Observations made at a ground site east of Vancouver, Canada, were used in a principal component analysis (PCA) to derive (1) the emission ratios (ER) SO2, NOx, HNO2, HNO3, ketones, and aldehydes relative to CO and (2) the photochemical production efficiencies (PPE) of O-3, HNO2, HNO3, PAN, and several ketones and aldehydes relative to the formation of NOz. It is shown that derived ER(SO2) and ER(NOx) are consistent with the mobile emission sources in the Greater Vancouver Regional District (GVRD) emission inventory. Combining the ER data for other species with the GVRD CO emission inventory, the daily emission rates for these species have been estimated, indicating significant sources for these species. The derived PPE values for O-3, HNO2, HNO3, PAN, the ketones, and the aldehydes are as follows: PPE(O-3) is approximately 8-11 depending on the time of day; PPE(acetone) is between 1 and 2; PPE(MVK), PPE(MACR), and PPE(MEK) are approximately the same between 0.1 and 0.2; PPE(HCHO) is between 1 and 2, but PPE(propanal) and PPE(butanal) are lower than PPE(HCHO) by a factor of 2-10. It is shown that the daily photochemical production rates of O-3 and the carbonyl species are approximately linear functions of the NOx daily emission rates. When the photochemical production rates are compared with counterpart daily emission rates, it is shown that for the organic compounds, the contributions from photochemistry were more important than daytime emissions; for HNO2, there is more emission contribution than photochemistry, opposite to the case for HNO3. (C) 1997 Elsevier Science Ltd.

Daily air measurements at one midcontinental and two coastal sites on the Atlantic Ocean and Pacific Ocean in Canada during April 1992 to June 1993 are used to show the seasonality of methanesulfonate (MSA), non-sea-salt (nss) SO4=, and SO2. At the Atlantic site, Kejimkujik National Park, Nova Scotia (44 degrees 22'N, 65 degrees 12'W), the concentration ranges were <0.002-0.13, 0.2-26, and <0.2-10 mu g m(-3) for MSA, nss SO4=, and SO,, respectively, and the annual means were 0.02, 2.2, and 1.3 mu g m(-3). MSA had a monthly median peak of 0.036 mu g m(-3) in June, versus an SO2 peak of 2.7 mu g m(-3) in February, and no discernible nss SO4= seasonal peak. The MSA/nss SO4= ratios were low compared to ratios in other parts of the world but exhibited a seasonality dominated by that in MSA. At the midcontinental site, Experimental Lake Area, Ontario (49 degrees 39'N, 93 degrees 43'W), the concentrations were usually lower than at the coastal sites. The ranges were <0.002-0.066, 0.04-12.5, and <0.16-16.6 mu g m(-3) for MSA, nss SO4=, and SO2, respectively, and the annual means were 0.01, 1.6, and 0.93 mu g m(-3). The monthly median seasonal peaks at this site were 0.01 mu g m(-3) for MSA in September, 2.1 mu g m for nss SO4= in March, and 2.8 mu g m(-3) for SO2 in January. The MSA/nss SO4= ratio at this site was similar to that found at Kejimkujik. At Saturna Island, British Columbia (48 degrees 47'N, 123 degrees 08'W), the ranges were 0.002-0.2, 0.14-6.4, and 0.12-21 mu g m(-3) for MSA, nss SO4=, and SO2, respectively, and the annual means were 0.06, 1.1, and 3.0 mu g m(-3). The monthly median peaks were 0.12 mu g m(-3) for MSA in August, 5 mu g m(-3) for SO2 in February, and a m nominal peak of 1.5 mu g m(-3) for nss SO4= in August. The MSA/nss SO4= ratio had a seasonal pattern dominated by that of MSA with a peak of 0.1 in August. These data should enhance the existing databases for testing global sulfur models.

