►
From YouTube: NUG Meeting 2014: Collins
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
So
our
next
speaker,
Bill
Collins,
is
a
recognized
world
expert
in
climate
research
and
a
lead
author
on
the
fourth
assessment
of
the
intergovernmental
panel
on
climate
change
or
IPCC,
which
was
awarded
the
2007
Nobel
Peace
Prize
build
a
senior
scientist
in
climate
science
department
head
at
berkeley
lab
and
a
professor
at
UC
Berkeley.
So
I'd
like
to
welcome
Bill
Collins.
B
Thanks
to
close
collaboration
between
earth
scientists
like
myself
and
people
in
the
coalition
research
division
here
at
Berkeley
Lab,
so
they
applied
mathematicians
and
computational
scientists,
because
that's
really
I
think
an
emerging
area
where
new
algorithms
are
being
demanded
to
take
on
some
of
the
challenges
that
we
have
in
predicting
the
Earth's
future
and
we're
fortunately
I.
This
partnership
is
a
lot
of
step
up
to
the
challenge.
B
So
I
wanted
to
start
briefly
by
getting
you
depressed
events
of
lunch.
We're
just
worried.
I'm
wanted
I
repeatedly
tell
audiences
that
I'm
an
optimist
and
for
good
reason,
but
I
think
we
also
have
to
be
sober
about
the
challenges
that
we're
facing
and
despite
the
fact
that
I've
managed
to
time
all
my
trips
to
the
east
coast
in
January
to
correspond
with
outbreaks
the
polar
vortex
it
is,
the
u.s.
is
getting
hotter
and,
in
fact,
depending
on
how
you
look
at
the
the
planetary
temperature,
it's
also
been
getting
hotter.
B
Despite
all
the
recent
buzz
about
the
hiatus,
2012
was
actually
the
hottest
year
on
record
according
to
the
national
climate
data
center.
So
all
the
states
are
enough
middle
part
of
the
country
that
are
known
as
red
states
are
in
fact
read
their
read
for
a
reason,
which
is
that
they're,
the
hottest
they've
been
in
hundred
twenty
years
according
to
the
records
kept
by
NCDC,
so
go
figure.
But
anyway,
this
was
the
hottest
year
on
record
and
I'll.
Come
back
to
the
fact
that
this
is
a
true
elsewhere
on
the
planet.
B
As
you
know,
we're
in
the
middle
of
a
quite
severe
drought,
it's
not
clear!
Yet,
whether
or
not
there
is
a
fingerprint
of
anthropogenic
influence
here,
so
this
may
just
be
due
to
natural
fluctuations,
and
we
do
know
that
the
California
has
gone
through
quite
long.
Droughts,
Centennial
length
droughts,
because
we
have
lakes
that
are
landlocked
no
outlet.
B
That
would
be
bad,
so
this
is
the
challenge
that
we
face.
Currently,
this
is
2013
2014.
According
to
data
from
the
NASA
motor
satellite.
The
thing
I
want
to
call
it
to
your
attention
to
is
the
absence
of
white
on
the
left
and,
if
you've
been
up
to
Tahoe
this,
this
January
you'll
know
what
I'm
talking
about,
and
if
this
continues
in
the
future,
we
have
a
challenge.
B
This
is
the
reason
why
your
comment-
modeling,
has,
has
become
sort
of
this
tool
for
policymakers,
try
to
figure
out
the
challenges
that
we're
facing
and
if
you
were
to
take
these
hot
conditions
and
run
them
forward,
regardless
of
whether
or
not
precipitation
of
the
California
increases
or
decreases,
and
actually
at
this
point
we
don't
know
the
answer
to
that
question.
What
we
are
quite
confident
in
predicting
is
that
the
phase
of
the
precipitation
will
change
from
snow
to
rain,
and
that
impacts
directly.
B
So
it's
for
this
reason
that
climate
modeling
has
become
sort
of
a
central
tool
for
predicting
the
future,
and
this
is
the
mode
in
which
we've
we've
used.
These
models
to
contribute
to
to
international
assessments,
as
we
heard
in
the
introduction
to
this
talk
both
internationally
and
also
nationally.
B
That's
interesting,
of
course.
It's
a
great
way
of
doing
you
know
it's
a
great
way
of
keeping
yourself
busy,
but
I
won't
be
around
really
to
be
confronted
with
whether
or
not
my
projections
failed
or
not,
and
so,
as
I
I
spend
most
much,
I'm
actually
using
these
two.
These
models
to
understand
the
president
of
the
past,
because
those
are
situations
in
which
one
can
do
falsifiable
hypothesis
testing.
So
much
of
my
group's
work
is
basically
built
on
on
these
latter
two
applications
of
the
models,
and
then
we
let
other
people
do
the
prediction
work.
B
Okay,
so
first
I
want
to
cover
basically
two
topics
in
this
talk,
and
these
are
areas
where
the
interaction
of
our
work
with
nurse
and
with
computational
research
division,
Erica
lab,
have
proven
to
be
quite
central.
