1
Downscaling storm surge models for
engineering applications
John Baugh
A. Altuntas, J. Rutledge, T. Dyer
Department of
Civil, Construction, and Environmental Engineering
North Carolina State University
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
2
Starting Point
• How do we assess the effects of storm surge
on civil infrastructure?
• Engineering design scenarios imply
topographic changes, as do failures.
• Hundreds of cases may be worth considering.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
3
Relevance to DHS
• Gap being addressed: No computationally
tractable approach exists for assessing storm
surge effects on proposed infrastructure
changes and improvements.
• The DHS mission includes managing risks to
critical infrastructure, ensuring resilience to
disasters, and mitigating hazards.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
4
Interface Conditions
• May be formulated in a variety of ways:
– One-way nesting (Spall and Robinson 1989)
– Two-way nesting (Debreu and Blayo 2008)
– Full coupling (Cailleau et al. 2008)
• Used to resolve smaller spatial and temporal
scale processes
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
5
Levee Failure Scenarios
• Nesting for infrastructure assessment:
(3 different failures, flooding outlined in white)
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
6
Year 5 Activities
1. Subdomain modeling of storm surge and
wave effects using ADCIRC and SWAN
2. Time-varying topobathy in ADCIRC using a
predetermined set of parameters
3. Additional means of reducing computational
cost in simulation approaches
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
7
1. Storm Surge and Waves
• SWAN is a phase-averaged spectral wave
model with two possible types of boundary
conditions:
– TPAR files contain significant wave height, wave
period, peak direction, and directional spread.
– 2D Spectra files contain N discrete frequencies
and M directions.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
8
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
9
Hurricane Fran and Cape Fear
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
10
Hurricane Fran and Cape Fear
TPAR Files
Parameter
2D Spectra Files
Threshold
Nodes
%
Nodes
%
Significant Wave Height
5 cm
7077
24.7
25
0.087
Wave Period
1 sec
7143
24.9
53
0.185
20⁰
1297
4.53
218
0.761
Peak Direction
Conclusion: Using 2D spectra files to enforce the boundaries of a
subdomain results in accurate and efficient computational
modeling for engineering purposes.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
11
2. Time-Varying Topobathy
• The result of a net transport of sand along the
ocean bottom or the overtopping and collapse
of a dune, or the failure of engineered
structures such as levees and geotubes.
• To accommodate, allow time-varying changes
in bathymetric depth DP using a
predetermined set of parameters.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
12
• When bathymetric depth DP
changes
– adjust water surface elevation
ETA to maintain constant
water column height
– mark node wet for
reevaluation
– remember prior DP values for
consistency in calculating
heights H0, H1, and H2
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
13
Test Case: Hatteras Subdomain
• Extracted from the western North Atlantic grid
with refinements for simulating Isabel Inlet
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
14
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
15
3. Subduration Modeling
• Subduration modeling is introduced as a means of
downscaling hurricane storm surge models in time.
• The hot-start feature of ADCIRC allows users to begin a
run from a specified timestep using initial conditions
obtained from a previously performed run.
• This feature is used to reduce the total runtime of
series of simulations where users have made
topographic or other changes to a model.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
16
Subduration Modeling in ADCIRC
Step 0 : Begin with an ADCIRC model
ETA: Surface Elevation, DP: Bathymetric Depth
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
17
Subduration Modeling in ADCIRC
Step 1 : Identify nodes that are part of the changed
terrain.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
18
Subduration Modeling in ADCIRC
Step 2 : Record a hot-start file before one of the
changed nodes becomes wet.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
19
Subduration Modeling in ADCIRC
Step 3 : Make changes to parts of the terrain (DP) in
the fort.14 file.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
20
Subduration Modeling in ADCIRC
Subdomain ADCIRC has been modified so that the
original ETA of a changed node is initially set to be
equal to changed DP, during a hot-started run.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
21
Subduration Modeling in ADCIRC
Step 4 : Perform the hot-started run. Modified
subdomain ADCIRC runs the hot-started simulation as it
normally would.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
22
Test Case: Cape Fear Subdomain
• Extracted from the western North Atlantic grid
with a protective structure to prevent flooding
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
23
Test Case: Cape Fear Subdomain
Simulation
Full Run:
Research Lead  The University of North Carolina at Chapel Hill
CPU Hours
%
1080
100%
Subdomain:
28
2.59%
Subduration:
7
0.64%
CHC-R 5th Annual Meeting
January 31-February 1, 2013
24
Products
• Software tools for subdomain modeling,
modifications to ADCIRC, user scripts,
visualization tools, user guide already being
distributed.
• Theses, technical reports, conference talks,
articles under review and in preparation.
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
25
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
26
Prospective Collaborators and End Users
Researchers
– Yoonhee Park / Professor Art Rice, Landscape Architecture, NCSU, Raleigh, NC
– Fernando Magarinos Lamas, Mathematician, Physical Oceanography,
Universidad Nacional Autonoma de Mexico, Mexico City, Mexico
– Celso Ferreira, Assistant Professor, George Mason University, Fairfax, VA
Consulting
– Mark Prater, Research Analyst, WeatherPredict Consulting, Kingston, RI
Emergency Management
– Sandy Sanderson, Director, Dare County Emergency Management, Manteo, NC
Federal Agencies
– Tucker Mahoney, Coastal Engineer, FEMA Region IV, Atlanta, GA
– Max Agnew, Hydraulic Engineer, USACE New Orleans district, New Orleans, LA
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
27
Prospective Collaborators and End Users
International Agencies
– Professor Weihua Fang, Academy of Disaster Reduction and Emergency
Management, Ministry of Civil Affairs and Ministry of Education, Beijing, China
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013
28
Looking Forward
• Case studies that draw on approaches developed
in preceding years, demonstrating their benefit in
enhancing the resilience of coastal communities
• Interactive software tools that support modeling
activities for the actual case studies being
performed
• Decision-making approaches that constitute best
practices from the operations research
community for coastal infrastructure design
Research Lead  The University of North Carolina at Chapel Hill
CHC-R 5th Annual Meeting
January 31-February 1, 2013