<p>Natural disasters pose significant challenges to traditional data-driven approaches for disaster management due to the inherent constraints of data availability, privacy, and computational resources. Federated learning (FL), a distributed Machine Learning (ML) paradigm, has emerged as a promising solution to address these challenges by enabling collaborative model training without the need to centralize sensitive data. This survey paper provides a comprehensive overview of recent advances in the application of FL for natural disaster management. The paper first introduces key concepts and algorithms in FL, highlighting its relevance and potential benefits in the context of natural disaster scenarios. Then it explores the use of FL for various aspects of disaster management, including disaster detection and classification, disaster response and mitigation, disaster preparedness and resilience, pandemic prediction and monitoring, and disaster management in social computing networks. For each application, the paper discusses the unique challenges, proposed solutions, and the impact of FL on improving the overall disaster management process. Furthermore, the survey examines the current limitations and future research opportunities in the field of FL for natural disaster management. These include addressing technical challenges in federated optimization, enhancing the robustness and security of FL systems, and expanding the integration of FL with other emerging technologies, such as edge computing and Internet of Things (IoT). This comprehensive review aims to serve as a valuable resource for researchers, practitioners, and policymakers interested in exploring the potential of FL to enhance the resilience and effectiveness of natural disaster management systems.</p>