Dynamic multimodal transport planning is vital for enhancing flexibility and responsiveness in emergency logistics. We propose a dynamic planning approach that integrates drones into the multimodal system with trains, trucks, and aircraft, introducing dual-role drones that can be transported as cargo and later operate as carriers. A Mixed Integer Programming (MIP) model, optimized via a rolling horizon approach, supports real-time route planning. Given the problem’s complexity, we develop an Adaptive Large Neighborhood Search (ALNS) algorithm with problem-specific operators. The model accounts for mode coordination, routing constraints, and cargo heterogeneity. It dynamically replans routes under disruptions such as road damage, considering mode availability and delivery requirements. Numerical experiments are conducted based on a real disaster scenario. A comparison with an exact method shows improved computational efficiency and solution quality. Further comparisons with a drone-free and static approach highlight gains in service rate and disruption resilience. We also examine the impact of cargo heterogeneity. These results, across instances from 5 to 400 orders, provide practical insights for optimizing drone deployment, transport mode selection, and cargo management in disaster response.