5.4. GC Measurements at Two Tall Towers in the U.S.

Automated, four-channel GCs have been in operation at the 610m WITN tower in eastern North Carolina (NC) since November 1994 and at the 447-m WLEF tower in northern Wisconsin (WI) since June 1996. Every hour these instruments measure 12 trace gases (CFC-11, CFC-12, CFC-113, CH₃CCl₃, CCl₄, CHCl₃, C₂Cl₄, N₂O, SF₆, H₂, CH₄, and CO) at 51, 123, and 496 m above ground on the NC tower and at 30, 76, and 396 m on the WI tower. The GCs are calibrated hourly with two standards of dried, whole air stored in Aculife-treated aluminum cylinders, one of which has been diluted by 10% with zero air. The design and operation of the GC at the NC tower were described previously [Elkins et al., 1996a; Hurst et al., 1997a].

Trace gas mixing ratios at the two towers are variable on diurnal, synoptic, seasonal, and longer time scales. Diurnal variations result primarily from the daily development of the planetary boundary layer (PBL), which defines the mixing depth of ground-based source emissions. At night, local and regional emissions augment mixing ratios beneath a shallow (100-200 m) inversion and create significant vertical gradients. During the late morning and afternoon, this vertical structure disappears as emissions are rapidly mixed into a 1-2 km deep PBL by convection [Hurst et al., 1997a, 1998]. Diurnal variability at the lower two sampling levels on each tower is greater than at the top because the nocturnal inversion consistently lies between the middle and top sampling levels.

Synoptic-scale variability is driven predominantly by the transport of pollution plumes from regional urban centers to the measurement sites. Daily mean mixing ratios (and daily standard deviations) at 496 m on the NC tower during 1996-1997 illustrate day-to-day (and diurnal-scale) variability (Figure 5.35). Daily means 5-10% above the majority of the data are regular features, especially for C₂Cl₄. Significant long-term trends in CH₃CCl₃ and SF₆ mixing ratios at the NC tower are also evident (Figure 5.35). Linear fits to regional “background” mixing ratios during 1996-1997 at the NC tower imply trends of -16.5 ± 0.5 ppt yr^-1 for CH₃CCl₃ and 0.24 ± 0.01 ppt yr^-1 for SF₆ [Hurst et al., 1997b] which are in good agreement with background trends at remote northern hemispheric sites (Figure 5.4) [Montzka et al., 1996; Geller et al., 1997].

Fig. 5.35. Time series of daily mean mixing ratios of CH₃CCl₃ (circles), C₂Cl₄ (crosses), and SF₆ (inverted triangles) at the NC tower during 1996-1997. The mean mixing ratio ± one standard deviation for each day are represented by vertical bars (solid for CH₃CCl₃ and SF₆, dashed for C₂Cl₄).

Synoptic-scale variability of trace gases at the NC tower was analyzed to identify regional-scale emission ratios [Bakwin et al., 1997]. C₂Cl₄ was chosen as the reference compound because of its high ratio of atmospheric variability to measurement precision at the NC tower and its reasonably well-known emissions [McCulloch and Midgley, 1996]. Several statistical approaches were used for the analysis, including a method where “scores” for each trace gas were computed as the sum of its mixing ratios during pollution events >4 hours in duration. For December 1994 through August 1996, 211 pollution events of 4 to 273 hours duration were identified [Bakwin et al., 1997]. Event scores for each trace gas were plotted against those for C₂Cl₄ and fit with a linear, orthogonal distance regression (Figure 5.36) to determine a regional-scale emission ratio. C₂Cl₄ correlated well with CFC-12, -11, CH₃CCl₃, SF₆, and several other gases (r >0.85). North American source strengths estimated from this “score” method and a time-domain analysis of the data are in good agreement with industry estimates for CH₃CCl₃ and CO but are 35-75% lower for CFCs. Discordance in CFC estimates may indicate that CFC emissions from the region surrounding the NC tower are not representative of North America (although CH₃CCl₃ and CO emissions appear to be) or that industry estimates of CFC emissions during 1995-1996 are too high. The latter scenario is supported by Elkins et al. [1993] and Cunnold et al. [1997] who demonstrated that global-scale observations of CFC-11 and CFC-12 mixing ratios during the 1990s are significantly lower than those calculated from emission inventories.

Fig. 5.36. Pollution event scores for CFC-12, CFC-11, CH₃CCl₃, and SF₆ versus scores of C₂Cl₄ at the North Carolina tower for December 1994 through August 1996. The y-axis of each panel represents event scores of the compound listed in the panel. The x-axes are scores for C₂Cl₄. Slopes of lines fit to each panel using an orthogonal distance regression reflect regional-scale emission ratios of the y-axis compound to C₂Cl₄ [from Bakwin et al., 1997].

The synoptic-scale variability of several halogenated trace gases recently decreased at both the NC and WI towers reflecting reductions in regional-scale emissions [Hurst et al., 1998]. Mixing ratio variability at all sampling heights on the towers was examined, but trends were deduced using only nighttime data from the top sampling level of each tower. This was done to minimize the influences of local sources and diurnal-scale variability, leaving regional emissions as the primary source of variability. Monthly variability, calculated as one standard deviation of monthly-binned mixing ratios with measurement errors subtracted in quadrature, was plotted as a time series and fit with a linear least-squares regression (Figure 5.37). Variability at the remotely-located WI tower was generally lower than at the NC tower, which is closer to large urban centers.

Fig. 5.37. Trends in monthly nighttime variability at the top sampling level of the North Carolina (NC) tower (circles) and Wisconsin (WI) tower (crosses). Variability was calculated as one standard deviation of nighttime mixing ratios during each month with random measurement noise removed. Lines fit to variability time series at the NC (solid) and WI (dashed) towers with least-squares regressions reflect trends in regional-scale emissions [Hurst et al., 1998].

Significant decreases in the synoptic-scale variability of CFC-12, CFC-113, CH₃CCl₃, and C₂Cl₄ were observed at both towers (Table 5.10). With the exception of CH₃CCl₃, variability trends at the two towers agreed to within their quoted uncertainties. The variability trend for CH₃CCl₃ at the NC tower, -1.31 ± 0.19 ppt yr^-1, represents a 72 ± 11% decrease in regional emissions between early 1995 and late 1997. Trends for CFC-11, CCl₄, and SF₆ at both towers did not differ significantly from zero. However, because of the low ratios of atmospheric variability to measurement precision for these gases, only trends >25% over the entire measurement period at each tower could have been detected at a 75% level of confidence. Reductions in CFC-12, CFC-113, and CH₃CCl₃ emissions are attributed to production restrictions imposed by the Montreal Protocol. Reduced emissions of C₂Cl₄ are probably the result of recent requests to industry by the U.S. Environmental Protection Agency (EPA) to voluntarily reduce emissions of this compound because of its toxicity.

TABLE 5.10. Trends in Nighttime Atmospheric Variability at the Top Sampling Levels of Tall Towers in North Carolina and Wisconsin

[BACK] [CONTENTS] [NEXT]