Observations of aerosol chemistry and microphysics were made at Chebogue Point, Nova Scotia, from August 16 to September 8, 1993 as part of the North Atlantic Regional Experiment (NAPE) intensive. Most of the aerosols were classified into two groups according to the geometric mean volume diameter (D-gv) of the particles which contributed the greatest volume (sub-0.5 mu m) The group-1 aerosols; representing 33% of the data, are characterized by D-gv of 0.18-0.19 mu m; the group-2 aerosols, representing 50% of the data, are characterized by D-gv of 0.20-0.22 mu m; and the remaining aerosols bear similarities to either groups 1 or 2 but lie outside the D-gv ranges. The differences between these aerosol groups are consistent with the addition of sulfate to the group-2 aerosols via recent processing through cloud. Factors supporting this possibility include the presence of low marine stratus upwind of the site only on days when the group-2 aerosol was observed, the higher D-gv for the group-2 aerosols consistent with the observed size threshold for activation in these clouds, and the association of non-sea-salt SO4= (nssSO(4)(=)) with larger particle sizes for the group-2 aerosols. In general, the masses of the most abundant inorganic and organic ions, nssSO(4)(=) and oxalate, were associated with the main volume of the sub-0.5-mu m particles. Cloud condensation nucleus (CCN) concentrations active at 0.4% supersaturation (CCN0.4) were highly correlated with the concentrations of particles >0.01 mu m and oxalate and moderately correlated with nssSO(4)(=). Concentrations of CCN active at 0.06% supersaturation (CCN0.6) correlate very well with the concentrations of particles >0.19 mu m diameter. In the case of the recently cloud-processed aerosols, for which nssSO(4)(=) is more strongly associated with particles >0.19 mu m, the CCN0.06 also correlate well with nssSO(4)(=) CCN spectra computed using the measured size distributions and aerosol chemistry agree well with the measured CCN.

Aerosol chemical and physical measurements were made at altitudes from 0.27 to 3 km near the coast of southern Nova Scotia, Canada, during the 1993 North Atlantic Regional Experiment. The volume distributions of aerosol with diameters between 0.005 and 3 mu m were dominated by accumulation mode particles. The mass and volume ratios (R(m) and R(v)) of the sum of soluble organics (SumOrg) to non-sea-salt (nss) SO4= were relatively invariant for estimated total particle mass (TPM) in excess of 13 mu g m(-3) (high TPM) but increased sharply with decreasing TPM below 13 mu g m(-3) (low TPM). Overall, the relationships between TPM and R(m) and R(v) were found to be R(m) = -(0.17 +/- 0.44) + (3.5 +/- 1.0)TPM-((0.69+/-0.39)) and R(v) = -(0.35 +/- 0.70) + (5.7 +/- 1.7)TPM-((0.68+/-0.39)). The high TPM aerosols originated in eastern North America and had average composition of 46% nssSO(4)(=), 8% NH4+, and 8% SumOrg. In contrast, low TPM aerosols were found to be of background continental or marine tropospheric origins and had average composition of 23% nssSO(4)(=), 9% NH4+, and 20% SumOrg. The aerosols in both TPM regimes were separated into two groups based on the mode of the volume distributions. The correlation between the mass of each species and the particle volume distribution was investigated for these groups.

Airborne observations from 14 flights in marine stratus over the Gulf of Maine and Bay of Fundy in August and September of 1993 are examined for the relationships among the cloud droplet number concentrations (Nd), the out-of-cloud aerosol particle number concentrations (N-a), the major ion concentrations in the cloud water, and turbulence in cloud. There was a wide range of aerosol concentrations, but when low stratus and the main anthropogenic plume from eastern North America were in the same area the plume overrode the cloud. The N-d increased with increasing N-a and cloud water sulfate concentration (cwSO(4)(=)), but the relationships were very weak. The separation of the data between smooth and lightly turbulent air substantially improved the ability to explain the variance in the N-d by either of these two quantities. Also, the relative increase in N-d for increases in N-a and cwSO(4)(=) was greater for lightly turbulent air than for smooth air. The estimated minimum size of particles activated in these clouds ranged from 0.14 mu m to 0.31 mu m, corresponding to average supersaturations of <0.1%. The minimum size tended to be lower for lightly turbulent air and smaller N-a. The results for lightly turbulent air agree well with previously reported parameterizations of the impact of aerosols on N-d, but the results for smooth air do not agree. In general, more knowledge of the physical factors controlling the N-d in stratiform clouds, such as turbulence, is needed to improve not only our ability to represent N-d but also to increase our understanding of the impact of the aerosol particles on the N-d and climate.

A new technique for condensing detailed organic reactions into more compact forms is introduced, and compared to methods which have been used in the ADOM and RADM air quality models. The previous methods made use of integrated reactivity weighting, designed to predict the correct product concentrations on time scales of one to two days or longer. The latter methods are shown to be subject to high errors in the prediction of initial hydrocarbon product concentrations. These errors occur when the hydrocarbon chemistry is the most reactive. The new method is a hybrid scheme which uses the reaction rates at the initial time as a further constraint on the lumping. This results in improved prediction of product concentrations on shorter time scales while retaining the long-term accuracy of the earlier techniques. The hybrid reactivity weighting technique is shown to have higher accuracy than the earlier ADOM or RADM techniques in the regions of the most reactive chemistry, and is therefore very relevant to regional oxidant models.

