Fusion Field Trials Research Articles

Source term estimation for continuous plume dispersion in Fusion Field Trial-07: Bayesian inference probability adjoint inverse method

In scenarios involving sudden releases of unidentified gases or concealed pollution emergencies, source control emerges as a critical procedure to safeguard residential air quality. Appropriate inverse source tracking methodology depending on diverse measurement data could be utilized to promptly identify pollutant source parameters. In this study, source term estimation (STE) method, i.e., jointly combining probability adjoint method with the Bayesian inference method, has been proposed. General form of the pollutant inverse transport equation was firstly established. Subsequently, the pollution source information, assumed from single continuous point releases during Fusion Field Trials 2007 under an unsteady wind field, was identified using the Bayesian inference probability adjoint inverse method. Metropolis-Hastings Markov Chain Monte Carlo (MH-MCMC) and Differential Evolution Markov Chain Monte Carlo (DE-MCMC) were then compared as sampling methods for Bayesian inference. Results indicated that the DE-MCMC algorithm has superior convergence and could present higher accuracy of pollutant source information than that of MH-MCMC algorithm, particularly for highly nonlinear and multi-modal distribution systems. Furthermore, the integration of Union standard Adjoint Location Probability (UALP) as prior information into the Bayesian inference probability adjoint inverse method effectively narrowed the sampling range, enhancing both the accuracy and robustness of the proposed approach. Finally, the impact of the covariance matrix on the inverse identification accuracy was explored. Overall, this research has provided insights into the future applicability of this Bayesian inference inversion technique for point source identification.

An accurate strategy to retrieve multiple source emissions in the atmosphere

Abstract We propose a new methodology to retrieve multiple point emission sources in the atmosphere. To allow more realistic modelling, the dispersion problem is solved by a stabilized finite element formulation. The simultaneous estimation of the source strengths, their locations and the total number of sources is made employing Tikhonov-type regularization, with a composite data misfit function. The corresponding minimization of the Tikhonov-type functional is carried out by a combination of a splitting strategy algorithm with a gradient-descent method. We illustrate the accuracy of the proposed technique by using Fusion Field Trials 2007 (FFT07) data.

Characterizing Variations in Concentration Data Measured From Unknown Releases

The statistics of concentration data, measured during unknown atmospheric dispersion events, are not fully understood although they are required in modeling, assessment, uncertainty analysis, and information fusion. The concentrations measured over a field of monitoring network is regarded as a vector which contains both magnitude and direction. Traditional statistics (mean, standard deviations, etc.) based on magnitude of concentration data summarize data properties but are limited in characterizing variations in data and their modeling quality. Comparatively, directions are efficient in providing valuable information to address these issues. Here we propose a statistical framework which transforms concentration measurements into directions projected on a hypersphere and analyzes their orientation and distribution. The directional data measured in identical conditions are expected to be rotationally symmetric around its principal axis and follow Watson distribution. The clustering parameter of Watson distribution measures tightness of directional data and, thus, can measure indirectly variations in observed data. It is shown that the clustering parameter is able to summarize an overall variation in data and modeling quality of data in a dispersion trial. The study analyzes real data taken from continuous release experiments, called “Fusion Field Trials,” conducted at Dugway Proving Ground, Utah, United States.

Performance evaluation of dispersion parameterization schemes in the plume simulation of FFT-07 diffusion experiment

Application of atmospheric dispersion models in air quality analysis requires a proper representation of the vertical and horizontal growth of the plume. For this purpose, various schemes for the parameterization of dispersion parameters σ′s are described in both stable and unstable conditions. These schemes differ on the use of (i) extent of availability of on-site measurements (ii) formulations developed for other sites and (iii) empirical relations. The performance of these schemes is evaluated in an earlier developed IIT (Indian Institute of Technology) dispersion model with the data set in single and multiple releases conducted at Fusion Field Trials, Dugway Proving Ground, Utah 2007. Qualitative and quantitative evaluation of the relative performance of all the schemes is carried out in both stable and unstable conditions in the light of (i) peak/maximum concentrations, and (ii) overall concentration distribution. The blocked bootstrap resampling technique is adopted to investigate the statistical significance of the differences in performances of each of the schemes by computing 95% confidence limits on the parameters FB and NMSE. The various analysis based on some selected statistical measures indicated consistency in the qualitative and quantitative performances of σ schemes. The scheme which is based on standard deviation of wind velocity fluctuations and Lagrangian time scales exhibits a relatively better performance in predicting the peak as well as the lateral spread.

Comparative interpretations of renormalization inversion technique for reconstructing unknown emissions from measured atmospheric concentrations

The study highlights a theoretical comparison and various interpretations of a recent inversion technique, called renormalization, developed for the reconstruction of unknown tracer emissions from their measured concentrations. The comparative interpretations are presented in relation to the other inversion techniques based on principle of regularization, Bayesian, minimum norm, maximum entropy on mean, and model resolution optimization. It is shown that the renormalization technique can be interpreted in a similar manner to other techniques, with a practical choice of a priori information and error statistics, while eliminating the need of additional constraints. The study shows that the proposed weight matrix and weighted Gram matrix offer a suitable deterministic choice to the background error and measurement covariance matrices, respectively, in the absence of statistical knowledge about background and measurement errors. The technique is advantageous since it (i) utilizes weights representing a priori information apparent to the monitoring network, (ii) avoids dependence on background source estimates, (iii) improves on alternative choices for the error statistics, (iv) overcomes the colocalization problem in a natural manner, and (v) provides an optimally resolved source reconstruction. A comparative illustration of source retrieval is made by using the real measurements from a continuous point release conducted in Fusion Field Trials, Dugway Proving Ground, Utah.

