Abstract:
Predicting sea water level is important as an increasing of sea level may
lead to inundation, flood and coastal erosion. This study is the application
of chaotic approach for Malaysian west coast sea level. Studied data is a
time series of observed sea level hourly at benchmark station located in
the district of Georgetown in Penang and Kukup, Johor respectively. This
research has two objectives, which is to detect the presence of chaos in
data and to predict it. To achieve objective number 1, multidimensional
phase space is reconstructed using the parameters of the delay time
and , in Penang with embedding dimension d = 6 and d = 4.
For Kukup, and with embedding dimension d = 8 and d = 7
were used respectively. Both time delay and embedding dimension
values was derived from the Average Mutual Information (AMI) and Cao
method. The results from the phase space diagram and parameter plot of
Cao method reveals that the data are chaotic. To achieve objectives
number 2, 1 hour ahead forecasting for an hourly sea level time series is
carried out using the Local Linear Approximation (LLA) method and
compared with the Auto Regressive Linear (ARL) method for testing the
prediction performance. However, this method used MATLAB to forecast
the data. Correlation coefficient (cc) value between the actual and
foretasted data is only 0.88. This shows the reliability of the local
approximation method to forecast the time series of sea level and it’s a
positive sign that this chaotic approach is applicable to the time series of
sea level in Malaysian West Coast.
