理解极端降雨事件中人类移动的韧性对于提升灾害应对能力与城市韧性至关重要。然而，现有研究大多忽视了韧性模式在不同尺度上的复杂性，未能揭示空间异常现象及其潜在成因。为弥补这一空白，本文提出一种基于大规模个体轨迹数据的框架，以解析多尺度下人类移动的韧性模式。通过动态网络模型量化人类移动流量，并利用韧性曲线分析都市圈与区域尺度下的韧性特征。本研究聚焦于台风玛娃引发的极端降雨事件，覆盖日本的大阪与名古屋地区。研究发现，尽管人类活动显著减少，但移动网络结构保持相对稳定。基于流入与流出量的象限分布分析，人类移动中异常与正常韧性模式的比例约为3:2，且该比例在两个尺度上均保持一致。有趣的是，异常韧性模式与建成环境的局部地理特征密切相关，呈现出收入、性别与年龄层面的差异。这些发现对政策制定者优化灾后恢复策略及指导未来城市基础设施建设以增强韧性具有重要价值。

Understanding human mobility’s resilience during extreme rainfall is paramount for enhancing disaster response and urban resilience. Most studies, however, have overlooked the complexity of resilience patterns across scales, missing out on the varied spatial anomalies and their underlying causes. To bridge this gap, we propose a framework using massive individual trajectory data to dissect resilience patterns of human mobility across scales. By leveraging a dynamic network model, we quantify human mobility flows and employ resilience curves to determine resilience patterns at urban-agglomeration and regional scales. Our study, centering on the extreme rainfall from Typhoon Mawar, covers Osaka and Nagoya in Japan. The findings reveal a marked reduction in human movement, although the structure of mobility networks remains relatively unchanged. Based on the quadrant distribution of inflows and outflows, we reveal that the ratio of abnormal to normal resilience patterns in human mobility stands at approximately 3:2, a consistency maintained across both scales. Interestingly, abnormal resilience patterns are intricately linked to local geographical settings of the built environment, revealing disparities based on income, gender, and age. These insights are invaluable for policymakers to improve postdisaster recovery efforts and guide future urban infrastructure development toward greater resilience.