European Tracer Experiment Research Articles

Localisation of atmospheric release of radioisotopes using inverse methods and footprints of receptors as sources

Releases of radionuclides to the atmosphere occasionally occur with no warning and with first observation at radioactivity monitoring stations. The Chernobyl accident of 1986 was first detected at Forsmark, Sweden, long before the official announcement by the Soviet Union, and the release of Ruthenium 106 detected across Europe in 2017 still has no official release location. The current study details a method based on footprint analysis of an atmospheric dispersion model to locate the source of an atmospheric release. The method was applied to the European Tracer EXperiment of 1994 to validate the method and to the Ruthenium observations of autumn 2017 to determine likely release locations and time characteristics of this release. The method can readily utilise an ensemble of numerical weather prediction data which improves the localisation results by taking into account meteorological uncertainties compared to only using deterministic weather data. In applying the method to the ETEX scenario, the most likely release location improved from a distance of 113 km from the true release location when using deterministic meteorology, to a distance of 63 km when using ensemble meteorology data, although such improvements may be scenario dependent. The method was constructed to be robust with respect to the choices of model parameters and measurement uncertainties. The localisation method can be useful for decision makers to enact countermeasures to protect the environment against the effects of radioactivity when observations are available from environmental radioactivity monitoring networks.

Inversion method for multi-point source pollution identification: Sensitivity analysis and application to European Tracer Experiment data

Fast and accurate identification of unknown pollution sources plays a crucial role in the emergency response and source control of air pollution. In this work, the applicability of a previously proposed two-step inversion method is investigated with sensitivity experiments and real data from the first release of the European Tracer Experiment (ETEX-1). The two-step inversion method is based on the principle of least squares and carries out additional model correction through the residual iterative process. To evaluate its performance, its retrieval results are compared with those of two other existing algorithms. It is shown that for those cases with richer measurements, all three methods are less sensitive to errors, while for cases where measurements are sparse, their retrieval accuracy will rapidly decrease as errors increase. From the results of sensitivity experiments, the new method provides higher estimation accuracy and a more stable performance than the other two methods. The new method presents the smallest maximum location error of 18.20 km when the amplitude of the measurement error increases to 100%, and 22.67 km when errors in the wind fields increase to 200%. Moreover, when applied to ETEX-1 data, the new method also exhibits good performance, with a location error of 4.71 km, which is the best estimation with respect to source location.摘要快速并且准确地识别未知污染源, 在大气污染应急响应和源头控制过程中起着至关重要的作用. 本文利用敏感性试验及欧洲示踪物测场试验(ETEX-1)数据研究了新提出的两步反演算法的实用性, 并将其反演结果与现有的两种算法进行了对比分析. 敏感试验表明, 在观测数据较为丰富的情况下, 三种算法对观测误差和风场误差的敏感性均较低; 而当观测数据较为稀疏时, 所有算法的估计精度都将随着误差的增加而下降, 但与其他两种算法相比, 两步反演算法具有更高的估计精度以及更稳定的估计性能. 此外, 欧洲示踪物测场试验的源项估计结果也表明, 在三个算法中, 两步反演算法具有最小的位置估计误差.

Evaluation and Uncertainty of Radioxenon Transport with a Mesoscale Model after the February 2013 Underground Test in North Korea

Abstract The transport of radioxenon released from the February 2013 underground nuclear weapons test in North Korea was analyzed at two receptors—one at the Comprehensive Nuclear Test Ban Treaty site Rn58 in Russia (400 km downwind) and a second at Rn38 in Japan (1000 km downwind). Transport was modeled with two ensembles of mesoscale simulations, one generated with varying initial and lateral boundary conditions taken from the Global Forecasting System uncertainty ensemble, and a second created from different parameterizations and surface conditions. The wind variability was similar for the two ensembles and consistent with observations at 925 mb (1 mb = 1 hPa) but not at the surface. Biases in calculated surface winds and the radioxenon concentration in the ensembles were attributed mainly to poor simulation of the sea breeze at both locations and mountain lee affects at Rn38 in Japan. These wind regimes affected the timing of the surface radioxenon plume at Rn58 and its duration at Rn38. Surface wind variability induced by terrain and land–sea contrast (the sea breeze) also had a significant effect on the surface winds and plume dynamics, including blocking of flow approaching elevated terrain near Vladivostok and the west side of Japan. Increased plume uncertainty was seen at night because of surface wind variability. Measured surface chemical variability was larger than found in the first European Tracer Experiment in central Europe. The study found that horizontal model resolution contributes to uncertainty but not as much as vertical resolution, boundary layer parameterizations, and assimilation of surface meteorological data near the receptor.