The
first
is
the
issue
of
climate
extremes
and
the
second
is
the
issue
of
sea
level
rise,
I'm
going
to
I'll
start
with
climate
extremes,
so
this
is
extreme
events
in
the
climate
jargon,
so
this
is
sort
of
now
reserved
word
in
common
community.
These
are
events
that
are
highly
localized
and
represent
stressors
to
either
the
environmental
to
society.
B
So
these
could
be
things
like
tornadoes.
Tropical
cyclones,
like
the
the
typhoon
haiyan,
that
with
the
philippines,
these
could
be
heat
waves,
and
these
are,
if
you
look
at
the
another
part
of
the
IPCC,
the
working
group
that
deals
with
impacts.
Climate
extremes
revives
the
second
most
important
form
of
climate
change,
because
they
essentially
push
other
environmental
systems
are
their
environment
or
they
presume
in
systems
to
the
brink
of
breaking,
because
the
events
are
getting.
B
We
predict
you're
going
to
get
stronger
for
reasons,
I'll
come
back
to
in
a
moment,
and
so
this
has
now
become
a
major
issue
both
in
terms
of
assessing
the
recent
climate
record.
This
is
a
special
issue
from
the
American
of
neurological
society
of
the
bulletin
for
that
society.
Looking
at
all
the
extreme
events
that
happened
just
in
2012,
we
were
the
IPCC
of
course
discussed
this
at
length
and
one
of
the
little
postage
stamp-sized
heads
in
this
picture.
B
This
is
taken
down
in
Hobart
in
Tasmania,
at
the
same
time
as
Australia
and
Tasmania
were
experiencing
an
in-person
a
deadly
hot
year,
so
2013
before
the
the
government
of
Australia
sort
of
flipped
sign
and
decided
to
go
from
being
quite
aggressive
on
the
issue
of
solving
climate
change
to
not
being
climate
deniers,
so
they've
they've,
not
the
government
has
changed
hands
since
this
report
is
released,
but
they
released
a
report
called
the
angry
summer
dealing
with
the
hunter
tweed
record
33
records
that
were
broken
in
90
days.
This
was
covered
extensively
in
the
news.
B
This
is
the
period
of
time
when
the
the
Bureau
of
Meteorology
in
Australia
had
to
add
additional
colors
to
their
temperature
graphs
because
they
never
experienced
temperatures
of
that
level
before
55
degrees,
Celsius,
frequently
unprecedented
lehigh
outbreaks
of
fires,
etc,
etc.
So
this
is
a
hot
topic
on
the
pun
and
we
need
to
be
able
to
predict
these
events
and
they're
they're
challenging
for
climate
models
be
for
a
couple
reasons.
These
events
occur
at
small
link
scales.
B
I've
already
pointed
out
to
you
that
those
are
areas
where
the
comment
waters
are
traditionally
sort
of
a
week
because
we
have
to
solve
for
the
whole
climate
and
the
the
other
reason
why
these
are
challenging
is
that
climate
has
to
integrate.
For
long
of,
for
its
primary
job
is
to
integrate
over
long
periods
of
time
at
low
resolution.
And
yet
we
need
to
uncover
these
events
because
they
are
distressing
events
in
the
climate
record.
So
how
do
we
both
integrate
for
long
periods
of
time
and
also
achieve
ultra
high
resolution?
B
B
A
resolution
of
about
one
kilometer
using
machines
of
this
elk,
which
would
allow,
is
to
begin
resolving
the
changes
in
micro,
climates
down
inside
the
the
greater
bay
area
of
San
Francisco,
which
would
be
nice,
but
we're
impatient
people,
and
we
don't
want
to
to
wait
now
we're
going
to
show
you
a
couple
things
we
can
do
with
conventional
computers
now
at
25
kilometer
resolution
and
then
discuss
how
we're
going
to
move
forward
a
little
bit
faster.
So
this
is
look.
It
uses
the
simulation
that
we
did
locally
using
nurse
resources.
B
This
is
a
movie
thanks
to
a
bot
and
michael
wayne
of
the
computation
research
division,
the
white
patterns
on
here
water
vapor,
these
localized
centres
of
high
water,
vapor
or
hurricanes.
We
can
track
these
hurricanes
now
we
can
do
these
disturbing
tracking
in
a
massively
parallel
mode.
So
we've
used
three
thousand
cores
on
the
mirror.
B
Ibm
a
blue
jean
machine
at
are
gone
and
can
gang
up
these
statistics
and
then
compare
our
models
against
present-day
statistics
for
hurricanes
and
the
model
is
spectacularly
well
for
the
Atlantic.
We
do
less
well
Suffolk.
The
Japanese
do
very
well
on
the
Pacific,
so
I
guess
it
depends
on
which
Basin
your
downstream
from
so
we're
downstream
from
the
atlantic's.
We
do
great
in
the
black
and
the
Japanese
sort
of
the
receiving
end
of
whatever's
happened.
The
Pacific's
is
that
you
very
well
the
Pacific
Ocean,
that's
great
as
I
mentioned.
This
is
a
you
know.
B
This
is
a
from
our
perspective
and
the
previous
speaker
was
talking
about.
