本研究以前人降雨誘發山崩率之研究為基礎，於台灣各個流域內以水利署所劃分之子集水區為選取研究區之基準，分析降雨誘發山崩發生率與雨量因子之間的關係，探討崩壞比是否隨著雨量值改變呈某一趨勢之變化，並求得誘發山崩之降雨門檻值。 各子集水區內之總雨量與崩壞比進行線性一次迴歸顯示，各研究區R2值結果良好，隨著雨量增加崩壞比亦增加，而有正相關性之物理意義。在進一步進行二次迴歸時，發現當有極端高降雨事件時，大部分子集水區之二次迴歸R2值結果優於一次迴歸者，顯示了二次迴歸較適用於解釋極端降雨事件。 迴歸曲線於橫軸截距定義為降雨門檻值。線性一次迴歸及二次迴歸結果發現，二次迴歸之降雨門檻值較一次迴歸低很多。這可能是由於雨量事件分布不平均及資料點不足之故，但此結果仍可用於降雨門檻值之訂定之參考，可推估門檻值位於一次迴歸及二次迴歸兩門檻值之區間。 ;Based on previous rainfall-induced landslide studies, this research analyzed the relationship between the occurrence rate of rainfall-induced landslides and a rainfall factor for several sub-catchments in Taiwan. To investigate how the landslide occurrence rate increases with rainfall depth, and whether there is a rainfall threshold for landslides. In each sub-catchment, first-order linear regression analysis indicates that evet total rainfall and the landslide ratio shows good relation with high value of R2. The landslide ratio increases linearly with rainfall after a certain threshold value. In the further quadratic regression, it shows that the performance is also good and the R2 value is high, when an extreme high rainfall event is involved. This means that the landslide occurrence rate has increased in an extreme rainfall event and a quadratic form would be more valid in describing it. The intercept of the regression line at the horizontal axis is defined as a rainfall threshold for landsliding. The threshold value is commonly larger after the linear regression than that after the quadratic regression. This is due to the concave nature of the quadratic regression line and lack of data near the intercept. If we have added more data near the intercept, the distance of two intercepts may become closer. This indicates that an idea threshold may locate between the threshold after linear regression and the threshold after quadratic regression.