Probabilistic Inverse Method for Source Localization Applied to ETEX and the 2017 Case of Ru-106 including Analyses of Sensitivity to Measurement Data

In recent years, cases of unexplained, elevated levels of radioactive particles have demonstrated an increasing need for efficient and robust source localization methods. In this study, a Bayesian method for source localization is developed and applied to two cases. First, the method is validated against the European tracer experiment (ETEX) and then applied to the still unaccounted for release of Ru-106 in the fall of 2017. The ETEX dataset, however, differs significantly from the Ru-106 dataset with regard to time resolution and the distance from the release site to the nearest measurements. Therefore, sensitivity analyses are conducted in order to test the method’s sensitivity to these parameters. The analyses show that the resulting source localization depends on both the observed temporal resolution and the existence of sampling stations close to the source. However, the method is robust, in the sense that reducing the amount of information in the dataset merely reduces the accuracy, and hence, none of the results are contradictory. When applied to the Ru-106 case, the results indicate that the Southern Ural region is the most plausible release area, and, as hypothesized by other studies, that the Mayak nuclear facility is the most likely release location.

Source term determination with elastic plume bias correction

Estimation of a source term, i.e. release rate, of atmospheric radionuclide emissions is of key interest for nuclear emergency response and further accident analysis. The source term estimate is, however, often very inaccurate due to biases in atmospheric transport and used meteorological analysis. We propose a method for atmospheric plume bias correction which uses not only concentrations modeled at a measuring site but also the information on concentration gradient from the neighborhood of each measuring site, i.e. information already available from the atmospheric transport model. To properly regularize the model, we propose an elastic model of the plume bias correction based on regularization with the use of known topology of the measurement network. The proposed plume bias correction method can be coupled with an arbitrary source term estimation algorithm and can be instantly applied to any other atmospheric release of hazardous material. We demonstrate the method in two real cases. First, we use data from the European Tracer Experiment to validate the methodology. Second, we use data from the 106Ru occurrence over Europe in 2017 to demonstrate the methodology in a more demanding case where agreement with state-of-the-art estimates is shown with much better reconstruction of measurements.

Method of Source Identification Following an Accidental Release at an Unknown Location Using a Lagrangian Atmospheric Dispersion Model

A computationally efficient source inversion algorithm was developed and applied with the Lagrangian atmospheric dispersion model DIPCOT. In the process of source location estimation by minimizing a correlation-based cost function, the algorithm uses only the values of the time-integrated concentrations at the monitoring stations instead of all of the individual measurements in the full concentration-time series, resulting in a significant reduction in the number of integrations of the backward transport equations. Following the source location estimation the release start time, duration and emission rate are assessed. The developed algorithm was verified for the conditions of the ETEX-I (European Tracer Experiment—1st release). Using time-integrated measurements from all available stations, the distance between the estimated and true source location was 108 km. The estimated start time of the release was only about 1 h different from the true value, within the possible accuracy of estimate of this parameter. The estimated release duration was 21 h (the true value was 12 h). The estimated release rate was 4.28 g/s (the true value was 7.95 g/s). The estimated released mass almost perfectly fitted the true released mass (323.6 vs. 343.4 kg). It thus could be concluded that the developed algorithm is suitable for further integration in real-time decision support systems.

Detection of radioactivity of unknown origin: Protective actions based on inverse modelling

The detection of radioactivity of unknown origin necessitates the use of models that can quantify unknown corresponding source term parameters. In this work, a method for solving this inverse problem is described. The main goal of the method is that it can be used in emergency response. Therefore, the full modelling chain dealing with the collection and pre-processing of measurement data, source term estimation, (forward) dispersion modelling, and consequence assessment are discussed. Firstly, to verify this inverse model SHERLOC, the part of the modelling chain concerning the source term estimation based on measurement data, is applied to the first episode of the European Tracer Experiment (ETEX). Secondly, the complete model chain is applied to a release that is still unaccounted for; the 106Ru measured in the atmosphere of Europe in September and October of 2017. It is estimated that during the night of the 25th to the 26th of September 2017 approximately 1.33 PBq (1.33×1015 Bq) of 106Ru was emitted at a location in the region of the Southern Urals in the Russian Federation. Statistical indicators show that the modelled levels of concentration are in good agreement with the measurements. The radiological consequences of the release are estimated to be minor at distances farther than 22 km from the estimated source. However, in the vicinity of the emission the maximum committed dose received by the public may have exceeded 100 mSv. Since the presented approach can be executed within few hours after the collection of measurement data it can be used in the emergency response following the detection of radioactivity of unknown origin.

