Characteristics and Sources of Carbonaceous, Ionic, and Isotopic Species of Wintertime Atmospheric Aerosols in Kathmandu Valley, Nepal

To investigate the air pollution from aerosols in Kathmandu during winter, bulk aerosol samples were collected during winter 2007–2008 to characterize carbonaceous and ionic species and carbon and nitrogen isotopes. This study illustrates the applications of carbon and nitrogen isotope data for characterizing aerosols and their implications for identifying sources that were inconsistent with the results for the carbonaceous and ionic aerosols. Mean concentrations of organic carbon (OC), elemental carbon (EC), and water soluble organic carbon (WSOC) in Kathmandu during the period were 20.02 ± 6.59 (1σ), 4.48 ± 1.17, and 10.09 ± 3.64 μgC/m, respectively. Elemental carbon and OC were correlated (R = 0.56), likely indicating common sources for both species, as well as for the precursors that led to the formation of secondary organic carbon (SOC). The mean estimated SOC contribution to OC was 31%, suggesting that local emission is more important than transport and processing during winter in Kathmandu. On average, 50% of the OC was water soluble, and the correlation of SOC with WSOC (R = 0.66) suggests that the majority of SOC and some primary organic carbon (POC) were water soluble in Kathmandu. The mean δC of –25.74 ± 0.19‰ observed in aerosols of Kathmandu confirms consistent anthropogenic sources such as fossil fuel combustion. Heavier carbon also was observed to be associated with the water-soluble fraction of OC in aerosols. The mean δN of 9.45 ± 0.87‰ suggests the limited influence of biomass burning and its strong correlation with crustal cations Ca (R = 0.74, p < 0.05) and Mg (R = 0.71, p < 0.05) indicates distant sources. Principal component analysis revealed four major sources/pathways for particles: local and vehicular emissions, secondary gas-to-particle conversion, aqueous processing, and dust transport, each explaining ~39, 23, 11, and 9% of the variance.


INTRODUCTION
Respiratory ailments, including cough and bronchitis (Folinsbee, 1992), have been associated with particulate matter (PM), which also affects climate directly and indirectly and reduces visibility.Visibility reduction and the direct effect are related to absorption and scattering of radiation by aerosols (Charlson, 1969;Charlson et al., 1992).The indirect effect is related to alteration of cloud albedo (Twomey, 1974) and lifetime through particles acting as cloud condensation nuclei and ice nuclei.
Carbonaceous aerosol species include elemental carbon (EC) and organic carbon (OC), where this distinction typically is based on thermal analysis (Birch and Cary, 1996).However, it should be stressed that EC is not equivalent to black carbon (BC), which is defined and measured based on optical properties and absorbs light more strongly than EC.Elemental carbon is a primary pollutant emitted directly from combustion sources that absorbs light efficiently (because a significant fraction is BC) and has a direct effect on the radiative balance of the Earth surface.Organic carbon aerosol is emitted directly (known as primary OC or POC) and formed secondarily (known as secondary OC or SOC) from the partitioning to the condensed phase of semi-or non-volatile products of the oxidation of volatile organic compounds or by aqueous-phase processing.POC and SOC are generated from both anthropogenic and biogenic sources (Jacobson et al., 2000).
In addition to carbonaceous material, aerosols also include inorganic species, often as cations and anions in an aqueous phase.For example, oxidation of sulfur dioxide and nitrogen oxides leads to the formation of sulfate (SO 4 2-) and nitrate (NO 3 -) in aerosols via sulfuric and nitric acids, respectively.Ammonium (NH 4 + ) ions buffer these acidic species, which typically exist in ambient aerosol as partially or fully neutralized salts.Inorganic species may also be emitted directly to the atmosphere in the particulate phase.
Stable carbon isotope ratios of 13 C to 12 C (δ 13 C) reveal the chemical fractionation associated with processes involved in PM formation such as photosynthesis, atmospheric oxidation (Jacobson et al., 2000), and phase changes.This ratio also reveals the relative contributions of C 3 (Calvin-Benson cycle) or C 4 (Hatch-Slack cycle) plants (Cachier, 1989).Similarly, nitrogen isotope ratios can be used to characterize atmospheric aerosol (Turekian et al., 1998).
Kathmandu, the capital of Nepal, has a land area of 395 km 2 , a rapidly increasing population of 1.08 million, and a vehicle usage rate that is growing by approximately 10 percent per year (CBS, 2005;Faiz et al., 2006).The city is located inside a valley with restricted free wind movement, resulting in poor air quality, especially with regard to PM.For example, Aryal et al. (2009) observed concentrations of PM with diameters smaller than 2.5 micron (PM 2.5 ) of 90 ± 24 (1σ) µg/m 3 during the 2006-2007 winter in Kathmandu.
The main objective of this study is to characterize the winter ambient aerosols of Kathmandu in terms of OC, EC, watersoluble organic carbon (WSOC), inorganic ions, and isotopes of carbon and nitrogen.

