Abstract:

Despite global efforts to reduce earthquake disaster risk, earthquakes remain one of the most destructive natural disasters in the world. Since precise earthquake prediction remains beyond human capability, conducting accurate earthquake casualty risk assessment has become one of the most feasible and effective strategies to reduce human losses. Kashgar and Artux are located in the southwestern part of the Xinjiang Uygur Autonomous Region, on the western edge of the Tarim Basin. These are two important cities in northwest China, situated in the Pamir tectonic knot region of the India-Eurasia continental collision zone, one of the world’s most seismically active regions due to intracontinental subduction. Historically, this region has experienced multiple major earthquakes, with the most representative example being the 1902 Artux earthquake, which had a magnitude of M_S8¼. This earthquake caused severe casualties and highlighted the region’s high susceptibility to seismic hazards. In this study, we used the historical seismic intensity scenario of the 1902 Artux earthquake as a deterministic earthquake scenario to understand the potential impact of a similar event under contemporary conditions. First, using time-series mobile signaling data and machine learning methods, we extracted the function types of buildings in the study area and developed a thematic dataset of building function types. Utilizing the extracted building function types, high-resolution population heatmap data, and a mapping method between grids and individual buildings, we allocated the population within the grids to specific buildings. Subsequently, by considering the relationships among building function types, temporal characteristics, local daily activity patterns, and indoor occupancy rates, we determined the spatial distribution of the indoor population in the study area. Understanding the potential damage levels of buildings under specific seismic scenarios is crucial for predicting casualty risks and formulating effective emergency response strategies. Therefore, a quantitative analysis of building damage levels under the deterministic earthquake scenario was conducted to clarify damage distribution for buildings in specific intensity scenarios. Subsequently, based on building vulnerability analysis methods, we estimated the daytime and nighttime fatality risks under the recurrence of the Artux earthquake at a 30″grid scale. The results indicate that if the 1902 M_S8¼ Artux earthquake were to recur, areas with high risk of fatalities would mainly be concentrated in the densely populated urban centers of Kashgar and Artux, where buildings are densely packed, populations are concentrated, and some structures lack adequate seismic resistance. The risk of casualties is higher at night than during the daytime. To evaluate the effectiveness of population heatmap data in earthquake casualty assessment, this study compared the evaluation method based on population heatmap data with the method using the Seventh National Census data. The results demonstrate that traditional census data can only provide a rough estimate of casualties based on administrative divisions. However, its reliance on administrative boundaries means it lacks the spatial detail necessary to accurately depict casualty distribution and presents scale inconsistencies that hinder effective risk comparisons across different areas. In contrast, the assessment method based on population heatmap data enables spatial visualization of fatality risks through optimized scaling and establishes a uniform spatial comparison benchmark, thereby providing strong support for the precise allocation of emergency rescue resources. By introducing high-resolution population heatmap data and a thematic dataset of building function types, this study has enhanced the accuracy and practicality of earthquake casualty risk assessment. The results not only reveal the potential for high human casualties if a historical earthquake were to recur today but also demonstrate the distribution characteristics and temporal differences in high-risk zones, emphasizing the importance of considering temporal factors in disaster mitigation strategies. These insights provide robust scientific support for developing precise disaster mitigation and preparedness strategies in earthquake-prone urban areas.

Key words: Artux M_S8 1/4 Earthquake, casualty risk, building function types, population heatmap data

摘要：

喀什市和阿图什市位于新疆维吾尔自治区西南部、 塔里木盆地西缘, 是中国西北地区2座重要城市。该区域地处印度-欧亚大陆碰撞带的帕米尔构造结, 是全球陆内俯冲作用最强烈、 地震活动最频繁的地区之一。历史记录显示, 1902年该区域曾发生M_S8¼大地震, 造成了严重的人员伤亡。文中以该历史地震的烈度情景作为确定性地震情景, 基于时序手机信令数据和机器学习方法提取研究区建筑物功能类型, 并结合高分辨率人口热力数据及网格与单体建筑之间的映射关系, 基于面积加权将网格内人口分配至建筑物中。随后, 基于建筑功能类型、 时间特征及其与人员在室率之间的关系, 确定研究区室内人口的空间分布。通过对确定性地震情景下建筑物震害程度的量化分析, 明确研究区在特定地震情景下的建筑物损伤情况。采用建筑易损性分析方法, 在30″网格尺度上分别估算了阿图什地震重现时喀什市和阿图什市白天与夜间的人员死亡风险。此外, 为分析人口热力数据在地震人员伤亡风险评估中的作用, 文中对基于人口热力数据方法与基于人口普查数据方法所得结果进行了对比分析。研究表明, 在阿图什M_S8¼历史地震情景下, 研究区人员死亡高风险区主要集中在喀什市和阿图什市城区等人口经济密集区, 夜间的人员死亡风险高于白天。

关键词: 阿图什M_S8 1/4地震, 人员伤亡风险, 建筑物功能类型, 人口热力数据