Industrial sources influence air concentrations of hydrogen sulfide and sulfur dioxide in rural areas of western Canada.

ABSTRACT

A survey of monthly average concentrations of sulfur dioxide (S[O.sub.2]) and hydrogen sulfide ([H.sub.2]S) at rural locations in western Canada (provinces of Alberta, British Columbia, and Saskatchewan) was conducted in 2001-2002, as part of an epidemiological study of the effects of oil and gas industry emissions on the health of cattle. Repeated measurements were obtained at some months and locations. We aimed to develop statistical models of the effect of oil and gas infrastructure on air concentrations. The regulatory authorities supplied the information on location of the different oil and gas facilities during the study period and, for Alberta, provided data on [H.sub.2]S content of wells and flaring volumes. Linear mixed effects models were used to relate observed concentrations to proximity and type of oil and gas infrastructure. Low concentrations were recorded; the monthly geometric mean was 0.1-0.2 ppb for [H.sub.2]S, and 0.3-1.3 ppb for S[O.sub.2]. Substantial variability between repeated measurements was observed. The precision of the measurement method was 0.005 ppb for both contaminants. There were seasonal trends in the concentrations, but the spatial variability was greater. This was explained, in part, by proximity to oil/gas/bitumen wells and (for S[O.sub.2]) gas plants. Wells within 2 km of monitoring stations had the greatest impact on measured concentrations. For [H.sub.2]S, 8% of between-location variability was explained by proximity to industrial sources of emissions; for S[O.sub.2] this proportion was 18%. In Alberta, proximity to sour gas wells and flares was associated with elevated [H.sub.2]S concentrations; however, the estimate of the effect of sour gas wells in the immediate vicinity of monitoring stations was unstable. Our study was unable to control for all possible sources of the contaminants. However, the results suggest that oil and gas extraction activities contribute to air pollution in rural areas of western Canada.

INTRODUCTION

According to Environment Canada, the industrial sectors responsible for the largest emissions of sulfur dioxide (S[O.sub.2]) in Canada are smelting of metal concentrates and power generation. In Alberta, the industrial activities leading to the largest releases of S[O.sub.2] are upstream oil and gas activities (which include natural gas processing), electric power generation, and oil sands activities. Among Canadian provinces and territories, Alberta is the second largest emitter of S[O.sub.2], with emissions of 608 kt in 1995 and 548 kt in 1999. Alberta's S[O.sub.2] emissions represented approximately 21% of Canada's total S[O.sub.2] emissions. (1,2)

Considerable controversy surrounds the issue of the impact of low-level emissions from normal operations (rather than operation upsets such as spills and blow-outs) in primary oil and gas facilities on animal and human health in western Canada (reviewed by Scott et al. (3)). The principal emissions of concern include sour gas (hydrogen sulfide [[H.sub.2]S]), but S[O.sub.2] has often been measured as its surrogate. However, it is not certain that the use of S[O.sub.2] as a surrogate for [H.sub.2]S adequately reflects sour gas emissions because there are more environmental sources of S[O.sub.2] than [H.sub.2]S. Concentrations of S[O.sub.2] per se are a commonly used air quality indicator in environmental epidemiology. In response to the earlier animal health studies by Scott et al. (3-5) and Waldner et al. (6-8) the Western Interprovincial Scientific Studies Association (WISSA) was formed and initiated a study to evaluate the impact of routine emissions from oil and natural gas field facilities on animal health. As part of the study, exposure measurements for multiple agents have been collected at fixed locations throughout the Canadian provinces of British Columbia, Alberta, and Saskatchewan. RWDI Air Inc., a firm contracted by WISSA, collected samples. WISSA oversaw the design, funding, and implementation of the overall project, but only provided funding for the work that led to results presented in this manuscript.

