Implementation of an Implicit Solver in Adcirc Storm Surge Model
Hurricanes are among the worst natural disasters, and storm surges caused by these hurricanes are the deadliest and most exorbitant contributors in the resulting destruction. The current state of science does not offer any remedy to stop a hurricane from occurring. Therefore, accurate storm surge models capable of predicting water velocity and elevation are indispensable. In this study, profiling and evaluation of the current implicit model, namely Computation and Modeling Engineering Laboratory (CaMEL), against Advanced Circulation (ADCIRC) model is performed to investigate the performance of these models. This investigation helped to understand the models’ strengths and weaknesses, including bottlenecks from various factors such as the following: execution time, computational intensity, Million Instructions Per Seconds (MIPS) and Mega Floating-Point Operations Per Second (MFLOPS), cache memory miss rates, Table Lookaside Buffer (TLB) misses, Input and Output (IO), and others. After evaluating the models, the implementation of the CaMEL implicit solvers to the ADCIRC storm surge model is planned and executed. Implicit solvers are generally more stable than typical explicit or semi-implicit solver, and hence they are capable of entertaining large timesteps. The implemented implicit solver uses hybrid finite element and finite volume techniques for solving shallow water equations. Objectives of this research include enhancing numerical stability, providing the option of using large timesteps, and the usage of a relatively easier mathematical formulation than that currently existing in ADCIRC. The storm surge hindcast of Hurricane Katrina that hit Louisiana and Mississippi in 2005 is used as a case study. Stability of the solver, comparison of water elevation and velocity against observed data, impact of timestep sizes, and execution times of solvers are thoroughly investigated in this study. Results of the implemented implicit solver are compared with those of existing lumped explicit and semi-implicit solvers of ADCIRC, and the findings appear promising.
Engineering|Ocean engineering|Computer science
"Implementation of an Implicit Solver in Adcirc Storm Surge Model"
ETD Collection for Tennessee State University.