Measurement and deduction of emissions of short-lived atmospheric organo-chloride compounds

Abstract

Atmospheric studies of halogenated organics have centered on long lived halocarbons due to their effect on stratospheric ozone. Now that controls have been put in place to curb emissions of longer lived halocarbons through the Montreal Protocol, and speculation about the safety of many short-lived chlorinated organic molecules has been raised, there has been more consideration given to the efforts aimed at determining the levels of human exposure to all types of halogenated organics. Most previous studies of reactive chlorine compounds have focused solely on quantifying their ambient levels in urban and rural regions. However, for many of these organo-chlorine molecules a detailed knowledge of emissions levels, transport, and final environmental disposition still does not exist. The present work was designed to aid in understanding the emissions patterns for several reactive halogenated organic compounds including trichloromethane (chloroform, CHCl3), trichloroethene (TCE, CHClCCl2), and tetrachloroethene (perchloroethylene, CCl2CCl2). A high temporal frequency (hourly) measurement campaign in Nahant, Massachusetts (approximately 10 km. northeast of Boston) provided automated gas-chromatographic measurements for these species as well as the somewhat more stable 1,1,1 trichloroethane (methyl chloroform, CH3CCl3). Cryogenic preconcentrations, daily calibrations, and weekly linearity tests insure high precision (</= 5%) measurements using electron capture detection. Calibration gases used for these tests, initially manufactured at MIT, have been corrected by intercomparison with gas standards used by the AGAGE program (produced at Scripps Institution of Oceanography) as well as those used at the National Center for Atmospheric Research. The absolute accuracy of our corrected MIT standard is estimated to be 10%. Over 12,000 measurements of the selected species were made between March, 1998 and January, 1999. These data show wide variability for the shortest lived species ranging from our detection limits (4.5 ppt for trichloroethene, 4.2 ppt for tetrachloroethene, and 7.8 ppt for trichloromethane) up to several hundred ppt during periods of local pollution. Data analysis combines the measurements with back trajectory information obtained from the HYSPLIT4 model (HYbrid Single-Particle Lagrangian Integrated Trajectory model, NOAA Air Resources Laboratory, Silver Springs, Maryland). Using a Kalman filter inverse method and an analytical solution of the continuity equation to estimate the effect of diffusion, we calculate the surface emissions for the selected species necessary to optimally match the observations. These emissions are compared with the estimates determined by the Reactive Chlorine Emissions Inventory (RCEI) working group of the IGAC (International Global Atmospheric Chemistry Program) Global Emissions Inventory Activity (GEIA). RCEI estimates are primarily derived from point source emissions in the US Toxic Release Inventory (TRI) and population-based distribution of residual national consumption from sales records. The new emissions scenarios computed here provide an observation-based assessment for comparison with the emissions inventories produced by RCEI for the northeastern United States and southeastern Canada. Results are statistically consistent with the RCEI estimates given the currently rough accuracy (±47-67%) achievable through this observation-based technique. We note, however, that the best estimate of corrections factors for land-based grid cells presented here indicate that the RCEI emissions for trichloroethene and tetrachloroethene need to be increased by a factor of ~2 to explain the observations. Only anthropogenic sources of trichloromethane were gridded in the initial (RCEI) inventory representing roughly 11% of estimated global emissions. We find that these emissions are, as expected, too low to explain the observations and that a land-based correction factor ~12 is required to produce emissions estimates that are consistent with natural sources (e.g. soil emissions). We also note that very large correction factors are calculated over the oceanic grid cells resulting in revised emissions estimates of the same order of magnitude as many land-based grid cells, consistent with a large oceanic source for this compound inferred from oceanic observations. The 47-67% uncertainty in the estimates of emissions correction factors increases with distance from the observation site due to both the increase in trajectory error as a function of total trajectory length and the decrease in the number of trajectories which have passed through a particular grid cell as one moves further from the observation site. These and other sources of uncertainty can be reduced by providing a realistic weighting of each trajectory's accuracy thus minimizing the impact of the trajectories which are likely to be most inaccurate, increasing the total number of measurements so that all grid cells have greater trajectory coverage, and improving estimates of the effective mixed layer height.