They
say
this
is
a
big
data
from
the
climate
precision
Clement
users
perspective.
This
is
no
longer
a
big
data
from
perspective
of
the
financial
services
industry
of
Google's
perspective.
They
they're
analyzing
20
petabytes
a
day
of
data
on
us,
of
course,
but
from
the
from
the
comedy
community's
perspective.
B
This
is
big
data
and
thanks
to
the
collaboration
that
we've
been
able
to
engineer
locally,
we
can
now
reduce
this
data
fairly
quickly
to
look
for
extreme
events
and
the
other
the
another
type
of
event
that
we've
been
looking
out
in
the
climate
data
record
or
atmospheric
rivers.
So
again,
this
is
moisture
when
fields
rainfall.
We
can
now
look
at
the
impact
of
atmospheric
rivers.
High
moisture,
high
rainfall
events
that
impact
the
coast
of
California
and
you
can
see
them
so
propagate
down
the
coast-
dropping
a
lot
of
rainfall
over
the
coastal
ranges.
B
If
that's
nice,
that's
what
we
can
do
with
their
current
climate
models,
but
we're
still
missing
the
action
with
respect
to
where
was
the
rainfall
occurs
and
here's.
Why?
So?
This
is
a
graph
of
cloud
cover.
This
is
a
sort
of
which
you
get
from
conventional
satellite
pictures,
etc
of
the
coverage
of
clouds
over
the
planet.
The
scale
runs
from
0
to
1
red
indicates
that
you've
got
a
lot
of
cloud
cover
in
the
storm
tracks
around
Antarctica
in
the
strong
tracks
in
the
northern
hemisphere.
Very
little
cloud
cover:
this
is
nothing
new.
B
B
Statistical
mechanics
for
clouds,
so
we're
sort
of
the
mercy
of
our
ability
to
represent
the
parametric
play,
and
this
is
a
major
source
of
uncertainty
in
these
common
models.
In
the
fifth
assessment
that
we
just
concluded-
and
this
is
now
in
the
public
domain
as
of
October-
we
brought
this
down
by
about
a
factor
of
two
still
five
factor
of
five
out
from
the
native
resolution.
B
The
runs
have
just
showed
you,
or
these
ultra-high
resolution
runs,
are
25
kilometers
and
we've
bought
new
runs
from
another
project
that
have
bought
this
town
of
another
factor
of
two.
Of
course,
the
fact
that
we
have
grid
space
and
it's
comparable
to
the
feature,
try
and
resolve-
is
not
adequate
to
resolve
the
feature.
As
you
all
know-
and
this
looks
a
little
bit
like
the
classical
Zeno's
paradox,
where
your
arrow
keeps
advancing
by.
You
know
a
factor
of
2
close
to
the
target,
but
you
never
quite
hit.
B
So
we're
not
we're
not
willing
to
wait
for
this
to
converge,
and
we
would
like
to
move
to
an
alternate
alternate
way
of
tackling
this
problem,
and
this
is
thanks
to
a
collaboration
that
we've
developed
between
my
department,
hans,
Johannsen,
focal
allah
and
company
in
the
competition
research
division,
where
we're
now
going
to
use
adaptive
atmospheric
dynamics
to
go
after
these
interesting
phenomena.
So
the
idea
is
that
many
of
the
meteorological
features
like
the
rivers
I
showed
you
that
the
hurricanes
are
highly
localized.
B
It
makes
no
sense
to
refine
the
grid
globally,
if
you're
after
these
localized
phenomena,
why
not
use
adaptive
mesh
techniques?
And,
oddly
enough,
there
are
no
existing
models
used
for
global
commerce
projections
that
use
adaptive
mesh
0.
There
are
models
that
have
embedded
techniques
for
doing
refinement.
I
can
assure
you
that
many
of
those
refinement
techniques
are
very
poorly
formulated,
so
they
can
sedate
it's
insured
designed
to
serve
conserved
scalars,
but
not
fluxes,
so
they're,
inherently
non-conservative
techniques
and
hence
unsuitable
for
doing
climate,
and
we
need
a
technique
that
is
strictly
conservative
of
fluxes.
B
There
are
the
key
phenomena
that
we
want
to
study.
These
clouds
are
occurring.
It
scales
where
the
atmospheric
dynamics
is
non,
how
to
static
and
so
for
all
the
virtually
all.
The
models
that
have
been
used
for
climate
change
are
in
the
hydrostatic
regime.
This
is
another
frontier
for
us,
where
we
would
like
to
be
able
to
get
the
interaction,
for
example,
of
convection
with
fast
buoyancy.
B
Boats
in
the
atmosphere,
this
fast
buoyancy
modes
are
very
important
for
the
momentum
budget
at
the
atmosphere,
and
this
is
a
class
of
physics
and
so
far
we've
been
unable
to
touch
with
conventional,
with
conventional
models
that
can't
look
at
the
interaction
of,
for
example,
atmospheric
convection
and
gravity
waves.
So
again,
there's
they're
important
reasons
why
we
want
to
go
both
adaptive
and
non
hydrostatic.
