Bayesian estimation of airborne fugitive emissions using a Gaussian plume model

Abstract

A new method is proposed for estimating the rate of fugitive emissions of particulate matter from multiple time-dependent sources via measurements of deposition and concentration. We cast this source inversion problem within the Bayesian framework, and use a forward model based on a Gaussian plume solution. We present three alternate models for constructing the prior distribution on the emission rates as functions of time. Next, we present an industrial case study in which our framework is applied to estimate the rate of fugitive emissions of lead particulates from a smelter in Trail, British Columbia, Canada. The Bayesian framework not only provides an approximate solution to the inverse problem, but also quantifies the uncertainty in the solution. Using this information we perform an uncertainty propagation study in order to assess the impact of the estimated sources on the area surrounding the industrial site.

Well-Posed Bayesian Inverse Problems: Priors with Exponential Tails

“…In this example we consider the problem of estimating the source term in a parabolic PDE from linear measurements of the solution. This problem is closely related to the inverse problem of estimating the sources of emissions in an atmospheric dispersion model [26] and we shall present this example in that context. Let D ⊂ R 3 be a smooth and connected domain and define Ω := D × (0, T ] for some constant T > 0.…”

“…The elaborate construction of the M i corresponds to a common method of measurement in the study of deposition of particulate matter where a number of plastic jars (also known as dust-fall jars) are left in the field for a given period of time [26,32]. At the end of this period the jars are taken to the lab and the concentration of deposited material in each jar is measured.…”

Estimating airborne particulate emissions using a finite-volume forward solver coupled with a Bayesian inversion approach

“…The work of Keats et al [23] is more closely related to this article, since they used a finite volume solver to construct the forward map within a Bayesian framework in order to infer the location and emission rate for a point source. A similar approach was employed by Hosseini and Stockie [19] to estimate the time-dependent behavior of emissions for a collection of point sources that are not operating at steady state. Here, we use a finite volume solver that was developed in [17] within a hierarchical Bayesian framework in order to infer the rate of emissions of multiple sources in an industrial site.…”

Optimal Design of Air Quality Monitoring Network for Pollution Detection and Source Identification in Industrial Parks

“…However, few have focused on the source reconstruction performance of AQMN in optimal design. Recently, many studies in the literature have explored how to reconstruct source characteristics based on the measurements from a dense AQMN and have analyzed the influence of the AQMN distribution on the back-calculation [17][18][19][20], while only single emission episodes were considered as the concerned objectives in these studies.…”