|
Swell and wind wave, as well as the southwest air current, induced by typhoons moving around Taiwan in summer influence severely the near shore structures and marine environment. Therefore, it is important to discuss the mechanism of typhoon wave’s deforming process while typhoon is moving on surroundings and to develop the typhoon wave height forecasting model to meet the requirement of disaster prevention during the typhoon season.
A forecasting model of typhoon wave height developed by the GMDH (Group Method of Data Handling) structure of Self-Organization Algorithm with four parameters of wave height (H), wind speed (V), distance (L) and azimuth (θ) between the target location and typhoon center is proposed in this paper. Data of these 4 parameters observed at Hsin Chu data buoy and obtained from CWB were be used to construct the prior 6hrs typhoon wave height forecasting model first to provide the necessity of warning facility with fifteen typhoon events data during 2006 to 2009. Then, a recursive GMDH model could be organized by using the update data to match the time variant properties in forecasting steps to improve the predict accuracy. In addition, a prior 1hr typhoon wave height forecasting model could be set up by cooperate prior 6hrs predict results with the log-logarithm correlation between progressive every 6hr’s and every hour’s measured wave height data to enhance the model advanced applications.
The modeling approach shows that the Stepwise regression GMDH (SGMDH) algorithm is better than GMDH on typhoon wave height forecasting in typhoon events modeling, and the Jangmi typhoon event modeling with the RMSE 22.67cm and CC 97.16％ result the best simulation. Hence, taking this optimum model to forecast wave height of other 14 typhoon events and results show that the average RMSE, CC and error scale ratio (RMSE/H(1/3)) are 20.37cm, 90.55% and 17.32% respectively. Due to the necessity of disaster prevention, a prior 6hrs wave height forecasting model was build up by taking the data of 11 typhoon events and the predicted results reveal that the average RMSE, CC and error scale ratio are 41.34 cm, 80.09% and 26.41% respectively. A prior 1hr wave height forecasting model base on the log-log relation as mentioned above was then applied to calculate the progressive every 1hr’s wave height by using the predicted values of every 6hrs step. Simulation tests show that the average RMSE, CC and error scale ratio are 29.53 cm, 90.76% and 18.42% respectively. In consequence of typhoon wave height forecasting with the data of 8~14 typhoon events, both prior 6hrs and 1hr typhoon wave height forecasting appear the trend of forecasting results in agreement with measuring materials of the data buoy and result reasonable predicting accuracy in this research with the average RMSE, CC and error scale ratio in between 20.37cm~41.34cm, 80.09%~90.76% and 17.32%~26.41% respectively. So, the GMDH forecasting model possesses the practical usage of typhoon wave height forecasting at the specific surroundings.
|