Uncertainty characterization in the retrieval of an atmospheric point release

The study proposes a methodology in a recent inversion technique, called as Renormalization, to characterize the uncertainties in the reconstruction of a point source. The estimates are derived for measuring the inversion error, the degree of model fit towards measurements (model determination coefficient) and the confidence intervals for the retrieved point source parameters (mainly, location and strength). The inversion error is reflected through an angular estimate which measures the deviation between the measured and predicted concentrations. The uncertainty estimation methodology is evaluated for point source reconstruction studies, using real measurements from two field experiments, known as Fusion Field Trials 2007 (FFT07) in flat terrain and Mock Urban Setting Test (MUST) in urban like terrain. In FFT07 and MUST experiments, the point source location is retrieved with an average Euclidean distance of 22 m and 15 m respectively. The source strength is retrieved, on average, within a factor of 1.5 in both the datasets. The inversion error is observed as 24o and 21o in FFT07 and MUST experiment, respectively. The 95% confidence interval estimates show that the uncertainty in the retrieved parameters is relatively large in approximately 50% FFT07 and 30% MUST trials in spite of their closeness towards true source parameters. For a comparative analysis, the interval estimates are also compared with a more general method of uncertainty estimation, Residual Bootstrap Sampling. In most of the trials, we observed that the intervals estimates with the present method are comparable (within 10–20% variations) to bootstrap estimates. The proposed methodology provides near accurate and computationally efficient uncertainty estimates in comparison to the methods based on Hessian and sampling procedures.

Simulation of plume dispersion of multiple releases in Fusion Field Trial-07 experiment

Abstract For an efficient source term estimation, it is important to use an accurate dispersion model with appropriate dispersion parameters. This is examined by simulating the dispersion of plumes resulted from the available multiple releases conducted at Fusion Field Trials, Dugway Proving Ground, Utah. The simulation is carried out with an earlier developed IIT (Indian Institute of Technology) dispersion model using the dispersion parameters in terms of measurements of turbulent velocity fluctuations. Simulation is discussed separately in both stable and unstable conditions in light of (i) plume behavior of observed and predicted concentrations in the form of isopleths, (ii) peak/maximum concentrations and (iii) overall concentration distribution. Simulated results from IIT model are compared with those obtained using AERMOD. Both, IIT model and AERMOD, predicted peak concentrations within a factor of two in all the releases and tracer transport is mostly along the mean wind direction. With IIT model, the higher concentrations are predicted close to observations in all the trials of stable conditions and with in a factor of two in the trials of unstable conditions. However, the relatively smaller concentrations are under-predicted severely in stable conditions and over-predicted in unstable conditions. The AERMOD exhibits the similar prediction of concentrations as in IIT model except slightly over-prediction in stable conditions and under-prediction in unstable conditions. The statistical measures for both the models are found good in agreement with the observations and a quantitative analysis based on F-test shows that the performance from both the models are found to be similar at 5% significance level.

An adaptive Bayesian inference algorithm to estimate the parameters of a hazardous atmospheric release

In the eventuality of an accidental or intentional atmospheric release, the reconstruction of the source term using measurements from a set of sensors is an important and challenging inverse problem. A rapid and accurate estimation of the source allows faster and more efficient action for first-response teams, in addition to providing better damage assessment. This paper presents a Bayesian probabilistic approach to estimate the location and the temporal emission profile of a pointwise source. The release rate is evaluated analytically by using a Gaussian assumption on its prior distribution, and is enhanced with a positivity constraint to improve the estimation. The source location is obtained by the means of an advanced iterative Monte-Carlo technique called Adaptive Multiple Importance Sampling (AMIS), which uses a recycling process at each iteration to accelerate its convergence. The proposed methodology is tested using synthetic and real concentration data in the framework of the Fusion Field Trials 2007 (FFT-07) experiment. The quality of the obtained results is comparable to those coming from the Markov Chain Monte Carlo (MCMC) algorithm, a popular Bayesian method used for source estimation. Moreover, the adaptive processing of the AMIS provides a better sampling efficiency by reusing all the generated samples.

