Interactive comment on “ Multi-year ( 2004 – 2008 ) record of nonmethane hydrocarbons and halocarbons in New England : seasonal variations and regional sources ”

Russo et al. present several years of hydrocarbon and halocarbon measurements from New Hampshire. The dataset will be a valuable one for testing models and emission inventories, and is one of a very few which includes high quality VOC measurements sustained over such a long period. The subject is appropriate to ACP and is clearly written. I recommend publication once the following points have been addressed:


Introduction
Volatile organic compounds (VOCs) (including nonmethane hydrocarbons (NMHCs), alkyl nitrates, oxygenated and halogenated hydrocarbons) are ubiquitous and important chemical constituents in the atmosphere.The reaction of VOCs with various oxidants (e.g., hydroxyl radical (OH), ozone (O 3 ), nitrate radical, halogens) produces organic (RO 2 ) and hydro (HO 2 ) peroxy radicals which react with nitrogen oxides (NO x ) to produce secondary species, such as tropospheric ozone, organic nitrates, and peroxides, and regulates the oxidation capacity of the atmosphere.The relative concentrations and speciation of NO x and NMHCs in a particular region determine whether ozone production or destruction occurs (e.g., Carter, 1994;Sillman and He, 2002;Kleinman et al., 2005).In addition, the partitioning of low volatility VOC oxidation products into the condensed phase produces secondary organic aerosols (e.g., Odum et al., 1997;Ng et al., 2007;Kroll and Seinfeld, 2008).Ozone and aerosols are components of photochemical smog, are respiratory lung irritants, and are harmful to vegetation and crops.Furthermore, several VOCs, such as benzene, toluene, xylenes, and tetrachloroethene, are classified as toxic air pollutants and are subject to federal regulations (US EPA, 2008).Therefore, it is necessary to identify and characterize the atmospheric distributions and sources of VOCs in order to Published by Copernicus Publications on behalf of the European Geosciences Union.R. S. Russo et al.: Multi-year (2004-2008) record of nonmethane hydrocarbons and halocarbons develop and validate emission inventories, reduce the levels of hazardous air pollutants, and to predict and control O 3 and aerosol concentrations.
The regional distributions of VOCs are highly variable because of several confounding factors, including different atmospheric lifetimes and removal mechanisms, varying meteorological conditions, and the wide range of potential sources.In order to minimize these complications and to eliminate site-to-site differences, long-term continuous measurements from the same location are necessary.Multi-year measurements of NMHCs and halocarbons at remote and urban North American sites have been reported (e.g., Jobson et al., 1994;Hagerman et al., 1997;Kang et al., 2001;Mohamed et al., 2002;Gautrois et al., 2003;Swanson et al., 2003;McCarthy et al., 2006;Qin et al., 2007), but not for New England since 1994-2001 (Goldstein et al., 1995;Kleiman and Prinn, 2000;Barnes et al., 2003;Lee et al., 2006).These studies have provided baseline data from which to monitor future changes in sources and ambient mixing ratios.
Air masses containing urban and industrial emissions from southern New England, the US East Coast and mid-Atlantic corridor, and the Midwest are transported to New Hampshire.Previous research has indicated that the trace gas measurements (including O 3 , carbon monoxide (CO), nitric oxide (NO), mercury, select VOCs) made at the UNH AIRMAP monitoring site at Thompson Farm (TF) in Durham, New Hampshire are representative of both inland and coastal New England (e.g., Talbot et al., 2005;Chen et al., 2007;Sive et al., 2007;Mao et al., 2008;White et al., 2008;Zhou et al., 2005Zhou et al., , 2008)).This suggests that the TF results can be applied to regional analyses of the short and long-term temporal variability and sources of VOCs for this region.This is particularly valuable because southern New England, including the seacoast region of New Hampshire, and extending to the southwest through New York and New Jersey is classified as an O 3 nonattainment area (US EPA, 2003, 2008).
We have been conducting VOC measurements at Thompson Farm since 2002.Multi-year VOC data published thus far includes measurements of C 1 -C 5 alkyl nitrates (Russo et al., 2010), methyl iodide (Sive et al., 2007), and oxygenated VOCs (OVOCs) and select NMHCs (Jordan et al., 2009).Summertime measurements (2002)(2003)(2004) of NMHCs, OVOCs, and marine-derived halocarbons at TF were discussed in Talbot et al. (2005), Chen et al. (2007), White et al. (2008), and Zhou et al. (2005Zhou et al. ( , 2008)).In this work, four years (2004)(2005)(2006)(2007)(2008) of ambient C 2 -C 8 NMHC and anthropogenic halocarbon data from daily canister samples collected at the Thompson Farm site in southeastern New Hampshire are presented and discussed.The primary objectives of this study are to characterize the seasonal to interannual trends and to identify the possible sources of C 2 -C 8 NMHCs and halocarbons.Additionally, emission rates of several NMHCs were estimated and compared with the EPA National Emissions Inventory.

