These are the slides from our May 23, 2014 Friday Forum workshop entitled 'Predicting and projecting the frequency of extreme marine events on time scales of days to decades with a focus on coastal flooding' led by Dalhousie University Professor Keith Thompson. The marine environment presents humankind with great economic opportunity but also major risks. It is a dangerous place to extract resources, and a particularly challenging environment for transportation, construction and human development. Our relationship with the marine environment is evolving due to climate change (e.g., global sea level rise, reduced pack ice in the Northwest Passage) and also shifts in economic and societal use (e.g., deep ocean drilling, marine recreational activities). In 2012 a new national network was established to bring together researchers and partners in a multi-sectoral partnership in order to improve Canada’s capabilities in Marine Environmental Observation, Prediction and Response (MEOPAR). In this talk Keith first provided an overview of this new network and then described some of its research, focusing mostly on coastal flooding. He then described how MEOPAR is making extended-range predictions of east coast storm surges, and the probability of coastal flooding, with lead times of hours to about 10 days. He also described a new statistically-based method for estimating the probability of coastal flooding over the next century, taking into account uncertainty in projections of sea level rise and storminess. Keith Thompson is a Professor at Dalhousie University with a joint appointment in the Department of Oceanography and the Department of Mathematics and Statistics. He holds a Canada Research Chair in Marine Prediction and Environmental Statistics. His research interests include ocean and shelf modelling, data assimilation, sea level variability, the analysis of extremes. New interests include the Madden Julian Oscillation and the Kuroshio Extension current system. He is presently a theme lead for the Marine Environmental Observation Prediction and Response (MEOPAR) network, a large national network established recently to help Canada respond more effectively to marine emergencies and change.

1.
Keith
Thompson
Natacha
Bernier
Dalhousie
University
Environment
Canada

2. Overview
of
Talk
 Overview
of
MEOPAR,
a
new
naDonal
network
 PredicDng
storm
surges
with
lead
Dmes
up
to
10
days
 ProjecDng
flood
probabiliDes
over
coming
decades

3. MEOPAR
in
a
Nutshell
 New
network
of
centers
of
excellence
 Marine
Environmental
ObservaDon
PredicDon
and
Response
 Reducing
vulnerability
to
marine
hazards
and
emergencies
 Established
in
2013,
headquartered
at
Dalhousie
 $25M
over
5
years
from
NCE
program
 May
be
renewed
twice
 Involves
50
researchers
from
12
universiDes
 Partners
include
EC,
DFO,
DND,
DRDC,
Lloyds
Register,
ICLR,
...

5. Dr.
Harold
Ritchie,
Environment
Canada/
Dalhousie
University
A
relocatable
atmosphere-­‐
wave-­‐ocean
forecast
system
that
can
be
set
up
within
hours
of
a
marine
emergency.
Provide
forecasts
(hours
to
days)
of
physical
properties
of
ocean
and
atmosphere
to
help
guide
response
to
an
emergency.
System
to
be
transferred
to
Environment
Canada
for
operational
use.
A
Relocatable
Atmosphere-­‐Ocean
Prediction
System
Who:
What:
Impact:
Photo
credit:
ArcticNet

6. Dr.
Jinyu
Sheng,
Dalhousie
Dr.
Susan
Allen,
UBC
Build
an
integrated
observation
and
prediction
system
for
Halifax
Harbour
and
Strait
of
Georgia.
Real-­‐time
forecasts
of
sea
level,
waves,
currents,
bio-­‐
geochemical
properties
for
ports,
municipalities,
and
the
oil
and
gas
sector.
Building
Network
of
Fixed
Coastal
Observing
&
Forecast
Systems
Who:
What:
Impact:

7. Dr.
Dany
Dumont,
UQAR
Improve
surface
drift
forecasts
in
seasonally
ice-­‐
infested
seas.
Some
buoys
deployed
by
the
UQAR
ice
canoe
team.
Respond
to
emergencies
along
Canadian
coasts
e.g.,
a
person
or
oil
patch.
Time
is
key
in
ice-­‐infested
water.
Improving
Surface
Drift
Forecasts
Who:
What:
Impact:

8. Dr.
Andrea
Scott,
University
of
Waterloo
Method
to
use
radar
(SAR)
satellite
images
to
improve
the
monitoring
of
sea
ice.
Accurate
information
about
sea
ice
conditions
is
critical
for
weather
forecasting
and
safe
navigation
in
ice-­‐
covered
regions.
Improving
Sea
Ice
Forecasts
Who:
What:
Impact:

9. Dr.
Gregory
Flato,
Environment
Canada/
Uvic
Develop
ways
to
assess
and
visualize
changes
in
the
marine
environment
and
the
associated
risks
on
climate
time
scales.
The
fishing
industry
and
coastal
communities
could.
e.g.,
use
risk
maps
to
manage
their
exposure
to
extreme
weather
events.
Climate
Change
and
Extreme
Events
in
the
Marine
Environment
Who:
What:
Impact:
Photo:
CC
Sam
Beebe

10. Dr. Katja Fennel,
Dalhousie University
Develop biogeochmical, predictive
models of the ocean and make
climate projections.
Assist planning by, e.g., fishing
industry, oil and gas industry, and
coastal communities.
Biogeochemical
Projections
Under
a
Changing
Climate
Who:
What:
Impact:

11. Photo
credit:
ArcticNet
Dr.
David
Atkinson,
UVic
Assess
how
large-­‐scale
weather
patterns
adversely
impact
marine
transport
and
industrial
activity
in
eastern
Beaufort
Sea.
Ensure
marine
operators,
coastal
communities
and
emergency
response
operators
have
access
to
weather
forecast
information
to
help
plan
operations.
User-­‐Driven
Monitoring
of
Adverse
Marine
and
Weather
States
in
the
Eastern
Beaufort
Sea
Who:
What:
Impact:

12. MEOPeople
Training
highly
qualified
personnel
is
one
of
MEOPAR’s
most
important
objectives.

13. INFORMED
SOCIETY
• More
people
using
research
results
• Information
about
the
ocean
readily
available
COORDINATED
CANADIAN
APPROACH
• Bringing
together
researchers,
industry,
and
NGOs
• Better
techniques
&
policies
• Hazard
management
TRAINED
PEOPLE
• Ocean
skills
• Student
mentoring
MEOPAR’S
Outcomes

14. PredicDng
Storm
Surges
With
Lead
Times
up
to
10
Days
Storm
surges
are
an
ever
present
danger
in
eastern
Canada
Home
damaged
by
the
storm
surge
of
December,
2010
Sainte
Luce,
Quebec
hZp://joansullivanphotography.com/STILLS/Climate-­‐change

15. Flooding
is
Caused
by
Tide
and
Surge
€
η =ηT +ηS
Halifax
February
1967

16. ForecasDng
Storm
Surges
Surge
models
are
usually
based
on
two
simple
physical
principles
expressed
by
the
following
equaDons:
DiscreDze
on
a
grid
with
realisDc
coastlines
and
water
depths.
Integrate
through
Dme
with
forecast
wind
to
forecast
surge.
€
Du
Dt
= − f ×u − g∇(η −ηp )+
τ
H
−
cd u u
H
∂η
∂t
+ ∇ • (uH ) = 0

17. Our
Surge
Model
and
Domain
•
Model
is
2D,
based
on
POM
•
Shelf
and
deep
water,
Labrador
to
Gulf
of
Maine
•
Driven
by
10
day
forecast
winds
and
air
pressure
•
DeterminisDc
(1/30°)
•
Ensemble
(1/12°)
•
1
March
2013
to
31
March,
2014

18. Typical
DeterminisDc
Forecasts
Rimouski
ObservaDons
in
black
3
day
forecasts
5
day
forecasts
7
day
forecasts

19. How
Good
are
the
DeterminisDc
Forecasts?
€
γ2
=
var(ηobs −ηmod )
var(ηobs )
=
error
obs
For
each
of
the
22
Dde
gauges
calculate
γ2

20. Allowing
for
Uncertainty
in
Wind
Forecasts

21. Visualizing
Ensemble
Surge
Forecasts
5d
forecast
for
22
March
2013

22. 5
Day
Forecasts
of
Total
Water
Level
€
η =ηT +ηS
Sea
Level
(m)

23. ProjecDng
Flood
ProbabiliDes
Over
Coming
Decades
Such
informaDon
is
needed
for
sensible
adaptaDon
strategies.
Problem
is
conceptually
similar
to
predicDng
total
water
levels
10
days
into
future.
Let’s
start
by
looking
at
some
observaDons
from
the
long
Halifax
sea
level
record.

24. Annual
Means
and
Maxima
for
Halifax
Halifax
1920-­‐2001
Offset
due
to
Ddes

25. Annual
Maxima
About
Annual
Means

26. Probability
of
Flooding
Today
Halifax
return
level
about
mean
(m)
+0.3m
Return
period
(years)

27. 100y
ProjecDons
of
Flood
ProbabiliDes
Simplest
approach:
Assume
mean
sea
level
will
increase
by
fixed
amount
and
just
raise
return
levels.
“DeterminisDc”.
But
sea
level
increase
over
next
century
is
highly
uncertain
(e.g.,
uncertain
emission
scenarios,
model
errors).

28. Projected
Sea
Level
Rise
Over
Next
Century
IPCC,
2013:
Summary
for
Policymakers.
Figure
SPM.9
“medium
confidence”

29. ProjecDng
Probability
of
Total
Water
Level
Write
annual
maximum
as
sum
of
annual
mean
and
a
deviaDon:
Assume
pdfs
for
these
two
components
are
of
form:
The
pdf
of
annual
maximum
is
convoluDon
of
these
two
pdfs.
€
η = ηA + ηD
€
p(ηA ) = w1δ(ηA −ηS1)+ w2δ(ηA −ηS2 )+...
p(ηD ) = φG (ηD )

30. Idealized
Example
Assume
there
are
only
possible
SLR
scenarios:
S1:
Sea
level
rises
at
0.3m
per
century
P(S1)=0.8
S2:
Sea
level
rises
at
1.0m
per
century
P(S2)=0.2

31. Impact
of
Uncertainty
on
Return
Levels
Return
level
for
Idealized
Example
(m)
Return
period
(years)

32. Average
Dme
between
floods
(years)
What
Should
Halifax
Expect
Today?
1.9m
300y

33. Impact
of
1.9m
on
Downtown
Halifax
Charles
et
al.,
2011
Expect
one
every
300y
if
present
condiDons
prevail

34. Flood
level
(m)
What
Should
Halifax
Expect
in
2100?
1.9m
4y
Probability
of
exceeding
high
flood
levels
is
determined
by
more
extreme,
but
less
likely,
scenarios

35.  Trend
toward
probabilisDc
predicDons
and
projecDons
of
sea
level,
based
on
ensembles
and
expert
knowledge.
 Uncertainty
is
not
a
sign
of
bad
models
or
science.
 Surge
predicDons
are
improving
(known
unknowns).
Expect
rapid
improvements
over
next
five
years.
 Climate
projecDons
more
complex
(unknown
unknowns?)
BeZer
understanding
may
lead
to
greater
uncertainty.
 Work
presented
here
illustrates
a
small
part
of
the
research
being
conducted
by
MEOPAR.

36. Impact
of
Uncertainty
on
Probability,
and
Number,
of
Floods
with
Time
Critical
level
Is
2
m