Sampling
Sampling was performed on the roof of a five-story building in urban Kathmandu where air quality likely is influenced mainly by vehicular emissions, as the vehicular traffic in the nearby roads remains heavy throughout the day.The sampling site is located in Sorhakhutte, Thamel, a main tourist area in Kathmandu.The volume of traffic on the roads of Kathmandu is estimated to be 100 to 500 vehicles per hour (CBS, 2009).
Bulk aerosols were collected on quartz fiber filters (Whatman, QM-A) using a pump with a flow rate of 50 liters per minute.To reduce background contamination, all filters were pre-baked at 600°C for 24 hours and Teflon®sealed in polyethylene dishes prior to sampling.The samples were collected for periods ranging from 18 to 24 hours depending on scheduled power outrages and rain events.Collected samples were also stored inside dishes that were Teflon®-sealed.Twenty-five samples and eight field blanks were collected from 26 December 2007 through 30 January 2008.The field blanks were collected after every two subsequent sample collections, and all data described subsequently are corrected by subtracting average values of field blanks.The storage, transportation, and analysis of field blanks were identical to those of samples.All the samples were stored in a freezer prior to shipment to the University of New Hampshire for analysis.

EC and OC
Elemental and organic carbon were analyzed by a thermal/optical method using a 1.5-cm 2 filter punch in a Sunset EC/OC analyzer, as described by Birch and Cary (1996).Instrumental calibration was performed using succinic and phthalic acid standards.The associated uncertainties, calculated as two times the standard deviation of the eight field blanks using the average volume of air drawn for the sampling, were 0.24 and 0.12 µgC/m 3 for OC and EC, respectively.

Secondary Organic Carbon (SOC)
Because EC can be used as an indicator of primary combustion emissions, the smallest observed ratio of OC to EC, (OC/EC) minimum , provides an estimate of a typical ratio of POC to EC assuming no or very low contribution from non-combustion POC and residual SOC (Castro et al., 1999).This POC/EC ratio is assumed to be constant for a specific location for a specific period of time (Turpin and Huntzicker, 1991).The secondary organic carbon (SOC) is then defined as: where POC is the product of the observed EC and (OC/EC) minimum .The contribution of SOC in the sample with minimum OC/EC is assumed to be negligible; the main source for EC and POC in this sampling site is assumed to be the emissions from fossil fuel combustion.Noncombustion related POC is assumed to be negligible (Castro et al., 1999).
Water Soluble OC (WSOC) Three punches, each 1.5 cm 2 , were cut from each filter and extracted with 15 mL of milli-Q water.The extract was shaken manually, allowed to sit passively for 10 minutes, and centrifuged for 10 minutes to separate the filter and aqueous solution.The aqueous solution was divided into aliquots, each of 5 mL.One aliquot was used to analyze WSOC using a total organic carbon analyzer (Model 800 TOC, Sievers).The related uncertainty, calculated as two times the standard deviation of the eight field blanks, using the average volume of air drawn for the sampling was 0.65 µgC/m 3 .

Statistical Analyses
The statistical software package JMP version 7.0.2 was used for statistical analyses.These analyses included linear regression and a principal component analysis (PCA) including factor rotation.

Meteorological Data
Meteorological data (maximum and minimum temperature and relative humidity (RH)) were taken from the archived report of the meteorological forecasting division of the Nepalese government (MFD, 2008).These parameters are included in the PCA.Temperature data are included in Fig. 1.

OC and EC
The daily concentrations of EC and constituents of OC during the sampling campaign are shown in Fig. 1 and summarized in Tables 1 and S1.Mean concentrations of OC and EC (with its standard deviation) were 20.02 ± 6.59 and 4.48 ± 1.17 µgC/m 3 , respectively.On average, OC accounted for approximately 81% of the carbonaceous aerosol.Overall, there was a decreasing trend for both EC and OC during the sampling period.Despite a stronger decreasing overall trend (Fig. 1) in OC, OC and EC showed a correlation with R 2 of 0.56 (Fig. 2), suggesting a common source for EC and OC.The OC/EC ratio ranged between 3.01 and 6.59, with an average of 4.47 ± 0.93.

