Performance of Back-Trajectory Statistical Methods and Inverse Modeling Method in Locating Emission Sources

Abstract

Back-trajectory statistical methods, e.g., potential source contribution functions (PSCF) and concentration-weighted trajectory (CWT) methods, have been widely used in previous studies to locate emission source regions of air pollutants or greenhouse gases. Inverse modeling methods have been developed and used in an increasing number of applications. To this date, there are no comparisons of performance between back-trajectory statistical and inverse modeling methods. This study evaluates the performance of PSCF, CWT and inverse modeling methods by taking advantage of precisely-known locations of trifluoromethane (CHF3; HFC-23) sources. Results show poor performance of the PSCF and CWT methods and good performance of the inverse modeling method. This study suggests that in studies with the purpose of locating emission source regions, the PSCF and CWT methods should be applied with caution in future studies, and that the inverse modeling method is encouraged to be used much more widely.

Analysis of air pollution characteristics, transport pathways and potential source areas identification in Beijing before, during and after the COVID-19 outbreak

“…We also conducted PSCF and CWT analysis at the release height of 500 m AGL (Supplementary Figure S4 and S5) to clarify the impact, and found that the distribution of potential source regions was very similar, like the results of previous studies (Zhu et al, 2011;Fang et al, 2018). Due to the longer retracing distance of the partial backward trajectory at 500 m altitude, the WPSCF and WCWT maps were larger at the release height of 500 m than that of 100 m, which may be helpful to identify more source areas.…”

“…It is worth mentioning that WCWT values were notably higher in the Bohai Bay region, which may be caused by the pollutants transported from their adjacent sea areas through the sea-land winds (Wang et al, 2022). Additionally, the PSCF and CWT model adopt the same treatment for air masses at various locations (e.g., urban, suburban, rural, forest, and ocean areas), that is the contributions of unit emission from each grid cell along a trajectory are equally weighted in calculating the mole fraction at the receptor sites (Fang et al, 2018), which may cause the overestimation in the contributions from remote areas such as ocean and Mongolia. The considerable difference of potential source areas between ACO and COB period could be explained by the seasonality or the changes in precursors emissions caused by the epidemic.…”

Emissions of HCFC-22 and HCFC-142b in China during 2018–2021 Inferred from Inverse Modeling

“…The reasons for choosing the FLEXPART model are as follows. First, the FLEXPART model has been widely used in estimating ozone-depleting substances emissions in China and has obtained relatively reliable results. ,,− Second, this study hopes to investigate the trend of HCFC emissions, and the conclusions obtained by using the same method as in previous study are more reliable. Third, Rigby et al and Park et al have previously compared multiple models (NAME-HB, NAME-InTEM, FLEXPART-MIT, and FLEXPART-Empa) and demonstrated the overall consistencies of trend and magnitude among multiple models.…”

Airborne Observations of CFCs Over Hebei Province, China in Spring 2016

“…Here, n ij is the total number of endpoints that fall in the ijth cell (Figure ) and m ij is the number of endpoints for which measured values exceed a specified criterion value in the ijth cell. Previous studies used the mean (Hui et al., 2019; Lin et al., 2019), 50th percentile (Li, Ye, et al., 2016; Polissar et al., 1999), 75th percentile (N. Gao et al., 2012; M. Yang et al., 2021; Zhao et al., 2015), and 90th percentile (Cheng & Lin, 2001; Fang et al., 2018) as criterion values. In this study, criterion values were selected based on the mean, 50th percentile, 75th percentile, and 90th percentile of measured CFC‐11 data to identify potential emission source regions of higher CFC‐11 mixing ratios (Figure ).…”