numerical modeling and uncertainty analysis for effects of ...oct 31, 2011 · florida state...

Ming Ye ([email protected]), Heng Dai, and Dejun FengDepartment of Scientific Computing, Florida State University

Joseph Donoghue ([email protected]) and Steven Kish ([email protected]) Department of Earth, Ocean and Atmospheric Science,

Florida State University

Alan Niedoroda ([email protected]) and Bikash SahaURS Corporation

Numerical Modeling and Uncertainty Analysis for Effects of Near-Term Sea-Level

Rise on Barrier Islands

(http://www.defense.gov/qdr/images/QDR_as_of_12Feb10_1000.pdf)

“… the National Intelligence Council judged that more than 30 U.S. military installations were already facing elevated levels of risk from rising sea levels. DoD’s operational readiness hinges on continued access to land, air, and sea training and test space. Consequently, the Department must complete a comprehensive assessment of all installations to assess the potential impacts of climate change on its missions and adapt as required. In this regard, DoD will work to foster efforts to assess, adapt to, and mitigate the impacts of climate change. Domestically, the Department will leverage the Strategic Environmental Research and Development Program, a joint effort among DoD, the Department of Energy, and the Environmental Protection Agency, to develop climate change assessment tools.“

Project Overview

2

U.S. Military Bases

Eglin AFB, FL

Norfolk Naval Base, VA

Camp LejeuneMCB, NC

Camp Pendleton MCB and Coronado Naval Base, CA

DoD-SERDP Studies on the Effects of Sea-Level Rise and Climate Change (2009-2014)

Project Overview

3

Study Area

Project Overview

4

Historic Storm Record, 1850-Present

Ocean

(Lambert, 2003)

Prehistoric Storm Record from Coastal Lake Cores

Santa Rosa Island

Choctawhatchee Bay

Western Lake

Analyzing Paleostorm Record in Coastal Sediments

Coastal Storms

5

Fig. 4

Fig. 2Fig. 1

Fig. 3

Figure 1. Aerial imagery of infrastructure located on Santa Island, Eglin Air Force Base, protected by a sea wall before Hurricane Ivan, 2004. Yellow arrows indicate location of features common to all photos.

Figure 2. Same location, after Hurricane Ivan (2004) and before Hurricane Dennis (2005).

Figure 3. Same location, after Hurricane Dennis (2005).

Figure 4. Historic shorelines of 1870 and 1934 are superimposed on a recent aerial image. Shoreline borders represent uncertainty in shoreline position measurement, +/- 15 m.

Shoreline Change on Santa Rosa Island, Site A-13

Coastal Change

6

7-17-2001

2-23-2007

9-17-2004

7-12-2005

Project Overview

Ø Identify and quantify the responses of coastal system components to sea-level rise and increased hurricane activity out to 2100 AD.

Ø Develop guidelines for using existing techniques and developing new methods for evaluating risk to coastal military installations.

Ø Develop probability methods for quantifying uncertainty in coastal risk analysis.

Ø Evaluate mitigation and adaptation strategies for near-future climate change.

Research Objectives

7

191980 80

RaRangnge e ofofSe

a Le

vel (

mm

)