The final report on the epidemiological study is freely available from http://www.wissa.info (under "WISSA Study Reports-May 18, 2006"). Briefly, it was observed that [H.sub.2]S and S[O.sub.2] were not associated with most studied health outcomes in cattle. However, mortality in the first 90 days of life was slightly (and statistically significantly) elevated among calves in the highest prenatal (3 months before calving) S[O.sub.2] exposure category (>1.3 ppb vs. [less than or equal to]0.7 ppb), with a significant dose-response trend (pages 8.38 and 8.55 in the report). (9)

The specific aims of this article are to describe the levels of air concentrations and identify factors that predict concentrations of [H.sub.2]S and S[O.sub.2] in the air.

METHODS AND MATERIALS

Sampling Strategy

The sampling aimed to assess exposure to [H.sub.2]S and S[O.sub.2] in cattle in three western Canadian provinces during a period of time critical to the animals' reproductive success. Air samplers were located wherever cattle from the study herds were managed or pastured from April 2001 (205 herds) to June 2002 (203 herds) and, for a subset of 50 herds, to December 2002. S[O.sub.2] was measured from April 2001 to December 2002, and [H.sub.2]S was measured from September 2001 to December 2002, because of the delays in development and implementation of [H.sub.2]S monitoring technology. The distributions of [H.sub.2]S and S[O.sub.2] sampling sites are shown in Figures 1 and 2, respectively. Sampling devices were located to account for all "management groups" within herds. In a random ([greater than or equal to]10%) sample of locations, replicate measurements were collected for a given month. Field blanks were used to identify shipping and handling practices that may influence measurements. For every 10 samples, one field blank was collected and analyzed.

[FIGURE 1 OMITTED]

All monitors were set 1.5-1.8 m above the ground, at sites chosen according to the following criteria: (1) adjacent to areas within the pasture where study cattle spent most of their time; (2) away from farm equipment operated by internal combustion engines, fuel and farm equipment storage areas; (3) more than 10 m from roadways and other areas where vehicles are expected (e.g., gates where vehicles stopped and idled); (4) more than 100 m from fuel and farm equipment storage areas; (5) outside the immediate area of local oil and gas facilities, to avoid "worst-case" sampling; (6) at least 20 m from the nearest tree canopy, as defined by drip line; (7) away from buildings, hay storage, and other objects that may obstruct airflow; and (8) in flat terrain (i.e., not at the tops or bottoms of hills). Each monitoring site was described in a site documentation database.

[FIGURE 2 OMITTED]

There were 116-914 [H.sub.2]S and 115-928 S[O.sub.2] monitoring sites (some with more than one measurement) per month. The numbers of monitoring sites and measurements peaked in summer and declined in winter, primarily because monitoring sites tracked the movement of cattle herds, which were dispersed into a number of management groups (subsets of herd) at different locations during the summer and concentrated in a small number of locations during the winter. The number of sites sampled also declined toward the end of the survey because of budgetary constraints. For S[O.sub.2], the proportion of sites with repeated measurements was approximately 90% until August 2001, but then declined to approximately 10%. Repeated [H.sub.2]S measurements were collected at 10% of locations within each month of the study.

Air Quality Measurements

Monthly average S[O.sub.2] and [H.sub.2]S concentrations were measured using PASS S[O.sub.2] and [H.sub.2]S passive monitors manufactured by Maxxam Analytics Inc., which also analyzed the samples. All samples were shipped in containers sealed with Teflon tape for analysis. The S[O.sub.2] sampling medium was a filter impregnated with sodium carbonate/sodium bicarbonate. The sulfate ion was extracted from the medium with a solution of hydrogen peroxide in ultrapure distilled/deionized water and ion chromatography, following U.S. Environmental Protection Agency method 300.1. Tang et al. (10) provide additional information for this method. The [H.sub.2]S sampling medium was a silver nitrate (AgN[O.sub.3])-impregnated filter. [H.sub.2]S was extracted using a sodium hydroxide/sodium cyanide solution and the resulting silver sulfide and/or AgSH was determined by a fluorometric procedure to measure sulfide ([S.sup.-2]). (11,12)