Measurements of Atmospheric Methyl Bromide and Bromoform

We have measured gaseous methyl bromide (CHaBr) and bromoform (CHBra) in air samples that were gathered approximately weekly from five ground-level sites: Point Barrow, Alaska; Mauna Los Observatory and Cape Kumukahi, Hawaii; Matatula, Samoa; and Kaitorete Spit, New Zealand. Approximately 750 samples have been analyzed for CHaBr between January 1985 and October 1987 and 990 samples have been analyzed for CHBra between early 1984 and September 1987, all by gas chromatography/mass spectrometry. Methyl bromide concentrations are typically 10-11 parts per trillion (ppt) by volume; there are no clear indications of temporal increases. Bromoform concentrations are typically 2-3 ppt, but large seasonal variations are seen at Point Barrow.

Arctic and Antarctic profiles of CH3Br and CHBr3, which will be included in a future publication.
To further our understanding of atmospheric bromine, both tropospheric and stratospheric, a much more detailed inventory is needed for the source gases, tropospheric RBr species: concentrations, spatial patterns, and temporal variability and trends. Recent exploratory studies [Berg et al., 1984;Penkerr et al., 1985;Class et al., 1986] have presented measurements of several bromomethanes and bromoethanes, including CHBr3 and some mixed Copyright 1988   The temperature profile for sample analysis is -25øC for 1 min, programed temperature elevation at 15øC/rain up to 150øC, and maintenance of 150 ø for 20 min. Approximately 1000 mL of sample is preconcentrated in the sample injection loops by pumping away the Oa and N2 while holding the glass-bead-packed stainless steel loops at -185.9øC (liquid Ar). To prevent hysteresis effects from carryover of one sample to the next, the sample loops are heated during evacuation between sample runs. Initially, blank tests were run by purging with purified helium using the same preconcentration techniques employed for samples. However, since September 1984, we have used moist zero air, which better simulates a clean atmospheric sample and has proven to be more efficient in lowering the system background.
A standard mixture containing five organic bromine g•es including CHsBr and CHBr3 (prepared by Scott Specialty Gases) h• been used for sample quantification.
The manufacturer allowed the mixture to equilibrate for approximately 1 month before analyzing and assigning absolute concentrations. Our own attempts to prepare absolute mixtures of CH3Br

Results of CHsBr measurements are summarized in
Tables 1 and 2. In Table 1  For each of five sampling sites, the measured monthly averaged CHsBr mole fraction is in Table 2, listed for each month from January 1985 through October 1987. These monthly averages were calculated from all available data points for each month, generally two to eight points for each site. A full listing of the individual data points is to be published elsewhere and will be available upon request. Where •ND n appears in Table 2 I  I  I  I  I  I  I  I  I  I  I  I  I  I   0  2  4  6  8  I0 12  With this kind of smoothing of the data, best fit regression lines yield correlation coefficients of only 0.22 and 0.42, respectively. Similar graphs of 12-month running means for CHjBr at Mauna Loa, Samoa, and New Zealand all display similar positive-slope regression lines; slopes are 0.05, 0.08 and 0.05 ppt per month, respectively, with higher correlation coefficients of 0.55, 0.79 and 0.50. It should be noted that all five stations display more convincing positive trends if the data had stopped in June or earlier in 1987. This is because rather low CHsBr mixing ratios were measured at all five stations in the most recent months; see Table 2. When we analyze data only through May of 1987, the 12-month running mean graphs give slopes of +0.08 to +0.14 ppt per month, with correlation coefficients between 0.78 and 0.94.
A second way to look for possible CHsBr trends is to examine the deviations from the overall mean over the measurement period. In Figure 2 Table  1 shows that when we divided the data into the first half of the 34-month period and the second half, the differences between the means for the two subperiods were  Table 2. insignificant. Also, the analysis of residuals in Figure 3 shows