Urban vulnerability to emergencies has become a relevant issue as cities get bigger and the negative impacts of climatic changes become more prominent. In recent years, smart city systems devoted to detecting, alerting, and mitigating emergency situations have gained momentum, with different complexities and expected outcomes. Although such systems have become more common, their proper configuration and adoption should rely on a more comprehensive perception of urban vulnerabilities to critical events. This article defines a data-driven approach to numerically evaluate the vulnerability of each region of a city, taking real-world data from open databases as a reference for a proposed clustering-based assessment algorithm. Additionally, the temporal dynamics within a city are modelled through the definition of time frames, each one comprising an associated risk assessment factor based on the expected flow of inhabitants over time. This clustering analysis categorises urban areas with similar vulnerability profiles by modelling the temporal dynamics of urban infrastructure, capturing their fluctuating nature and impact on vulnerability. By leveraging temporal urban perceptions, this approach may contribute to more effective emergency management in urban areas since regions with higher population density may be assumed as more vulnerable to emergencies, potentially supporting the optimisation of smart cities and general urban planning. Experimental results for the Portuguese cities of Porto and Lisbon demonstrate the practical applicability of the proposed approach by accurately identifying regions with higher temporal vulnerability to urban emergencies.