<p>With the increasing frequency of natural disasters worldwide, urban safety is under severe threat, making scientific and efficient emergency decision-making critical to maintaining stable urban operations. Urban natural disasters often exhibit complex, cascading effects and dynamic evolution, posing significant challenges for timely and effective emergency responses. To address these issues, this study presents a dynamic emergency decision-making method that integrates complex network analysis, case-based reasoning (CBR), and hierarchical task network (HTN) planning. The disaster development process is segmented into multiple key decision points, each associated with a corresponding cascade scenario network. First, a temporal scenario risk assessment model is developed by incorporating scenario dynamics, network topology metrics, and escalation probabilities to identify priority scenarios requiring immediate action. Second, for these critical scenarios, the method innovatively combines CBR and HTN planning: historical response strategies are matched based on scenario similarity, tasks are decomposed structurally through HTN, and task priorities and execution paths are dynamically updated according to changes in the scenario network and feedback from response outcomes. Finally, the methodology is validated using the “7.20” catastrophic rainstorm in Zhengzhou as a case study. Results demonstrate that the proposed method effectively addresses cascade scenarios and their dynamic evolution in urban natural disaster events. It further demonstrates enhanced flexibility and comprehensiveness in identifying critical scenarios and developing adaptive phased strategies.</p>