Using a principal component analysis technique and data on atmospheric gases and aerosols at a rural site in Ontario, Canada from the Eulerian model evaluation field study (EMEFS), the Eulerian acid deposition and oxidant model (ADOM) is evaluated. Seventy-nine and 76% of the variances in the data and model output, respectively, are explained by three principal components. They are a chemically aged/transported component, a diurnal cycle component, and an area emission component, all characterized by their ratios of gases and temporal variation patterns. The ADOM component contributions to sulphur species are in general agreement with the EMEFS components, but with notable differences for key photochemical species including O-3. The temporal variations of the ADOM components are close to those of the EMEFS components. The EMEFS chemically aged/transported component shows a high degree of photochemical processing, with the ratios [NOx]/[TNOy]=0.3 and [O-3]/([TNOy]-[NOx])=9+/-1. The corresponding ADOM component predicts lower [NOx]/[TNOy] and [O-3]/([TNOy]-[NOx]) ratios, probably caused by a chemical mechanism in the model that is too fast, and lower contributions to O-3, NO2, TNO3, PAN, TNOy, and HCHO, probably caused by model grid dilution or lower model emissions. The EMEFS diurnal component owes its variance to the daily photochemistry and nighttime dry deposition of the chemical species. In comparison, the matching ADOM component underpredicts the ratio [O-3]/([TNOy]-[NOx]) and the NO2 consumption and O-3 production but overpredicts the contributions to the other species. The EMEFS emission component represents emissions from local/regional area sources. The corresponding ADOM component underpredicts TNOy by 44% and the fraction of TNOy as NOx compared to the EMEFS component, suggesting that the model has lower emissions of NOx and a photochemical mechanism that converts NOx faster than indicated by the EMEFS results.

Based on atmospheric measurements of multiple species at Egbert, a rural site in Ontario, Canada, during summer 1988, the emission ratios of HCHO/CO and HCHO/SIGMANO(y) for area sources and secondary production of HCHO have been estimated using a modified principal component analysis technique. The technique yields three principal components that represent a photochemically aged air mass, a diurnal cycle, and fresh area emissions. The area emission component has an emission ratio CO/SIGMANO(y) = 9 +/- 3 and SO2/CO = 0.005 +/- 0.003, in agreement with NAPAP area emission data for the eastern US [Buhr et al., 1992]. The emission ratios of HCHO/CO and HCHO/SIGMANO(y) in this component are 0.0056 +/- 0.0022 and 0.05 +/- 0.007, respectively. If these ratios are typical of eastern North American area emissions, the total primary HCHO emission for this region will be 8 x 10(9) moles HCHO annually based on the NAPAP CO emission inventories. Evidence of secondary HCHO production can be found in the photochemically aged component which has considerably higher HCHO/CO (0.016 +/- 0.004) and HCHO/SIGMANO(y) (0.29 +/- 0.03) ratios than the emission ratios. It is estimated that for every 1 ppb NO(x) converted to NO(y), 0.4 ppb HCHO are produced for the ratio (1-NO(x)/NO(y))<0.6; after which the relative HCHO production rate becomes smaller. Using this relative rate, the maximum total HCHO production over the eastern North America is estimated to be 1.3 x 10(11) moles year-1, or approximately 16 times that from primary emission.

Surface ozone, particulate bromine and inorganic and organic gaseous bromine species were measured at Barrow, AK, during March and April 1989 to examine the causes of surface ozone destruction during the arctic spring. Satellite images of the Alaskan Arctic taken during the same period were also studied in conjunction with calculated air mass trajectories to Barrow to investigate the possible origins of the ozone-depleted air. It was found that during major ozone depletion events (O-3 < 25 ppbv) concentrations of particulate bromine and the organic brominated gases bromoform and dibromochloromethane were elevated. Air mass trajectories indicated that the air had crossed areas of the Arctic Ocean where leads had been observed by satellite. The transport time from the leads was typically a day or less, suggesting a fast loss mechanism for ozone. A similarly fast production of particulate bromine was shown by irradiating ambient nighttime air in a chamber with actinic radiation that approximated daylight conditions. Such rapid reactions are not in keeping with gas-phase photolysis of bromoform, but further studies showed evidence for a substantial fraction of organic bromine in the particulate phase; thus heterogeneous reactions may be important in ozone destruction.