SOC
Using the observed (OC/EC) minimum of 3.01, which occurred on the day following the one significant rain event during the sampling period, Eq. ( 1) yields SOC concentrations in the range of 0 to 18.45 µgC/m 3 , with an average of 6.81 ± 4.48 µgC/m 3 .Daily POC and SOC concentrations are also shown in Fig. 1 and Tables 1 and S1.On average, SOC constituted 31% of OC, with POC accounting for the remaining 69%.Overall SOC also has a decreasing trend during the sampling campaign.Total organic aerosol mass concentrations were calculated by summing the POC and SOC after multiplying each by 1.2 and 2.0, respectively (Turpin and Lim, 2001).The mean organic aerosol mass concentration was 29.26 ± 11.24 µg/m 3 .
The average ionic concentration was 9.06 ± 2.49 µg/m 3 , and the mass of NH 4 + , SO 4 2-, and NO 3 -accounted for 64% of the total observed ionic mass concentration.Ammonium was correlated to both SO 4 2-(R 2 = 0.48, p < 0.005) and NO 3 -(R 2 = 0.39, p < 0.001).There was very little difference between median and mean.On average, ions accounted for 23% of the total observed aerosol concentration (ions and OM).On average, the aerosol [H + ] (Fig. 6) was calculated to be 49.99 ± 18.66 nmol/m 3 and exhibited a large range of 9.14 to 84.32 nmol/m 3 .

Carbon and Nitrogen Isotopes
The average δ 13 C in the samples was -25.74 ± 0.19‰, and ranged from -26.05 to -25.51‰.The average δ 15 N was 9.45 ± 0.87‰, and it varied from 8.13 to 10.56‰ (Tables 1  and S1, Fig. 7).Isotopic content is compared to that measured elsewhere in Table 2.

OC and EC
During the study period, OC exhibited greater temporal variability than EC, where Fig. 1 indicates an overall decreasing trend in OC.A small decreasing trend in EC suggests, in general, consistent local primary emissions sources of aerosols and boundary layer dynamics.Because the majority of OC is POC, total OC follows trends similar to those for POC and EC.The strong correlation between EC and OC (R 2 = 0.56, p < 0.0001) at the sampling site likely indicates a common source for EC and OC including the precursors that lead to SOC.
The average EC and OC of 4.48 and 20.02 µgC/m 3 , respectively, in Kathmandu are comparable to those measured during winter in other urban Asian cities.Average EC and OC during winter of 2005 in Guangzhou, China, were 4.2 and 23.9 µgC/m 3 and in Hong Kong were 2.5 and 12.4 µgC/m 3 , respectively (Duan et al., 2007), and during winter of 2006 in Xi'an, China were 23.7 and 4.6 µg/m 3 , respectively (Shen et al., 2009).Carrico et al. (2003) observed EC of 1.0 ± 0.7 µgC/m 3 during October 1999 to January 2000 in Nagarkot, a site inside Kathmandu valley that is approximately 20 kilometers from the site of the current study.
In general, OC to EC ratios depend on emission sources and secondary organic aerosol formation.Hildemann et al. (1991) found OC/EC in fine particles of 2.2 for light-duty gasoline vehicles and 0.8 for heavy-duty gasoline vehicles.The OC/EC values in Kathmandu were larger than these ratios on all days.Cao et al. (2005) found an average OC/EC ratio of 4.1 for vehicle exhaust, a value close to that observed in Kathmandu.Thus, the common source is most likely vehicular emissions.The variation of OC/EC during the sampling period and the values larger than 3.01 indicate the importance of SOC, assuming no fluctuations in emission ratios.
Emission from vehicles appears to be one of the most important sources of aerosols.Kondo et al. (2005) also concluded that the transportation-related emissions are the main sources of air pollution in Kathmandu valley, which is further corroborated by the observations of Panday and Prinn (2009) that showed peaks in concentrations of PM 10 during the mornings from 2004 to 2005.Further, they observed more concentrated air pollution during winter compared to summer.In Kathmandu, numerous poorly maintained and old vehicles and low-grade and adulterated fuel are often the norms (Faiz et al. 2006).Poor road conditions also contribute to emissions related to vehicle use.The total length of road in Kathmandu is 1279.09km, of which only about 40% is paved (CBS 2009).In 2005, PM 10  Open waste burning practices are also an important source contributing to the aerosol loadings in Kathmandu.Refuse burning was estimated to emit approximately 172 tons of PM 10 in 2005 in Kathmandu valley (ICIMOD, 2007).In addition, protests called by various groups often include the burning of vehicle tires on the streets, emitting CO, SO 2 , and NO 2 (Shakya et al., 2008); uncontrolled tire fires may also be significant aerosol sources in Kathmandu.