On the tuning of atmospheric inverse methods: comparisons with the European Tracer Experiment (ETEX) and Chernobyl datasets using the atmospheric transport model FLEXPART

Abstract. Estimation of the temporal profile of an atmospheric release, also called the source term, is an important problem in environmental sciences. The problem can be formalized as a linear inverse problem wherein the unknown source term is optimized to minimize the difference between the measurements and the corresponding model predictions. The problem is typically ill-posed due to low sensor coverage of a release and due to uncertainties, e.g., in measurements or atmospheric transport modeling; hence, all state-of-the-art methods are based on some form of regularization of the problem using additional information. We consider two kinds of additional information: the prior source term, also known as the first guess, and regularization parameters for the shape of the source term. While the first guess is based on information independent of the measurements, such as the physics of the potential release or previous estimations, the regularization parameters are often selected by the designers of the optimization procedure. In this paper, we provide a sensitivity study of two inverse methodologies on the choice of the prior source term and regularization parameters of the methods. The sensitivity is studied in two cases: data from the European Tracer Experiment (ETEX) using FLEXPART v8.1 and the caesium-134 and caesium-137 dataset from the Chernobyl accident using FLEXPART v10.3.

A robust adaptive Lasso estimator for the independent contamination model

The Lasso has become a benchmark method for simultaneous parameter estimation and variable selection in regression analysis. It is based on the least-squares estimator and, therefore, suffers from the presence of outliers. Robust Lasso methods combine the objective function of a robust estimator with ℓ1-penalization. We address robustness for cases in which the number of observations is smaller (or not much larger) than the number of predictors. Further, we assume that the regression matrix may contain cellwise outliers. In such settings, even a few highly contaminated predictors can cause existing robust methods that are based on the commonly used rowwise contamination model to break down. Therefore, we propose a new adaptive Lasso type regularization. It takes into account cellwise outlyingness in the regression matrix and uses this information for robust variable selection. The proposed regularization term is integrated into the objective function of the MM-estimator, which yields the proposed MM-Robust Weighted Adaptive Lasso (MM-RWAL). A performance comparison to existing robust Lasso estimators is provided using Monte Carlo experiments. Further, the MM-RWAL is applied to determine the temporal releases of the European Tracer Experiment (ETEX) at the source location. This sparse and ill-conditioned linear inverse problem contains cellwise and rowwise outliers.

The Application of an Evolutionary Programming Process to a Simulation of the ETEX Large-Scale Airborne Dispersion Experiment

AbstractAirborne tracer simulations are typically performed using a dispersion model driven by a high-resolution meteorological model. Besides solving the dynamic equations of momentum, heat, and moisture on the resolved model grid, mesoscale models must account for subgrid-scale fluxes and other unresolved processes. These are estimated through parameterization schemes of eddy diffusion, convection, and surface interactions, and they make use of prescribed parameters set by the user. Such “free” model parameters are often poorly constrained, and a range of plausible values exists for each. Evolutionary programming (EP) is a process to improve the selection of the parameters. A population of simulations is first run with a different set of parameter values for each member, and the member judged most accurate is selected as the “parent” of a new “generation.” After a number of iterations, the simulations should approach a configuration that is best adapted to the atmospheric conditions. We apply the EP process to simulate the first release of the 1994 European Tracer Experiment (ETEX) project, which comprised two experiments in which a tracer was released in western France and sampled by an observing network. The EP process is used to improve a simulation of the RAMS mesoscale weather model, with weather data collected during ETEX being used to “score” the individual members according to how well each simulation matches the observations. The meteorological simulations from before and after application of the EP process are each used to force a dispersion model to create a simulation of the ETEX release, and substantial improvement is observed when these are validated against sampled tracer concentrations.

A random walk model to simulate the atmospheric dispersion of radionuclide

To investigate the atmospheric dispersion of radionuclide in large-medium scale, a numerical simulation method based on random walk model for radionuclide atmospheric dispersion was established in the paper. The route of radionuclide migration and concentration distribution of radionuclide can be calculated out by using the method with the real-time or historical meteorological fields. In the simulation, a plume of radionuclide is treated as a lot of particles independent of each other. The particles move randomly by the fluctuations of turbulence, and disperse, so as to enlarge the volume of the plume and dilute the concentration of radionuclide. The dispersion of the plume over time is described by the variance of the particles. Through statistical analysis, the relationships between variance of the particles and radionuclide dispersion characteristics can be derived. The main mechanisms considered in the physical model are: (1) advection of radionuclide by mean air motion, (2) mixing of radionuclide by atmospheric turbulence, (3) dry and wet deposition, (4) disintegration. A code named RADES was developed according the method. And then, the European Tracer Experiment (ETEX) in 1994 is simulated by the RADES and FLEXPART codes, the simulation results of the concentration distribution of tracer are in good agreement with the experimental data.

