碩士 / 國立中山大學 / 海洋環境及工程學系研究所 / 104 / In February 2008, a lot of fishes were killed by cold water at Penghu, which was caused by the exceptional cold water intrusion of China Coastal Current. The event was called Cold Disaster, which brought huge losses to the local aquaculture and serious impact on the local marine ecosystem. According to statistics, Cold Disaster happened every 30 years from 1937 to 2011, which did great damage to the local economy and ecosystem at Penghu. Therefore, to study the process of Cold Disaster will help us to more understand the whole process of oceanic dynamics of the Taiwan Strait in winter and make a good disaster prevention.
In this study, a correlation analysis is chosen to predict the sea temperature. Analysis data come from stations at Penghu and at the western coast of Taiwan. The strong correlation factors are selected by Penghu SST, and these factors are such as sea surface temperature, air temperature, wind speed, air pressure, etc. However, this study only considers winter data from December to February. Regression is done by the Fourier or quadratic regression analysis. Predicted values are fed by the stepwise regression to build the predictive equation of sea surface temperature at Penghu.
Next, the numerical model of ocean is used to simulate the phenomenon of exception cold water intrusion in 2008. And then temperature and velocities from model are analyzed in order to understand the process of Cold Disaster.
By using the process of cold disasters in 2008 and in 2011 to testify the numerical models, the result shows that the low temperature of sea surface at Penghu was mainly due to the northeast monsoon. However, in 2011, the wind was still strong but there was no low temperature of sea surface at Penghu. The main cause was due to the northeast monsoon that brought the Kuroshio into the Taiwan Strait in the northeast of Taiwan, so that the sea surface temperature at Penghu in 2011 was not so low as usual.
By combining the numerical model and regression model to SST forecast, the root mean square error is about 1-2 degrees. Thus, the numerical models and regression models can be combined to do SST forecast.
Identifer | oai:union.ndltd.org:TW/104NSYS5282005 |
Date | January 2015 |
Creators | Che-Yuan Liu, 劉哲源 |
Contributors | Jason Yu, 于嘉順 |
Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
Language | zh-TW |
Detected Language | English |
Type | 學位論文 ; thesis |
Format | 132 |
Page generated in 0.0128 seconds