Assimilation of concentration measurements for retrieving multiple point releases in atmosphere: A least-squares approach to inverse modelling

The study addresses the identification of multiple point sources, emitting the same tracer, from their limited set of merged concentration measurements. The identification, here, refers to the estimation of locations and strengths of a known number of simultaneous point releases. The source–receptor relationship is described in the framework of adjoint modelling by using an analytical Gaussian dispersion model. A least-squares minimization framework, free from an initialization of the release parameters (locations and strengths), is presented to estimate the release parameters. This utilizes the distributed source information observable from the given monitoring design and number of measurements. The technique leads to an exact retrieval of the true release parameters when measurements are noise free and exactly described by the dispersion model. The inversion algorithm is evaluated using the real data from multiple (two, three and four) releases conducted during Fusion Field Trials in September 2007 at Dugway Proving Ground, Utah. The release locations are retrieved, on average, within 25–45 m of the true sources with the distance from retrieved to true source ranging from 0 to 130 m. The release strengths are also estimated within a factor of three to the true release rates. The average deviations in retrieval of source locations are observed relatively large in two release trials in comparison to three and four release trials.

Reconstruction of an atmospheric tracer source in Fusion Field Trials: Analyzing resolution features

AbstractReconstruction of unknown atmospheric releases using measured concentrations is an ill‐posed inverse problem. Due to insufficient measurements and dispersion model uncertainties, reliable interpretation of a retrieved source is limited by lack of resolution, nonuniqueness, and instability in the inverse solution. The study presents an optimality analysis, in terms of resolution, stability, and reliability, of an inverse solution given by a recently proposed inversion technique, called as renormalization. The inversion technique is based on an adjoint source‐receptor framework and construction of a weight function which interprets a priori information about the unknown release apparent to the monitoring network. The properties of weight function provide a perfect data resolution, maximum model resolution, and minimum variance (or stability) for the retrieved source. The reliability of the retrieved source is interpreted in view of the information derived from the geometry of the monitoring network. The inversion technique and resolution features are evaluated for a point source reconstruction using measurements from a recent dispersion experiment (Fusion Field Trials 2007) conducted at Dugway Proving Ground, Utah. With the real measurements, the point release is reconstructed within an average distance of 23 m from the true release where the average distance of the nearest receptor from the true source was 32 m. In all the trials, the point release is retrieved within 3–60 m Euclidean distance from their true location. The source strength is retrieved within a factor of 1.5 to the true release mass. The posterior uncertainty in the release parameters is observed to be within 20% of their mean value. The source localization features are resolved to its maximum extent feasible with the design of the monitoring network. The sensitivity studies are conducted to highlight the importance of receptors reporting zero concentration measurements and variations in the resolution features of the source retrieval with respect to the various arrangements of the receptors.

Application of genetic algorithm for the simultaneous identification of atmospheric pollution sources

A computational model is developed for retrieving the positions and the emission rates of unknown pollution sources, under steady state conditions, starting from the measurements of the concentration of the pollutants. The approach is based on the minimization of a fitness function employing a genetic algorithm paradigm. The model is tested considering both pollutant concentrations generated through a Gaussian model in 25 points in a 3-D test case domain (1000m × 1000m × 50 m) and experimental data such as the Prairie Grass field experiments data in which about 600 receptors were located along five concentric semicircle arcs and the Fusion Field Trials 2007. The results show that the computational model is capable to efficiently retrieve up to three different unknown sources.

A least-squares inversion technique for identification of a point release: Application to Fusion Field Trials 2007

Identification of a point release is a parametric estimation problem associated with the estimation of its parameters namely, location and strength. A least-squares inversion algorithm, free from initial guess of release parameters, is utilized here for the source identification in eleven trials of single continuous point releases conducted during Fusion Field Trials 2007. The source locations are retrieved within an average error of 23 m from their true locations. The maximum and minimum errors in the retrieval of the source location are obtained as 57.3 m and 3 m respectively. The source strength is retrieved within a factor of 1.6 in all the trials. The sensitivity of the source estimation is analysed with respect to (i) variation of grid sizes in discretized space, (ii) inclusion of zero measurements and (iii) addition of new measurements. Posterior uncertainty is mentioned in terms of variance of the source parameters, approximated by using the Hessian of the cost function. In addition, an attempt is made to obtain the minimum number of measurements for a successful source inversion. The study explores the future applicability of this least-squares inversion technique for point source identification.

Simulation of plume dispersion from single release in Fusion Field Trial-07 experiment

Accurate description of source–receptor relationship is required for an efficient source reconstruction. This is examined by simulating the dispersion of plumes resulted from the available ten trials of single releases conducted at Fusion Field Trials, Dugway Proving Ground, Utah. The simulation is addressed with an earlier developed IIT (Indian Institute of Technology) dispersion model using the dispersion parameters in terms of measurements of turbulent velocity fluctuations. Simulation is described separately in both stable and unstable conditions, characterizing the peak as well as overall observed concentration distribution. Simulated results are compared with those obtained using AERMOD.With IIT model, peak concentrations are predicted within a factor of two in all the trials. The higher concentrations (>5 × 10−4 g m−3) are well predicted in stable condition and under-predicted (within a factor of two) in unstable condition whereas relatively smaller concentrations (<5 × 10−4 g m−3) are severely under-predicted in stable conditions and over-predicted in unstable conditions.The AERMOD exhibits the similar prediction of concentrations as shown by IIT model in most of the trials. Overall, both the models predict 70–80% concentrations in stable conditions and 85–95% concentrations in unstable conditions within a factor of six. The statistical measures for both the models are found well in agreement with the observations.