Optimizations of COAWST for a Large Simulation on the Earth Simulator

Tropical cyclones are one of the major reasons for catastrophic damage in low-altitude coastal regions. In the last few decades, the world has experienced several devastating damages because of intense cyclones influenced by climate change in the tropical, the sub-tropical and the mid-latitude regions. Therefore, it is necessary to research on the changes in the frequency and intensity of the cyclones influenced by the global warming and to predict the possible damage caused by them using computer simulations. Our aim is to study the characteristics of the high-intensity tropical cyclones using a scientific application, named COAWST, designed to understand the coastal changes caused by the natural processes and to develop a faster model to predict the probable damage with proper warning and with adequate accuracy on the Earth Simulator which is known for one of its kind in the efficient simulation of the natural phenomenon using its high-performance vector processor. In this paper, we have proposed some vectorization strategies on the Earth Simulator to improve the computational performance of COAWST up to 38% to that of its original version for a large simulation. It also aims to showcase the importance of the vectorization technology in order to simulate tropical cyclones related to coastal disasters timely and accurately.