In the Mind of the Modeller

AWI Director Antje Boetius (left) and Thomas Jung (right) handed over the Helmholtz Fellow Award certificate to Peter Bauer (middle). Photo Credit: Luisa Cristini

A conversation with ECMWF Deputy Director of Research Peter Bauer on supercomputing, his work to advance weather prediction and the future prospects of technology and science

Peter Bauer is Deputy Director of Research and Manager of the Scalability Programme at the European Centre for Medium-Range Weather Forecasts (ECMWF). He has been one of the pioneers shaping the Polar Prediction Project and the Year of Polar Prediction as a Steering Group member from Day 1. In 2018, he has received the Helmholtz International Fellow Award. On 5 February 2019, Antje Boetius, Director of the Alfred Wegener Institute (AWI), handed over the Helmholtz Fellow certificate to Peter Bauer for his frontier work in the field of high-resolution modelling, prediction and computing. During a one-week visit, he had the chance to meet with researchers and early career scientists at the German Helmholtz Centre for Polar and Marine Research in Bremerhaven. In this lecture entitled "Why future weather and climate prediction needs to pay serious attention to computing?", Peter Bauer shared insights on the current challenges of predictive capabilities limited by computing skills, in view of developing more realistic Earth-system models.

Receiving the Helmholtz Fellow Award is a great honour for Peter Bauer. "I am very grateful for that as it emphasises the need and opportunity for international collaboration. An international fellow is, by definition, based on collaboration between Helmholtz as a whole, with its different institutes and disciplines, and an international organisation like ECMWF."

Peter’s work brought significant advancements in many fields of Earth system research, such as satellite remote sensing in weather and climate prediction, the assimilation of observations in numerical models, climate monitoring, weather and climate model development and, more recently, the preparation of models and forecasting systems for extreme-scale computing. In the different roles he had since he first joined ECMWF in 2000, Peter Bauer has significantly supported the Year of Polar Prediction (YOPP), as well as large-scale projects such as the EU Horizon-2020 APPLICATE and Helmholtz Earth System Modelling (ESM) projects.

"With the apparent challenges that we are facing for global Earth system modelling, for enhancing predictive skill and technological challenges, tight science-technology collaboration is even more important", he says.

We have met Peter for an interview during his stay at AWI to ask him more about his work and research, the challenges that come with it, and the potential for future advancements in the field. The entire interview is available on video here.

What does it mean for you to be an International Helmholtz Fellow?

It will allow me to visit the different Helmholtz Centres, to engage in dialogues with people in polar sciences here at AWI but also in computing science at the Jülich Supercomputing Centre, about Earth-system science with colleagues in Potsdam, or on environmental science at the Centre for Environmental Research in Leipzig. I will have the opportunity to initiate dialogues and come up with a common strategy that will not only help internationally, but also within Helmholtz. With the challenges that we are facing for global Earth system modelling, it is fundamental that scientists get together and work shoulder-to-shoulder.

In your awarding lecture at AWI, you talked about the problems that numerical weather prediction has in terms of computing capacity. What do you think is the biggest challenge in representing physical processes in numerical models?

Ultimately, we want to create better and more realistic Earth-system models. This means that we account for all the physical processes that are relevant for predictive skill and this will need better computing in order to be more precise and realistic. We also need to understand better some of the processes, for example the exchange of heat and energy between atmosphere, sea ice and ocean. Even land surfaces, vegetation, or urban effects have an impact on the atmosphere. With this level of complexity, enhanced computing performance is a must.

Even if the computing challenges you explained in your lecture were to be overcome, what would be the biggest issue related to data management?

More complex models will inevitably produce more data. Even if  we could run a 1-km global simulation today, what would we do with the data? We cannot even plot it on our current system! Yet, our users would like to extract the full spectrum of the information contained in these data, at the same speed as today.

How do you think the work of ECMWF could help in solving these issues?

It is all part of research. ECMWF is not alone – the weather and climate prediction  community is large and very well-organised. ECMWF has been very effective in bringing things together and turning research into operations. This is because we centralise European excellence and resources in one place but also have a very focused mission on medium-range forecasting, that allows us to concentrate all our efforts on that goal.

What do you think about the involvement of the public in weather prediction? Do you see issues in the communication of this subject?

People are affected by weather every day, and the impact of weather and climate extremes is becoming more and more clear. Events like the violent storms or heatwaves over  Europe last summer affect people’s lives, and I think this is communicated quite well. Perhaps not quite as well communicated is the enormous investments that are still required to make the next step towards even more enhanced predictive skills. We are trying to do this through big international projects but there is still a lot of work to do.

Peter, you had a remarkable career since you completed your PhD in Meteorology at the University of Hamburg and spent about ten years as a researcher at the German Aerospace Center DLR. You have received many awards and fellowships overseas at the National Oceanic and Atmospheric Administration (NOAA) and NASA and joined ECMWF in the year 2000. What do you like most about your job?

It is very satisfying to work at an operational centre because you can see the immediate impact of your work every single day. This is what motivates most people at ECMWF, namely to work very hard and contribute to that success. In my current position, the satisfaction comes also from being able to define the future of where our research goes, how we interact with our member states, how we can cooperate with specific organisations like Helmholtz and then see how that turns into success.

What would you say is your “strongest suit”?

I am good at bringing people together, setting up a motivated team and doing something that nobody else has tried yet, like working on the interaction between science and technology, in this case atmospheric physics and computing.

…and what about your weakness?

Sometimes things advance too slowly for me and I become impatient, and that does not really help. If you are really ambitious, you want to have success fast and see the outcome of what you have been planning for years. Certain things take longer and others go faster, but you have to give things time and let researchers do their work.

What is, in your experience, the best strategy for a successful teamwork?

Communication, for sure. You have to accept that people are different and understand that this difference is what makes working in a team successful. You do not want ten or twenty times the same personality with the same technical skills, but rather the different types, the strengths and the weaknesses that people have to offer and that combination advances us further. For this to work, I think you need good transparency and communication.

What would you like to see realised in the coming five years?

I would like to see the ambitious science-technology programme that we are implementing now as being realised successfully. For example at ECMWF, we see now for the first time a real change of mind in the way we deal with the computing and data handling issues. We can see that we will be able to implement the scientific ambitions that we have for more realistic Earth system models, and I would like to see this brought to life.

In his awarding lecture at the AWI Nordsee Museum on 5 February 2019, Peter Bauer discussed the current limitations of computing skills to further improve weather and climate predictions. Photo credit: Luisa Cristini