2020 CESM Workshop Day 3
Description
2020 CESM Workshop Day 3 - Wednesday, 17 June 2020 - 8:30-12:00 - Land Model Working Group
The 25th 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 15 June 2020 with presentations on the state of the CESM. On 16-17 June 2020, working groups and cross working groups have half-day sessions, some with presentations and some that are discussion only.
More information about the CESM Land Model Working Group:
http://www.cesm.ucar.edu/working_groups/Land/
More information about the workshop:
http://www.cesm.ucar.edu/events/workshops/2020/
Talks include:
Summary of Tuesday discussions
Challenge for next discussion
Prioritize list of needs for meeting our goal of historical CTSM-FATES simulations
Identify specific tasks & interested people to implement this integration
Whole Working Group discussion
A
B
D
G
G
B
B
D
G
B
G
Thinking
some
about
what
we
wanted
to
accomplish
today
came
up
with
this
kind
of
revised
agenda,
which
you
can
see
hopefully
displayed
on
the
street
now
so
kind
of
broken
into
three
chunks.
But
the
three
main
topics
that
kind
of
came
out
from
the
discussions
yesterday
so
the
first
hour
or
so
we'll
go
over
some
technical
features
related
to
see
TSM
and
fates
and
creation.
Then
we'll
spend
an
hour
from
10:00
to
11:00
talking
about
benchmarking
and
evaluation
and
then
I
left
almost
an
hour
at
the
end
for
kind
of
governance
and
coordination.
G
I
think
no
more!
That
might
be
spent
on
wrap-up
and
kind
of
making
sure
we're
all
on
the
same
page
and
figuring
out
kind
of
who's
going
to
be
doing
what
and
on
what
timeline
related
to
first
two
parts
of
the
discussion:
that's
why
I
wanted
to
start
and
I
feel
like
you
know,
you
know
in
many
ways
this
whole
process
of
doing
the
meeting
remotely
has
been
I
would
say,
surprisingly,
productive
until
last
night,
and
it
was
like
last
night.
G
G
G
You
know
prioritize
that
needs,
identify
specific
tasks
and
then-
and
maybe
most
importantly,
the
interested
people
who
are
going
to
help
implement
the
see
TSM
fates
version,
zero
integration
and
I
added
the
version
zero
through
this,
because
in
reading
some
of
the
responses
from
different
discussion
groups
yesterday
I
think
emphasizing
like
this
is
just
version.
Zero
is
important
and
entirely
mention
this
in
the
discussion
yesterday
afternoon.
You
know
the
importance
of
avoiding
kind
of
a
loss,
averse
mindset
in
the
counter
op.
G
G
We
did
to
the
model
was
going
to
be
better,
and
so
we've
had
nearly
you
know
close
to
a
decade
of
being
able
to
just
always
make
improvements
in
the
model
and
I
think
you
know
in
many
ways
CTS
Evan
face
and
the
states
integration
are
at
a
challenging
point
in
the
sense
that
you
know
we
already
have
a
good
model,
and
so
anything
we
do
likely
won't
make
it
better,
at
least
in
the
short
term,
and
so,
but
the
counter-argument
to
that
is
you
know.
Really.
H
G
Of
things
that
face
needs,
you
know
it
needs
nutrients,
it
needs
isotopes.
It
needs
a
lot
of
trees
and
spin
ups,
and
you
know
there's
critical
resources
that
are
missing
my
favorite
ones
on
here,
or
it
needs
Jackie
and
other
individuals
were
actually
called
out
which
is
kind
of
fun,
but
I
think
the
other
thing
we
have
to
realize
the
CTS
mphase
version.
Zero
will
obviously
have
some
reductions
in
complexity.
We're
not
going
to
have.
G
You
know
every
tool
in
our
pocket
that
we
want,
but
we
also
are
going
to
have
some
new
features
that
are
being
added
and
so
I
think
it's
important
to
think
about
the
long-term
goal
of
where
we
want
to
go
and
I.
Think
many
of
the
discussion
points
that
were
raised
yesterday,
especially
related
to
questions
two
and
three
were
long-term
Arbor
and
long-term
goals
is
I'd
like
to
spend
the
discussion
you
know
for
the
next
hour
or
so
thinking
about
the
shorter
term
milestones,
and
so
dhammika
did
a
really
nice
job
synthesizing.
G
G
I
G
I
G
And
I
think
you
know,
I
put
this
up
kind
of
as
a
as
a
straw,
dog
and
I'm
like
concrete.
This
is
what
see
TSM
fates
version.
Zero
will
be
so
it
will
include
transient.
Land
use,
it'll,
be
run
offline,
both
offline
and
within
a
met
configurations.
I
think
there
was
a
resounding
statement
that
we
need
to
run
this
in
a
couple
of
mode.
G
I'd
suggest
that
we
have
it
with
three
configurations:
one
with
prescribed
le
I,
so
faked,
SP
version
fates,
biogeography
fix
it
with
fix
without
biogeography,
and
there
was
some
discussion
questions
that
that
would
be
the
default
because
it's
the
most
similar
typically
see
TSM
and
then
also
have
fully
prognostic.
These
crops
was
that
is
obviously
a
distinguishing
feature
of
see,
TSM
and
and
for
this,
these
simulations
I
suggest
that
we've
come
up
with
a
way
to
include
crops,
but
not
include
them
within
the
faith's
code
base,
just
keep
them
in
CTS.
G
So
I
don't
know
what
technical
aspects
that
that
involves
and
for
the
fates
version
you
know,
I,
guess
they
don't,
though,
by
default,
have
to
function
in
the
carbon
only
mode,
because
these
doesn't
have
nutrients
I.
Think
the
first
kind
of
topic
of
discussion
is
you
know
is
this:
is
this
our
target.
G
J
K
K
F
K
G
G
L
F
G
G
G
I
I
A
Up
yeah,
I
think
well,
Jackie
and
I
were
talking
about
the
fixed
biogeography,
but
maybe
like
a
loose
version
of
that.
So
you
would
maybe
not
allow
like
boreal
needle
either
need
a
leaf
evergreen
to
grow
in
the
tropics.
But
so
you
would
sort
of
loosely
determine
by
just
by
seed
source
like
where
PF
teas
are
available
to
colonize
yeah.
F
L
M
These
sort
of
square
square
bullets
these
second,
two,
your
bullets
of
scribe
daily
I,
mean
I,
think
I,
don't
really
see
it
as
an
or
so
much
isn't,
and
because
you
know
I
think
we're
always
gonna
be
it.
We
should
be.
One
question
is:
should
we
think
about
this?
Maybe
as
an
experiment,
I
mean
what
happens
if
we
compare
these
two
versions
of
the
model,
both
you
know
and
they're,
they're
gonna
be
things
that
they're
they're,
good
and
bad
about
both
of
them.
I
mean
the
fixed
by
geographies
me.
M
A
lot
faster
it'll
be
spin
up
a
lot
faster
because
you'll
have
less
of
the
you
know.
The
kind
of
community
assembly,
time,
scale,
spin
up
dynamics
and
so
I
think
it's
gonna
have
a
lot
of
features
that
are
particularly
amenable
to
fully
coupled
simulations,
particularly
the
reduced
spin
up
time,
and
so
I
think
you
know
it's
it's
not
that
we're
gonna
want
to
just
sort
of
there's
said:
I,
guess
it's
not
the
FIFO
geographies
judge
is
only
a
stepping
stone
on
the
way
to
the
so
fully
part.
M
I
G
E
It
seems
to
me
one
of
the
differences
between
a
fix.
You
know,
as
Charlie
said,
or
somebody
said,
the
fix
biogeography.
You
know
a
major
way
to
constrain.
That
is
just
what
seeds
are
available
or
what
initial
recruits
are
available,
as
the
model
is
spinning
up
and
I
think
this
raises
the
bigger
question
of
spin
up,
because
I
think
or
initialisation
really.
F
E
G
K
G
K
What
is
the
sort
of
consequence
of
that?
Does
that
mean
that
it's
going?
There's
gonna
be
a
testing
phase
prior
to
CES
m3
and
when
what
needs
to
be
working
and
at
what
level
when
CES
m3
is
released?
We
basically
need
this,
so
that
vegetation
is
not
doing
things
that
are
so
crazy
that
it's
going
to
counteract
the
the
work
that
somebody's
doing
in
the
atmospheric
in
an
atmospheric
run,
where
they
need
to
know
that
the
land
surface
is
reasonable.
At
that
point,
you
know
yes,.
G
K
G
I
Dream
it
you
kind
of
follow
up
on
that.
Well,
we
all
know
that
yet,
but
but
CSM
just
to
separate
itself
see
if
some
release
cycles
trying
to
separate
itself
from
the
cement
cycle.
So
that
may
not
be
the
right
gauge
it's.
The
question
is
when
it
sees
them
three
going
to
come
out,
and
we
don't
really
know
about
that.
It's
just
that.
Dokkan
is
currently
saying
that
we're
gonna
start
assembling
the
components
next
summer.
D
A
I
guess
going
back
to
Lara's
comment,
I
think,
that's
sort
of
what
jackie
and
I
on
and
I
mean
maybe
Josh.
That's
also
what
you're
talking
about
in
your
group
we're
thinking
with
the
sort
of
loosely
fixed
biogeography,
because
you
know
we,
you
know
we're
not
trying
I'm
assuming
we're
not
gonna
run.
You
know
simulations
from
like
Pangaea
onward,
because
so
we're
not,
we
can't
simulate.
A
There
are
trees
in
certain
places,
not
just
because
they
can
grow
there,
but
also
because
that's
just
where
they've
migrated
to
over
you
know
centuries
and
millennia,
and
so
that
into
the
model
that
I
use
that's
kind
of
how
I
get
around
the
other
model
that
I
use
get
around
sort
of
like
land
use,
because
it's
a
smaller
scale.
So
that's
how
I
get
around
kind
of
like
land
use
problems
that
the
model
is
not
able
to
simulate
it's
just
by
saying:
okay!
A
Well,
all
these
trees
can
these
species
can
potentially
colonize
here,
but
not
these
other
species
and
then
the
model
is
then
has
those
species
compete?
And
you
know
when
you
let
PF
tease
that
just
don't
realistically
occur,
they're
their
seed
source
isn't
even
there.
It
does
slow
down
the
model
quite
a
bit
and
they
could
grow
there
because
they're
able
to
not
just
because
they've
migrated
there.
G
M
Yeah,
you
say
that
in
principle
we
could
already
do
all
this
kind
of
stuff,
the
AG
on
Ramallah
talking
about.
If
we
you
know,
if
you
start
the
model
and
fix
biogeography
mode
sort
of
spin
up
the
vegetation
structure
and
then
use
this
feature
of
the
model
where,
in
addition
to
restart
files,
the
model
can
also
output
an
input
census
effectively.
A
census
found
it's
like
a
listing
of
all
the
trees
which
we
normally
use
for
initializing
from
from
actual
fara
census.