Daily canister sample collection and analysis
The University of New Hampshire AIRMAP Observatory at Thompson Farm (TF) is located in Durham, New Hampshire (43.11 • N, 70.95 • W, elevation 24 m) and is located approximately 20 km inland from the Atlantic Ocean and 100 km north of Boston, MA (www.airmap.unh.edu) (Fig. 1).TF is surrounded by agricultural fields and a mixed deciduous and coniferous forest.An ambient canister sample has been collected at some point between 10:00-15:00 (EST; UTC−5 h) each day at the top of the 15 m tower next to the manifold inlet for all the instruments housed in the TF building.Sample collection began on 12 January 2004 and continues through the present.The sample collection time window is representative of daytime conditions when photochemistry is most active and the boundary layer is likely well mixed.Prior to sampling, the 2-liter electropolished stainless steel canisters (University of California, Irvine, CA) were prepared by flushing with UHP helium that had passed through an activated charcoal/molecular sieve (13X) trap immersed in liquid nitrogen.The canisters were then evacuated to 10 −2 torr.
The canister samples were analyzed in the laboratory at UNH approximately every 1-3 months for C 2 -C 10 nonmethane hydrocarbons, C 1 -C 5 alkyl nitrates, C 1 -C 2 halocarbons, several OVOCs, and select sulfur compounds.A three gas chromatograph system equipped with two flame ionization detectors (FID), two electron capture detectors (ECD), and a mass spectrometer (MS) was used for analysis of each 1500 cc (STP) sample aliquot.The PLOT-FID, OV-1701-ECD, and OV-624-MS column-detector combinations have remained the same throughout 2004-2008 (see Sive et al. (2005) and Zhou et al. (2005Zhou et al. ( , 2006Zhou et al. ( , 2008) ) for additional information).In 2006, the FID channel used for C 4 -C 10 NMHC analysis was changed from a DB-1 column (60 m×0.32 mm I.D., 1 µm film thickness) to a VF-1ms column (60 m×0.32 mm I.D., 1 µm film thickness).Also, the OV-624-ECD channel was changed to a CP-PoraBond-Q column (25 m×0.25 mm I.D., 3 µm film thickness) coupled to a Restek XTI-5 column (30 m×0.25 mm I.D., 0.25 µm film thickness).The MS was operated in electron impact mode with single ion monitoring for measuring OVOCs and sulfur compounds, as well as duplicate measurements of several halocarbons and NMHCs.A 1500 cc aliquot from one of two working standards was assayed every ninth analysis.The measurement precision for the whole air standards (i.e., relative standard deviation (RSD) = standard deviation of peak areas/average of peak areas) was <1-4% for the C 2 -C 8 NMHCs and 5% for C 2 HCl 3 and C 2 Cl 4 at 0.50 and 6.0 pptv, respectively.