B
Unfortunately,
we
have
an
opportunity
to
do
so
as
I'll
describe
now
and
again.
This
is
a
collaboration
we've
been
able
to
pursue,
thanks
to
both
nurse
providing
us
with
the
computational
resources
and
the
way,
a
very
close
partnership
between
the
earth
sciences,
discipline
and
applied
mathematicians
and
computational
scientists.
So
we're
building
this
off
of
a
particular
form
of
adaptive
mesh
dynamics
called
Chombo,
which
is
now
housed
under
the
said
act
fast.
Math
Institute:
this
is
a
fully
known,
hydrostatic
former
dynamics,
its
fourth
order,
accurate
about
space
and
time.
B
It
can
be
fine
in
a
both
space
and
time
in
three
dimensions
in
space
as
excellent
scaling
as
I'll
show
you
in
a
moment,
and
this
genomics
has
been
bulletproof
tested
on
a
number
of
different
applications,
including
imagery
for
tokamaks
cosmology
space
plasmas,
mhd
I
used
to
be
an
astrophysicist
services
of
interest
to
me.
Microscope
fluids,
it's
been
used
to
even
model,
you
know,
dynamics
within
San,
Francisco,
Bay,
I.
Think
it's
been
used
in
the
human
circulatory
system
has
excellent
scaling.
So
this
is
a
code.
B
That's
been
used
for
wide
variety
of
applications
which
so
far
not
for
climate,
and
so
what
we've
done
now
is
establish
a
collaboration
to
bring
it
to
climate.
In
the
first
place
we
did
pull.
This
is
and
I
want
to
come
back
to
this
a
little
bit
later.
My
talk
is
actually
licking
two
ice
sheets,
and
it
is
the
second
half
of
my
talk.
B
Actually
deals
with
this
job
is
I'm
not
going
to
dwell
on
it
now,
but
we
have
now
successfully
implemented,
am
Chombo
as
Mr
technique
inside
a
traditional
Ice
Sheet
model
to
look
at
what
amounts
to
the
Mount
Everest
of
sea
level
rise
issues
which
is
what's
going
to
happen
to
erotica.
So
this
is
an
infomercial
for
the
second
half
of
my
talk,
but
this
is
already:
we've
already
have
a
proof-of-principle
calculation
that
AMR
could
be
very
useful
for
climate
and
I'll
talk
about
that.
B
A
little
bit
later,
John
Bowe
is
formulated
on
it
as
it
cubes
spheres.
We
take
the
the
sphere,
I
drop
it
on
to
an
embedding
cube
and
then
solve
the
dynamics
on
the
faces
of
this
cube,
it's
strictly
conservatives
with
the
finite
volume
technique
and
absolving
the
euler
equations,
and
I
won't
get
the
details
of
this
and
interest
of
time.
B
B
And
we
have
the
the
adaptive
mesh
technique
is
able
to
handle.
But
the
refinement
over
several
orders
of
refinement
up
to
four
levels
of
refinement
in
a
single
solution
and
sort
of
handle
the
consistency
between
the
coarse
and
fine
meshes
at
each
level
of
refinement
in
a
way
that
both
is
computationally
efficient
and
also
as
scales
extremely
well,
and
so
here's
a
demo
calculation.
This
is
a
work
thanks
to
Joshua
Elliott
in
the
conversation
research
division.
This
is
a
simplified
solution
of
the
what's
known
as
the
Hadley
circulation.
So
this
is
moisture
this.
B
This
looks
like
something
out
of
the
movie
alien,
but
it's
actually
moisture
being
infected
aloft
and
then
in
the
descending
branch
is
the
end.
This
is
the
earth,
of
course,
in
highly
simplified
form.
This
is
precipitation
and
the
down
wanting
branch,
the
Hadley
cell
and
the
this
meshing
that
you
may
be
able
to
see
are
the
various
levels
of
refinement
in
the
grid
before
these
demo
calculations.
B
This
is
essentially
where
we
are
at
right
now
in
terms
of
coupling
the
the
adaptive
mesh
dynamics
with
a
very
simplified
form
of
the
hydrological
cycle
inside
a
comet
model.
What
we're
now
doing
is
appealing
to
the
department
of
energy
for
support
to,
rather
than
taking
the
dynamical
cord
to
the
comet
model,
we're
going
to
take
the
physics
out
of
the
climate
model
and
bolt
it
on
to
chomp.
B
B
This
could
be,
in
my
view,
at
the
killer
application
to
go
after
her,
the
what
it
will
happen
with
hurricane
frequency
intensity
in
the
future.
This
is
what
this
will
allow
us
look
at
that
the
for
not
the
Hurricanes
at
their
native
scale
in
a
global
simulation
in
a
way
that
is,
that's
not
possible
currently,
so,
with
that
sort
of
information
about
the
direction
that
we're
going
I'm
going
to
now
switch
to
the
next
part
of
my
talk,
which
is
projections
of
sea
level
rise.
