應用手機信令軌跡資料推估通勤廊道之時空地震災害風險

Abstract

台灣由於地理位置的特殊性，地震、颱風等自然災害頻傳。統計自民國47年至108年為止，平均每年發生3.6次颱風，地震為0.5次。儘管地震發生次數較少，但平均每次所造成的社會經濟損失卻最為慘重。身處如此的環境之中，更是突顯了災害風險評估的重要性。在評估災害風險時，往往忽略了風險的時間動態特性，無法在更細緻的時空尺度上提供災害防救決策。而手機信令資料能夠以較低的成本取得真實的人口動態暴露數據。本研究藉由此數據在風險評估中加入人口移動的時空動態特性，切分不同的時間區段，藉此挖掘並觀察通勤廊道的地震災害風險時空模式，改善以往假定靜態災害風險的不足。 研究成果顯示: 廊道暴露度與風險值皆由06:00急遽上升，12:00稍趨緩，直到接近17:00時再度上升並達到最高值，隨後逐漸下降至隔日凌晨。空間分布則是以中正路、中山路、國道10號為主要風險高峰廊道。廊道的風險時空分布以新興熱點分析後得知，凌晨至通勤尖峰與中午至通勤尖峰這兩個時段呈現較高風險熱點強度，但時間趨勢上前者熱點較晚出現，後者熱點則是逐步增強。災害風險的時空模式探勘結果，能夠在減災階段上提升風險知覺；在整備階段，能夠協助兵棋推演腳本的擬定、並且以更細緻的時空尺度規劃設備物資的調度、以及交通的規劃，增強地方的災害應對能力。Due to the special geographical location of Taiwan, hazards such as earthquakes and typhoons are frequent. According to statistics, from 1948 to 2019, typhoons occur 3.6 times per year, and earthquakes occur 0.5 time per year. Although there are fewer earthquakes than typhoons, the average social and economic losses caused by earthquakes each time are more serious. Such an environment background highlights the importance of disaster risk assessment. When assessing disaster risk, the space-time dynamic characteristics of risk are often ignored, and it is impossible to provide disaster prevention decisions on a more accurate space-time scale. In order to integrate space-time characteristics of population movement into risk assessment, Call Detail Record(CDR) data was used because it can provide more samples of real population dynamics at a lower cost. In this study, earthquake risks of the traffic in different time periods corridor was assessed to observe the space-time pattern, so that the lack of time dynamics in the past static disaster risk assessment can be improved. The results of the research show that the corridor exposure and risk value both rise sharply from 06:00, until it slows down at 12:00. It rose again and reached the highest value near 17:00, and then gradually decreases to the early morning of the next day. Zhongzheng Road, Zhongshan Road, and National Highway no.10 are the main corridors where the peak of risk occur. The space-time risk distribution of corridors is analyzed based on emerging hotspots. The two time windows, “03:00 - 08:00” and “11:00 - 17:00” show higher risk hotspot intensity, but the former hotspot appears later in the time trend, while the latter hotspot gradually Intensified. The data mining results of the disaster risk space-time patterns can improve risk awareness in Mitigation phase, which can be benefitial in the preparation of Simulation Exercises in Preparedness phase. In addition, it can also help plan the dispatch of equipment and Supplies on a more accurate space-time scale, as well as transportation planning to enhance local capabilities when facing disasters.