In six aircraft flights of AGASP-II, 2-15 April 1986, from ca. 1300-8100 m altitude, the most abundant elements measured in size separated aerosol samples were silicon, chlorine, and sulfur. Concentrations were higher than at ground level (G), particularly at highest altitudes (HT, 5600-8100 m, upper troposphere to lower stratosphere) compared to mid troposphere (MT, 1300-4700m), especially for ultrafine particles <0.0625 mu m aerodynamic diameter. HT and MT median and G average concentrations, mu g m(-3) STP, respectively (1) Si = 3.64, 1.30, 0.092; (2) S = 1.44, 0.265, 0.087; (3) Cl=1.62, 0.36, 0.213. The weight ratio Al/Si was less than half that expected for Earth crust material (0.3), evidence against fine silicon originating mainly by dispersion of volcanic debris or other eolian dust particles. Instead, pollution from high rank (mainly bituminous) coal combustion, which can form SiO vapors from quartz in the ash and fine alkaline aerosol with low Al/Si ratio, is a more likely source of apparently widespread aerosol silicon contamination of the Arctic atmosphere. Chlorine and sulfur gases may be scavenged by coarse alkaline dust particles and acidic chlorine and sulfur may be derived from coal combustion processes, thus also accounting for their high concentrations.

In the spring of 1986 and 1989, particle nitrite was measured at Barrow, Alaska, by filter sampling and by ion chromatographic analysis. Particle nitrite concentrations averaged 2.9 +/- 3.4 and 2.6 +/- 2.0 ppt (molar ratio) in 1986 and 1989, respectively. Both seasons showed diurnal variations with higher concentrations during the day which might have been caused by daytime down mixing. In 1989, nitrite was determined in several snow samples with concentrations between 0 and 0.18 mu mol l(-1). Particle nitrite was probably in disequilibrium with gas phase, suggesting a heterogeneous source for gaseous HONO. A relationship between particle nitrite and sodium ions suggests that sea salt could be involved in nitrite formation, perhaps through hydrolysis of nitryl halides.

Interpretation of simultaneous measurements at three stations in different parts of the Arctic suggests that during winter air masses are forced into the Arctic from Eurasia in a surge towards Alaska and further return over the North Pole towards the European Arctic. On some occasions direct flow of the Eurasian air masses was detected in the European Arctic. Simple statistical methods and dispersion modelling proved useful in studying source-receptor relationships in the Arctic.

Measurements in the Arctic troposphere over several years show that MSA concentrations in the atmospheric boundary layer, 0.08-6.1 parts per trillion (ppt, molar mixing ratio), are lower than those over mid-latitude oceans. The seasonal cycle of MSA at Alert, Canada (82.5 degrees N, 62.3 degrees W), has two peaks of 6 ppt in March-April and July-August and minima of 0.3 ppt for the rest of the year. At Dye 3 (65 degrees N, 44 degrees W) on the Greenland Ice Sheet, a similar seasonal MSA cycle is observed although the concentrations are much lower with a maximum of 1 ppt. Around Barrow, Alaska (71.3 degrees N, 156.8 degrees W), MSA is between 1.0 and 25 ppt in July, higher than 1.5+/-1.0 ppt in March-April. The mid-tropospheric MSA level of 0.6-1 ppt in the summer Arctic is much lower than about 6 ppt in the boundary layer. Al Alert, the ratio of MSA to non-sea-salt (nss) SO42- ranges from 0.02 to 1.13 and is about 10 times higher in summer than in spring. The summer ratios are higher than found over mid-latitude regions and, when combined with reported sulfur isotope compositions from the Arctic, suggest that on average a significant fraction (about 16-23%) of Arctic summer boundary layer sulfur is marine biogenic. The measurements show that the summer Arctic boundary layer has a significantly higher MSA/nss-SO42- ratio than aloft.

Eight water-soluble organic anions were measured in 70 aerosol samples and 10 snow samples at Barrow, Alaska in March-April, 1989. The ranking of the ions in aerosols according to total (coarse+fine aerosol) median concentrations was acetate (44 ng m-3), oxalate (27), benzoate (23), formate (22), propionate (6), methanesulfonate (5), lactate (4), and pyruvate (4). When added up, the median organic anion mass was 156 ng m-3. The organic anions/nssSO4= mass ratio had a median of 0.18 and 0.07 in the coarse (>l mum) and fine (<1 mum) size fractions, respectively, but can be very high on occasions. On average, the organic anions made up more than 10% of the water-soluble aerosol mass. A similar ranking in concentration was also found for the organic ions in the snow pack samples. The organic anion/nssSO4= mass ratio in these samples was >0.5, substantially higher than in aerosols.