LS-APC v1.0: a tuning-free method for the linear inverse problem and its application to source-term determination

Abstract. Estimation of pollutant releases into the atmosphere is an important problem in the environmental sciences. It is typically formalized as an inverse problem using a linear model that can explain observable quantities (e.g., concentrations or deposition values) as a product of the source-receptor sensitivity (SRS) matrix obtained from an atmospheric transport model multiplied by the unknown source-term vector. Since this problem is typically ill-posed, current state-of-the-art methods are based on regularization of the problem and solution of a formulated optimization problem. This procedure depends on manual settings of uncertainties that are often very poorly quantified, effectively making them tuning parameters. We formulate a probabilistic model, that has the same maximum likelihood solution as the conventional method using pre-specified uncertainties. Replacement of the maximum likelihood solution by full Bayesian estimation also allows estimation of all tuning parameters from the measurements. The estimation procedure is based on the variational Bayes approximation which is evaluated by an iterative algorithm. The resulting method is thus very similar to the conventional approach, but with the possibility to also estimate all tuning parameters from the observations. The proposed algorithm is tested and compared with the standard methods on data from the European Tracer Experiment (ETEX) where advantages of the new method are demonstrated. A MATLAB implementation of the proposed algorithm is available for download.

Development of the Eulerian atmospheric transport model GEARN-FDM: Validation against the European tracer experiment

In this study, we developed an atmospheric dispersion model based on the finite difference method (GEARN-FDM) to simulate long-range radiological dispersion. The model includes a mass conservation and monotonic advection scheme and a horizontal diffusion scheme based on the deformation of the resolved horizontal wind. Meteorological fields in calculating dispersion were simulated by the Advanced Research Weather Research and Forecasting (WRF) model. ERA-interim reanalysis was used for the WRF nudging calculation. By using tracer gas concentrations from the European tracer experiment, the GEARN-FDM performance could be tested. The results indicated high performance with factors of 2 and 5 of 37% and 72%, respectively. To investigate the impact of horizontal diffusivity on the concentration distribution, we examined an additional simulation run using GEARN-FDM with a constant value of horizontal diffusivity. In comparison with the additional run, the plume calculated in the original GEARN-FDM run was locally and temporally diluted due to large horizontal diffusivity in a mountainous region. Therefore, geographical heterogeneity of horizontal diffusivity may affect regional-scale atmospheric dispersion simulations.

Estimation of the radionuclides emission region using trajectory and atmospheric dispersion models

A three-dimensional trajectory model has been developed to estimate the emission area of radioisotopes measured at the international monitoring stations of the Comprehensive Nuclear-Ban Treaty Organization (CTBTO). Calculated backward trajectories were compared with measurements of the European Tracer Experiment (ETEX). These trajectories generally moved toward the original release point and detected within the release duration of the tracer experiment. However, the comparative results had uncertainties with regard to starting time and the height of the tracer in the backward trajectory model. An atmospheric dispersion model based on the Fields of Regards (FOR) technique was used to reduce the uncertainties of trajectory model and to improve the accuracy of detective technology. The simulated results generated by the trajectory and atmospheric dispersion models together were agreed better with the measurements compared to those obtained from the trajectory model alone.

Sparse optimization for inverse problems in atmospheric modelling

We consider inverse problems in atmospheric modelling represented by a linear system which is based on a source-receptor sensitivity matrix and measurements. Instead of using the ordinary least squares, we add a weighting matrix based on the topology of measurement points and show the connection with Bayesian modelling. Since the source-receptor sensitivity matrix is usually ill-conditioned, the problem is often regularized, either by perturbing the objective function or by modifying the sensitivity matrix. However, both these approaches may be heavily dependent on specified parameters. To ease this burden, we propose to use techniques looking for a sparse solution with a small number of positive elements. Finally, we compare all these methods on the European Tracer Experiment (ETEX) data where there is no apriori information apart from the release position and some measurements.

