Technical note: An improved estimate of uncertainty for source contribution from effective variance Chemical Mass Balance (EV-CMB) analysis. (January 2015)

• Record Type: Journal Article
• Authors: Shi, Guo-Liang; Zhou, Xiao-Yu; Feng, Yin-Chang; Tian, Ying-Ze; Liu, Gui-Rong; Zheng, Mei; Zhou, Yang; Zhang, Yuan-Hang
• Is Part Of: Atmospheric environment. Volume 100(2015)
• Page Start: 154
• Page End: 158
• Publication Date: 2015-01
• Subjects: CMB -- Source apportionment -- Improved uncertainty -- Particulate matter -- Effective variance-least squares (EV-LS); Air -- Pollution -- Periodicals; Air -- Pollution -- Meteorological aspects -- Periodicals; 551.51
• DOI: 10.1016/j.atmosenv.2014.09.035
• Languages: English
• ISSNs: 1352-2310

Abstract

Abstract: The CMB (Chemical Mass Balance) 8.2 model released by the USEPA is a commonly used receptor model that can determine estimated source contributions and their uncertainties (called default uncertainty). In this study, we propose an improved CMB uncertainty for the modeled contributions (called EV-LS uncertainty) by adding the difference between the modeled and measured values for ambient species concentrations to the default CMB uncertainty, based on the effective variance least squares (EV-LS) solution. This correction reconciles the uncertainty estimates for EV and OLS regression. To verify the formula for the EV-LS CMB uncertainty, the same ambient datasets were analyzed using the equation we developed for EV-LS CMB uncertainty and a standard statistical package, SPSS 16.0. The same results were obtained by both ways indicate that the equation for EV-LS CMB uncertainty proposed here is acceptable. In addition, four ambient datasets were studies by CMB 8.2 and the source contributions as well as the associated uncertainties were obtained accordingly. Highlights: We propose an improved EV-LS CMB uncertainty for the modeled contributions. Difference between modeled and measured values is added in EV-LS uncertainty. The synthetic tests show the EV-LS CMB uncertainty might be accepted. The ambient dataset from four cities were analyzed by the EV-LS CMB uncertainty.