SOC
The average observed percent contribution of secondary OC to carbonaceous aerosol in Kathmandu (31%) is smaller than but comparable to values observed in other urban Asian cities, though it should be noted here that the values presented in this study likely represent a lower bound because some SOC could have been present during the period associated with the minimum OC/EC ratio.Lin and Tai (2001) reported that the SOC contribution to OC was larger in PM 2.5 (40%) compared to PM 10 (32.4%) during November 1998 to April 1999 in Kaohsiung City, Taiwan.In Guangzhou, China, the average SOC concentration during winter 2005 was 11.5 µgC/m 3 (Duan et al., 2007), which is larger than that observed in Kathmandu.During winter 2006-2007in Tianjin, China, Li et al. (2009) observed SOC contribution to OC of 37.3-50.3%compared to mean of ~34% observed in Kathmandu.The majority of OC was primary in nature in Kathmandu, comparable to the measurements made in the San Joaquin Valley of California, where secondary contributions to OC were smaller in winter due to decreased mixing heights and photochemistry (Strader et al., 1999).The smaller percentage of SOC relative to OC suggests that motor vehicles and other local emissions (as opposed to transport and processing) are the major factors controlling concentrations of carbonaceous aerosols.However, the strong relationship between OC and EC (R 2 = 0.56) indicates that any SOC likely was derived from volatile organic compounds co-emitted with POC and EC.

WSOC
The correlation of WSOC with SOC (R 2 = 0.66 and slope of approximately unity, Fig. 4) indicates the likelihood that a large fraction, if not all, of the SOC is WSOC.Like OC, there was an overall decreasing trend in WSOC during the sampling; no such trend was observed for WSOC/OC.WSOC/OC was observed to be the smallest (0.39) during the rain event (4.9 mm) on 19 January 2008, likely indicating favorable wet deposition of WSOC and decreased photochemical processing.WSOC did not show any significant correlation with SO 4 2-and only a weak correlation with NO 3 -(R 2 = 0.31, p < 0.005).The lack of significant correlation of WSOC with SO 4 2and NO 3 -indicates that WSOC likely is formed primarily via pathways that are different than those for SO 4 2-and NO 3 -.This is in stark contrast to the strong correlation between WSOC and SO 4 2-and NO 3 -observed in urban and suburban sites in Nanjing, China (Yang et al., 2005).
The non-zero intercept (i.e.-3.57) in Fig. 4 underscores that some fraction of the POC in Kathmandu is likely water-soluble.If it is assumed that all SOC is WSOC, the fraction of POC that is WSOC can be found by (WSOC-SOC)/POC.This value, on average, is 0.26, likely a lower bound because not all SOC will definitively be WSOC.In Houston, Anderson et al. (2008) also noted an increase in particulate WSOC associated with rush hour, suggesting a primary source for WSOC.A large WSOC fraction is usually assumed to be indicative of processed aerosol or SOA (Sullivan et al., 2004;Weber et al., 2007).The average ratio of WSOC to OC (~0.18) in Lahore, Pakistan during spring 2006 with very large average PM 10 concentrations (459 µg/m 3 ) mainly as a result of vehicular activities (Zhang et al., 2008) was smaller than the observed ratio of ~0.50 in the present study.