M
D
F
It
occurs
to
me
that
a
way
of
doing
a
really
easy
way
of
doing
the
loose
constraint
over
by
geography
is
just
to
alter
the
surface
dataset,
so
you
have
because
in
the
fixed
by
drug
remote
with
competition,
the
only
thing
that
matters
is
whether
the
seeds
are
present
or
not
right.
So
you
could
say:
okay
well,
we'll
make
a
sort
of
a
PFT
map
where
basically
at
all
temperate
areas
or
the
temperate
things
are
present
and
in
all
tropical
areas,
all
the
tropical
things
are
present
right.
F
E
A
Yeah
I
mean
I,
guess
in
terms
of
dispersal,
so
I've
done
dispersal
with
the
gap
model
that
I
use
and
it's
not
I
mean
you
can
do
it
without
it
being
terribly
computationally
intensive.
You
could
also
I
mean
without
having
to
do
grid
cell
interaction.
You
could
just
have
the
surface
data
set
like
so
the
get
the
biogeography
get
looser
over
time.
So,
if
you're,
you
know
you
could
spin
it
up
with.
A
This
is
sort
of
like
a
like
a
I
guess,
a
test
case
where
you
could
spin
it
up
with
very
sort
of
more
fixed
and
then,
as
if
you
decide.
Okay,
now
we've
reached
current
conditions
and
maybe
we're
interested
in
how
PFDs
might
move
under
climate
change.
You
could
then
just
make
over
subsequent
years,
just
have
the
seed
source
be
like
larger
and
larger
for
all
the
PFT
s.
But
if
you
didn't
want
to
deal
with
doing
seed,
direct
seed,
dispersal
right
away.
G
L
Maybe
I
do
have
a
comment
or
question
about
the
bio
geography
which
is
maybe
is
it?
Is
it
useful
to
think
about
I
mean
there's
Rosie
pointed
out
at
the
beginning
there
you
know,
there's
some
question
about
what
we'd
actually
be
able
to
accomplish
in
a
year,
and
and
and
maybe
it's
too
early
to
decide
that,
but
is
it
useful
to
talk
a
little
more
about
an
approach
like
I
mean
I,
see,
I,
see
two
possibilities
in
terms
of
where
does
the?
L
Where
does
the
bulk
of
the
kind
of
scientific
testing
and
getting
things
looking
right
happened?
Do
we
start
with
a
dynamic
biogeography
and
then
just
with
the
you
know,
in
the
back
of
our
mind
that
we
can
back
off?
If
that's
not
working
or
do
we
try
to
get
things
working
well
with
good
fluxes
of
carbon
and
water
and
energy
with
a
fixed
biogeography
first
and
then
once
we
have
that
working
then,
if
that's
working
in
time
then
expand
the
scope
to
now.
L
So
I
guess
I
had
a
question.
I
have
a
question
about
crops,
and
this
ties
back
in
with
discussion.
We
had
was
it
in
the
winter
meeting
I
believe,
which
is
to
what
extent
do
we
feel
that
it's
important
to
share
some
of
the
science
parameterizations
between
the
big
leaf
model?
The
standard
see
TSM
and
fates.
L
Is
that
a
critical
or
important
aspect
of
of
this
I
like
if
crops
are
handled
in
standard
CTS
M
then
do
we
feel
it's
important
that
they're
using
the
same
photosynthesis
code,
the
same
whatever
other
parameterizations
as
fates,
so
that
when
you
transition
a
grid
cell
between
the
amount
between
natural
vegetation
and
crop
you're,
not
getting
this
kind
of
accidental
differences?
Just
because
there's
differences
in
how
photosynthesis
is
parameterised,
I.
G
L
Not
familiar
enough
with
the
faith's
code
to
you
know,
so
we
talked
in
February
about
like
pulling
out
some
of
it.
I
guess,
in
my
mind,
like
pulling
out
some
of
some
of
the
fates
codes,
so
that
it's
almost
like
a
low-level
library
that
you
could
call
photosynthesis
routine
with
kind
of
bare
bare
arrays
that
are
kind
of
agnostic
to
whether
you're
doing
a
big
leaf
model
or
or
something
more
fate's,
ish
and
I.
Don't
I
have
no
idea
of
what
what
technically
that
would
take.
L
I
I
mean
I
think
just
to
follow
up
in
that
bill.
I
think
we
discovered
a
small
meeting
yesterday
afterwards
too
started
to
try
to
synthesize
and
I
think
we
all
agree
that
that's.
That
is
the
target,
but
I
don't
think
we
agree
upon
is
whether
that's
a
target
for
v-0
v1.
You
know
I
kind
of
felt
like
we
were
saying
it
wasn't,
because
it
just
seemed
unrealistic
on
the
timescale,
but
I
do
feel
like
it.
I
What
I,
personally
would
like
is
a
list
of
to
see
at
some
point
is
a
list
of
things
that
that
could
be
done
too.
So
we
have
this
list
in
front
of
us
that
if
we
do
see
opportunities
to
start
working
on
them,
we
can
start
working
on
them
in
some
sort
of
priority
order
and
Rosie
I
think
was
suggesting
yesterday.
That
photosynthesis
should
be
that
the
first
priority
in.
F
F
D
J
N
The
clm
see
TSM
configuration
for
for
a
little
while,
because
I
think
that
that's
gonna
be
one
of
them.
The
big
shifts
for
crop
for
crop
growth,
but
you
know
maybe
we
start
with
something
like
photosynthesis
as
Rosie
was
suggesting
I
think
it's
it's
gonna,
be
much
better
for
things
to
be
consistent
across
all
vegetation,
so
things
like
photosynthesis
than
it
is
to
have
two
separate.
You
know
totally
separate
models.
O
O
No,
it
hasn't
been
tested,
okay
and
I
guess
what
I
was
thinking
was
that
maybe
so
I
think
Dave
or
somebody
brought
up
the
fact
that
if
one,
if
the
crop
model
is
running
with
active
nitrogen
and
fades
isn't,
then
there
will
be
nitrogen
balance,
issues
that
will
pop
up
and
I'm
I'm.
One
I'm
wondering
if
there
are
simple
hacks
around
that
such
as
you
know,
the
nitrogen
cycle
overall
in
the
model
is
active.
L
So
so
I
think
the
I
think
the
issue
might
have
arise
in
that
when
you
have
dynamic
land
use
as
long
as
the
column
areas
are
fixed,
I,
don't
think
that
I'm
not
sure
that
I
see
a
nitrogen
balance
issue
arising,
but
when
crop
areas
expand,
you
would
need
to
assume
something
about.
What
is
the
nitrogen
content
of
the
soil
into
which
they're
expanding,
and
you
could
do
something?
L
L
G
M
We
just
need
to
do
that
or
we
could
we'd
not
I
mean
need
to,
but
we
could
do
that
and
that
would
solve
this
problem
and
so
I
guess
I
would
I
would
advocate
that
we
plan
on
if
we
want
fates
running
with.
You
know,
next
to
the
carbonyl
that
we
plan
on
doing
something
like
that
and
until
such
time
we
have
scientific
confidence
in
the
representation
of
nitrogen
in
feeds.
M
The
second
point
that
I
wanted
to
make
about
this,
this
issue
of
like
this
sort
of
photosynthesis.
You
know,
and
this
sort
of
like
the
proliferation
of
photo
said
this
is
code,
throw
them
all
anything.
You
know,
I
agree,
you
don't
want
lots
of
different
photos,
codes,
I,
think
there
are
benefits.
You
know
just
purely
within
CLM
outside
of
fades
to
doing
this,
I
mean
you
know.
The
my
understanding
is
the
the
the
nature
of
this
recent
polygon.
M
D
N
That's
gonna
have
on
the
actual
crops
as
far
as
if
they
have
access
to
that
fertilizer
or
not,
but
that
might
be
important
feedback
to
atmospheric
chemistry,
and
so
is
that
something
that
since
we
don't
have
you
know
if
the
nitrogen's
and
that
turned
on
and
in
fates,
are
we
aware
of
what
you
know?
Do
we
just
abandon
the
the
fan
model?
Is
it
going
to
be
important
for
the
the
atmospheric
chemistry
or
is
that
something
that
it's
more
long-term
I
guess
thinking.
G
J
C
G
M
I
Think
that
there
is,
they
are
atmospheric
chemistry
group.
He
is
pushing
for
kind
of
closing
the
nitrogen
cycle
and
getting
into
emissions
as
a
as
a
relatively
high
priority
for
them.
Louisa
brought
it
up
in
her
overview
presentation,
so
we
should
just
be
aware
that
that
could
be
one
sticking
point.
D
J
G
G
G
And
then
you
know,
with
all
of
these
kind
of
meeting,
to
stack
up
before
we
can
run.
You
know
a
fully
couple
of
simulation
with
expired
geography
or
SP
mode,
really
about
the
illegal
flexes
and
that
evaluation
piece
I
want
to
save
until
you
know
after
ten
o'clock.
That's
that's
the
benchmarking
and
valuation
side,
so
I
wonder
if
folks
have
thoughts
on
what's
involved
in
putting
this
together,
if
there's
more
things
that
need
to
be
added
to
the
list,
just
under
more
technical
aspects
of
what
we
need
to
do
here,
Dave.
I
I
P
P
P
F
D
I
F
I
P
J
Yes,
it
was
about
PFT
performance
investigations
with
that
I
think
just
highlighting
the
need
to
benchmark
coexistence.
Poly
has
done
some
really
good
work.
Looking
at
that
in
California,
just
it's
a
it
can
be
a
challenge.
So
I
just
want
to
add
that
as
a
bullet
point,
coexistence
I
think
it
comes
up
later
in
benchmarking,
but
I
don't
know
if
we
need
to
so
we
can
talk
about
it
within
the
benchmarking.
L
Should
two
isotopes
both
carbon
and
water-
need
to
be
added
to
this
list,
I
mean
probably
probably
people
on
this
called
Charlie
and
others
can
comment
on
the
carbon
isotopes
water
isotopes
is
viewed
as
a
critical
need
for
for
CES
m3.
So
if
this
is
going
to
be
our
standard,
configuration
I
imagined
that
water
isotopes
need
to
be
included
in
fates
as
well,
for
others
can
maybe
comment
on
that,
and
it
might
be
that.
O
D
M
For
the
carbon
isotopes,
we
ought
to
be
able
to
get
pretty
easily.
We
you
know
we
can
try.
We
can
trace
how
all
the
different
carbon
pulls
through
through
all
the
pine
poles
already
in
Chalmette
azzam
code,
to
do
the
isotopic
discrimination
during
photosynthesis.
First,
either
team,
you
don't
have
radioactive
decay
for
radiocarbon,
but
that's
you
know
trivial
thing
to
do
so
bridge.
Well,
that's
that
we
could.
M
All
it
does
is,
is
GABA
can
be
conductance
back
to
CLM,
which
then
computes
the
transpiration,
and
so
it
would
effectively
use
for
whatever
I
stopped
with
discrimination
is
having
water
associate
with
transpiration
there.
