The aim of the two-day workshop held in Bonn on 9 and 10 April 2025 was to approach the challenges of coupled numerical weather prediction (NWP) by bringing together different modelling communities working across the Earth system. In the past, many of the Earth system components used in NWP were developed in isolation and then brought together. With that in mind, discussion was a key element of the workshop, with talks at the start of each discussion session providing initial food for thought about challenges and issues. Discussion groups were formed with diverse ranges of scientific backgrounds to exchange ideas across disciplines. These groups covered various topics across the two days, revolving around the energy and water cycles, and coupling technologies and strategies for both physical and data-driven models. The poster session, held in conjunction with concurrent workshops, enabled lively and cross-thematic discourse in an informal atmosphere, and it was great to see such interest across all the workshops' attendees.
Main workshop themes
The workshop discussions focused on what had been achieved and what we saw as the main challenges going forward. Talks on the first day of the workshop were about the current state of the art in coupled systems and the coupling technologies that are being used or currently developed. ECMWF alongside national meteorological services, such as the UK Met Office, the German National Meteorological Service (DWD) and Environment and Climate Change Canada (ECCC), shared details of their operational systems and the progress that has been made so far in coupled NWP. The day continued with presentations discussing the methodology and tools for coupling and an example of the impact of ocean surface observations to help develop prediction systems. Finally, we considered the role of coupling within machine-learned forecasting systems, new developments in representing Earth system components, and what we could learn from this rapidly developing field. The second day had presentations looking at energy and water cycles. Talks concerned the model components of ocean surface waves and land, which have historically been tightly coupled to the atmosphere in ECMWF's Integrated Forecasting System (IFS), as well as the ocean and sea ice. This work showcased what processes could be important to include in our systems in the future and elements that could be improved. The workshop concluded with discussion about next steps and possible avenues for future collaboration.
Lively discussion
A common theme we kept returning to was conservation between coupled components. It was also emphasised that, when we couple different elements, we must ensure consistency across our systems in terms of how we parametrize physical aspects or make assumptions about how the components interact. Conservation properties of the system, which were perhaps considered less important in the past within the framework of NWP, are now becoming more important as we run our models at higher resolutions.
The discussion also recognised the challenge of modelling a complex system and the fact that we still need fundamental research in many areas. We will not only have to understand the processes we are trying to capture that govern exchanges between the interface of the atmosphere and the underlying surface, but also to represent missing processes. This also highlighted the need for pragmatic choices in the ongoing development of modelling systems. It was recognised that increasing the complexity within our system does not necessarily lead immediately to better scores as we have compensating biases in our system. We noted the time and domain-specific expertise required to retune parametrization schemes within the coupled system when we bring in new developments. We also need to have appropriate observation-based datasets which describe the processes we are trying to capture. We saw value in the community coming together for intercomparison studies focused on observational campaigns that target surface processes. Beyond field campaigns, the discussion also highlighted the need for coupled assimilation systems to initialise forecasts.
Feedback from participants was that they found the interdisciplinary nature of the meeting useful and that it was important to openly discuss the various issues that were highlighted. There is still a lot of work and interesting science to be done to improve our coupled systems.
More details of the talks and discussion can be found on the workshop page on the ECMWF website: https://events.ecmwf.int/event/427/.