B
This
is
a
place
where
we
already
have
the
the
adaptive
mesh
functioning
and
I
can
show
you
some
initial
simulations
and
the
consequences
of
the
simulations
for
sea
level
rise.
So
this
is
a
picture
of
a
calving
event
from
I
think
this
is.
This
is
actually
from
a
nautica
if
I
hope,
all
of
you
saw
the
movie
chasing
ice.
B
If
you
haven't
I
highly
recommend
that
you
see
it
it's
a
striking
movie,
it
makes
it
clear
that
these
ice
sheets
are
that
they
are
the
great
ice
sheets
on
the
planet,
namely
Greenland
and
Antarctica
are
highly
dynamic
systems
and
they've
actually
represented
one
of
the
real
challenges
for
projecting
sea
level
rise.
This
is
the
this
is
a
figure
taken
from
the
current
IPCC
report,
the
fourth
assessment.
B
This
was
released
to
the
the
global
community
in
September,
2030s
of
2013
sea
level,
rise
in
meters
going
from
pre-industrial
all
the
way
or
just
prior
to
industrial
times,
up
to
the
present
day,
roughly
the
projections
following
what
we
was
currently
an
unmitigated
or
business-as-usual
scenario,
which
is
red
curve,
end
up
being
about
a
meter
for
the
21st
century.
That's
the
central
estimate:
I
should
note
that
there's
a
fat
tail
that
goes
up
to
2
meters
off
the
top
of
this,
so
I
could
get
a
lot
higher.
B
This
is
three
times
higher
than
the
estimate
that
appeared
in
the
fourth
assessment,
and
the
reason
is
that
the
fourth
assessment
essentially
just
included
the
expansion
of
seawater,
which
occurs
when
you
heat
it,
which
is
a
one
on
property
of
the
equation
state
of
cielo
of
water,
it's
known
as
stearic
sea
level
rise.
So
if
you
heat
water,
it
gets
expand
slightly.
That
was
the
basis
for
the
production
in
the
fifth
and
the
fourth
assessment,
because
we
didn't
know
how
to
project
what
was
happening
with
ice.
B
If
we
continue
on
an
unmitigated,
a
pathway,
45
degrees
centigrade
right,
we're
looking
at
we're.
Looking
at
a
major
change
in
sea
level,
that's
compal
hold
to
the
paleoclimate
record.
In
fact,
if
you
look
at
the
paleoclimate
record,
it's
sort
of
you
know
it's
grab.
It
unfortunately
didn't
exceed
twitter
meters,
but
20
meters
is
kind
of
a
steep
steep
hill
to
climb,
and
this
is
with
a
fairly
moderate
amount
of
temperature
increase.
So
this
has
motivated
us.
B
The
patio
climatic
record-
and
there
are
a
lot
of
people
now
going
on
looking
at
clothes
lines
around
the
world-
try
to
figure
this
out
as
motivate
assisted
us,
meaning
Berkeley,
and
our
collaborators
at
low
slot
Los
Alamos
see
if
we
could
build
a
model
that
was
really
tailored
to
looking
at
sea
level
rise
and
the
Grand
Challenge
here
has
been
nuts,
which
Greenland
Green
has
been
thoroughly
examined.
The
comic
community,
Grand
Challenge,
is
been
a
nautica,
the
and
and
they're.
The
reason
is
that
Antarctica
has
a
piece
of
physics.
B
It's
quite
different
from
that
of
Greenland
Antarctica
has
these
floating
ice
shelves,
so
these
I
shells
are
in
this
mist
by
the
way
the
aspect
ratio
has
been
greatly
compressed,
so
the
it
this
sixth,
the
system
is
as
much
greater
extent
laterally
than
it
does
vertically.
These
floating
ice
shelves
extend
for
some
cases,
tens
to
hundreds
of
kilometers
off
the
ship,
the
young
of
the
underlying
bedrock
underneath
Antarctica,
and
these
ice
shelves
are
subject
to
an
instability
where,
if
you
warm
up
this,
and
so
what
you're
looking
at
is
a
plan
view.
B
So
height
of
the
vertical
axis
is
here:
this
is
horizontal.
Here's
the
so
here's
the
ice,
restaurant
bedrocks.
This
is
the
continent
and
here's
the
floating
ice
shelf,
extending
off
to
the
right
here
over
an
overlying
cold
ocean
in
orica
has
warm
a
rut
walt
abuses,
all
relative,
that
relatively
warm
water
upwelling
around
it.
They
can
flow
over
cells
and
then
melt
the
ice
from
below,
and
this
instability
has
been
well
known
since
the
70s.
B
If
you're
going
to
do
it
with
a
global
unit
with
a
uniform
Nash
and
you
need
to
run
this
model
for
millennia,
so
that
turns
out
to
be
a
heavy
left
computation
way
and
that's
reason
why
we
think
this
system
is
particularly
amenable
to
using
adaptive
techniques
where
we
drop
the
computational
resources
down
at
the
front
where
we
need
it
and
for
the
rest
of
the
ice
sheet.
That's
largely
quiescent.
We
use
very
coarse
mesh,
so
we're
doing
this
on
the
top
of
a
fair
amount
of
work.