Water Soluble Inorganic Ions
Because K + , a tracer of biomass burning, was found in small concentrations, biomass burning seems to have a small influence on aerosol mass loadings in Kathmandu in winter.However, K + increased significantly after 25 January 2008, possibly because of the influence of biomass burning after this day.The average K + /EC before 01/25/08 was very small (0.03).A K + /EC ratio of 0.03 was also observed in a location influenced by vehicular emissions in Auckland, New Zealand (Wang et al., 2005).The average ratio of K + to total ionic aerosols was less than 0.14%, compared to 7.3% observed during winter 2006-2007 in Tianjin, China (Li et al., 2009).However, the ratio increased unusually to ~24% on 28 January 2008.Fires made by street vendors for heat during winter and open refuse burning may have played a role in increased K + concentrations at the end of the campaign.
The strong correlation of Ca 2+ with Mg 2+ (R 2 = 0.59, p < 0.0001) suggests crustal sources of aerosols in Kathmandu.Carrico et al. (2003) also reported the influence of dust transport inside Kathmandu valley from a Saharan region.The influence of dust transport was highly variable, with Ca 2+ and Mg 2+ contribution to total ionic aerosols ranging from 6 to 22%.
Sulfate was significantly correlated with NH 4 + (R 2 = 0.48, p < 0.005).The average winter SO 4 2-concentration in Kathmandu (3.16 µg/m 3 ) was smaller than the annual mean SO 4 2-observed in Seoul (8.70 µg/m 3 ) (Lee et al., 1999) and larger than in Nagarkot, Kathmandu valley (2.5 µg/m 3 ) (Carrico et al., 2003).The neutralization ratio ranged from 0.15 to 0.92 during the campaign and had a mean of 0.45 ± 0.19.Clearly, NH 4 + was not sufficient to neutralize the aerosols, which were most likely acidic unless they were neutralized by compounds other than NH 4 + .Daily aerosol acidity ranged from 9.14 to 84.32 nmol/m 3 in Kathmandu during winter.This is larger than the mean H + concentration observed in Seoul during fall (5.19 nmol/m 3 ) (Lee et al., 1999).Peak [H + ] concentration coincided with those of SO 4 2-and NO 3 -.The molar ratio of H + to SO 4 2-was found to range from 0.31 to 2.35. Lee et al. (1999) observed a range for the H + to SO 4 2-ratio of 0.013 to 0.184 during fall in Seoul with the presence of (NH 4 ) 2 SO 4 and (NH 4 ) 3 H(SO 4 ) 2 .

Carbon and Nitrogen Isotopes
The aerosol data indicate that Kathmandu is polluted and that the dominant source most likely is fossil fuel combustion.In such an area, smaller δ 13 C values are expected (Cachier 1989).Furthermore, Cachier (1989) found mean δ 13 C values of -25.5 and -26.5‰ for industrial and vegetation combustion, respectively, compared to -22.5‰ for vegetative emission.The mean δ 13 C of -25.74‰ in the samples is consistent with a dominant fossil fuel combustion source, and is comparable to the values observed in aerosols collected from urban areas (Table 2).The minimum δ 13 C (-26.05‰) was observed on 8 January 2008, with the maximum WSOC/OC, while the maximum δ 13 C (-25.51‰) was observed on 19 January 2008, with the minimum WSOC/OC.Though δ 13 C was not correlated to WSOC, it had a strong correlation with WSOC/OC (R 2 = 0.79; p < 0.01) indicating its association with water-soluble fraction of OC and that the variation in δ 13 C may be linked to processing.The little variation in these samples, however, indicates the consistent source of carbonaceous material.The mean δ 13 C observed in Kathmandu is comparable to the observations made in other urban and anthropogenically influenced locations around the world (Table 2).
The mean δ 15 N (9.45 ± 0.87‰) observed in this study is comparable to the aerosol samples collected from Paris during winter (10.00 ± 3.4‰) (Widory, 2007).The maximum δ 15 N (10.56‰) was observed on 27 December 2007, associated with maximum EC and POC.The minimum δ 15 N (8.13‰) was observed on 30 January 2008 with minimum EC, OC, and WSOC.There was a strong correlation of δ 15 N with NH 4 + /NO 3 -(R 2 = 0.84, p < 0.005) (Fig. 8).Yeatman et al. (2001) observed the larger δ 15 N (-1-7‰) in aerosol NO 3 -(10-12‰) compared to that in aerosol NH 4 + from the samples collected next to vehicle sources in London.The correlation of δ 15 N with crustal cations such as Ca 2+ (R 2 = 0.74, p < 0.05) and Mg 2+ (R 2 = 0.71, p < 0.05) indicates that heavier nitrogen isotopes and some associated nitrates may have originated from distant sources related to dust.Because larger δ 15 N values in aerosols indicate residual material from the combustion of vegetation (Turekian et al., 1998), the smaller δ 15 N values observed in this study indicate the lack of an influence of biomass burning.Further, the aerosol sources from fuel oil and coal (δ 15 N < 0‰) are less likely because the aerosols in this study were enriched in 15 N (Widory, 2007).

