本研究探討2008年至2018年間阿里山公路34K到97K崩塌事件相關資料,應用時間序列分析將雨量與崩塌次數及體積(土方量)依時間序列來探討。分析可知降雨量,崩塌次數及體積都具有季節性趨勢,且每年第一季及第四季的土方崩塌體積低於四季平均,第二季與第三季則高於四季平均,降雨量與土方崩塌體積成正相關。雖然時間序列分析顯示崩塌次數與體積具有季節性的波動,但消除季節因子後的崩塌體積時間序列不具有長期趨勢,無法有效以時間序列預測方法來估計崩塌土方量,因此在估計崩塌體積應該要考慮季節因子。每季的崩塌體積可依時間序列長期趨勢的回歸模式預測崩塌體積,針對防救災資源作更有效的分配。
This study collected landslide data along the Alishan Expressway between kilometers 34 and 97 from 2008 to 2018. A time series analysis was conducted to monitor the landslide volume in the study area. We report that the annual precipitation, number of landslides, and landslide volume varied seasonally. The landslide volume in Spring and Autumn was lower than the four-season average; those of the Summer and Winter were higher than the four-season average. Precipitation was positively correlated with landslide volume and not directly related to time. After the elimination of seasonal factors, the time series did not exhibit a long-term trend, and yearly cumulative rainfall was a positively correlated with landslide volume. Seasonal factors must be considered in landslide volume estimation to improve disaster mitigation.