Daily canister sample data set
Data for several classes of NMHCs and two halocarbons from the canister samples collected during 12 January 2004 to 8 February 2008 were used for this study.Collection of the daily samples is ongoing.The specific compounds, which represent a wide range of chemical reactivities and sources, are C 2 -C 6 alkanes (ethane, propane, ibutane, n-butane, i-pentane, n-pentane, n-hexane), C 2 -C 4 alkenes (ethene, propene, 1-butene), C 6 -C 8 aromatics (benzene, toluene, ethylbenzene, m+p-xylene, o-xylene), ethyne, isoprene, tetrachloroethene (C 2 Cl 4 ), and trichloroethene (C 2 HCl 3 ).The data for each year and for the combined four year data set was separated into four seasons which are defined as winter: December, January, February; spring: March, April, May; summer: June, July, August; and fall: September, October, November.Note that the data encompasses five winter seasons (2004)(2005)(2006)(2007)(2008) and four spring, summer, and fall seasons (2004)(2005)(2006)(2007) with the exceptions of winter 2004 which only includes 12 January to 29 February and winter 2008 which is only through 8 February.Mixing ratios higher than the 95th percentile value for each month were removed in order to ensure that the results were representative of typical conditions and were not skewed by outlying or spurious data points.

Thompson Farm automated gas chromatograph
Hourly measurements of C 3 -C 6 alkanes, ethyne, propene, benzene, toluene, ethylbenzene, m+p-xylene, and o-xylene from an automated GC system during December 2005-January 2006 at TF were also used in this analysis.The sample size of the in situ GC system was also 1500 cc.Details of the four channel (2 FIDs, 1 ECD, 1 MS) GC system, MMR preconcentrator, sample trapping and splitting, calibrations, and instrument control are given in Sive et al. (2005).The system deployed at TF also contained four channels, but VOC detection was made with two FIDs and two ECDs.The channel which replaced the MS was a PoraBond-Q/OV-1 column coupled to an ECD for measuring C 1 -C 2 halocarbons.A 1500 cc aliquot from one of two working standards was assayed every tenth analysis.The precision (i.e., RSD) for each of the hydrocarbons discussed in this work ranged from 3-10%.

Standards and calibration
In order to ensure that the VOC mixing ratios in samples analyzed at different times are comparable, whole air and synthetic standards were routinely analyzed and calibration scales were cross referenced and validated.When conducting the canister sample analysis, one of two whole air standards was analyzed after every eighth sample in order to monitor changes in detector sensitivity and measurement precision and to quickly detect any analytical problems.The whole air working standards employed for this work have mixing ratios representative of clean free tropospheric air and suburban air, thus bracketing the low and high ranges for the measurements at TF.Moreover, linearity studies are regularly conducted to evaluate the detector response over the observed mixing ratio ranges for all classes of compounds.The working standards are part of the larger network of whole air standards maintained by B. Sive at UNH as part of the AIRMAP program.In total, there are currently ten high pressure cylinders, six 36-liter electropolished low-pressure pontoons (∼350 psi), and three 34-liter electropolished highpressure pontoons (∼900 psi) containing whole air standards that have been filled and calibrated by UCI (D. Blake) and UNH.We estimate the upper limit of the absolute accuracy of the calibrated standards to be on the order of ±1-5% for the gases reported here.In addition to the whole air standards, calibrations are conducted using five different high-pressure cylinders containing synthetic blends of selected NMHCs, OVOCs, and halocarbons at the ppbv level (Apel-Reimer Environmental, Inc.).The absolute accuracy for all of the gases in the synthetic standards is less than ±5%.
Response factors (RF) for each compound in a particular standard were calculated by dividing the detector response (peak area=A) by the mixing ratio (MR) of that compound in the standard (RF=A/MR).Mixing ratios for each gas to be quantified in the ambient samples were subsequently calculated using the average RF determined from the whole air standards during each set of individual analyses.In order to monitor the response of NMHCs with different numbers of carbon atoms, it is useful to evaluate the per-carbon response, particularly when dealing with long-term calibrations.The per-carbon response provides information regarding analytical system performance and standard integrity with time.The per-carbon response factors (PCRF) were determined by dividing the RF for each NMHC by the number of carbon www.atmos-chem-phys.net/10/4909/2010/Atmos.Chem.Phys., 10, 4909-4929, 2010 R. S. Russo et al.: Multi-year (2004-2008) record of nonmethane hydrocarbons and halocarbons  atoms (C) in each particular hydrocarbon (PCRF=RF/C).
While the response for hydrocarbons containing the same number of carbon atoms should be uniform, this should always be verified by analyzing different classes of compounds (e.g., alkanes, alkenes, alkynes).Examples of the PCRFs for several NMHCs are shown in Fig. 2 as representative examples for the analyses when a new standard began to be used and every ∼3-6 months when the same two standards were being analyzed throughout 2004-2008.While there is some scatter about the mean PCRF over the four years, the PCRF of the two standards analyzed during each analysis period are self-consistent.Moreover, this illustrates the long-term stability of our analytical system.The PCRF of the C 2 -C 4 NMHCs measured with the PLOT-FID column-detector pair decreased with increasing carbon number, but remained approximately the same over the four years and did not vary with standard.Additionally, the PCRFs remained constant for the C 3 (mean ± standard deviation = 4.12±0.19)and C 4 (3.70±0.16)compounds (Fig. 2c-f).The PCRFs of the C 4 -C 10 NMHCs for the DB-1/VF-1ms-FID column-detector pair have remained essentially constant (1.43±0.03)for compounds with different carbon numbers and for different compound classes since 2005 (Sive, 1998;Zhou, 2006).Periodically, the standards used by the automated GC system at TF were returned to the laboratory and analyzed on the canister analysis system.The PCRFs for the TF standards (e.g., DC2) agreed (within ±5%) with the laboratory standards ensuring that the measurements made by the two independent systems are comparable.

