Large urban areas are vital for the economic and social development of modern cities. The mobility in these areas strongly depends on a seamless and available multi-modal transport infrastructure network. Since many of these networks operates at capacity limits, any disruptions of a single edge can have negative effects on the mobility. Hence, modeling and quantifying the resilience of large-scale urban transport networks will allow cities guaranteeing higher quality of service, even in the presence of unpredictable disruptive events. This paper aims to develop a new methodology linking together topological data with traffic dynamics for resilience analysis. To do so, we adopt a dynamic weighted graph approach where weights are generated according to prevailing traffic conditions. Weight-extended centrality measures are leveraged in order to capture the evolution of vulnerabilities. The paper also aims at studying the impact of area-wide disruptions, such as snow storms or floods, insufficiently explored in the literature. This is based on a stress testing approach. Our graph-based approach is also supported by other analyses performed in the paper by leveraging more traditional traffic indicators. Results demonstrate the importance of both the spatial and temporal dimension in the assessment of transport network criticality, thus stressing the importance of jointly considering the topological properties and traffic dynamics in the study of transport resilience. Consequences of area-wide disruptions are also proved to be different with respect to single-link failures, therefore deserving a deeper characterization in the study of transport resilience