Are urban residents trapped by high temperatures? A travel-based heat segregation analysis

Rapid urbanization has profoundly reshaped urban thermal environments and mobility patterns. While socioeconomic segregation has been widely studied, it remains unclear whether daily mobility connects places with similar or dissimilar thermal conditions. This study proposes a travel-based heat segregation index and applies it to dockless bike-sharing trips and diurnal land surface temperature data in Shenzhen, China. Spatial analysis and explainable machine learning reveal widespread heat segregation, with high-temperature and high-segregation clusters concentrated mainly in peripheral districts. Inflow-based segregation varies more between daytime and nighttime, whereas outflow-based segregation differs more between weekdays and weekends. Mobility features are the strongest predictors: greater mobility diversity, more variable flows, and greater spatial extent of OD flows are generally associated with lower heat segregation. Urban environmental factors also show nonlinear, context-dependent effects. These findings support spatially targeted and temporally adaptive heat-mitigation strategies, particularly where high LST coincides with strong OD-based heat segregation.