Seasonal and interannual variability of VOCs at
Thompson Farm

General characteristics of NMHCs and comparison with previous studies
The highest monthly mean and median mixing ratios of NMHCs (excluding isoprene, Sect.3.2) at TF were observed in the winter (Fig. 3, Table 1).This reflects the slow removal rates from the atmosphere caused by minimum OH radical concentrations at this time of year.Lower boundary layer heights in winter are conducive to the build up of trace gas concentrations and may also contribute to the wintertime peak mixing ratios.In general, the lowest NMHC mixing ratios were observed in spring to summer when the maximum OH concentrations occur and the photochemical removal of NMHCs is the most rapid.It should also be mentioned that there is a variation in dominant air mass transport pathways throughout the year which may contribute to the seasonal variation in NMHC mixing ratios.For example, in the winter, the transport of clean, Canadian air masses to New England, which are representative of background conditions, is more frequent (Munger et al., 1996;Moody et al., 1998;Shipham et al., 1998).In contrast, transport from the south and west is more frequent during the summer (Moody et al., 1998;Fischer et al., 2004;Mao and Talbot, 2004b).Overall, the seasonal variation at TF is consistent with the general tropospheric trend observed at other North American sites (e.g., Jobson et al., 1994;Bottenheim and Shepherd, 1995;Goldstein et al., 1995;Hagerman et al., 1997;Gautrois et al., 2003;Swanson et al., 2003;Lee et al., 2006;Qin et al., 2007).

