[2B10-13] Nuclear Emergency and Preparedness

[2B11] Uncertainty estimation of plume directions in atmospheric dispersion predictions: application of Bayesian machine learning

*Masanao Kadowaki¹, Haruyasu Nagai¹, Toshiya Yoshida², Hiroaki Terada¹, Katsunori Tsuduki¹ (1. JAEA, 2. Japan Wind Energy Consulting Inc.)

Keywords:Bayesian machine learning, atmospheric dispersion prediction, uncertainty estimation, numerical simulation, WSPEEDI-DB

We have developed a method to quantitatively evaluate uncertainties in the dispersion direction of a radioactive plume in atmospheric dispersion predictions using an analysis model obtained by applying Bayesian machine learning to a database that has accumulated long-term prediction calculation results. The validity of this method is verified by the results of atmospheric dispersion calculations for hypothetical releases from the site of JAEA.