Principal Component Analysis (PCA)
The PCA found four factors with Eigenvalues greater than unity that explained a combined 82.2% of the variance in the data (Table 3).Factor 1 explained the most of the variance, 38.7%, and had strong loadings of EC, POC, SOC, WSOC, Cl -, and maximum temperature.Factor 1 is attributed to the primary local sources, mainly from vehicular emissions.Interestingly, it had a high negative loading of K + , suggesting the role of biomass burning when local vehicular emission sources decrease.During the end of the sampling campaign, the smallest values for carbonaceous aerosol concentrations were observed, and K + values peaked.Factor 2, with strong loadings from NH 4 + and NO 3 -explained an additional 23.0% of the variance.This factor appears to be associated with secondary aerosol formation via gas-to-particle partitioning.Factor 3, with strong loadings from SO 4 2-, RH, and minimum temperature likely represents aqueous processing and accounts for 11.4% of the variance.Interestingly, minimum temperature and RH   Moreover, the dominance of SO 4 2-together with RH in factor 3 may mean that the major pathway for SO 4 2formation in this locality is mainly through heterogenous reactions (Kai et al., 2007).The strong loadings of mineral aerosols (Ca 2+ and Mg 2+ ) indicate that Factor 4 is associated with transport of dust.This factor explained an additional 9.1% of the variance.However, the loading of NH 4 + in this factor is minor, and thus, the transport of NH 4 + with dust also appears to be minor.In summary, the PCA performed on this data shows that primary local emissions from vehicular activity had the strongest influence on particle concentrations at this site during winter, followed by secondary processing and long-distance dust transport.

CONCLUSIONS
Kathmandu, the capital of Nepal, experiences very large levels of air pollution due to urbanization and being located inside a valley.In winter, the capacity of aerosols to accumulate in ambient air increases in the absence of major deposition processes such as precipitation.Thus, this study was carried out to gain information regarding the composition, concentration, and sources of aerosols in Kathmandu winter.The daily OC/EC ratio varied between 3.01 and 6.59.Correlation of EC and OC (R 2 = 0.56), large POC contribution to carbonaceous aerosol, and smaller δ 13 C (-25.74 ± 0.19‰) all imply that local vehicular emission is the single most important source in Kathmandu.Principal component analysis revealed that other important sources/pathways include dust transport and secondary processing.The strong correlation of Ca 2+ and Mg 2+ (R 2 = 0.59) supports some crustal aerosol sources for the transported dust.Biomass burning did not play a significant role in aerosol formation/emissions, except at the end of the sampling period.
Most (69%) of the observed OC was estimated to be primary.About half of the OC observed in Kathmandu was water-soluble, with contributions from both SOC and POC.Ionic concentrations are approximately a factor of two smaller than those of organic carbon and are dominated by

Fig. 2 .
Fig. 2. Regression of EC versus OC during the sampling period.The darker line indicates the regression through the observed values, while the lighter line indicates the minimum ratio of OC/EC used for estimating SOC.

Fig. 5 .
Fig. 5. Anion (top) and cation (bottom) concentrations observed in aerosols of Kathmandu during the sampling period.

Fig. 7 .
Fig. 7. Carbon and nitrogen isotope ratios in aerosols in Kathmandu during the sampling period.
ratio indicated that the aerosols in Kathmandu are acidic in nature, as NH 4 + was not sufficient to neutralize the acidity in fine particles.Sulfates were the dominant ion, confirming the anthropogenic source of aerosols.Analysis of carbon isotopes indicates fossil fuel combustion sources, while nitrogen isotopes indicate transport from distant sources and minimal influence from biomass burning.

Table 1 .
Statistics for aerosol concentrations of carbonaceous species, inorganic ions, and carbon and nitrogen isotopes and for derived parameters in Kathmandu during winter.
* Number of samples.For those less than 25, see TableS1to see if samples were below detection limit or if analyses were not performed.For statistics, only those samples for which measured values were above detection limits are included.

Table 2 .
Carbon and nitrogen isotope ratios in aerosol collected in different locations.

Table 3 .
Principal component analysis of carbonaceous components, water-soluble organic, water-soluble inorganic ions, and meteorological data.Only factors with eigenvalues exceeding unity were included.The largest absolute factor loading for each variable is shown in bold (if the original loading was positive).category, possibly suggesting the sensitivity of aqueous phase oxidation of SO 2 to RH levels and the relationship between minimum temperature and RH.