Abstract:
Most of the countries around the Bay of
Bengal are threatened by storm surges associated
with severe tropical cyclones. The destruction along
the coastal regions of India, Bangladesh, and
Myanmar are serious due to the storm surge. To
mitigate the impacts of tropical storm, the prediction
of storm surge need to be accurate. Traditional
process-based numerical models have the limitation
of high computational demands to make timely
forecast and deterministic numerical models are
strongly dependent on accurate meteorological input
to predict storm surge. In this work, a Multilayer
perceptron (MLP) and a Radial Basic Function
Network (RBFN) used to predict the maximum
sustained wind speed in knots (VMAX) of storm in
coastal areas of Bay of Bengal. The ANN network
model provides fast, real-time storm surge estimates
at Bay of Bengal. Simulated and historical storm data
are collected for model training and testing,
respectively. North India Ocean Best Track Data
from Joint Typhoon Warning Center (JTWC) used to
perform experiments. The result of MLP is predicted
VMAX value closer than in RBFN prediction.