Bayflats

channels

InletEbb shoalThroatFlood shoals

BeachSurf zoneDune

Shoreface/Shelf

Longshoreone

-line

‘Gensis’-type Cross shoreCowell/Roy-type

ASMITA

2D advection/ diffusion

HAZUSSLOSH MOCCS SLAMM SEAWATX-BEACH

Modeling Strategy

Inputs and forcing Calibration/verification

New

MODELING ANALYSES

Historic

COASTAL DATA

Lidar imageryShoreline change-

1780-2007

Project Overview

0 5,000 10,0002,500 Meters¹

Explanation

De and Unde Dry Land

Hardwood Swamp

Cypress Swamp

Inland Fresh Marsh

Tidal Fresh Marsh

Transitional Salt Marsh

Regularly Flooded Marsh

Mangrove

Estuarine Beach

Tidal Flat

Ocean Beach

Inland Open Water

Riverine Tidal Open Water

Estuarine Water

Open Water

Irregularly Flooded Marsh

Inland Shore

Tidal Swamp

Explanation


Hardwood Swamp

Cypress Swamp

Inland Fresh Marsh

Tidal Fresh Marsh



Mangrove

Estuarine Beach

Tidal Flat

Ocean Beach

Inland Open Water


Estuarine Water

Open Water


Inland Shore

Tidal Swamp

Explanation


Hardwood Swamp

Cypress Swamp

Inland Fresh Marsh

Tidal Fresh Marsh



Mangrove

Estuarine Beach

Tidal Flat

Ocean Beach

Inland Open Water


Estuarine Water

Open Water


Inland Shore

Tidal Swamp

8

Sea-Level Change

Pensacola, FL, tide gage data, 1923-2010. Mean sea-level rise: 2.1 mm/yr.

(Global data adpted from Burke et al. (2008); original data from Church and White (2006), and Cazenave and Nerem (2004))

0.8 mm/yr

2.0 mm/yr

3.3 mm/yr

Satellit

e DataTide Gauge Data:

Long-Term Mean 1.7 mm/yr Global mean sea-level rise, from tide gage and satellite data

9

(U.S. Army Corps of Engineers, 2009, Water Resources Policies and Authorities Incorporating Sea-Level Change Considerations into Civil Works Programs: USACOE Circular No. 1165-2-211)

Sea-Level ChangeProjecting Future Sea-Level Rise

U.S. Army Corps of Engineers (2009) Sea-Level Rise Scenarios for Civil Works Project Planning

10

Recent Published Model Estimates of

Year 2100 Sea-Level Rise (m)

0.6IP

CC

(20

07)

Hor

ton

et a

l (2

008)

Jeve

rjeva

et a

l. (2

010)

Grin

sted

let a

l (2

009)

Rah

mst

orf

(200

7)

Verm

eer a

nd R

ahm

stor

f(2

009)

1.9

1.4

1.6

0.9

Pfef

fere

tal.

(200

8)

2.0

1.3

Sidd

alle

t al.(

2008

)

0.8

Sea-

Leve

l R

ise

(m)

Long-Term Tide Gage Data

1870-present (1.7 mm/yr)

Satellite Altimetry Sea-Level Data 1993-present(3.3 mm/yr)

1880 2000

Range of SLR rate over the past 2000 yr

Projecting Future Sea-Level RiseSea-Level Change

11

Shoreline retreat, eastern portion of Eglin AFB facilities on Santa Rosa Island, 1871-2006

0 250 500125 Feet

Explanation2006

2004

1999

1995

1985 Shoreline

1901-1979

1871

Site A-13EGLIN AFB

Santa Rosa Island2007

Site A 13

0 100 20050 Meters

Shoreline retreat: 100 meters (330 feet) (1871-2006)

1871

2006

Santa Rosa Island

Effect of Storms on Coastal RetreatCoastal Storms

12

ResultsGulf of Mexico

Eglin AFB

SouthNorth

Dune elevation pre- Hurricane Opal (1995)

Coastal StormsEffect of Storms on Coastal Morphology

13

Flow Chart of Dune Modeling (ACUTE)

● Total simulation time is 100 years.● Consider storm, sediment transport, dune growth, and sea-level rise.● Model simulation under random storm events and different sea-level rise

scenarios.

Characterization of Dune Geomorphology

Characterization of Dune Geomorphology(a) Dht

ZoYoff

DBofffBackofff

BSoff

OWoff

Dht

ZoYoff

DBofff Backofff

BSofff

OWoff

DunePasss

DL

DunePass

AFace ARepose

Single Dune

Dune FieldAFace

ARepose

(b)

Dune Erosion and Eroded Volume Partition

Calculation of the volume of storm erosion requires running SLOSH (Sea, Lake, and Overland Surges from Hurricanes).

