Uncovering Spatiotemporal Patterns of Travel Flows Under Extreme Weather Events by Tensor Decomposition (Short Paper)

Extreme weather events have caused dramatic damage to human society. Human mobility is one of the important aspects that are impacted significantly by extreme weather. Currently, focus on human mobility research during extreme weather is often limited to the transport infrastructure and emergency management perspectives, lacking a systematic understanding of the spatiotemporal patterns of human travel behavior. In this research, we examine the structural changes in human mobility under the severe rainstorm that occurred on July 20th, 2021 in Zhengzhou, Henan Province, China. Innovatively applying a tensor decomposition approach to analyzing spatiotemporal flows of human movements represented by the mobile phone big data, we extract the characteristic components of human travel behaviors from the spatial and temporal dimensions, which help discover and understand the latent spatiotemporal patterns hidden in human mobility data. This study provides a new methodological perspective and demonstrates that it can be useful for uncovering latent patterns of human mobility and identifying its structural changes during extreme weather events. This is of great importance to a better understanding of the behavioral side of human mobility and its response to external shocks and has significant implications for human-focused policies in urban risk mitigation and emergency response.