You
know
there.
Okay,
that's
different
for
the
case
when
planted
relics
is
turned
on
what
we
do
then
have
prognostic
water
pools
that
we
apply
your
own
source,
sink
to
the
to
the
water
pools
in
the
soil
and
our
own
transpiration
flux
out
of
the
leaf
water
pool,
and
so
that
would
be
much
more
involved.
L
E
F
M
F
I
Maybe
the
other
thing
on
the
list
here
is
is
actually
the
cost
question.
I
would
be
nice
to
have
a
cost
estimate
to
know
whether
or
not
we're
talking
I
don't
have
no
idea
how
expensive
fates
is,
and
it's
different
configurations
relative
to
CLM,
and
if
it's
going
to
be,
you
know
a
factor
10
more
expensive.
We
we
kind
of
need
to
know
that
if
it's
gonna
be
about
the
same,
that's
then
we
don't
need
to
really
know
it.
D
I
M
I
I
M
E
D
P
P
J
J
If
you
change
the
decomposition
rate
of
the
fuels
which
then
transfer
to
the
host
land
model
and
decompose,
that
does
have
an
impact
on
how
long
the
fuels
are
retained
and
how
much
retention
you
get,
and
so
we
initially
we
had
like
way
too
much
carbon
down
in
the
soil
carbon
down
in
the
tropics
so
yeah.
We
need
to
look
at
this
and
then
gonna.
J
Was
it
was
excessive
and
so
right,
so
the
the
parameters
were
adjusted
so
that
it
would
decompose
more
rapidly?
Okay,
but
it's
worth
considering
that
this
is
you
know
it's
kind
of
vary,
regionally
and
so
settling
on
a
litter
fuel
decomposition
that
is
going
to
be
reasonable
globally,
but,
like
it's
gonna,
be
a
balance
and
testing
the
fire
model
is
going
to
help
determine
how
that
plays
out,
because
that
parameter
is
used
in
the
fire
model,
but
it
sort
of
determines
how
things
are
added
to
the
post
land
of
model
decomp.
So.
J
M
Couple
things
that
last
thing:
basically
it's
a
substitution
of
the
course
where
you
be
pooled,
so
when
phases
on
there
is
no
host
model.
Of
course,
we
recall
freights
uses
that
what
I
was
going
to
say
there
was
to
Eric's
point
I
mean
I,
wonder
if
you
know
I
think
the
because
more
sir
structurally
complex
grid
cells,
you
know
cost
more
to
run
I
wonder
if
there
is
some
way
to
do,
maybe
not
dynamic
load-balancing.
M
But
spatial
you
know
include
more
spatial
information
when
you
went
for
the
load
balancing
problem
on
you
know
these,
you
know:
grid
cells
in
the
middle
of
Sahara,
don't
have
any.
You
know,
have
one
one
patch
and
zero
cohorts.
You
know
therefore
great
cheap
to
run
versus
ones,
the
middle
of
the
Amazon
or
wherever
are
more
expensive
and
so
I
wonder
if
there's
a
way
to
mitigate
that
that
issue
through
sort
of
more
dynamic
load-balancing.
L
Well
so
the
way
I
mean
dynamic
load,
balancing
is
would
be
a
very
hard
problem.
First
I
mean
solvable
but
very
hard
to
like
in
terms
of
actually
a
over
time,
but
in
terms
of
smart
load-balancing
I
think
we
could
do
that.
I
mean
right
now.
It's
done
as
effectively
kind
of
a
round-robin
decomposition,
so
so
so
which
which
helps
with
that
right
so,
like
different
grid
cells
in
the
Sahara
would
be
distributed
to
different
processors
that
wood
as
opposed
to
them
all
being
on
the
same
processor.
But
you
know.
L
P
F
So
so
the
the
the
emerald
project
that
I
mentioned
there
is
kind
of
spinning
up
I,
don't
know
if
we
is
on
the
core:
either
Yeley
users
on
tweeze
and
O's
on
there
just
spinning
up
and
have
lots
of
interest
in
adding
new
public
functional
types.
For
various
reasons,
and
so
you
know
you
could
imagine
kind
of
like
having
a
mini
working
group
to
think
about
shrubs
or
mosses
or
whatever
we
felt
was
necessary.
There
wasn't
if
one
of
you
guys
wants
to
comment
on
that,
a.
F
Ongoing
craziness
of
the
c4s
algorithm
I
plan
to
look
at
drought
deciduous
grass,
and
this
isn't
an
exhaustive
list.
I
feel
like.
Maybe
we
sort
of
get
the
word
out
that
you
know
these
investigations
would
be
useful.
Then
we
could
sort
of
bring
together
a
team
of
people
to
work
on
their
areas
of
interest.
I,
didn't
kind
of
write
down,
obviously,
Jackie's
work
on
the
fire
tree
grass.
Coexistence
in
savannas
is
really
important,
but
yeah.
This
is
at
this
definition.
People
can
like
raise
their
hands
and
be
like
I
want
to
think
about
this.
G
J
Just
want
to
add
yeah
Thank
You
Rosie
I
have
been
looking
at
c4
grasses
in
the
tropics,
but
then
I
just
want
to
add
that
Nick
Smith
from
Texas
I
can't
remember
which
University
he's
at,
but
he
has
expressed
interest
in
being
involved
with
updating
the
c4
grass
photosynthesis
routine
and
that's
sort
of
yeah,
so
Nick
Smith
is
interested
in
pairing
up
and
doing
that.
I
know
he's
talking.
He's
talked
about
a
couple
of
folks,
so
we
could
add
his
name
to
it.
You
could
put
my
name
on
the
c4
grass.
J
I
can
follow
up
with
people
needs
chatted
with
Danica.
So
there's
a
few
of
us
and
then
also
mosses
folks
brought
up
mosses
yesterday
and
so
moss
Audrianna
and
I
have
been
talking
about
or
we're
going
to
add
moss
as
part
of
capturing
fire
dynamics
in
the
boreal,
and
so
we
can
stay
in
touch
with
the
folks
of
the
enrolled
project
to
coordinate
on
Moss,
or
we
can
see
the
two
different
variants.
I
J
So
the
within
the
big
leaf
clm
humans
are
included
as
a
component
of
fire
through
the
population
density
and
GDP
data
sets,
and
so
as
part
of
the
work
that
Sam,
Levis
and
Eric
did
to
get
the
lightning
data
set
in
there's
sort
of
a
framework
to
add
in
those
those
other
two
data
sets,
and
so
yes
I've
been
thinking
about
them
and
yeah.
It's
on
my
plan
to
do
that
this
year,
and
so
in
terms
of
fire
happening
globally.
J
It's
there
might
be
some
parameters
that
are
not
currently
spatially
variant,
that
maybe
we
need
to
consider
having
some
spatial
variance
or
adding
in
a
latitude
melt
like
temperature,
precipitation
controls.
Specifically
this
this
decomposition
parameter
I
didn't
realize
that
the
the
litter
pools
in
dates
were
a
pure
substitution
of
the
course.
What
do
you
debris
pool
that
actually
makes
things
a
little
bit?
J
J
J
Then
we
could
add
an
additional
data
set
that
would
improve
on
those
modified
and
it
you
know
it's,
but
the
the
the
links
in
the
in
the
fire
code
are
already
there,
and
so
we
have
an
idea
of
how
that
would
link
in
it's
just
a
matter
of
adding
the
functionality
on
the
data
set
side.
So
that
fans
can
read
that
and.
I
F
I
That
it
does
feel
like
the
decent
mount
of
work
is
gonna
need
to
go
into
mean.
Maybe
Jackie's
right,
it's
gonna
be
pretty
straight
forward,
but
I
know
Fong
had
to
do
a
lot
of
work
with
the
fire
model
to
get
it
with
a
much
simpler
file
model
to
get
it
to
work
globally,
and
she
was,
you
know,
painstaking
tuning
and
everything
like
that.
So
I
guess
we'll
just
see
where
we
get,
but
that
does
feel
like
that's
a
potential.
J
It's
going
to
need
a
lot
of
testing.
Polly
and
Lara
are
testing
fire
in
California
I've
done
a
lot
of
testing
in
the
tropics,
and
things
have
really
come
a
long
way
with
fire
photo
and
so
I'm
cautiously
optimistic
Audrianna
and
our
testing
fire
in
the
boreal.
And
so
you
know
we
have
a
good
representation
of
the
ecosystems
and
so
I
expect
well
we'll
have
a
lot
more
information
in
the
fall.
J
I
P
I
F
F
P
I
I
C
C
L
Creating
actually
creating
the
short
name,
cops
and
aliases
is
easy.
The
the
hard
part
which
is
eric
is
kind
of
identifying
is
what
capabilities
would
be,
would
be
needed
for
that
I,
don't
know
if
there's
anything
else
that
would
be
needed
like
if
we
can
get
it
working
in
an
eff,
cops
that
I
think
it
would
be
just
as
easy
to
do
a
because
I,
don't
think
there's
any
additional
functionality
that
would
be
needed
at
that
point
is
that
is
that
right,
yeah.
C
I
You're
right
it
doesn't
it's
just
it's
a
question
of
whether
or
not
the
eff
comps
that
really
the
cam
CLM.
If
there's
anything
that
is
we're,
gonna
run
it
any
barriers
which
is
again
a
good
reason
to
start
trying
now
and
and
probably
we're
getting
out
of
ourselves.
But
but
a
promise
for
your
chemistry
simulation
would
be
the
real,
the
real
test
cuz.
Then
it
uses
everything
or
wacom.
K
Yeah
I
guess
this
is
coming
back
from
a
fire
question
so
right
now,
fire
in
CLM
is
just
on
or
off
and
I'm
wondering
if,
if
there'll
be
more
granular
control
over
fire
in
the
you
know
in
in
faiths
like
can
you
control
whether
certain
patches
or
you
know,
do
our
are
or
are
not
expired,
exposed
to
fire
or
anything
like
that?
I'm?
J
So
we
lost
your
audio
for
a
second
there,
but
the
the
comment
that
Dave
was
making
about
the
human
fire
interactions,
so
the
population,
the
the
data
sets
that
come
in
that
account
for
variability
in
suppression
as
well
as
activation
of
fire
sort
of
handle.
Some
of
what
you're
talking
about
it's
handled
more
at
a
regional
scale,
so,
like
all
of
the
United
States,
has
a
suppression
term
applied
to
the
potential
ignitions,
and
so
there's
it's
not
the
type
of
granularity
that
you're
talking
about
I.
J
Think
there
is,
you
know,
but
limiting
it
so
that
it
does
not
burn
down
all
the
crops.
That's
something
to
consider.
So
at
this
point
you
know
fades
is
just
natural
vegetation,
so
we
haven't
had
to
think
about
that.
But
yeah,
that's
something
we
should
consider.
Are
there
certain
vegetation
types
or
you
know,
but
cells
that
should
not
be
exposed
to
fire,
but
I
know
that
slash
and
burn
is
an
important
component
in
some
areas
so
yeah.