B
That's
been
done
in
the
past,
so
idealized
models
of
various
kinds,
realistic
models
of
low-resolution,
typically
not
involving
coupling
the
place
where
we
want
to
bend
now.
Some
idealized
models
with
highly
simplified
geometries,
where
we
want
to
be,
is
realistic.
High
resolution
fully
coupled
I,
so
shun
models
and
I'll
show
you
an
example
of
a
pioneering
simulation
with
that
system
right
at
the
very
end
of
this
section
of
the
talk.
B
So
this
work
is
built
on
two
components.
The
first
is
a
model
called
bicycles,
which
is,
of
course,
the
preferred
mode
of
birth
of
transportation
in
Berkeley.
For
a
couple
of
reasons,
one
is
that
we
continue
to
take
away
parking
spaces
and
the
second
is,
of
course
it's
a
green
mode
of
transportation
and
in
conjunction
with
the
we
were
trying
to
get
a
bike
rack
installed
after
the
last
IPCC
report.
You
know
sort
of
in
honor
of
that
of
the
of
the
Nobel
Prize,
and
we
never
got
it
done.
B
But
anyway,
bicycles
is
the
name
of
the
model.
That's
been
developed
by
colleagues
here
in
the
competition
research
division,
so
as
mundane
Dan
Martin
in
conjunction
with
collaborators,
the
University
of
Bristol,
who
built
the
underlying
underlying
model
and
collaborators
at
the
at
Los
Alamos
and
this
model
is
solving,
is
essentially
trying
to
solve
the
stokes
equation.
B
This
is
a
cup
of
plum
we've
coupled
these
two
models
together
to
try
to
solve
or
begin
trying
to
solve,
the
impact
of
global
warming
on
Antarctica
and
subsequently
on
sea
level
rise.
So
let
me
show
you
some
proof
of
principle
calculations
using
the
system.
So
what
you're
going
to
look
at
is
this
is
this?
Is
a
lot
going
to
be
there's
a
lot
going
on
in
this
movie?
B
So
let
me
step
revisit
before
I
run
the
movie,
so
the
melt
rate
in
meters
per
year
is
shown
in
the
upper
left-hand
diagram
and
the
westerner
directly
ice
sheet,
which
is
the
region
of
a
great
deal
of
interest.
There
are
a
couple
of
areas
that
have
been
blown
up
here,
so
a
couple
of
the
areas
that
are
subject
to
provide
high
melt
rates
have
been
magnified.
B
You
know
the
large
annual
cycle
associated
with
the
heating
of
the
ocean
from
sunlight
during
the
summer
during
the
day
are
the
southern
summer,
our
winter,
and
then
the
temperature
of
the
ocean
bottom
and
you'll
notice
that
these
temperature
scales
are
rather
I
mean
they're.
Both
these
are
insider'
grades,
and
these
are
both.
What
we
would
regard
is
chilly,
but
it's
this
has
been
sort
of
highlighted
in
pink
I
guess
to
exaggerate
the
fact
that
it's
relatively
warm,
so
this
is
this
solution,
is
one
in
which
we
have
now
taken.
B
B
Is
just
exercising
the
chombo
dicor
in
the
ice
sheet
model,
but
again
this
is
the
one
of
the
very
few
models
in
existence
that
using
adaptive
mesh
techniques
to
solve
for
ice
dynamics.
So
you
can
see
the
annual
cycle
and
sea
surface
temperatures
the
ocean
currents.
You
can
see
the
melt
rates
evolving
over
here,
so
they're
sort
of
low
and
they
will
they'll
gradually
increase
a
few
hot
spots
where
the
moat
rates
are
a
really
high
like
five
meters
a
year
and
the
ocean
temperature
at
the
bottom
isn't
doing
very
much
right.
B
This
is
just
sort
of,
although
you
can
see
places
words
especially
for
this
particular
glacier,
where
you
can
see
the
how
relatively
warm
it
is
right
at
the
face
of
the
underneath
the
glacier,
so
we've
now
been
able
to
run
whole
Ice
Sheet
models
using
adaptive,
mesh
techniques,
and
let
me
zoom
in
now
to
show
you
what
what
the
adaptive
mesh
is
doing
for
us.
This
is
a
particularly
important,
a
glacial
system
that
appears
to
be
evolving
kind
of
quickly
this
the
ice
philosophy.
This
is
in
meters
per
per
year,
or
meters
per
annum.
B
This
is
again
as
the
reason
why
you
should
see
the
movie
chasing
ice,
because
in
some
cases
these
glaciers
are
really
of
what
what
involving
quite
quickly
so
here's
the
you'll
see
the
the
face
of
the
ice
retreat
as
you
warm
it.
It
goes
so
the
face
of
the
glacier
is
your
2100.
This
is
a
run-out
22
and
50
meter
resolution,
so
this
is
inaccessible
using
can
do
conventional
techniques
and
see
how
rapidly
these
glaciers
will
evolve
and
now
you're
looking
at
the
adaptive
Nash.