SurgeHeight

WaveRun-upHeight

DunePass

24 ( )E stV C R Sht ZoT

Lidar DEM

Model Output

ACUTE Component Description

Model Simulation with a Test Island

18

30 km

0.6 kmTRUE SCALE

10 20 30 40 50 60 70 80 90 100 110 1200

5

Dun

e he

ight

(m)

10 20 30 40 50 60 70 80 90 100 110 1200

50

Grid indices

Dun

e le

ngth

(m)

0 20 40 60 80 100 1200

500

Isla

nd

Wid

th (m

)

Grid Indices

Year 6 16 21 52 70 70 87

Storm number 1 1 1 1 2 2 1

Storm track 2 1 3 3 2 2 2

Storm size 2 4 3 1 3 2 1

0 20 40 60 80 1000

1

2

3

4

5

6

Time (year)

Dun

e he

ight

(m)

Dune height at cross-section 15Dune height at cross-section 106

Storm Sequence

Effect of Storm on Dune Height Evolution

Storm track 1

Storm track 2

Storm track 3

0 20 40 60 80 100 1200

200

400

600

Isla

nd W

idth

(m)

Grid Indices

Quantification of predictive uncertainty is needed for coastal management and decision-making.

Predictive Uncertainty

Analyzing Uncertainty

20

Uncertainty Sources Classification and Uncertainty Quantification

Scenario S1 Scenario S2

AddressScenario

Uncertainty

Model M1 Model M2

Parameter P1

AddressModel

Uncertainty

AddressParametricUncertainty

UQ(P2|M2,S1)

UQ(P,M|S1) UQ(P,M|S2)

UQ(P,M,S)

QuantifyParametricUncertainty

QuantifyParametric and Model

UncertaintyQuantify

Parametric, Model, and ScenarioUncertainty

UQ(P1|M1,S1)

Same or Different Models and Model

Parameters

Parameter P2


21

Uncertainty Quantification in Modeling of Barrier Island Evolution: Uncertainty Sources

● Scenario Uncertainty: consider five scenariost Four SERDP SLR scenariost Baseline scenario with the current rate of sea-level rise

● Parametric Uncertainty: parameters of stormst Storm numberst Storm trackst Storm size

0 10 20 30 40 50 60 70 80 90 1000

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Time (year)

Sea

-leve

l ris

e (m

)

scenario 1scenario 2scenario 3scenario 4baseline scenario


22

Uncertainty Characterization: Storm Number

Storm number follows the Poisson distribution

( ; )!

keP kk

λ = 6: on average six storms occur in 100 years.k: random number of storms over 100 years


23

0 5 10 150

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

Storm number in 100 years

Pro

babi

lity

Den

sity

Storm number dataFit with Poisson distribution

Uncertainty Characterization: Storm Track

● Storm track follows the uniform distribution● A storm hits the island along one of the three tracks with equal

probability (1/3)


24

Storm track 1 Storm track 2 Storm track 3

0 20 40 60 80 100 1200

200

400

600

Isla

nd W

idth

(m)

Grid Indices

Uncertainty Characterization: Storm Size

Determined empirically based on the Okaloosa County flood study (FEMA, 2002)

Coastal Surge Elevation

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 1 2 3 4

Surge Height

Cum

ulat

ive

Prob

abili

ty

Series1

0.3 m 0.64 m 1.16 m2.1 m


25

Example: Realization and Quantities of Interest

● Generate 1,000 parameter realizations

● Run the Acute model for the 1,000 realizations

DHT

DunePass

ZoYoff

DBoff BackoffBSoff

OWoff

AFace ARepose

Single Dune

Quantities of interest:• Dune height• Backshore position• Backflat elevation


26

Year 6 16 21 52 70 70 87

Storm number 1 1 1 1 2 2 1

Storm track 2 1 3 3 2 2 2

Storm size 2 4 3 1 3 2 1

Predictive Uncertainty of Dune Height

● Temporal variation of mean prediction of dune height at the center of the island under the five sea-level rise scenarios.

