US Army Corps of EngineersCoastal and Hydraulics Laboratory - ERDC SWIMS Hawaii Hurricane Inundation Fast Forecast Tool Jane McKee Smith, Andrew B. Kennedy,

US Army Corps of Engineers Coastal and Hydraulics Laboratory - ERDC

SWIMS Hawaii Hurricane Inundation Fast Forecast Tool

Jane McKee Smith, Andrew B. Kennedy, Alexandros A. Taflanidis, Joannes J. Westerink,

Kwok Fai Cheung, and Thomas D. Smith


Introduction• Island communities are vulnerable to storms

• Nowhere to evacuate

• Infrastructure within hazard zone

• Unique/Important Island Features & PhysicsSteep slopes

Reef roughness

Reef flat dynamics (ponding, wave reformation)

Importance of waves

Growth of long waves


Pacific Islands Land-Ocean Typhoon (PILOT) Exp.

Surge & Wave Island Modeling Studies (SWIMS)

Products• Field data sets:

• Guam• Saipan• Oahu, Hawaii• St. Croix, VI

• Laboratory Data Sets (2D/3D)• Model Improvements• Fast Forecasting System

Objectives:

•Collect coastal processes data under typhoon and high-wave conditions in island environments

•Improve model physics and tools for island application for emergency management and design


SWIMS Fast Forecasting System• Pre-run storms with high-fidelity models

• ADCIRC

• unSWAN

• BOUSS-1D

• Create database of response

• Develop surrogate model to forecast inundation• Deterministic

• Probabilistic

• Hurricane Evacuation Studies Mass Management System • Interface for Emergency Managers (MMS)


Potential Impacts

• Cat 4 storm on Oahu• Severe damage to air & sea ports• Island-wide power and communications outages (1 month

or longer)• 80% of homes destroyed• 650,000 people seeking shelter

• Since 1950:• Nina (1957)• Dot (1959)• Iwa (1982)• Estelle (1986)• Iniki (1992)

Hurricane InikiHurricane Iniki


•Five base storm tracks from hurricane climatology (NWS)

•Tracks shifted to give different landfall locations

•Tracks and parameters varied to give a matrix of potential storms (bound most possible landfall scenarios)

Storm Selection: Tracks

A (120°)

B (150°)

C (180°)

D (210°)

E (240°)

165° W 160°W 155

°W 150

° W 145° W

10° N

15°N

20°N


Storm Selection

• Base Tracks• 120,150,180, 210, 240

deg

• Central Pressure• 940, 955, 970 mb

• Radius of Maximum Winds• 30, 45, 60 km

• Forward Speeds• 7.5, 15, 22.5 kts

• 15 Landfall Locations (Oahu and Kauai)

A (120°)

B (150°)

C (180°)

D (210°)

E (240°)

165° W 160°W 155

°W 150

° W 145° W

10° N

15°N

20°N


Wave and surge prediction (high-fidelity model)• High-resolution grid Oahu and Kauai

• SWAN+ADCIRC wave and circulation models

• Validation with tides and Hurricane Iniki

Wave runup prediction

• BOUSS-1D

Wave, surge, and runup inundation database and predictions for new events

• Surrogate modeling

Wave, Surge, and Runup Inundation Database


• Domain incorporates Hawaiian Islands and north central Pacific Ocean

• Grid resolution ranges from 30 m on land and in the nearshore to 5000 m in deep water

• Incorporates high resolution features, channels, coral reefs and wave breaking zones

• 1,590,637 nodes

• 3,527,785 elements

Grid Domain



• ADCIRC solves for water surface elevations and currents in two dimensions

• SWAN solves the wave action density and is a phase-averaged wave model with wave energy represented by a spectrum

• ADCIRC passes water elevation and currents to SWAN

• SWAN passes wave radiation stresses to ADCIRC

• Models run on in parallel on the same grid

SWAN+ADCIRC Model – Coupled Waves and Currents on Unstructured Grids


Hurricane Iniki (1992)


Hurricane Iniki Water Levels

NOAA water elevation gauges


Storm Atlas


Wave Height (ft) for Category 4 storm, forward speed of 7.5 knots


Wave Height (ft) for Category 4 storm, forward speed of 22.5 knots


• SWAN+ADCIRC gives wave heights and still water levels near shore

• Wave runup (intermittent wave inundation at the shore) can be dominant in some storms

• Hundreds of meters inland, several meters more elevation than still water level

• During Hurricane Iniki (6-8m)

• Two approaches to wave runup

• Parameterized relations predict runup given the significant wave height, wave period, and basic nearshore bathymetry

