Bayesian Estimation of Volatile Organic Emissions from Wood Furniture Coating

Abstract

Volatile organic compound (VOC) emissions from industrial coating operations typically are estimated using use rate–based models. Sample size considerations often require that model parameters such as coating use rates and emission factors be developed from regional or national data. Although these data are valuable, they may not be current or reflect local use patterns. In this Bayesian estimation of VOC emissions from wood furniture coating in Los Angeles County, California, statewide data are combined with local data using Bayes' theorem. Statewide data and expert judgment are used to formulate maximum entropy prior probability distributions. The coating emission factor prior is updated with local survey data. Bayesian estimation can reduce the cost of estimating emissions by updating prior estimates with small samples of contemporary regional data. The method provides minimally biased distributions of VOC emissions that are more informative than estimates obtained with other statistical methods.