The amount of data and computer code used in meteorological applications is growing very rapidly, both due to technological advances that provide new instruments capable of measuring meteorological quantities with greater precision, the proliferation of sensors in industry and among citizens, and due to the focus on modeling and observations shifting to processes at shorter scales.
This requires the adaptation of existing, and the development of new, software methods that enable the efficient handling of the increasing amount of data, to process them, explore them, visualise them and introduce them into existing and new meteorological applications, including Earth Observation, NWP, reanalysis and nowcasting methods.
The research area focuses on:
- Durable, maintainable, flexible software. This can be achieved through the promotion of high coding standards and the use of systematic multi-level testing, standardisation, interoperability, monitoring and quality control for new and existing code. It ensures the highest possible quality of data produced both for internal use and for external customers, and guarantees reproducibility of research.
- Optimisation of data management for large data sets. Many aspects are involved, including data storage, data exploration, data processing, data transfer and data quality. Solutions can be found for example through non-relational databases, distributed computing, design of software architecture based on GPU or parallelisation, out-of-memory computations and so on.
- Further development of existing and new methods in meteorological applications with strict time constraints, such as data assimilation for NWP, AI and machine learning for data processing and monitoring.
- Coordination of research on visualisation methods for meteorological quantities suitable for pre-exascale data volumes.