• Boussinesq modeling along one-dimensional transects

• Hawaiian topography too complex to use parameterized results

Wave Runup Analyses


• Use one dimensional Boussinesq model BOUSS-1D to compute runup over 750 transects on Oahu, 443 on Kauai

• Phase resolving - represents each individual wave through time and space

• Forced by waves and water level at offshore boundary of transect

• Precompute runup for a matrix of storms and water levels

• SWAN+ADCIRC computations of waves, water level projected onto pre-computed results to give runup for any storm

Wave Runup Analyses


Time(s)

Time(s)

Time(s)

Run

up

Ele

vatio

n (m

)C

ross

-sho

re

Dis

tanc

e (m

)O

ffsho

re w

ave

time

serie

s (m

)

Wave Runup Analyses: Boussinesq Model Output along a Transect


160 W 158

W 156

W 154

W

18 N

20 N

22 N

• Pre-run suite of basis hurricane scenarios

• Moving least-squares response surface surrogate model

• Predict the output for any new hurricane scenario

Basis hurricane scenarios

New hurricane scenario

Surrogate Model


1

1

1

3.5

3.5

3.5

6

6

6

8.5

8.5

11

11

11

13.5

13.5

13.5

16

16

16

16

16

16

18.5

18.5

18.5

18.5

18.5

21

21

21

21

21

23.5

23.5

23.5

23.5

26

26

2626

26

28.5

28.5

28.5

28.5

28.5

28.5

28.5

31

31

31

31

31

31

33.5

33.5

33.5

33.533

.5

33.5

33.5

36

36 36

36

26

26 26

38.5

38.5

38.5

23.5

23.5

21

21

18.5

18.5

16

16

13.5

13.5

11

11

8.5

8.5

6

6

3.5

3.5 13.5

13.5

1

1

11

11

3636

8.5

8.566

3.5

3.511

13.5118.563.51

13.5118.56

28.5

3.51

36

41

-159.5 -159 -158.5 -158 -157.5 -157 -156.5

20.5

21

21.5

22

22.5

Feet

5

10

15

20

25

30

35

40

Comparison of Hurricane Output Predictions

Surrogate model predictions

1

1

1

3.5

3.5

3.5

6

6

6

8.5

8.5

11

11

13.5

13.5

13.5

16

16

16

16

16

18.5

18.5

18.5

18.5

18.5

21

21

21

21

21

23.5

23.5

23.5

23.5

23.5

26

26

26

26

26

28.5

28.5

28.5

28.5

28.5

28.5

28.5

31

31

31

31

31

31

33.5

33.5

33.5

33.5

33.5

33.5

33.5

36

36 36

36

38.5

38.5

38.5

26

26 2623.5

23.5

21

21

18.5

18.5

16

16

13.5

13.5

11

11

8.5

8.5

6

6

3.5

3.5 13.513

.5

11

36

36

11

118.5

8.566

3.5

3.511

13.5118.563.51

1613.51

18.5

28.5

63.51

26

41

1

-159.5 -159 -158.5 -158 -157.5 -157 -156.5

20.5

21

21.5

22

22.5

Feet

5

10

15

20

25

30

35

40

High-fidelity model predictions


Cone of potential tracks

Expected hurricane track

165 W 160

W 155

W 150

W 145

W

10 N

15 N

20 N

• Uncertainty in all hurricane parameters (described by probability models)

• Use the surrogate model to predict•Deterministic scenario

•Average (expected) output

•Output with certain probability of being exceeded

•Probability (%) of exceeding a specific threshold

Current hurricane location

Probabilistic Prediction of Hurricane Output (Risk)


Graphical User Interface




Summary• Fast Forecasting Tool provides framework

for dynamic and fast evaluation of waves, surge and inundation

• High-fidelity, high-resolution models to simulation hundreds of hurricanes

• Query the database for deterministic or probabilistic estimates

• Results in seconds to minutes

• Simulations for the Big Island and Maui County are running now (~800 runs)

• Ongoing work in Guam, too…


Guam• Jackson State U.: Himangshu S. Das and Hoonshin Jung

• Grid: circular domain (6 deg radius ~ 666 km)

• Resolution 14 km to 25 m

• 239,664 nodes and 473,083 elements


Guam --Validation (Nockten)


Guam --Validation

(Nida)


Synthetic Storms for Guam (~750)


Questions

Documents

US Army Corps of EngineersCoastal and Hydraulics Laboratory - ERDC SWIMS Hawaii Hurricane Inundation Fast Forecast Tool Jane McKee Smith, Andrew B. Kennedy,