We
should
think
about
that
aspect.
So.
L
P
O
Haven't
visited
that
in
a
while,
so
I'll
I'll,
try
to
remember
right
now,
I
think
that
it's
I
don't
believe
that
it's
going
through.
Is
it
going
through
the
hole?
It's
definitely
an
option
within
the
within
within
families
fire
model,
but
it's
it's
very
specific
and
I.
What
I
don't
remember
is
it's
coming
up
with
how
much
biomass
I
guess
is
being
removed
in
agricultural
fires
are
really
what
they're
burning
the
the
residue
that's
left
behind
and
so
yeah
I
guess
that's!
So
that's
what
I
remember
right
now.
I
know
I
can't
say
more.
O
O
I
O
If
I
could
jump
in
on
that,
I
think
what
maybe,
what
you're
referring
to
bill
is
actually
the
the
whole
lightning
issue,
because
it
used
to
be.
That
was
one
of
the
recent
changes
that
Eric
and
I
made,
which
is
that
we
removed
the
the
requirement
that
the
lightning
data
set
can
only
be
read
in.
If
you
see
n
was
true.
If.
L
M
Yeah
so
I'm,
just
trying
to
in
my
head,
sir
wrap
up
the
wraps
to
move
towards
that
kind
of
like
the
the
who's
who's
actually
gonna
do
stuff,
or
what
how
much
efforts
can
be
required
for
it.
For
some
of
these
more
important
things
and
I'm,
just
trying
to
figure
out
like
like
a
lot
of
these
things
seem
like
they're,
relatively
minor,
and
the
one
thing
that
I
feel
like
is
just
be
really
big
or
is
sort
of
unknown
scope.
M
I
Some
my
suggestion,
charlie,
is
that
is
that
we,
the
tss
group
Eric,
made
myself,
maybe
Keith,
maybe
bill
to
since
he's
got
some
familiar,
do
a
start
to
sometime
in
the
next
month
or
so
to
try
to
scope
that
out.
I
think
that's
about
all!
We
can
do
that
may
or
may
not
how
hard
it's
going
to
be
until
we
scope
it
out,
I,
don't
think
we
can
say,
but
Bill
do
you
or
Eric?
Do
you
have
any
other
idea,
I
think.
I
I
agree,
I
think
it's
I
think
I
kind
of
put
in
this
category
if
Eric
starts
to
work
on
on
comp
sets
and
where
that
really
means
getting
all
these
configurations
to
work.
That's
within
that
set
of
tasks
and
with
that
most
likely
being
the
most
challenging
one.
It's
getting
a
lot
of
the
work,
but
then
I
mean.
M
Actually
modular,
possibly
some
of
the
some
of
the
history
fields,
I
think
or
we've
been
we've
had
a
thing
up
that
we
wanted
to
fault
them.
No,
no
data
over
non
fades
columns
other
than
zero
one
fades
columns,
but
you
know
that's
the
gun.
I
think
all
the
other
stuff
is
pretty
much
I
believe
it's
worked
out.
Okay,
great.
I
M
There
are
two
classes
of
variables:
they're,
the
ones
they're
handled
wholly
within
fates,
which
are
says
don't
sort
of
super
structure
built
on
top
of
this
history.
Infrastructure
into
those
variables
sort
of
by
definition,
are
only
defined
on
fades
columns,
and
then
there
are
the
variables
that
are,
you
know
coming
out
of
CLM
like
soil
temperature,
or
you
know
other
things
that
are
they're
sort
of
say
and
be
here
like
or
like.
Less
obviously,
fates
like
you
know,
III
in
ela
are
coming
out
of
CLM
because
those
are
actually
being
defined.
M
Fades
is
giving
that
information
to
CLM,
and
so
those
does.
Those
other
ones
are
just
thirty
cz.
You
know
those
are
defined
on
all
the
columns.
You
know
so
so
they'll
be
for
those
ones
either.
My
et
there
might
be
some
it'll,
be
yeah
in
principle
should
work,
but
you
know
third
call.
If
they're
those
variables
are
being
calculated
in
a
given
column,
they
should
be
history.
Structure
should
work
for
that,
but
I
haven't
tested
is.
M
Right
now,
there's
it's
only
sort
of
the
only
way
to
know
is
by
if
the
variables
are
defined
within
fates
and
within
the
fades
history
interface
mod.
That's
how
you
know,
but
we
have
been
talking
about.
You
know
names
facing
all
the
variables
to
say
you
know
put
fades
or
something
like
that
in
from
we
haven't
done
that
yet.
But
it's
thinking.
I
Transpiration
depends
on
whether
you're
using
my
jobs
or
not,
but
if
there
was
transpiration
from
fates
but
transpiration
from
the
crop
and
increase
the
leverage
you
would
have
them
both
anyway.
It
sounds
like
we
might
need
a
data
subgroup
at
some
point
too,
unless
you
guys
are
confident
that
it's
all
working
really
well.
J
G
G
G
L
I
I
There
will
be
one
thing,
but
hill
slopes
doesn't
work
with
land
use
and
it
doesn't
work
with
fates
and,
although
I
think
it
might
work
with
face
pretty
quickly
so
anyway,
it
just
needs
to
be
part
of
the
whole
hopper,
but
I
don't
think
we
should
try
to
let
that
whole
thing.
We
should
let
that
stop
us
and.
I
M
Also,
the
interest
of
having
names
for
the
land
use.
You
know
W
breaking
III
SM.
So
so
you
can't
really
be
involved
in
the
in
the
clm
stuff
and
and
it
would
be
nice
to
get
an
in-car
name
next
to
Charlie.
You
know
Alan's
involved
right,
but
but
but
it'd
be
nice
to
get
to
have
a
third
name.
That's
an
in-car
name.
There.
M
I
G
It's
getting
close
to
10
and
close
to
a
break.
One
thing
that
didn't
come
up
in
any
of
the
summaries
of
things
that
we
want
for
questions
2
&
3,
but
they
came
up
in
questions
to
come
to
through
sakes.
I'm
sorry,
4
through
6
were
single
point
capabilities,
and
then
you
know,
presumably
that's
trivial
to
add.
Simply
capabilities
to
these
global
transit
Oracle
runs
but
I
also
know
how
commonly
that's
done
in
it's
obviously
kind
of
things
and
sudden
CTS
I'm.
Making
special
needs
be
done.
D
F
D
G
G
D
G
G
D
G
D
G
G
G
G
G
D
G
M
G
F
G
Great,
let's
go
ahead
and
get
get
started
again.
So
the
purpose
of
this
I,
let's
try
to
keep
to
45
minutes.
That
might
be
a
challenge,
we'll
see
how
we
do
and
for
the
folks
that
are
joining
us.
You
know
for
the
first
hour
and
a
half
we
kind
of
talked
about
these
short-term
milestones
versus
the
long-term
vision
and-
and
you
know
maybe
the
long-term
vision
is-
is
Faith's
CES
m
CS
m
with
fates
in
it
kind
of
this
red
flag
down
here
and
really
what
I'm
trying
to
focus
on
more
is
this
blue
flag?
G
So
Jackie
went
through
and
summarized
a
bunch
of
the
responses
to
questions
four
and
five
from
the
summary
and
I
put
some
of
the
highlights
on
here,
and
so
you
know
for
this
kind
of
version.
Zero
global
scale,
fully
transient,
see
TSM
fates,
run.
The
kind
of
high
level
thing
is
grid
cell
level,
comparisons
of
kind
of
a
standard
benchmarks
that
are
used
in
a
Liam,
so
things
like
GP
ete,
but
you
flex
everything
else
at
the
global
scale.
G
M
G
There's
a
time
on
the
list
and
I
have
no
idea
how
to
prioritize
any
of
these.
You
know
the
one
to
me
that
makes
the
most
sense
like
Rosie.
Had
this
in
her
talk
when
you
have
a
fully
free
running
fates,
you
know
you
get
these
obvious
biases
and
where
plants
are-
and
we
can
tell
that
by
our
eye,
but
it
would
be
nice
to
have
something.
That's
a
metric,
that's
an
island
that
you
can
get
a
score
or
whether
you're,
better
or
worse
than
another
version
of
fates.
As
we're
developing.
G
You
know
the
parameterization
of
things
P
of
T's,
so
I
put
that
at
the
top
of
the
list.
Like
you
know,
using
our
current
PFD
distribution
map,
as
opposed
to
an
input
model
is
using
it
to
benchmark
to
the
model
seems
like
to
me
the
first
level
kind
of
benchmark
that
we
need,
and
then
a
number
of
groups
had
a
ton
of
ideas
about
other
benchmarks
and
so
I
guess
the
the
question.
N
D
L
On
that
first
point,
it
strikes
me
that
that
you
want
more
than
just
a
yes/no
for
each
kind
of
grid
cell
and
PFT,
but
like
if
you
have
a
broadleaf
deciduous
versus
broadleaf
evergreen
tree
in
the
tropics.
That's
not
quite
as
bad
as
if
you
have
like
a
grass
where
you
should
have
a
tree.
How
do
you
quantify
that?
How
bad
you
are
in
terms
of
your
PF
tees.
G
A
Yeah
I
guess
this
might
be
separate,
but
I
was
thinking
of.
Maybe
I
was
like
a
first
step
we
could
have.
You
could
come
up
with
a
set
of
like
super
sites
where
we're
running
like
site
level
or
very
small
region
level
simulations
with
both
based
CLM
and
CLM
faiths
and
we're
comparing
you
know.
So
these
super
sites
would
be
where
we
have
really
good
data
on
energy
and
water
fluxes,
as
well
as
vegetation
characteristics
and
structure.
G
The
super
centers
you
have
made
it
under
this
list,
but
yeah
I
think
that's
a
great
idea.
Yeah
I,
don't
know
what's
happening
with
ng
tropics.
You
know
this
in
Kearney
on
partnership.
That
I've
talked
about
is
hopefully
going
to
come
to
fruition
in
the
next
year.
So
that
to
me
strikes
me
as
a
sweet.
G
You
know
nyan
tower
sites
or
a
suite
of
sites
that
have
a
lot
of
data
to
potentially
bring
to
bear
the
specific
fates
relevant
data
that
neon
has
I'm
a
little
I,
don't
know
what
it
would
be
other
than
the
LP
and
I.
Don't
know
how
to
translate
AOP
data
into
something
that
mean
is
being
pulled
for
fates,
but.
G
Sorry
AO
p
is
the
neon
airborne
observatory
and
it's
airborne
observatory
platform,
so
they
fly
it
with
lidar
and
hyperspectral
I
know.
Kyle
has
worked
with
that
data.
Quite
a
bunch
I,
don't
know
how
involved
Kyla
is
with
fates,
work
and
then
in
the
chatty
gentleman
looks
like
a
salsa
segment.
He
on
his
inventory
data
so
that
you
know
they
have
on
the
ground
measurements
that
have
the
airborne
Observatory,
but
in
the
inventory,
data
is
sometimes
co-located.
