Wave Height Prediction Based On Wind Information By Using General Regression Neural Network, Study Case In Jakarta Bay

Abstract

Abstract—Information about ocean wave is very important for
naval navigation, port operations, offshore or nearshore activities
around the sea waters. Moreover prediction of wave condition is
necessary for design of harbour, coastal and offshore structures.
Variations in wave heights are caused by wind pressure on free
waves which make it random and uncertain, so that become
difficult to predict. In previous studies, wave prediction have been
carried out by using semi-empirical methods and conventional
methods that require high resolution simulations and high
computation. In this paper, we propose a method for prediction
wave height from wind data by using a variant of Artificial Neural
Network (ANN) with single pass associative memory-forward, so
called General Regression Neural Network (GRNN). To obtain
a set of training data, we perform numerical wave simulation
by using SWAN (Simulating Wave Nearshore) model by using
wind data obtained from ECMWF ERA-5. As a study area, we
choose a rather shallow bathymetry and complex geometry, in
Jakarta Bay, Indonesia. Results of prediction by using GRNN
show a good agreement with wave data.
Keywords – General Regression Neural Network, Wave
Prediction, Wind Waves, SWAN model.