This master’s thesis focuses on developing a modern, cloud-based architecture to make environmental forecasting services more robust, scalable, and continuously available. As climate change increases the frequency and severity of floods, wildfires, and marine hazards, forecasting systems must handle growing volumes of environmental data while ensuring uninterrupted operation across multiple regions. Conducted externally at KISTERS AG in Aachen, the research investigates how data management, processing, and system design can be combined into a single, highly resilient forecasting framework.

The thesis explores how improved data storage strategies, efficient processing of large environmental datasets, and geo-redundant cloud infrastructures can work together to ensure reliability and performance even under failure scenarios. It aims to design a system capable of maintaining data consistency, reducing downtime, and optimizing operational costs while supporting 24/7 forecasting availability. The outcome will be a reference architecture and practical design guidelines for building next-generation environmental forecasting services that are efficient, fault-tolerant, and sustainable — contributing to more dependable decision support in the face of increasing environmental risks.