Urban flood risk is becoming a severe issue in most cities and has been a critical global concern of the twenty-first century. Flooding events can trigger significant failures in public transport networks (PTNs) within urban areas. With a growing population in urban areas, demand for transportation increases. Consequently, the dual forces of climate change and rapid urbanization have intensified the size and frequency of floods. Furthermore, the increasing threat of targeted attacks, natural hazards, and compound failures to PTNs has called for an urgent need to improve the resilience of critical lifeline systems. Here, we propose a new synthetic criterion for identifying the critical nodes of PTNs. Additionally, a hypothesis-driven framework for resilience analysis of PTNs against deliberate and compound failures is developed, which we demonstrate on Mass Transit Railway (MTR) in Hong Kong to obtain a quantitative understanding of resilience. Besides, the quantitative comparison of the viability of multiple recovery strategies suggests that the proposed method ($EWM-TOPSIS$) performs the best for the MTR network to recover after a disruption. Our findings can generalize across urban critical infrastructure systems as well as natural and human systems.