探討強風是否為崩塌致災因子與建立崩塌機器學習模型;Investigating Strong Winds as a Risk Factor for Landslides and Establishing a Landslide Machine Learning Model

NCU Institutional Repository > 工學院 > 土木工程研究所 > 博碩士論文 > Item 987654321/94420

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/94420

题名:	探討強風是否為崩塌致災因子與建立崩塌機器學習模型;Investigating Strong Winds as a Risk Factor for Landslides and Establishing a Landslide Machine Learning Model
作者:	謝芮云;Hsieh, Jui-Yun
贡献者:	土木工程學系
关键词:	風雨效應;強風;崩塌;颱風;隨機森林;Wind and Rain Effect;Strong Wind;Landslide;Typhoon;Random Forest
日期:	2024-06-28
上传时间:	2024-10-09 14:42:09 (UTC+8)
出版者:	國立中央大學
摘要:	台灣因天氣型態與地質地形條件，導致常有坡地災害的發生，造成人員大量傷亡以及財產損失。傳統上，針對由颱風引發崩塌的研究主要探討的致災因子包含強降雨、地質和地形條件，而強降雨已被證明是最重要的因子之一。然而，颱風常伴隨著瞬間強風，並劇烈搖晃樹木導致土壤受到擾動，降低了坡地的穩定性。此外，有部分崩塌事件發生於有樹木覆蓋的邊坡上，且只受颱風帶來強風的影響，而非特別受到強降雨的影響。因此本研究採用資料驅動(data-driven)的方式，將雨和風同時納入探討，以另一個角度證明颱風帶來的強風是不可忽視的崩塌致災因子，特別是持續數小時的強風。我們透過三維直方圖和曼-惠特尼U檢驗，探討了結合強風和降雨對崩塌的影響。結果顯示，崩塌事件發生時的風雨指標皆顯著高於未發生崩塌事件時，且隨著強風持續時間增加，崩塌發生的機率亦會增加。另外，我們建立了機器學習隨機森林模型，將強降雨、強風、地質和地形條件等因素作為訓練特徵變數進行崩塌預測，結果表明將強風納入考量可以提高預測準確度，因此，在颱風侵襲之下，除了強降雨，強風亦是提高或引發坡地災害的致災因子之一，在考量由颱風誘發的崩塌地時，若忽略強風的影響可能導致嚴重的損失。 ;Landslides, which result in numerous casualties and significant property losses, are a major natural disaster in Taiwan. Traditional landslide studies focused on heavy rainfall, geological condition and topographical condition as trigger factors, often overlooking the impact of strong winds. However, typhoons often bring intense wind gusts that can severely sway trees, leading to the soil disturbance which decreasing the slope stability. Additionally, some landslide events occurred on broad-leaved forest along the slopes where were primarily affected by strong winds of the typhoon rather than its heavy rainfall. We examined the significance of the combined rain-wind influence on landslides by Three-dimensional (3D) Histogram and Mann-Whitney U test. The Mann-Whitney U test results reveal that wind and rain conditions are both significant differences at a significant level of P≤0.001 between typhoon-induced landslide events and non-landslide events. And 3D Histogram results demonstrate a positive skewed distribution similar to the rainfall direction in strong wind duration axis, which implies that the probability of landslides occurring increased with an increase in the duration of strong wind. Furthermore, we construct machine learning Random Forest models based on factors, such as heavy rain, strong wind, traditional geological conditions, and topographical factors. The model that includes strong wind factors shows better accuracy than the model that does not include strong wind factors. Therefore, apart from heavy rainfall, strong wind is also one of the important factors that may increase or trigger the risk of landslides. Ignoring the effect of strong wind when investigating typhoon-induced landslides might lead to severe damage.
显示于类别:	[土木工程研究所] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	36	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....