Dr Florence Rabier first became inspired by the meteorological sector when she was a young girl living by the sea in southwest France. “There were a lot of storms rolling in off the sea and the weather was very variable,” she says. “We always needed to be prepared for changeable weather conditions. That is where my interest in the weather and forecasting began.”
Dr Rabier has been director general of ECMWF since January 2016, after two years leading the center’s Forecast Department. Her career so far has taken her back and forth between Météo-France and ECMWF. “I studied mathematics, physics and meteorology before joining Météo-France in the 1980s. Since then I have switched roles between the French met office and ECMWF, working on wave modeling, data assimilation, numerical weather prediction, forecast verification and model diagnostics,” she explains.
According to Dr Rabier, her career highlight is her role in the development of a data assimilation method, called 4D-Var, in 1997, which was a first worldwide. This contributed to the optimal use of satellite observations in weather forecasting and led to substantial improvements in ECMWF’s forecasts. “This is what I worked on for my PhD and 25 years later it is still used today. Other meteorological centers are also using it as it really was a step change in the quality of the forecast.”
In her current role as director general of ECMWF, Dr Rabier and her team are working on blending the use of machine-learning technology with physics-based modeling. “I hope that within the next few years we will have managed to incorporate machine learning at the right level so that we can really make the best of both traditional and new techniques to deliver high-quality services to our member states,” she says.
She also highlights how “new space” – the emergence of the private space industry – and its contribution to Earth observation via private-sector satellites is set to disrupt the system. “We have to embrace this,” she says. “At ECMWF we use all sorts of satellite data, including operational and research data, blended to create forecasts. We need to ensure that data coming from new space systems is accurate. This will mean we need to be much more flexible in the way we use data, which is a manual process now, but maybe machine learning could help us to characterize bias and do quality control in a cleverer way in the future.”
In terms of gender equality, Dr Rabier notes that ECMWF is gradually becoming more balanced, with more members of the leadership team now female. “We have a gender, diversity and inclusion plan and we do a lot of training to ensure that there is no unconscious bias during recruitment or operations in the business,” she explains.