Journal
JOURNAL OF MARINE SCIENCE AND ENGINEERING
Volume 5, Issue 3, Pages -Publisher
MDPI
DOI: 10.3390/jmse5030035
Keywords
shallow water equation; storm surge; parallel model; hybrid finite volume and finite element method; implicit matrix-free solver; spherical coordinate system
Categories
Funding
- HBCU-UP Research Initiation Award from the National Science Foundation [HRD-1401062]
- Enrichment Program for Minority Serving Institution Faculty through the Department of Homeland Security's Coastal Hazard Center of Excellence
- Direct For Education and Human Resources
- Division Of Human Resource Development [1401062] Funding Source: National Science Foundation
Ask authors/readers for more resources
The Computation and Modeling Engineering Laboratory (CaMEL), an implicit solver-based storm surge model, has been extended for use on high performance computing platforms. An MPI (Message Passing Interface) based parallel version of CaMEL has been developed from the previously existing serial version. CaMEL uses hybrid finite element and finite volume techniques to solve shallow water conservation equations in either a Cartesian or a spherical coordinate system and includes hurricane-induced wind stress and pressure, bottom friction, the Coriolis effect, and tidal forcing. Both semi-implicit and fully-implicit time stepping formulations are available. Once the parallel implementation is properly validated, CaMEL is evaluated against ADCIRC, an established storm surge model, using a hindcast of storm surge due to Hurricane Katrina. Observed high water marks are used to verify that both models have comparable accuracy. The effects of time step on the stability and accuracy of the models are investigated and indicate that the semi-and fully-implicit solvers in CaMEL allow the use of larger timesteps than ADCIRC's explicit and semi-implicit solvers. However, ADCIRC outperforms CaMEL in parallel scalability and execution wall clock times. Wall times of CaMEL improve significantly when the largest stable time step sizes are used in respective models, although ADCIRC still is faster.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available