GLOWPP is a project aimed at optimizing the global meteorological model GLOBO for use in HPC, cloud, and hybrid environments. GLOWPP seeks to provide a reliable and scalable tool for large-scale atmospheric simulations, particularly emphasizing I/O performance, parallelization, and code maintainability.

Reconsidering the original scope, the research team, alongside a deep discussion with the spoke representatives, agreed to refocus the project effort on deeply analyzing the GLOBO source code with the aim of code refractory and restructure, overall improvements targeting scalability, and the implementation of new features in terms of HPC architectural enhancements.
Summarizing: although we refer to the project by the official name GLOWPP, our effort is to improve the GLOBO source code to reach a production-grade prototype, as stated in the project submission. Since the mid-term review documents, GLOWPP has to be intended as “GLOBO with imProvements”.

As a direct consequence of the project refocusing, this deliverable is about designing the actions that must be implemented to achieve the overall improvements that match the spoke representative’s needs.

The GLOWPP project stems from the need to bring the GLOBO model to a “production” level, capable of efficiently handling global meteorological simulations under various operating conditions. The evolution of supercomputers and cloud platforms offers new opportunities for distributed computing but also requires appropriate code restructuring and a well-defined parallelization strategy. Moreover, the vast amount of data to be processed makes optimized Input/Output (I/O) management essential.

The topics to be addressed include:

• Describing the architecture and design choices adopted during the GLOWPP design phase.
• Defining the system’s key components, highlighting how they interact to ensure efficiency and flexibility.
• Providing a reference framework for the subsequent implementation and optimization phases, outlining the guidelines for code development.