With
the
Flex
tower.
Aop
includes
the
Flex
footprint
of
the
Flex,
our
plus
a
much
broader
area.
M
So
the
larger
questions
are
gonna
ask.
Is
that
that's
exactly
the
point
is
is,
is
that
you
know
I
think
we're
as
big
leaf.
Clm
is
really
you
know
been
tested
against.
You
know.
Any
ad
covariant
side
is
really
useful,
I
think
for
fades.
It's
really
the
combination
of
any
covariance
and
inventory
some
door
lidar,
some
some
structural.
You
know
that
the
pairing
of
those
two
things
so
I
think
really
focusing
on
where
the
sites
where
we
have
both
both
fluxes
and
structure
and
make
that
our
superset.
G
G
C
We
talked
about
this
in
our
group
yesterday
to
me.
Those
are
two
really
different
things
right.
If
you
can
get
tree
high
data,
that's
distributed,
that's
one
test
of
fates
across
space,
but
these
site
level
tests
would
be
you'd
a
be
able
to
bring
a
lot
more
of
the
inventory
data
or
the
biometry
data
that
might
be
available
at
a
site.
C
But
we
can
still
do
some
of
these
things
across
space
like
tree
height
and
maybe
Jedi,
oh
yeah,
that
would
be
the
tree
height
Dana
anyway,
so
some
distributed
things
would
sort
of
fit
into
I
Liam,
as
it
is
now
and
others
would
be
looking
in
more
detail.
It
put
specific
sites,
we
didn't
call
them
super
sites,
but
they
should
be
in
the
list.
D
O
Yeah
what
I
wanted
to
bring
up?
It
goes
back
to
the
PFC
distributions,
which
is
more
of
what
I've
worked
on
over
the
years
with
the
old
generation.
The
gbm's
I
was
remembering
back
in
the
early
days
when
sandy
Harrison
was
involved
in
this
kind
of
work
that
she
had
proposed.
The
use
of
a
statistic
called
the
Kappa
statistic
which
looks
at
the
simulated
PFD
distributions
versus
the
observed
and
the
statistic
performed.
O
Q
Mean
we
have
kicked
the
can
down
the
road
a
little
on
this
superspy
idea,
but
nothing
has
kind
of
just
solidified
that
something
that
we
just
run
a
lot
frequently
as
we
do
like
the
globally
gridded
stuff.
But
it's
all
like
poised
and
ready
ready
to
go.
It's
more
I
think
about
getting
people
who
are
really
interested
in
looking
at
the
output
and
making
sure
that
we
code
stuff
that's
relevant,
like
the
things
you're
talking
about
and
and
I,
don't
think
it
will
be
it'll
be
hard
to
get
that
going.
D
N
Think
another
important
component
of
a
super
sight
idea
is
having
some
disturbance
history.
Since
you
know
the
dynamic
competition,
then
it's
important
to
know
what
has
happened
at
a
site,
so
we
can
impose
that
on
the
model
and
I
second
Sam
died,
eeeh
of
a
sort
of
like
an
image
classification
way
of
looking
at
PFT
distributions.
M
Yeah,
since
you
know,
since
neon
and
FIA
data,
it's
all
sort
of
you
know,
America
focused
and
we
were
trying
to
go
about.
You
know
I,
think
the
the
the
global
networks
to
really
work
towards
the
forest
geo
in
any
tropics.
We've
really
been
focusing
on
the
tropical
forest
geo
sites
as
sort
of
our
key
testbed
settings.
But
you
know
the
first
years,
but
not
just
tropics
is
they're
all
around
the
world,
so
I
think
that
and
and
and
you
know
not
all
them
have
coherency
in
it.
G
M
J
I
just
want
to
mention
the
bad
database,
so
it's
a
llama
tree,
but
it
is
distributed
globally,
and
so
you
could
it's
good
to
have
that
as
a
site
level,
bad
location
for
actual
data,
I
haven't
looked
at
it
extensively
myself,
so
somebody
else
who's
spent
a
lot
of
time.
Looking
at
it,
I
know,
audriana's
looked
at
it,
I
think
Polly's
used
it
to
Laura.
J
M
G
D
G
Yeah
I
put
it
at
the
bottom
of
the
list
because
it
was
maybe
outside
the
bounds
of
this
version.
Zero,
but
maybe
it's
a
version,
0.1
type
of
question:
if
there's
somebody
who's
interested
in
doing
the
work,
I
think
that's
the
other
challenges
on
everyone.
In
my
mind,
every
one
of
these
datasets
needs
somebody
to
kind
of
be
its
champion
for
sit
in
Diane
Lambert.
You
know
massages
the
dataset,
so
it's
usable,
Joshua
I,
think
your
hands
are
no
prove
it.
K
K
K
That
result
in
getting
the
P
of
T's
where
they
are
since
chances
are
the
pts
are
not
going
to
be
perfect.
They're
not
going
to
be
exactly
what
we
expect.
There's
a
gonna
be
a
lot
of
question
about
the.
Why
and
if
there's
so,
do
you
make
it
a
standard
part
of
the
evaluation
to
assess
against
a
you
know,
unknown
set
of
sites
that
actually
our
successional
and
test
the
competition
dynamics
in
a
different
way.
G
D
C
Think
all
of
these
data
sets
have
the
site
history
embedded
in
them,
and
so,
if
the
better
question
might
be,
what
is
are
not
exactly
spin
up.
But
what
is
the
sort
of
paces
that
we
put
fates
through
before
we
compared
to
modern
day
observations,
and
we
might
need
a
standard
for
how
we
get
fates
to
what
we
consider
to
be
the
point
we
compared
to
the
modern
day,
data
or
paleo
data,
because
that's
in
the
global
data
sets
as
well
as
the
site
level.
C
Data
sets
might
just
might
just
be
a
much
harder
bar
to
meet
at
the
site
level.
If
you're
trying
to
get
the
distribution
of
tree
size
is
correct,
but
some
of
these
globally
distributed
data
sets
they're
still
specific
to
things
that
are
emergent
from
faiths
that
wouldn't
be
emergent
from
CLM
and
are
still
going
to
depend
on
how
we
get
the
model.
To
that
point,.
G
J
Can
comment
yes,
so
just
in
reference
to
Joshua's
point
about
succession
and
Abi's
point
about
spin-up
and
what
the
current
state
is.
It's
important
to
keep
in
mind
temporal
scales,
as
well
as
the
spatial
scales
and
so
depending
on
the
spatial
scale
of
the
data.
You
will
capture
some
of
that
successional
variability,
so
different
size
distributions
different
competitive
success.
But,
as
you
come
up
in
spatial
scales,
you
it's
important
to
remember
that
you're
trying
to
capture
a
forest
landscape,
we're
not
trying
to
capture
a
specific
site
and
disturbance
history.
J
You
can
do
that
more
with
a
site
level.
Point
simulation
that
you
initialized
from
data,
but
when
you're
looking
at
the
broad
landscape
patterns,
it's
a
it's
a
different
type
of
comparison,
and
so
you
have
to
keep
that
in
mind.
The
spatial
scale
of
the
data
that
you're
using
as
well
as
the
spatial
scale
that
you're
simulating
at
you
know
and
so
resolution
is
important.
If
we're
talking
about
half
a
degree
or
one
degree,
you
know
it.
J
M
See
also
that
you
know
in
thinking
about
how
to
do
this,
being
you
know,
being
cognizant
of
the
sort
of
geographic
gradients
on
the
scale
of
disturbance
or
genes
where
you
know
boreal
forests,
we've
got
these
gigantic
fires
and
tropical
forests.
You've
got
these
relatively
much
smaller
blowdowns
is
the
dominant
disturbance
agents
kind
of
you
know,
create
and
members
of
with
a
whole
scale
between
or
a
whole
spectrum
in
between,
and
so
how
we
evaluate.
M
N
Yeah
very
similar
in
that
I
see
it
as
nested
from
the
point
through
regions
that
have
some
sort
of
common
disturbance
regime.
You
know
globally,
where
you
can
look
across
a
disturbance,
gradient
and
maybe
using
something
like
land
Trender
that
says.
What's
the
disturbance
frequency
do
we
get
a
relationship
between
what
whatever
the
metric
is
percent
of
pft
composition
in
a
grid
cell
and
the
disturbance
frequency?
You
know
that
we
can
take
it
across
a
spatial
scale
of
benchmarking,
where
we're
making
taking
advantage
of
where
the
data
exists.
G
B
B
But
we
expect
some
pattern
of
distribution
of
what
fates
produces
to
correspond
to
some
patterns
of
distribution
and
density
and
gap.
Enos-
and
you
know
all
of
those
things
on
on
the
landscape
from
observations
and
some
of
those
kinds
of
landscape
metrics
would
be
a
useful
addition
to
I'll
am
and
I.
Don't
think
it
would
be
very
difficult.
Most
of
them
are,
you
know
mostly
straightforward
statistical
quantities.
Q
It's
more
matter
discussing
what
does
it
make
sense
to
compute?
You
know,
because
we
can
talk
about
all
these
theta
sets
and
then
but
visit
reasonable
to
expect
a
model
to
be
able
to
capture
I
guess
we
can
compute
any
kind
of
statistic
very
easily.
It's
not
that
part's
not
hard,
but
trying
to
come
up
with
something
that
helps.
You
all.
Do
your
work
or
helps
capture
some
important
aspect
of
the
performance.
It
can
be
tricky.
E
Think
one
of
the
challenges
is
that
all
of
these
different
plot,
all
these
plot
networks,
there
are
multiple
of
them
and
they're
their
data
access
policies
are
different,
so
there's
some
real
like
there's
some
real
work
to
do
and
I'm
not
sure
it's
like
this
v-0,
it's
gonna
be
accomplished
for
this
v-0
evaluation.
I
guess
is
what
I'm
saying,
but
it
does
feel
like.
There's
really
some
sort
of
concerted
effort
needed
to
kind
of
move
toward
a
place
where
we
can
have
sort
of
globally
distributed
information
on
structure,
because
this
is
you
know
this.
D
I
D
I
F
I
was
just
gonna,
ask
Nate
again
about
like
this.
This
idea
of
getting
site
data
to
populate
the
eylem
tools
seems
really
important
to
me
and
I
wondered
if,
like
that's
a
sort
of
serious
actionable
item,
is
to
maybe
organize,
like
a
smaller
follow-up
meeting,
to
think
about
like
what
you
guys
are
looking
for
and
what
data
we
all
know
about.
It
can
populate
it
and,
like
I,
feel
like
that's.
That's
a
super,
exciting
thing
for
us
to
think
about
so
yeah.
What
are
your
thoughts
on
that.
Q
You
know
we've
done
currently
as
we've
looked
at
a
mere
flux
data,
that's
where
we
got
like
some
half-hourly
their
variables
and
and
looked
at
diurnal
cycles,
and
things
like
that,
but
I
think
for
me,
I
don't
have
a
very
good
handle
on
what
it
is
exactly
that
you
want
to
benchmark
right.
