ECMWF Fellow explores post-processed ensemble forecasts

Sándor Baran giving a presentation at ECMWF

Prof. Sándor Baran in June 2023 during a presentation at ECMWF on post-processed ensemble forecasts.

Prof. Sándor Baran became an ECMWF Fellow in January 2021. Since then he has worked on post-processing a range of ECMWF ensemble forecasts to increase their skill.

His most recent area of interest is the statistical post-processing of visibility ensemble forecasts, which are important in aviation as well as in water and road transport.

Sándor is the Head of the Department of Applied Mathematics and Probability Theory at the University of Debrecen, Hungary.

Our Fellowship programme aims to foster and formalise links with individuals who are carrying out pioneering scientific and technical research in areas relevant to the strategic goals of ECMWF.

There are about ten Fellowships at a time. They run for three years, with the possibility of being renewed once for another three years.

A long-standing interest

Sándor’s interest in ensemble forecasts, which are sets of forecasts that present the range of future weather possibilities, reaches back to 2011. At the time, he started to collaborate with Prof. Tilmann Gneiting in Germany, a previous ECMWF Fellow.

In 2017, while visiting Tilmann, ECMWF scientists Martin Leutbecher and Zied Ben Bouallègue presented to him the issue of post-processing dual-resolution ensemble forecasts to reassess the balance between resolution and ensemble size.

“I became interested in the topic,” Sándor says. “Together with a PhD student, we started working on it and published a paper in 2019.”

He subsequently spent some time at ECMWF to work on the statistical post-processing of heat index ensemble forecasts. As part of a paper published in June 2020, he also considered a machine learning approach, which showed comparable performance for heat index warning levels.

Man drinking water on a hot day

Prior to his Fellowship, Sándor worked on the statistical post-processing of heat index ensemble forecasts. (Photo: MonikM/iStock/Getty Images Plus)

Sándor’s Fellowship

Since becoming a Fellow, Sándor has repeatedly visited ECMWF, and he has been given access to ECMWF forecasts and computing facilities to pursue his work on statistical post-processing.

During his Fellowship, the horizontal resolution of ECMWF’s ensemble forecasts (ENS) went up from 18 km to 9 km. This was preceded by much research to find the best possible upgrade of ENS in terms of resolution and the number of ensemble members.

In that context, Sándor finalised a joint paper on parametric post-processing of dual-resolution precipitation forecasts while at the Centre, with the help of ECMWF scientist Estíbaliz Gascón.

Other work he has carried out is part of a larger-scale project with the Hungarian Meteorological Service (OMSZ). The idea is to work on aspects of ensemble forecasts that have direct economic significance.

In November 2022, for example, Sándor gave a talk at ECMWF on the statistical post-processing of 100-metre wind speed and solar irradiance forecasts for energy production, showing the models he developed with his wife, Ágnes, for operational implementation at OMSZ. And in June 2023, he presented his work on the statistical post-processing of visibility ensemble forecasts.

Wind and solar energy installations

Part of Sándor’s work has concerned the statistical post-processing of 100-metre wind speed and solar irradiance forecasts for energy production. (Photo: hrui/iStock/Getty Images Plus)

He is also collaborating with a member of ECMWF’s Destination Earth (DestinE) team, Ivana Aleksovska, working on post-processing.

“The Fellowship has been an excellent opportunity to meet people working in similar areas,” Sándor says. “People at ECMWF have given me a lot of ideas and inspiration.”

“It is also extremely useful to have access to ECMWF data, because getting reliable, good data is one of the most difficult parts of my job. Finally, I use my visits to inspire my team in Hungary, including PhD students.”