Scientist: uncertainty quantification for Destination Earth using statistical post-processing methods

You are here

ECMWF has an exciting opportunity for a Scientist to help shape and deliver DestinE developments in collaboration with partners throughout Europe, by making use of machine learning techniques and statistical methods. It is expected that machine learning and statistical methods will play an important role in DestinE, supporting efforts to increase computational and data handling efficiency, to augment the quality of the Digital Twins through fusion of data from different sources and to rigorously handle quality and confidence information.

The successful candidate will apply powerful data analytics techniques based on machine learning and statistical methods to support the uncertainty quantification for the Digital Twin for weather induced extremes. This Digital Twin will rely on high-resolution (km-scale) medium-range simulations produced with ECMWF Integrated Forecasting System (IFS) to drive much enhanced weather-induced extremes predictions. Uncertainty quantification is essential for assessing the confidence of these simulations. Ensemble predictions provide information on confidence and are key for decision making, because they quantify the probability of possible outcomes given the uncertainty in observations and models. Particularly in cases where (large) ensembles are not affordable, which is the case for the resolutions and data volumes envisaged in DestinE, machine learning and novel statistical methods offer new levels of complex data analysis to enhance the representation of uncertainty and complement ensemble methods. Before the data from the weather-induced extremes Digital Twin becomes available, the successful candidate would test the added value of such an approach by applying machine learning and statistical methods to blend low-resolution ensemble forecasts with a high-resolution forecast and estimate how this can support uncertainty quantification.

The candidate will join an existing group at ECMWF working on applying machine learning and statistics techniques to improve various aspects of the ECMWF operational workflows, and will make the link between these efforts and similar efforts in DestinE.

The successful candidate will also support the interactions with ECMWF’s DestinE partners ESA and EUMETSAT on the use of machine learning and statistics techniques, interact with external contractors as required and help with communications and outreach activities with dedicated partners of the consortia contributing to the DestinE implementation, stakeholders and the user communities. They will also contribute to regular progress reports to the European Commission.

Whilst the position will be based in Bonn, Germany, there will be strong collaboration with staff based at the ECMWF HQ in Reading, in the UK so it is anticipated that visits to the HQ in Reading will be required.