So
the
kind
of
data
that
you
collect
depends
on
what
it
is.
You
want
to
test
about
your
model,
so
some
of
the
discussion
just
previously
you
know
you
want
to
look
at
that
structure.
I.
Q
Think
yeah
Abby
had
a
comment
there
as
well.
So
I
think
like
there
are
some
data
sources
that
we
could
go
in
and
get
data
from
and
I
could
tell
you
lots
about
GPP
in
various
different
ways,
but
if
I
do
all
of
that?
Do
you
care
about
those
results?
You
know
it's
kind
of
like
you
got
to
go.
Look
for
the
data,
that's
going
to
give
you
some
notion
about.
Q
F
F
Know
that
make
sense.
It's
also
like,
like
sort
of
talking
back
to
what
Laura
was
saying
about
how
easy
it
is
to
get
these
scaled
relationships
after
the
plot
network.
I
think
Shawn
mentioned
yesterday.
The
sort
of
the
sort
of
variable
where
variable
two
variable
relationships
could
also
help
us
out
here.
You
know
not
trying
to
get
individual
sites
right,
but
just
trying
to
get
the
relationship
between
some
sort
of
climate
variable
and
some
aspect
of
sort
of
the
biotic
properties.
F
You
know
like
the
stuff
you
already
plot
a
nylon
for
the
glow
runs
and
so
that
flux
tape,
the
flux
data
that
you
have
is
that
when
you
go
to
I
low
and
you
click
on
like
GBP
and
there's
like
a
bunch
of
different
things,
and
one
of
them
is
flux
net.
Is
that
it's
that
what
you're
talking
about
it's
already
in
that.
Q
Dated
that
data
was
that
I
think
somebody
processed
that
where
was
a
monthly
product?
No,
no.
We
also
wrote
something
that
harvested
Ameri
flux
and
grabbed
all
of
their
half-hourly
data.
So
it's
much
much
finer
time.
Resolution
and
I
have
some
some
comparisons
that
Shawn
Serban
has
been
working
with,
but
that
don't
they
don't
sit
like
in
that
they
don't
sit
like
in
the
standard
analysis
because
yeah
the
model
results.
We
have
don't
have
that
kind
of
time
resolution
so
that
could
be
run
now
interesting.
F
Q
F
H
P
D
So
we
already
use
a
lot
of
observations
at
very
many
different
spatial
and
temporal
scales.
They
don't
have
to
be
gridded,
they
can
occur.
You
know,
as
the
light
hour
flies
over
the
forest
or
the
landscape.
We
could
use
them
in
their
native
resolution.
Trying
to
relate
those
observations
of
particular
locations
to
the
right
PFT
in
a
coelom,
see.
Tsm
grid
cell
is
kind
of
the
tricky
part,
and
we
would
need
to
build
on
that,
but
the
framework
is
there
and
you
don't
actually
have
to
modify
a
coelom
state
to
do
this.
D
You
could
run
dart
in
and
evaluate
only
framework.
So
if,
whatever
reason
you
want
to
just
evaluating
its
observations
without
actually
coercing
the
clm
state
with
the
information
in
those
observations
we
the
framework
is
basically
there
and
it
happens
at
different
time
scales.
We
already
use
Andy
Fox
has
already
done
experiments
with
the
Ameri
flex,
data
at
30,
minute
resolutions
and
we're
also
doing
stuff
with
GPP
monthly,
and
there
was
biomass
observations
at
an
annual
scale.
So
that's
that
you
know
that's
the
10-second
pitch.
I
Yeah,
so
on
thanks
Tim,
that's
that's
good
good
info.
It
feels
like
named
Rosie
that
won't
be
nice
to
maybe
you
guys
have.
This
already
is
just
a
site
level,
only
I,
Lam
configuration
and
and
seeing
what
that
looks
like
and
I
felt,
like
that's
what
happened
with
the
global
one
is,
and
once
people
started
seeing
it
I
can
start
seeing
how
you
can
tap
in
much
easier.
So
what
the
plus
look
like,
what
the
Diagnostics
are
and
then
you
can
start
thinking
about
these
stand
structure
how
it
builds
on
top.
I
So
it
sounds
like
care
to
have
that
I.
Think
like
what
Rosie
said
like
a
demonstration
if
it
exists
a
demonstration,
if
it
doesn't
exist
and
maybe
discussion
about
how
do
you
get
to
that-
would
really
kind
of
move
this
along
yeah
YG's
chemistry,
you
can
do
it
for
fluxes,
then
you
can
search
in
four
states,
and
so
that's
one
point.
The
second
point
I
want
to
make
was:
was
I
I
wonder
if
what
we
could
try
to
come
out
of?
This
is
like
you
know,
I
feel
like
an
island.
I
I
There,
like
can
we
identify
two
global
data
sets
that
would
be
new
and
unique
domain.
You
know
tackle
with
the
PFT
distributions,
like
seems
like
one
of
them
for
sure.
Is
there
a
second
global
data
set
that
could
hit
something
and
that
fates
is
doing,
does
not
capture
by
a
big
leaf
model
that
we
could
try
to
add
as
it
targeted?
You
know
next
several
month
in.
M
I
think
we,
these
come
up
a
bunch
in
our
California
core,
not
bunch.
It
came
released
when
three
California
calls
it.
You
know
the
the
classic
light.
Our
definition
of
height
is
this
concept
of
colores
high,
which
is
a
quadratic
mean
to
stem
down
there
waited
plant
height
and
that's
something,
that's
something
that
we
can
in
principle:
output
from
feeds.
We're
not
we're
not
currently
doing
that,
but
I
think
there's
a
broader
point
to
be
made,
which
is
that
face.
M
Has
all
this
structure
and
all
this
like
sort
of
sub-grid
scale
stuff,
and
we
need
to
think
about
how
we,
how
we,
you
know,
get
this
kind
of
the
this
kind
of
you
know
the
the
high.
Whatever
these
quantities
are,
how
do
we
best
do
this
mapping
I?
Think
you
know
outputting
and
lorries,
how
you
or
you
know
whatever
the
precise
weighting
kernel
that
lidar
sees
whether
that's
you
know
the
okay
they're
weighted
by
canopy
area
rather
than
basal
area
or
whatever
you
know,
I
think
that
could
go
a
long
way
toward
doing
this.
M
D
J
A
A
Well,
yeah:
well,
biomasses!
Yes!
So
where
will
I?
My
P
is
one
of
the
VI
ism
Jedi,
so
I'm
I'm
gonna
toot
its
horn,
but
so
there
are
lots
of
ongoing
work
for
turning
Jedi
waveforms
into
biomass,
but
you
can
also
get
different
structure
measurements,
so
we
can
get
like
the
full
canopy
composition.
So
we
can
by
looking
at
like
95th
percentile
height
and
10th
percentile
height,
you
can
get
the
full
wave
form
of
the
canopy
structure.
M
You
see,
we
can
output
all.
We
can
output
all
those
percentiles
as
well
from
fades
as
well,
so
we
can
do
it,
but
it's
okay.
So
as
well
as
being
a
blurry,
though,
just
because
some
of
the
some
of
this
you
know
percentile
stuff
might
be
a
little
bit
overkill
for
the
v-0
like.
What's
the
simplest
thing,
we
can
start
with.
K
M
K
G
G
Are
there
people
that
can
want
to
volunteer
to
help
and
I
champion
these
data
sets
or
analyses?
You
know
to
support
Nate
and
I
keep
looking
at
Nate's
picture
forests?
You're
lucky
you
don't
have
your
video
on.
You
know
these
guys
have
a
lot
going
on
already.
So
I
think
you
know.
If
we
want
this
stuff
to
get
into
island,
we
need
to
help
make
it
happen.
Q
I
Just
say:
I
mean
I've
worked
with
me
and
a
bunch
on
things
and
I
think
it's
a
pretty
good
works
really.
Well,
if
you
know
we
have
an
expert
who
understands
the
data
sets
and
what
the
Diagnostics
should
be
in
the
knee
can
can
usually
handle
the
technical
aspects,
and
really
it's
just
a
couple
of
iterations
out
these
things
may
not
be
that
hard.
Just
need
somebody
to
champion
it
and
I
would
because
I,
don't
think
I'm
the
right
person
to
do
it
because
I
don't
understand
the
stands
for
adidas
f50.
M
M
Q
And
that's
another
kind
of
obstacle
that
comes
up
when
we
start
trying
to
integrate
new
data
sets,
especially
for
sites
I've
got
a
lot
of
output.
That's
globally
gridded,
but
inevitably
I
will
get
some
site
level
output
at
sites
where
I
have
no
modeling
runs,
and
so
it
can
be
a
challenge
because
I
don't
run
those
models
myself
either.
So
that
can
be
a
kind
of
an
obstacle
to
kind
of
pushing
the
ideas
forward.
It's
somebody
who
can
you
know
get
some
results
turned
out
at
those
sites.
If
it's
site
level,
output.
C
G
E
N
E
G
E
E
N
You
can
definitely
put
me
on,
as
Laura
said,
you
know,
to
share
what
we're
going
through
and
learning
at
a
much
smaller
scale
and
Chon
gang
and
I
are
also
talking
about
this
statistical
comparison
today,
after
this
later
this
afternoon,
so
I
know
he's
been
thinking
about
this.
A
lot
too
so
yeah
we'd
love
to
be
involved,
although
even
with
the
caveat
that
I
probably
won't
do
any
global
simulations
in
the
near
term.
No.
D
H
N
H
Yeah
I
can
dig
up
a
reference
for
that
and
also
with
the
stand
structure,
Laurie,
psych,
Jedi,
I,
assume
that
means
Jedi
lidar.
What
I
think
we've
been
doing
with
with
with
Edie
is
thinking
about
how
to
compare
distributions.
So
you
know,
in
the
same
with
face
you
can
spit
out
the
size,
age,
class,
distribution
and
pft
by
earth.
Sorry
la
vie,
honey,
so
you
can
in
Jedi
provides
the
same.
H
So
you
could
one
things
we've
been
think
about
us
trying
to
compare
for
a
particular
column
or
plot
or
cite
the
comparison
between
what
Jedi
it's
terms
of
la
I.
Distribution
are
structural
distribution
and
what
fates
can
generate
and
I
think
I
think
you
could
automate
that,
to
the
extent
that
there's
data
available
for
any
one
location,
you're
modeling,
the
challenge
is
I-
think
right
now
with
Jedi's
discontinuous
observations,
so
it
might
require
a
bit
of
a
first
pass
at
figuring
out
where,
where
there's
good
coverage
and
then
targeting
model
simulations
for
those
locations,
I.
D
Was
gonna
say
also
for
canopy
height
we're
using
the
eye
set
data
in
CLM
as
in
the
satellite
phenology
mode?
That's
how
we
set
up.