B
So
this
is
the
the
beauty
of
the
adaptive
mesh
coarse
mesh
on
the
center
of
the
glacier
were
nothing
much
is
happening,
fine,
mesh
right
at
the
face
of
the
knowledge,
so
this
is
what's
made
it
possible
for
us
to
try
tackling
this
problem
and
we
were
to
get
quite
good
solutions
if
you
compare
the
the
melt
rates
from
observations.
These
observations
on
the
on
the
right-hand
panel,
taken
from
a
variety
of
different
field
stations
and
field
campaigns
versus
the
solution
coming
from
the
parallel
ocean
model.
B
I
mean
qualitatively
the
sar
we're
getting
rates
of
about
five,
where
we
should
be
getting
weights
of
about
five
and
I
realized.
This
is
an
extremely
qualitative
comparison,
but
at
least
we're
not
out
by
the
order
of
magnitude
and
the
my
colleagues
at
Los
Alamos
in
the
process
of
doing
a
much
more
rigorous
comparison
between
the
two
solutions.
We've
also
run
this
for
other
glaciers
of
interest.
This
is
a
slightly
longer
simulation
again
for
another
glacier
that
if
wall
paneling
is
a
particularly
important
one,
this
is
a
glacier.
That's
really
under
intense
study.
B
Now
we're
going
out
into
the
21st
century
in
really
watch
the
face.
The
glacier
sort
of
retreat
back
up
very
high
melt
rates
as
this
glacier
sort
of
degrades
again.
You
know
it's
really
nice
that
we
want
being
around
to
see
what
an
honor
simulations
are
actually
correct.
This
is
the
year
2400,
but
the
reason
why
this
is
of
concern
is
that
the
West
Antarctica
is
equivalent
to
about
seven
meters
of
sea
level
rise
and
just
these
systems
alone.
B
If
you
let
them
run
or
about
these
glaciers,
these
two
glaciers
alone
or
equivalent
to
about
oh
about
half
meter
of
sea
level,
rise
so
off
the
edge
of
this
graph.
So
these
this
system
is
highly
dynamic
and
will
have
an
important
impact
on
coastlines
around
the
world.
This
shows
you
just
drive
this
point
home
about
bicycles.
This
is
a
solution
for
the
ice
velocity
over
the
hole
over
the
whole
ice
sheet
relative
to
measurements
that
have
been
made
from
things
like
synthetic
aperture
radar.
B
So
people
are
now
using
since
that
ate
the
same
techniques
that
used
to
measure
the
changes
in
the
elevation
of
volcanoes.
So
these
techniques
will
example,
have
been
used
a
lot
in
the
Cascade
Range
to
look
for
the
risk
of
volcanic
eruption
by
comparing
radar
images
of
volcanoes
one
year
after
the
next
to
see.
If
the
cone
is
inflating,
you
can
use
exactly
the
same
technique
and
other
techniques,
including
GPS,
to
merge
the
velocities
I
sheets.
B
So
these
are
the
measurements
of
ice
sheets
over
polished
into
in
2011,
we're
in
the
process
of
doing
again
a
rigorous
comprised
of
these
from
from
the
bicycles
code.
But
again,
this
is
I
didn't
bring
along
a
figure
of
the
mash,
but
we
have
10
kilometer
mesh
here
in
the
sooner
the
glacier.
Where
not
much
is
happening.
The
velocity
is
essentially
zero
and
one
kilometer
two
kilometer
mesh
down
at
the
edges,
where
all
the
action
is
okay.
B
Let
me
one
of
the
time
for
questions
so
I
do
want
to
show
you,
though,
the
first
fully
coupled
simulations
that
we've
done
with
this
model.
These
are
now
simulations
where
we
are
having
the
ocean
heat,
the
glacier
from
below
the
line
dies
from
below
driving
this
instability,
so
the
the
land
ice
retreats
and
then
feeding
the
melt
rates
back
into
the
ocean.
So
these
are
the
first
calculations
that
we've
done
they're
quite
short,
they're
only
a
decade
long.
B
This
is
a
major
step
forward
for
us
in
terms
of
being
able
to
run
complete
regional
solutions
for
the
impact
of
this
physics,
and
this
coupling
on
sea
level
rise
again
the
as
far
as
we
know
where
the
first
group
in
the
world
to
be
able
to
use
adaptive
mesh
techniques
in
this
fashion
to
assimilate
sea
level
rise
from
Antarctica.
The
resolution
of
the
ocean
in
these
simulations
is
about
five
kilometers
down
around
the
edge
of
an
article's
we've
also
greatly.
B
D
E
B
I
mean
the
usual
suspects,
so
mass
momentum
and
energy,
but
we
also
there
are
the
system,
is
also
very
sensitive
to
the
strict
conservation
of
choice
constituents
in
the
Earth's
atmosphere,
because
it
turns
out
that
those
are
very
important
for
the
change
in
the
energy
budget
of
the
Earth's
atmosphere.
So
these
models
are
not
only
doing
physics
but
they're,
also
doing
chemistry
and
biochemistry
and
the
MOT.
B
We
can't
use
models
of
that
class
for
during
climate
yeah.