● Dune height increasesunder the baseline scenario and SERDP scenario 1 (m), but decreases under the other three scenarios.


27

Predictive Uncertainty of Dune Height

Probability density function (PDF) of dune height at the island center

25 years 50 years

75 years 100 years

• For all the scenarios, predictive uncertainty increases with time.

• At early time, predictive uncertainty is similar under different scenarios.

• At later time, predictive uncertainty is smaller for larger sea-level rise.


28

Scenario Averaging

● Consider I scenarios (five sea-level rise scenarios)● p(Δ|D,Si): probability density of Δ under scenario Si

● p(Si): probability of occurrence of scenario Si

● p(Δ|D): averaged over all scenarios● How to determine the scenario probability?

t Expert eliminationt Quantitative assessment

1

,I

i ii

p p S p S

D D

(Kopp et al. (2009, Nature): Probabilistic assessment of sea level during the last interglacial stage)


29

Scenario Uncertainty• Results of scenario averaging for dune height at the island center using

126 sets of scenario probability• Predictive uncertainty is between the uncertainties of the two scenarios

of the smallest and larger sea-level rise.


30

-1 -0.5 0 0.5 1 1.5 2 2.5 3 3.50

0.1

0.2

0.3

0.4

0.5

Dune height(m)

Pro

babi

lity

dens

ity

S4 resultBaseline result

-1 0 1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Dune height(m)P

roba

bilit

y de

nsity


-1 0 1 2 3 4 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Dune height(m)

Pro

babi

lity

dens

ity


-1 0 1 2 3 4 50

0.2

0.4

0.6

0.8

1

1.2

1.4

Dune height(m)

Pro

babi

lity

dens

ity


Scenario 4Baseline scenario




25 years 50 years

75 years100 years

Conclusions● Sea-level rise and storms have significant effects on barrier island

evolution and on military infrastructure on the island.● We have developed an ACUTE model for simulating barrier island

evolution under sea-level rise; a large-scale model for the coastal system, MoCCS, is under development.

● We have developed a method of quantifying predictive uncertaintydue to parametric and scenario uncertainty.

● Numerical modeling and uncertainty analysis were performed for atest island developed based on Santa Rosa Island, NW Florida.

● Predictive uncertainty is different at different simulation times and under different scenarios of sea-level rise.

● Parametric uncertainty dominates at early time, but scenario uncertainty becomes more important at a later time.

● Predictive uncertainty of scenario averaging is between that of the worst and best scenarios.

Model Component for Morphologic Change on a Barrier Island

Island platform freeboarddecreases

Dune height reduced by SLRbut also grows each yearBeach prism

trackssea level rise

Bay shorelineencroaches

Island platform freeboarddecreases

Dune erodedBay shorelineprogrades

Island platform freeboardincreases

Shoreline erodes OVERWASH

StormSurge

Years with sea-level rise and no storms

Year with sea-level rise and a major storm

Coastal Change

33

Barrier Island Response to Sea-Level Rise

Tracks Sea Level Does Not Track Sea Level

Dune Annual GrowthEquilibrium max height

Monte Carlo Simulation

● Identify random parameters X and their distributions p(x) (uncertainty characterization)

● Draw samples from the distributions ● Run the model for each sample● Obtain probability density function of desired predictions

X

p(x)


36

Predictive Uncertainty of Backshore Position

● At early time, predictive uncertainty is similar under different scenarios.● At later time, predictive uncertainty is larger for larger sea-level rise.

85 90 95 100 1050

0.1

0.225 years

70 80 90 100 1100

0.1

0.250 years

60 80 100 1200

0.1

0.275 years

40 60 80 100 1200

0.1

0.2100 years

Bay shore position (m)

Pro

babi

lity

dens

ity

SLR+0.15SLR+0.5SLR+1.0SLR+1.5SLR+2.0

Probability density function (PDF) of dune height at the island center


37

numerical modeling and uncertainty analysis for effects of ...oct 31, 2011 · florida state...

Documents