Can
it
behind
me
like
that
I
said,
because
as
a
product
and
their
global
product
to
tree
hides
that's
globally,
that's
it
what
so
it's
a
one-time
thing,
but
it's
a
global
data
set.
That's.
D
D
D
J
D
M
Yeah
I
guess
the
the
one
thing
so
I
was
there
all
this
discussion
of
sort
of
PFT
distributions.
You
know
I,
think
you
know
part
of
what
we've
been
fighting
against
with
with
fates.
Is
this
sort
of
idea
of
like
canonical
PF
T's
as
being
this
would
be
all
and
end
all
of
what
we
mean,
and
so
I
don't
want
to
go
too
far
down
this
relative
saying
well
now
we
actually
want
to
make
sure
we
can
get
these
sort
of
canonical
e
defined
PF
T's
as
our
metric
and
I
think
I
mean
oh
yeah.
M
You
know
to
some
degree
affair,
yeah
like
trees
and
grasses,
but
but
where
we
can
relate
into
less
ambiguous,
you
know
physically
to
find
things
like
canopy
height.
Well,
you
know
it
might
be
a
higher
higher
priority,
just
because
that's
then
extensible
when
we
move
beyond
these
canonical
PF
T's
towards
more
complex
P
of
T's.
G
N
Just
wonder
if
we
want
to
include
something
about
disturbance
frequency
in
this
v-0.
You
know
if
we
can
I,
don't
know
what
the
land
Trender
coverage
is
spatial
coverage,
but
if
we
want
to
it,
you
know
we
need
to
benchmark
fire
somehow,
and
can
we
do?
We
want
to
include
any
disturbance,
frequency,
benchmarking,
I.
F
M
Q
C
Wanted
to
follow
up
on
something
charlie,
just
said
and
I
had
an
example.
Data
set
of
that
may
not
be
this
specific
when
we
want
to
focus
on,
but
we
could
think
about
things
that
test
stuff
like
PFT
distribution,
but
our
morphate
specific
or
use
the
capabilities
of
fates
more,
which
is
more
like
testing
the
distribution
of
traits
or
functionality
of
plants
spatially,
and
one
idea
would
be
along
the
lines
of
the
map
of
ISIL,
Hydra,
City
and
I
suggest
II.
That
was
Alex,
koenings
derived,
but
I
know.
C
Someone
in
our
group
yesterday
said
they
didn't
like
that.
I
didn't
think
that
was
a
very
solid
product,
but
we
can
think
of
other
things
like
that
that
are
sort
of
showing
that
faith
could
show
an
emergent
distribution
of
some
functionality
of
plants
that
isn't
just
a
PFT
distribution
but
also
gets
at
what
PFD
distributions
might
look
like.
D
J
Just
want
to
follow
on
to
the
points
about
PFT
distribution
maps,
so
the
the
Kappa
statistics
really
just
compares
like
the
presence
of
a
set
of
P
of
T's
to
the
the
presence
of
a
specific
classification
of
vegetation
to
that
presence
in
another
map,
and
so
the
challenge
is
going
to
be
defining
those
sets
of
P
of
T's.
And
so
you
know
we
can
take
the
fades,
output
and
aggregate
it.
I
was
processing
it.
J
D
H
Yeah
I
wonder
if
there's
a
way
to
do
something
like
a
ROC
as
well,
where
you
you
look
at
how
often
you
are
you,
you
are
an
agreement
sort
of
spatially
on
an
inner
region,
so
you
could
look
at
that
ROC
curve
and
how
how
often
you're
capturing
the
true
the
true
state
versus
the
false
false
state.
It's
not
quite
the
same,
but
you
don't.
You
can
see
what
I
mean
where
you
gonna
be
able
to
aggregate
aggregate
that
statistic
and
it
was
sort
of
a
simple
metric
across
the
broad
region.
G
F
G
So
again,
I'm
thinking
about
short-term
milestones
and
long-term
vision
for
the
last
part
of
this
conversation,
which
is
more
about
kind
of
governance
and
and
next
steps,
so
Gregory
kindly
went
through
and
synthesized
a
bunch
of
the
answers
for
questions.
Six
and
seven
in
terms
of
support.
The
kind
of
high-level
things
that
came
up
were
creating
new
or
extended
capabilities
tools
to
deal
with
the
greater
complexity
of
a
CT
sfynxx
model.
I
didn't
know
what
this
meant,
but
offline,
modeling
capabilities.
G
G
Some
of
the
coordination
and
integration
issues
that
came
up
in
the
summary
were
reduced.
Complexity.
Modes
of
fates
will
allow
us
to
evaluate
in
duplicate
aspects
of
CLM
fates,
more
regular
meetings
between
fates
and
see
TSM
software
engineers
and
key
stakeholders
for
meeting
these
milestones
and
priorities
and
reducing
duplication
of
efforts.
G
Key
tests
for
CMM
simulations
that
can
drive
milestone
targets
using
github
project
boards
to
coordinate
tasks
between
organizations.
Oh
I,
liam
fates
and
cps
mm
all
around
github
now.
So
that
seems
like
the
natural
way
for
us
to
be
communicating
and
then
encouraging
members
of
all
the
teams
to
sign
up
for
mailing
lists,
slack
workspaces,
etc,
to
participate.
G
So
again,
with
this
goal
of
being
able
to
run
fully
trained
in
historical
steam
at
simulations
I
put
three
suggestions
up
here
about
things
that
we
will
do
so
it
seems
like
kind
of
digesting
the
information
that
we've
talked
about
into
github
eschews
projects
and
milestones
both
on
the
CTS.
The
fates
I'd
probably
make
some
sense
to
link
issues
with
I
am
I
didn't
realize.
I
Liam
was
even
on
github
until
yesterday,
and
then
I
put
this
up
as
a
question
mark
grease.
G
Can
be
during
fates
meetings
during
software
engineering
meetings,
I,
don't
know
what
the
best
wait.
I,
don't
know
what
the
best
mechanism
is
and
I
don't
want
to
add
more
meetings
for
folks,
just
maybe
putting
it
out
on
the
radar
of
something
that
we
need
to
be
talking
about.
More
often
go
back
to
the
slide,
maybe
and
Bill
you
had
a
had
a
kind
of
raised
hand
about
how
to
govern
in.
L
More
on
the
support
side,
which
is,
is
it
worth
thinking
about?
Do
we
need
to
discuss
sort
of
formally
or
informally
how
the?
How
user
and
developer
support
will
happen,
like
you
know,
for
CTS
M
in
general,
Eric
and
Keith
as
the
software
and
science
liaisons
and
a
lot
of
their
time?
I
perceive
helping
users
via
forums
and
emails,
etc,
and
when
people
start
to
encounter
issues
with
using
fates
within
C
TSM.
Is
that
something
that
see
TSM
liaisons
will
will
be
answering
or
their
fates
people
that
can
serve
that
role
or
some
combination.
M
You
know
right
now
you
know
faith
development
is
effectively
being
supported
by
the
ng
tropics
project.
In
you
know,
Co
EES
strength
is
not
in.
You
know:
community
outreach.
It's
in
support.
You
know,
that's
that's
just
not
it's
hard
to
argue
to
the
DV
program
managers.
You
know
that
that
when
we
try,
we
definitely
try
and
I.
Think
we've
gotten.
You
know
farther
making
that
argument
than
than
most
DV
projects,
but
I
think
we're
about
as
far
on
that
as
we
can
can
be
so.
L
I
I
E
E
You
know
be
kind
of
supporting
more
on
the
outreach,
and
you
know,
community
engagement
side
of
things
would
be
one
way
to
think
about
it
or
another
would
be
to
say
we
want
to
sort
of
have
some
parity
across
those
two
agencies
and
sort
of
you
know.
Bringing
this
to
the
level
of
you
know
a
request
to
do
a
into
NSF
to
kind
of
engage
with
this
question,
because
it's
been
a
bit
of
a
thorny
one,
but
we
also
have
some
turnover
in
program
managers
at
D,
OE
and
I.
F
I
B
B
M
Far
simpler,
also
at
the
same
time,
you
know
they're
they're.
Basically
the
do
is
basically
predicating.
You
know:
goe
funded
University
projects
on
using
USM
and
so
I
think
they
discover
that
there
needs
to
be
some
communities
report.
If
that's
the
strategy
they're
pursuing
so
I
think,
like
Laura,
says,
I
think
there
is
some
opening
there
that
we
need
to
push
and
see
what
we
can
make
happen.
I.
D
D
I
Yeah
we
will
have
to
I
want
to
I
want
to
have
some
more
conversations
about
that,
but,
but
it
may
be
not
for
a
public
forum.
The
Ivan
I
have
a
question
about
how
to
do
this.
Github
management
of
the
of
this
series,
of
whatever
targeted
you
know
all
stones
and
where
that
should
live,
because
I
I
think
we've,
both
groups
been
finding
it
pretty
useful
to
have
projects.
At
least
we
I,
don't
know
exactly
how
the
face
one
works.
I
M
F
M
I
P
I
F
P
F
Hypothesizing
that
the
majority
of
the
majority
of
users,
technical
issues,
are
to
do
with
things
around
getting
the
model
to
run
at
all.
You
know,
setting
up
driver
data
sets
and
comp
sets
and
and
the
environment
they
want
the
model
to
run
in
rather
than
specific
technicalities
with
individual
component
pieces.
Oh.
P
B
F
D
H
Data
input
because
the
images
are
fairly
large,
which
is
kind
of
an
asthma
to
what
doctor
ization
really
was
meant
to
be
in
the
first
place.
You
know
it's
supposed
to
be
really
small
I
also
want
to
bring
up
to
that
one
of
the
things
that
we're
we've
been
talking
about
when
you've
been
working
on
containerization
efforts
is
right.
Now
we're
focusing
really
on
just
fate
side
issues,
but
especially,
if
we,
you
know
as
we're
going
forward
with
this
v-0
short
term
implementation.
H
L
H
P
L
H
P
G
M
Yeah
I
until
I
had
it
before
to
say
that
you
know
I
didn't
want
to
make
my
last
Steven
sound
like
we're.
Not
you
know
like.
We
can't
support
you
because,
of
course
we
can
and
or
and
do
and
want
to
keep
do
so
right.
So
so
they
we're
definitely
keen
to
do
more
of
these
like
feeds
tutorials,
like
we've
done
in
the
past,
and
that
kind
of
thing
so.
G
I
guess
I,
don't
see,
I
see
that
kind
of
scientific
engagement
is
being
really
critical.
Obviously,
to
the
success
of
this
and
you
all
as
developers
and
leaders
are,
you
know,
wouldn't
want
you
to
come
to
the
tutorials,
but
I
also
recognize
that
the
best-
that's
not
something
ng
tropics
is
really
funded.
To
do.
I
G
G
C
I
Did
so,
but
just
you
know
getting
other
technicalities
here
about
but
I
think
in
terms
of
moving
the
project
forward
in
this
your
commnets,
you
get
some
fades
meanings,
those
already
the
faith's
meaning
but
absolutely
focused
on
science.