So
the
model
has
to
be
the
level
of
conservation
that's
required
for
climate.
They
is
completely
different
from
that
required
for
during
weather,
modeling
weather
essentially
does
one
those
models
do
not
have
to
conserve
energy,
ours
do
and
mass
and
momentum.
This.
A
B
The
atmosphere,
every
five
minutes
and
that
model
could
be
a
couple
at
join
assimilation
system
be
used
for
operational
weather
forecasts,
so
they
have
matt.
In
some
sense,
the
difference
between
the
weather
in
the
comic
community
is
the
clement.
The
the
meteorological
community
operates
at
higher
spatial
resolution,
so
they
routinely
operate
now.
They're,
there
weather
models
routinely
one
at
ten
kilometer
resolution
and
ours
are
more
like
at
fifty
or
25.
That's
the
primary
difference.
I
was.
F
Curious
that-
and
you
may
have
said
this-
I
apologize
and
have
time,
if
I
missed
it,
does
the
the
sea
level
rise
have
any
kind
of
significant
feedback
on
on
climate
change,
positive
or
negative.
B
No,
I
mean
there's
some
very
fast
feedbacks
in
the
system
that
that
will
be
manifest
a
lot
more
quickly
than
sea
level
rise
I'll,
be
there
fast
feedbacks
in
the
comment
system
that
manifest
themselves
and
timescales
of
days,
so
if
you
poke
them
in
a
suitable
way,
so
this
is
this
is
not
an
important
feedback.
It
does
have
some
there's
some
other
things.
It
does
to
the
planet
that
are
pretty
interesting
right.
So
it
starts
it
all
to
the
Earth's
gravitational
field.
B
It
starts
to
alter
the
geo
dynamics
of
the
mantle
because
you're
removing
major
amounts
of
mass,
so
there's
isostatic
rebound
yeah.
There's,
there's
some
pretty
complicated
interactions
that
structure
once
you
start
moving
ice
sheets
around
Alaska,
for
example.
You
know,
perversely,
because
of
one
of
our
former
vice
presidential
konica
candidates
is
busy
rising
out
of
the
ocean
because
of
ice
aesthetic
rebound.
So
it's
going
the
other
way,
because
it's
sort
of
bouncing
back
from
the
removal
of
the
Laurentide
ice
sheet.
B
No
well,
we've
talked
about
doing
so.
We're
going
to
have
to
that
actually
turns
out
to
be
an
order,
one
perturbation
to
the
solution
and
see.
If
you
do
not
account
for
isostatic
rebound,
you
can
get
really
badly
flawed
simulations
to
sea
level
rise.
It's
it's
actually
a
major
term
in
the
solutions,
so
we're
going
to
have
to
account
for
it
and
we'll
put
it
in
initially
in
a
fairly
crude
approximation,
but
they're
people
at
Harvard
who
work
very
hard
on
us.
C
B
B
That's
a
Freddie's
face
change
is
something
we
really
understand
very
well
in
the
atmosphere,
the
unfortunately,
the
phase
change
you
go
from
vapor
to
liquid
as
all
the
stuff
is
stuff
that
goes
along
for
the
ride
like
atmosphere,
convection,
so
honey
and
isotropic
turbulence
with
buoyancy
and
a
heat,
and
you
built
right
into
the
middle
of
it,
so
that
that
phase
change
in
the
atmosphere,
chances
would
be
a
lot
harder.
Condensing
turns
out
to
be
a
lot
harder
than
melting
in
the
climate
system.
Is
that
basically,
that
the
wolf
they're.
D
So
you
talked
about
like
extreme
weather
events,
considering
you're,
essentially
figuring
out
the
probability
of
something
extremely
unlikely
occurring.
How
good
are
your
models
today
and
how
do
you
communicate
to
the
public,
even
if
you're
out
by
a
factor
of
two,
that's
still
a
cheap,
pretty
damn
good.
D
B
You
an
example
from
our
25
kilometer
run.
We
we
nail
the
annual
cycle
and
number
of
hurricanes
in
Atlantic
compared
to
climatology,
so
we
get
exactly.
We
get
the
right
number
to
within
five
percent,
and
the
seasonal
cycle
is
spot
on
the
observations
again
to
within
roughly
ten
percent.
The
models
about
half
the
models
in
the
IPCC
assessment
get
the
right
number
of
a
mr.
rivers
of
a
year
about
half
of
them
overestimated
by
a
factor
of
two
the
models.
B
So
if
events,
but
that's
a
particularly
challenging
thing
for
the
models
to
get
heat
waves,
are
quite
easy
to
get
right,
the
models
do
a
very
good
job
of
that,
and
the
projections
about
their
rise
in
frequency
of
heat
waves
is
a
very
robust
projection
coming
out
of
that
PCC
report,
so
that
that
information
is
contained
in
the
IPCC
assessment.
For
what
it's
worth,
that's
not
a
that
PCC
report
is
not
something
that
you
know
the
general
public
tends
to
pick
up
and
read,
unfortunately,
so,
but
that's
the
issue
of
science.