孤立内波反射率影响:二元罗吉斯与顺序罗吉斯回归模型比较

Internal Solitary Wave Propagation:A Comparison of Binary Logistic Regression Model and Cumulative Logistic Regression Model

旨在运用二元罗吉斯回归模型(binary logistic regression model)及顺序罗吉斯回归模型(cumulative logistic regression model)分析孤立内波(internal solitary wave)于海脊地形的振幅反射率,并且藉由精确的回归模式,进而预测影响孤立内波传递的物理因子。根据顺序罗吉斯回归模型,以强、中、弱3种振幅反射率的分类形式,检验障碍物高度对振幅反射率的显著特性,具显著解释能力(P值皆小于0.001)。研究成果显示多元顺序罗吉斯回归模型较二元罗吉斯回归模型适合分析该实验数据之回归关系。

Experiments on internal solitary wave （ISW） reflecting from submarine ridge are performed in the laboratory. The appropriate logistic regression models are employed to analyze amplitude-based reflection rate, defined as ratio of reflected wave amplitude and incident wave amplitude. The regression models can predict physical factors, which play important roles on ISW propagation. The interaction was roughly classified as three groups by use of the amplitude-based reflection rate. Three classifications with weak, moderate and strong interaction are appropriate when we test the reflection rate, affected by the obstacle height. The results found that the goodness of fit on a cumulative logistic regression model is better than a binary logistic regression model.