You
know
you
should
probably
continue
to
be
like
that.
You
know
one
suggestion
would
be.
M
P
F
L
P
L
P
I
P
D
H
G
G
L
Mean
my
my
own,
my
main
serious
comment
and
support
for
Rosie's
one
instant
of
zero
was
that
calling
in
v-0,
and
especially
advertising
it
as
such
to
be
implies
that
this
is
super
preliminary
and
not
at
all
for
public
consumption.
V1
implies
that
it
is
vetted
and
somewhat
you
know,
supported,
or
at
least
we're
behind
it.
Yes,.
G
F
G
C
P
P
H
How
often
does
the
software
engineering
team
do
they
have
on
the
CTS
em
aside
and
do
they
have
multiple
meetings
or
sir
one
meeting
and
my
primary
question
I'm
trying
to
target
is
we're
moving
over
the
fates
and
X
API
to
track
the
you
know,
the
dev
tags
and
what
I'm?
Where
does
the
discussion
happen
for
the
next
updates
for
the
next
tag?
The
death,
the
incrementation
of
the
development
eyes.
D
I
L
I
So
I'm,
just
this
is
getting
in
the
weeds
here,
but
I'm
little
hesitant
about
like
trying
to
schedule
out
a
specific
time
where
I
have
a
fates
meeting
like
I'm,
making
it
every
two
weeks
every
month
or
some
like
that.
Because
then
you
have
to
kind
of
save
up
your
agenda
items
and
it
can
become
too
big.
You
know,
if
you
have
it
frequently,
you
just
start
ticking
off
things
all
the
time.
F
H
Right
and
I
had
talked
about
that
a
bit
as
whether
or
not
like.
Initially
when
I
joined
on
or
a
little
bit
after
I
joined
whether
or
not
we
should
distribute
the
labor
and
like
set
just
one
of
us
as
being
yeah
you're
gonna
do
you're
gonna
be
the
three
assemblies
on,
but
we
decided
that
we
because
we're
just
two
people
and
we
talked
to
each
other
fairly
frequently.
H
D
So
I'm
a
little
bit
confused
right
now,
I'm
in
terms
of
the
development
of
Sam
as
a
host
land
model.
Is
that
gonna
continually
develop
with
like
the
multi-layer,
canopy
and
different
photosynthesis.
Parameterizations,
that's
the
thing
or
is
it
gonna
be
sort
of
static
through
this
process?
I
wasn't
quite
sure.
So
then
you
can
clarify
that
we're
gonna
be
trying
to
put
more
functionality
into
the
host
Lam.
D
I
I
I
We
should-
and
it
won't
always
be
clear
and
then
eventually,
as
we've
sorta
discussed
at
least
tangentially
here,
we
would
try
to
separate
out
those
pieces
of
code
that
are
shared
so
that
they
were
both
using
the
same
code
technology
that
could
take
a
while
to
be
able
to
do
that
in
a
clean
way
with
like
photos
and
says
phenology
and
other
things
so,
I
think
the
only
way
that
that's
kind
of
the
way
to
do
it
is
to
try
to
just
focus
see
to
some
development
on
things
that
are
not
handled
by
fate's.
Right.
M
Mean
you
know,
you
totally
view
that,
but
I
guess
I
would
also
hope
that
people
who
you
know
are
sort
of
see
themselves
as
see
gsm
developers
are
also
contributing
things
to
fades
as
well.
You
know
if
it's
within
this
sort
of
process,
scope
of
you
know
if
you
feel
like
there
needs
to
be
a
new
representation
of
you
know,
leaf
angle,
you
know
or
whatever
process
you
know
the.
We
would
certainly
hope
and
welcome
that
those
contributions
right.
I
M
D
Dave
I
want
to
jump
in
on
the
photosynthesis
side
in
that
I
think
that
the
work
that
Danica
is
going
to
do
with
c4
photosynthesis
is
a
great
example
of.
Maybe
we
should
be
developing
code
independent
of
any
sort
of
host
land
model,
meaning
we
have
to
develop
an
understanding
and
a
marvelous
e
for
photosynthesis
at
the
reef
scale,
and
that
really
can
be
done
independent
of
a
standalone
code
itself
before
we
even
thinking
about
whether
it
goes
in
to
see
TSM
or
comm
or
fades.
D
I
Yeah
I
think
Courtney
missed
that
we
talked
about
earlier
that
yeah,
that's
that's!
The
goal
is
to
is
to
move
towards
that,
and
the
question
would
just
be
what
happens
in
the
interim
before
we
can
sort
of
build
that
separate
structure,
dude
and
I
think
there's
just
enough
to
be
in
a
case-by-case
basis
to
decide
what's
the
most
efficient
way
to
get
towards
that
vision
and
advance
the
science.
At
the
same
time,
okay,.
D
I
think,
most
of
the
focus
that
c4
photosynthesis
will
be
done,
sort
of
at
the
leaf
scale,
uncoupled
from
CLM
or
fades
or
anything,
because
we
need
to
evaluate
what
the
existing
c4
models
are
find
out,
what
they
are
evaluate
them
and
just
understand
that
before
we
even
thinking
about
coupling
to
a
host
model.
So
that's
probably
what's
going
to
be
happening
over
the
next
year
for
c4
photosynthesis.
N
Yeah
I
just
wanted,
because
part
of
the
reason
my
Nick
and
I
decided
decided
to
start
with
C
TSM
is
because
the
optimality
theory
is
much
harder
to
implement
in
fates.
He
sort
of
looked
into
it
and
thought
about
it,
and
so
I
think
that
we're
gonna
probably
run
into
some
of
these
issues
where
some
of
these
things
are
a
little
more
difficult
to
implement
in
Bates,
just
because
the
structure
is
more
difficult.
So
it's
a
thing
to
consider.
N
We
I
think
the
goal
that
we
had
talked
about
was
ultimately
to
get
both
the
c3
and
c4
photosynthesis
into
Bates,
his
optimality
photosession.
So
that's
you
know,
ideally
what
we'd
like
to
work
towards,
but
we'd
love
to
work
with
any
of
the
feebs
developers
to
to
try
and
make
this
process
easier,
but
that's
part
of
the
reason
why
we
were
starting
where
we
were.
F
Yeah,
it
would
be
I'm
just
watching
the
chat,
but
it
would
be
good
to
think
I
mean
so
I've
been
having
this
conversation
with
Nick
about
his
optimality
thing
and
it'd
be
good
to
think
about
what
he
perceives
the
roadblocks
to
be
we're
putting
in
fates
and
where
those
weeds
and
ways
we
can
overcome
them.
I
mean
yeah.
It
would
be
good
to
have
that.
That
thought
also
just
to
clarify
Gordon
Antony's
map
model
is
not
in
fates.
Thus
far.
D
F
And
Ryan's
done
a
bunch
of
work
on
this
sort
of
pulling
like
getting
python
wrappers
to
pull
out
parts
of
fates
so
that
you
can
test
them
in
that
kind
of
a
way
like
you
can
pull
out
the
the
nutrient
model
and
the
allocation
model
and
see
how
they
kind
of
perform
with,
like
I,
said
an
external
wrapper.
So
you
can
imagine
that
the
leaf
thing
could
work
in
any
case
fashion.
D
P
Is
talking
about
it?
He
just
did
that
to
some
degree,
so
you
you're
saying
so
Gordon
you're
saying
you
want
a
a
framework
that
you
could
run
photosynthesis
separate
from
suffer
from
see,
TSM
right,
just
on
its
own
like
and
like
it
is.
What
Rosie
is
talking
about
with
like
a
Python
wrapper.
Is
that
would
that
function
to
what
you
want
to
do.
D
Yeah
I
mean
ideally
you'd,
be
running
the
exact
same
code,
at
least
for
relief
simulations
that
would
you
be
running
for
a
CTS
mm
or
fate
simulation
I
think
that's
actually
harder
to
do,
because
the
codes
aren't
really
set
up
to
do
that.
That's
why
rosie
was
talking
about
this
system
that
Anthony
Walker
put
together,
isn't
actually
sort
of
part
of
fades,
but
what
it's
doing
is
it's
developing
the
photosynthesis
parameterizations
that
would
go
ultimately
into
things.
I've,
ultimately
adopt
yeah
I
came
about
this
strategy.
D
You
know
with
the
multi-layer
canopy
because
you
can't
actually
get
it
into
comm.
If
you
did,
you
were
just
fine.
You
spend
years
recoding
CLM
just
to
get
it
to
run.
Then
you'd
find
you're
compensating
biases
in
it.
You
really
wouldn't
understand,
while
you're
getting
the
answers,
you're
getting
and
so
I
actually
sort
of
created
a
code
that
runs
independent
of
CLM,
but
that
exact
same
code
can
actually
be
coupled
into
seal
and
it
compiles
I've
never
actually
run
it,
but
at
least
it
compiles
and
I
thought.
D
That
was
a
good
step,
so
you
actually
can
develop
code
independent.
You
know
in
a
modulized
code
that
can
run
in
different
configurations,
either
independent
of
co-member
feeds
or
as
part
of
comm
feeds,
and
it
can
be
the
exact
same
code.
It
just
takes
a
lot
of
effort
to
code
that
correctly
well,
that's
sort
of
what
I'm
advocating
for
work.
You
know
and
I
think
that
sort
of
should
be
a
goal
across
all
these
modeling
efforts
is
to
try
to
get
that.
It
just
takes
a
lot
of
effort
to
get
there
right.
P
G
So
I
guess
on
the
note
of
governance
and
coordination,
we've
kind
of
mentioned
it,
but
if
there's
folks
that
are
on
the
call,
but
don't
that
are
already
getting
emails
for
me
about
our
weekly
Thursday
meetings,
we
will
not
have
one
tomorrow
because
we've
all
spent
a
lot
of
time
together
already
this
week,
we'll
start
off
again
next
week
and
if
you,
if
you'd
like
to
get
included
on,
as
you
know,
let's
just
send
me
an
email
on
the
Fate
side.
Those
are
bi-weekly
fates
meetings.
G
So
similarly,
if
you're
interested
in
learning
more
go
ahead
and
just
maybe
offer
rooms,
Ian
Charlie,
that's
alright-
to
join
the
fates
development
community.
Those
calls
are
also
on
Thursdays.
You
can
just
pack
you
Thursday's,
but
fun,
science
talks
now
and
and
hopefully,
between
those
two
forums.
You
know
I
think
that'll
be
the
most
productive
waiting
coordinating.
You
know
this
effort
moving
forwards
is
to
kind
of
with
regular
updates.
You.
E
G
Other
points
or
other
you
know
and
I
think
then
the
software
engineering
team
is
small
enough
that
hopefully
we
can.
We
do
it
better
to
help
coordinate.
You
know
to
me.
Those
seem
like
the
biggest.
The
two
biggest
is
the
coordinating
on
the
science
side
and
the
softness
off
versus
I
can
more
regularly
moving
forward
to
improve
on
this
integration.