Ocean Wave Forecasting in the Gulf of Thailand during Typhoon Linda 1997 : WAM and Neural Network Approaches

Wattana Kanbua^a*, Seree Supharatid^b, and I-Ming Tang^c

ABSTRACT: This paper presents an investigation of wave field during the attack of typhoon Linda in 1997 in the Gulf of Thailand. Two modeling approaches are studied: The hard computing approach by the WAM cycle 4 model was used firstly to simulate wave heights and periods distribution covering the domain 95?E to 105?E and 5?N to 15?N. Then, the soft computing approach by the GRNN model was developed to predict the wave characteristics for lead times of 3, 6, 9, 12, and 24 hrs. The input wind data were obtained from NOGAPS model archives with 1 degree resolution and are linearly interpolated to specify wind components at each grid points. It was found that the WAM model underestimated the wave height as much as 20%. The root mean square errors (RMSE) and the mean absolute deviations (MAD) are 0.18 – 0.26 m and 0.13 – 0.18 m, respectively. The GRNN showed better forecasting results than the WAM model (RMSE < 0.15 m and MAD <0.10 m). The maximum wave height simulated by the GRNN model during the typhoon Linda 1997 event was found to be 4.0 m while the observed data was 4.06 m. This indicates that for short-term prediction within 24 hrs, the data-driven model such as the GRNN should be viewed as a strong alternative in operational forecasting.

Download PDF

a Department of Mathematics, Faculty of Science, Mahidol University, Bangkok 10400, Thailand.
b Natural Disaster Research Center, College of Engineering, Rangsit University, Pathumhani 12000, Thailand.
c Department of Physics, Faculty of Science, Mahidol University, Bangkok 10400, Thailand.

* Corresponding author, E-mail: watt_kan@hotmail.com, watt_kan@yahoo.com

Received 26 Jul 2004, Accepted 2 May 2005