Journal of Environmental Radioactivity 220-221 (2020) 106302 Available online 16 May 2020 0265-931X/© 2020 Elsevier Ltd. All rights reserved. Adaptation of the RODOS system for analysis of possible sources of Ru-106 detected in 2017 Ivan V. Kovalets a, b, * , Oleksandr Romanenko c , Roman Synkevych a a Institute of Mathematical Machines & Systems Problems, NAS of Ukraine, Prosp. Glushkova, 42, 03187, Kyiv, Ukraine b Ukrainian Center of Environmental & Water Projects, Prosp. Glushkova, 42, 03187, Kyiv, Ukraine c Rivne Nuclear Power Plant, National Nuclear Energy Generating Company “Energoatom”, 34400, Rivne Oblast, Ukraine A R T I C L E INFO Keywords: Source term estimation Atmospheric dispersion RODOS JRODOS Ru-106 Ruthenium ABSTRACT In this work, we present a method to use the European nuclear emergency response system RODOS for analysis of potential sources of airborne radioactivity of an unknown origin. The method is based on a solution of adjoint equations, without modification of the code of long-range atmospheric dispersion model MATCH used in RODOS. The method has been successfully applied to the Ru-106 accident of 2017. The obtained spatial distribution of the correlation between simulations and measurements which could be achieved with source located in a given place, is in a qualitative agreement with analogous results published in other works. The high correlation is centered on the Ural Mountains; this is explained by a very wide expansion of the plume. However, the location of the maximum correlation obtained in this work is in the northern part of Russia, close to a military test site on Novaya Zemlya. This location is far away from the reprocessing plant Mayak in the South-Eastern Urals mentioned in other investigations as the most probable location of the source. In the results presented here, the correlation at the source location corresponding to the Mayak plant is still quite high (0.49); release inventory from this source of about 300 TBq could explain the observed measurements. 1. Introduction In late September - early October 2017, European and other stations reported detection of ruthenium-106 (Masson et al., 2019). The detected cloud was very large, it covered a big part of Eurasia from Norway to Kuwait and from Germany to East Siberia (Fig. 1). The source of this radionuclide was unclear. At the time of the accident, the International Atomic Energy Agency (IAEA) collected measurements from many countries (IAEA, 2017) and coordinated exchange of information be- tween them. Immediately after the accident, several research teams looked for a potential source of the Ru-106 using mathematical simu- lations (Kovalets, 2017; Kovalets and Romanenko, 2017; Sørensen, 2018; Saunier et al., 2019; Hamburger and Gering, 2019; Bossew et al., 2019). The EU nuclear emergency response system RODOS is a software tool designed for forecasting atmospheric transport of radionuclides from a local to planetary scale (Landman et al., 2014a). It is used in many countries in Europe, and it was widely applied to forecast consequences of different radiological accidents such as Fukushima and others (Landman et al., 2014b). However, this system does not include software tools for inverse transport simulations, such as backward trajectories analysis or adjoint equations solving to identify unknown sources of radioactive contamination. Despite this, it is possibile to use RODOS for identification of unknown sources. It was used in several studies iden- tifying contamination sources. In particular, one of the first simulation results showing the map of potential sources of the Ru-106 accident in 2017 was obtained with the RODOS system (Kovalets, 2017; Kovalets and Romanenko, 2017; Hamburger and Gering, 2019). In this work, we describe the approach of solving adjoint equations with the RODOS system (without modifying its source code) and subsequently solving the source identification (or inverse) problem. We describe results of application of this approach to the analysis of possible sources of Ru-106 detected in 2017. 2. Method of source term identification with RODOS In this work, we use the redesigned Java-version of the RODOS system – JRODOS (Ievdin et al., 2010) that among others contains the long-range Eulerian atmospheric transport model (ATM) MATCH (Robertson and Langner, 1999; Robertson, 2010). Kovalets et al. (2014) * Corresponding author. Institute of Mathematical Machines & Systems Problems, NAS of Ukraine, prosp. Glushkova, 42, 03187, Kyiv, Ukraine. E-mail address: ik@env.com.ua (I.V. Kovalets). Contents lists available at ScienceDirect Journal of Environmental Radioactivity journal homepage: http://www.elsevier.com/locate/jenvrad https://doi.org/10.1016/j.jenvrad.2020.106302 Received 6 March 2020; Received in revised form 28 April 2020; Accepted 5 May 2020