A study on time series analysis and forecasting of wave heights near the coastal industrial park

• Main Authors: Yen-PinLin, 林演斌
• Other Authors: Mi-Chia Ma
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/03587836249339877785

碩士 === 國立成功大學 === 高階管理碩士在職專班(EMBA) === 100 === Trends and forecasts of a variety of parameters in the ocean are very important to marine management. This study gathers data of wave heights of Suao buoy which is located in the coastal area of Ilan Long De Lize Industrial Park from 2000 to 2011. This study converts data into information for marine management by utilizing the methodology of time series analysis and forecasting of statistics. The ADF test is utilized to analyze stationary and trends. ARMA(1,1), ARMA(2,2), ARIMA(2,1,1), ARIMA(2,1,2) and ARIMA(1,2,1) models are established and tested. Series of wave heights are transformed by Napierian logarithm before fitting ARMA(2,2) and ARIMA(2,1,2) models. Data are sampled by different ways and used to validate hypotheses of this study. Based on results of the ADF test, it shows that: First, both series of maxima of month and of typhoon are stationary. Secondly, series of annual maxima and averages of significant wave heights are non-stationary. There is no deterministic trend for the non-stationary series. The non-stationary is due to the stochastic trend. In addition, according to results of different ways of sampling, whether the series is stationary or not is also infected by the sampling ways. Based on analytical results of forecasting models, it shows that: First, the series of daily maxima of significant wave heights is stationary. Transformed series of annual maxima and averages are stationary too. Secondly, Both ARMA(1,1) and ARMA(2,2) models are not so suitable for untransformed and transformed series of daily maxima of significant wave heights of 2009. ARIMA(2,1,1) and ARIMA(2,1,2) model are suitable for untransformed and transformed series of annual averages of wave heights of 2000-2011. ARIMA(2,1,1) model is better than ARIMA(2,1,2) model. ARIMA(1,2,1) and ARMA(2,2) model are suitable for untransformed and transformed series of annual maxima of wave heights of 2000-2011. ARMA(2,2) model is better than ARIMA(1,2,1) model. In addition, series sampled by different ways is fitted by different model. Sampled wave heights data do not fit ARMA(1,1) well which is utilized by Agrawal ＆ Deo(2002). Transformed annual maximum wave heights fit ARMA(2,2) well which is used by Stefanakos(1999). Trend and forecast of waves is helpful to oil and electricity industries directly. These information may minify loss. It’s necessary for managers of industries located in coastal industrial parks to be aware of risk from natural hazards and to prevent them as more as possible. Managers have to minify loss from these hazards in order to increase profit.