Seasonal variation of anthropogenic C 2 -C 8 NMHCs
The C 2 -C 6 alkanes, ethyne, toluene, and benzene exhibited reproducible seasonal trends where the longer lived compounds had higher mixing ratios and reached minimum annual mixing ratios later in the year.Ethane mixing ratios peaked in winter-early spring and then decreased until mid to late summer when minimum mixing ratios were observed (Fig. 3a).Mixing ratios of the C 3 -C 6 alkanes, alkenes, ethyne, and aromatics began to decrease in mid to late winter and reached minimum levels 2-4 months later (Fig. 3; Table 1).Ethene, ethyne, and benzene remained at their annual minimum mixing ratios from late spring to late summer (Fig. 3d).Despite its order of magnitude shorter lifetime, ethene mixing ratios were often similar to or higher than ethyne in summer, fall, and winter indicating the importance of ethene emissions in NH.The monthly mean and median propane, i-butane, and n-butane mixing ratios were lowest in late spring (May-June) followed by an increase in early summer before reaching a second minimum in late summer (Table 1).In comparison, minimum mixing ratios of the shorter-lived C 5 -C 6 alkanes, propene, and toluene occurred earliest (April-May), increased in early summer, and then remained within a similar range through October-November (Fig. 3c, f; Table 1).The mean summer mixing ratios of propene, toluene, and the C 5 -C 6 alkanes were ∼15-90% higher than the April-May mean mixing ratios.These NMHCs react rapidly with OH (summer lifetime <1.5 days assuming [OH]=2×10 6 molecules cm −3 ), and thus would be expected to remain at minimum mixing ratios throughout the entire summer.The summer increase in mixing ratios is indicative of a strong influence from evaporative emissions on the NMHC distribution (discussed in more detail in Sect.4).
On average, 1-butene, ethylbenzene, and xylenes (m+p and o) were also highest in winter, lowest in early to mid spring, and increased in early summer (Fig. 3e-g, Table 1).Superimposed on the general alkene and C 8 aromatic seasonal patterns were unique interannual trends reflecting varying sources or emission rates.For example, a higher and narrower range of propene (∼50-80 pptv) and 1-butene (∼8-17 pptv) monthly mean and median mixing ratios were observed in spring 2006 through winter 2008 compared to the previous two years when distinct winter peaks and springsummer minimum mixing ratios occurred.Additionally, the highest monthly mean and median m+p-xylene and o-xylene mixing ratios of the entire four year study period (∼25-40 and 25-35 pptv, respectively), as well as the highest summer toluene (90-140 pptv) and ethylbenzene (14-22 pptv) mixing ratios, were observed in summer 2007 and elevated mixing ratios persisted through winter 2008.

Influence of enhanced NMHC emissions on summer photochemistry
The enhanced summer mixing ratios of reactive NMHCs will likely influence the ambient mixing ratios and regional budgets of secondary photochemical species, such as organic aerosols, OVOCs, and O 3 .NMHCs make variable contributions to the production of secondary species because of their different rates of reaction with OH.The incremental reactivity (IR) is a method for determining the various ozone formation potentials (OFP) of VOCs and takes into consideration the different chemical reaction pathways following the initial reaction of the NMHC with OH.The incremental reactivity is defined as the change in O 3 resulting from the addition or subtraction of an increment of VOC to/from an airmass divided by the amount of VOC added or removed (Carter, 1994;Bowman and Seinfeld, 1995).An estimate of the maximum possible amount of O 3 that could be formed resulting from the higher propene, C 5 -C 6 alkane, and toluene in summer was made using the following equation: O 3 = NMHC×MIR where NMHC = maximum summer mixing ratio -average spring mixing ratio and MIR is the maximum incremental reactivity factor (gram of O 3 produced/additional gram of NMHC) (Carter, 1994(Carter, , 2008)).For these calculations, we are assuming that conditions are favorable for O 3 production.The atmosphere over New England in summer is generally NO x -limited because of the abundance of biogenic VOCs.Nonetheless, summer NO x mixing ratios at TF have been observed to range from 0.21-17.5 ppbv (average 2.3 ppbv) (Griffin et al., 2007).These NO x mixing ratios are sufficient for the NO+peroxy radical reaction to be preferred over peroxy radical self reactions (e.g., Flocke et al., 1991;Roberts et al., 1998).These calculations indicate that 0.5-2.4ppbv, 0.46-0.65 ppbv, 0.15-0.24ppbv, 0.08-0.1 ppbv, and 1.2-2.3ppbv (total range 2.4-5.7 ppbv) of additional O 3 could be formed as a result of the enhanced propene, i-pentane, n-pentane, n-hexane, and toluene mixing ratios, respectively, in summer at TF.In comparison, the 24 h mean O 3 mixing ratios at TF are ∼30 ppbv in summer (Mao and Talbot, 2004a;Talbot et al., 2005).These results are not absolute levels of O 3 that will be produced because the MIR factors were derived for conditions that do not necessarily reflect New England (Carter, 1994).
Nonetheless, these results are intended to illustrate the potentially significant impact of the enhanced summer mixing ratios of reactive NMHCs on the ozone formation potential in this region.

