fema region 3 chesapeake bay storm surge project computational system
DESCRIPTION
FEMA Region 3 Chesapeake Bay Storm Surge Project Computational System. Brian Blanton Senior Research Scientist Renaissance Computing Institute University of North Carolina 12 December 2008 Chapel Hill, NC. RENCI Objective. - PowerPoint PPT PresentationTRANSCRIPT
FEMA Region 3Chesapeake BayStorm Surge ProjectComputational System
Brian BlantonSenior Research ScientistRenaissance Computing InstituteUniversity of North Carolina
12 December 2008Chapel Hill, NC
RENCI Objective
• Extend/augment an existing modeling system into the Chesapeake Bay/Region III geographic area:
– Model grids, DEM
• Incorporate STWAVE nearshore wave model into computational system
• Post-simulation statistical analysis (development of flood-frequency curves from simulations + JPM storm weights)
• Coordinate our work with the USACE and partners for data preparation and output data management, grid generation, storm and wind characterization, and wave model integration.
• I.e. : Communicate as much of our experience in the North Carolina Floodplain Mapping Project (NCFMP) as possible.
RENCI/IMS Personnel Involved
• RENCI– Dr. Brian Blanton : Coastal Oceanographer, Computational
Scientist– Margaret Blanchard : GIS/DEM– Ken Galluppi : Management– Howard Lander : Research Programmer– Lisa Stilwell : web design/support
• UNC Institute of Marine Sciences– Dr. Rick Luettich : Coastal Oceanographer, ADCIRC Developer– Crystal Fulcher : SMS/ADCIRC grid development
Three primary RENCI Phase 1 Tasks
Task 1 – Assembly of Digital Elevation Model (RENCI/Blanchard)Existing topo/bathy data into regional, consistent DEMNCFMP has identified potential issues regarding data quality,
bathymetric data, etc.
Task 2 – Grid Development and Modeling Consultation (UNC/IMS-Luettich)
Task 3 – Extend Computational System (RENCI/Blanton)
Computational System Overview
• Use NCFMP model system as starting point– “Flexible” use of different ADCIRC grids– Has been ported to different computer systems (IBM BG/L, Dell
Clusters)
• Incorporate project ADCIRC grids
• Incorporate OWI TC96 and STWAVE as modules– Develop flexibility of model choices through well-defined interfaces
• Execute validation simulations
• Execute probabilistic tropical and extratropical storms
Physical System
Computation of StillWater ELevation and wave SETUP component to total water level due to probabilistic Tropical and ExtraTropical systems.
ADCIRC
STWAVE
Computational SystemWind/Pressure Simulator OWI TC96 and IOKA
WaveWatch III for oceanic wave field
STWAVE for nearshore wave field and radiation stress
ADCIRC for SWEL+Setup
Water level
Wave Force
Numerical Model Suite
• Oceanic waves NOAA WaveWatch3• Storm Surge ADCIRC• Wave setup STWAVE
• Wind/Pressure OWI’s TC96 (tropical)• OWI’s IOKA (extratropical)
• Augmentation of existing system with STWAVE and TC96.
• Existing system already uses OWI format (NWS12/212) so incorporating TC96 easy.
• Wrap STWAVE to “look like” SWAN.
Current Work Items
100-500 individual simulations
Order 256 processors per simulation
2-5 million cpu-hours
Compute Resources TBA
Anticipated Computational Effort
Compile/test OWI TC96 (done)
Compile parallel STWAVE (done)
Test parallel STWAVE
in progress, with Jane Smith (ERDC) and Harry Friebel (NAP)
STWAVE Scaling•Test problem
•100x100, Gulf of Mexico•48 hour simulation•30 minute timestep•96 time evaluations
•Intel Xeon 2.6Gz cluster
•About Linear speedup
•Don’t have consistent comparison with SWAN yet
•Expecting STWAVE to be faster, mostly due to globalization of output. (Handled similar to ADCIRC’s globalization)
STWAVE Scaling
NOAA Ches/Del Bay ADCIRC grid
•Probably good grid-dev starting point for bathy
•Does not support inundation
•Bathymetry to be converted to MSL by NOAA
•Then to NAVD88 by RENCI for incorporation into DEM database.