Accurate estimation of atmospheric release sources and their time profiles is essential for assessing the environmental impacts of incidents such as the emission of radionuclides from forest fires in Chernobyl in 2020 or the release of ruthenium in 2017, whose source has not been officially confirmed to this day.

Our approach builds on the linear SRS (source–receptor sensitivity) transport model, leading to an inverse problem formulation. Not only is the inversion highly ill-posed, but the SRS coefficients themselves may contain errors due to inaccurate configuration of the atmospheric transport simulations or due to uncertainties in the input data.

In this talk, I will introduce a correction of the SRS coefficients using a Gaussian process prior, which incorporates assumptions of smoothness and sparsity in the correction and furthermore enables the prediction of corrections at new locations, providing an additional tool for the analysis of the results.

The speaker: Antonie Brožová

The seminar will be held on Monday, December 8, 2025 at 14:00 in the AS-meeting room 474.