Best practice for transforming an inter-institutional research on climate services into an operational system referring Technology Readiness Level (TRL)

Climate change has led to a high demand for an appropriate system for monitoring and forecasting climate extremes, which could support disaster risk reduction and climate change mitigation. This has also led to global initiatives like WMO’s Early Warnings for All Initiative, which aims to provide early warning systems to support decision-making processes by the end of 2027. In this context, there is an urgent need to accelerate the transition from research, primarily conducted in academia, to a sustainable application for developing long-term operational environmental services. Here, we argue that this transition can be enabled and accelerated through Open-Source software tools and libraries, containerization, and the professionalization of research software engineering. They play a crucial role at all stages of technology development, from early research and prototyping to system deployment and scaling. The Technology Readiness Level (TRL) is an effective and standardized measure to assess the maturity of such developments. However, it is still unclear how the TRL can be applied in research-based tools and services and what preparatory steps need to be taken to ensure a certain pre-defined TRL.

In this talk, we will discuss best practices in developing a climate service system, using the example of the ongoing OUTLAST project (operational, multi-sectoral global drought hazard forecasting system), in which an operational drought forecasting system will be developed. OUTLAST is one of the first attempts to build a ready-to-be-transferred system using a cloud-ready concept to seamlessly transfer research-based developments into an operational system among governmental institutions. The present work will show how the currently developed software tools can support researchers in overcoming the current obstacles in technology development. We use OUTLAST to demonstrate how the automated pipeline is executed, from downloading the newly released climate data (ERA5 and SEAS5) provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) to triggering models and generating drought hazard indicators to be pushed to a webpage. In this approach, each processing step and its dependencies in the model chain are encapsulated in a "container" at the research institution before being transferred to run in an infrastructure at an external government institution. The containers are then orchestrated to allow upscaling of the system based on computational requirements and availability of hardware resources. We will then discuss the obstacles in building such a system and how the flexibility and portability can be improved.

Our work highlights the benefits using cutting-edge research software engineering practices for facilitating a seamless transition from research to operational systems and propose best practices, including the necessary preparatory steps. We further present our work as a blueprint for similar initiatives to ultimately support the development and deployment of advanced environmental service systems, which can provide the urgently needed information for decision-makers, stakeholders, and other potential end-users.