Ensemble-based simultaneous emission estimates and improved forecast of radioactive pollution from nuclear power plant accidents: application to ETEX tracer experiment

The accidental release of radioactive materials from nuclear power plant leads to radioactive pollution. We apply an augmented ensemble Kalman filter (EnKF) with a chemical transport model to jointly estimate the emissions of Perfluoromethylcyclohexane (PMCH), a tracer substitute for radionuclides, from a point source during the European Tracer Experiment, and to improve the forecast of its dispersion downwind. We perturb wind fields to account for meteorological uncertainties. We expand the state vector of PMCH concentrations through continuously adding an a priori emission rate for each succeeding assimilation cycle. We adopt a time-correlated red noise to simulate the temporal emission fluctuation. The improved EnKF system rapidly updates (and reduces) the excessively large initial first-guess emissions, thereby significantly improves subsequent forecasts (r = 0.83, p < 0.001). It retrieves 94% of the total PMCH released and substantially reduces transport error (>80% average reduction of the normalized mean square error).

The European long–range tracer experiment, ETEX: a database for comparing model results and for harmonisation purposes

This report presents the layout and a summary of preliminary results of the European tracer experiment (ETEX), which was devised to serve several purposes, one of which was to create a database for long–range modelling.

APOLLO2, a new long range Lagrangian particle dispersion model and its evaluation against the first ETEX tracer release

APOLLO2 is a new version of the long range Lagrangian particle dispersion model integrated into ARIES (Accidental Release Impact Evaluation System), the system that I.S.P.R.A. (Italian National Institute for Environmental Protection and Research) developed to provide prognosis about the dispersion of radioactive clouds in case of nuclear accidents. In this paper the theoretical aspects of the APOLLO2 formulation are briefly described. Moreover, the model performances have been evaluated qualitatively and quantitatively against the observations of the first release of the European Tracer Experiment (ETEX). The global analysis showed a FA2 = 56%, a FA5 = 78%, and a good agreement for intermediate values within the Q–Q plot. Considering a total of 126 stations, the number of locations where NMSE < 10, PCC > 0.7 and FMT > 20% is respectively 91, 65 and 86. The number of stations where the Chang and Hanna (2004) performance evaluation criteria are satisfied is 33. The capability of APOLLO2 to predict arrival time, cloud duration and time of peak is comparable to the one shown by similar models.Finally, the sensitivity of the model to some input variables has also been investigated.

Bayesian design of control space for optimal assimilation of observations. Part I: Consistent multiscale formalism

AbstractIn geophysical data assimilation, the control space is by definition the set of parameters which are estimated through the assimilation of observations. It has recently been proposed to design the discretizations of control space in order to assimilate observations optimally. The present paper describes the embedding of that formalism in a consistent Bayesian framework. General background errors are now accounted for. Scale‐dependent errors, such as aggregation errors (that lead to representativeness errors) are consistently introduced. The optimal adaptive discretizations of control space minimize a criterion on a dictionary of grids. New criteria are proposed: degrees of freedom for the signal (DFS) built on the averaging kernel operator, and an observation‐dependent criterion.These concepts and results are applied to atmospheric transport of pollutants. The algorithms are tested on the European tracer experiment (ETEX), and on a prototype of CO2 flux inversion over Europe using a simplified CarboEurope‐IP network. New types of adaptive discretization of control space are tested such as quaternary trees or factorised trees. Quaternary trees are proven to be both economical, in terms of storage and CPU time, and efficient on the test cases. This sets the path for the application of this methodology to high‐dimensional and noisy geophysical systems. Part II of this article will develop asymptotic solutions for the design of control space representations that are obtained analytically and are contenders to exact numerical optimizations. Copyright © 2011 Royal Meteorological Society

Subgrid‐scale treatment for major point sources in an Eulerian model: A sensitivity study on the European Tracer Experiment (ETEX) and Chernobyl cases

We investigate the plume‐in‐grid method for a subgrid‐scale treatment of major point sources in the passive case. This method consists in an online coupling of a Gaussian puff model and an Eulerian model, which better represents the point emissions without significantly increasing the computational burden. In this paper, the plume‐in‐grid model implemented on the Polyphemus air quality modeling system is described, with an emphasis on the parameterizations available for the Gaussian dispersion, and on the coupling with the Eulerian model. The study evaluates the model for passive tracers at continental scale with the European Tracer Experiment (ETEX) and the Chernobyl case. The aim is to (1) estimate the model sensitivity to the local‐scale parameterizations and (2) bring insights on the spatial and temporal scales that are relevant in the use of a plume‐in‐grid model. It is found that the plume‐in‐grid treatment improves the vertical diffusion at local scale, thus reducing the bias, especially at the closest stations. Doury's Gaussian parameterization and a column injection method give the best results. There is a strong sensitivity of the results to the injection time and the grid resolution. The “best” injection time actually depends on the resolution but is difficult to determine a priori. The plume‐in‐grid method is also found to improve the results at fine resolutions more than with coarse grids by compensating the Eulerian tendency to overpredict the concentrations at these resolutions.