Isoprene
Isoprene is the only NMHC discussed in this work with a predominantly biogenic origin (deciduous plants and trees) (e.g., Guenther et al., 1995).Isoprene mixing ratios rapidly increased in the beginning of June, remained high through August, and gradually decreased in September-October of each year (Fig. 4a, Table 1).In July-August 2005, 2006, and 2007, isoprene was the most abundant NMHC (monthly mean mixing ratios=1000-2100 pptv) illustrating the importance of biogenic emissions in this region.In comparison, mean ethane (the longest lived NMHC) mixing ratios were ∼800-1250 pptv.Isoprene was positively correlated with the ambient temperature during each summer.The relationship can be expressed as log(isoprene)=0.074T-1.9(r 2 =0.57) for all of the available June-September data (Fig. 4b) (T in • C corresponding to the hour the sample was collected).The highest isoprene mixing ratios (>3 ppbv) were observed in the warmest summers (2005 and 2006) while the lowest (<1600 pptv) mixing ratios occurred during the coolest summer ( 2004).The temperature dependence of isoprene was nearly the same each summer, on sunny/clear days, and on cloudy/rainy days.Furthermore, this relationship is consistent with previous studies (e.g., Fehsenfeld et al., 1992;Jobson et al., 1994;Goldan et al., 1995;Gong and Demerjian, 1997;Hagerman et al., 1997;Kang et al., 2001) indicating a similar temperature dependence of ambient isoprene mixing ratios at various North American sites.

Halocarbons
Trichloroethene and tetrachloroethene are primarily used as dry cleaning and degreasing solvents and are thus tracers of industrial sources (e.g., Wang et al., 1995;McCulloch and Midgley, 1996).It is necessary to monitor the atmospheric trends of C 2 HCl 3 and C 2 Cl 4 because they are used as replacements for CFCs and in the production of HFCs and HCFCs and because they are classified as toxic air pollutants and are regulated by the EPA (US EPA, 2007).A wide range of C 2 Cl 4 (3-65 pptv) and C 2 HCl 3 (<1-23 pptv) mixing ratios were observed at TF (Fig. 5).The shorterlived (days-weeks) C 2 HCl 3 had a similar seasonal variation as the NMHCs with a winter maximum (monthly mean mixing ratios=5-8 pptv), late spring minimum (1-2 pptv), and an early summer increase (2-5 pptv).Similar to propene, the C 5 -C 6 alkanes, and toluene, the mean mixing ratios in summer were 50-75% higher than in spring illustrating that evaporative emissions of C 2 HCl 3 are important to its atmospheric distribution in this region.In contrast, the monthly mean mixing ratios of the longer-lived (months) C 2 Cl 4 were fairly uniform.A seasonal variation was apparent in the C 2 Cl 4 background (monthly 10th percentile) mixing ratios which were highest in winter (8-9 pptv) and lowest in late summer (3-5 pptv).
The monthly mean halocarbon mixing ratios at TF are higher than observed at remote sites, such as Mace Head, Ireland (Simmonds et al., 2006), along the US west coast (Simpson et al., 2004), NOAA CMDL sites (Thompson et al., 2004), Chebogue Point, Nova Scotia (Millet et al., 2006), and over the North Atlantic Ocean (Dimmer et al., 2001), reflecting the closer proximity to industrial sources and their continued use in North America, but are lower than observed in heavily urbanized areas, such as Pittsburgh, PA (Millet et al., 2005), Houston, TX (Gilman et al., 2009), and Mexico City, Mexico (Velasco et al., 2007).The C 2 Cl 4 mixing ratios at TF are similar to background levels in Massachusetts during 1996-1999 (Kleiman and Prinn, 2000;Barnes et al., 2003) which suggests that C 2 Cl 4 emission rates in more populated areas in the US did not change considerably between the late 1990's and 2004.However, we have observed a decrease in the magnitude of peak mixing ratios, and in the annual mean, median, and background mixing ratios throughout 2004-2008 at TF (Fig. 5, Table 2).More specifically, the background mixing ratios of C 2 Cl 4 and C 2 HCl 3 were 30% and 65% lower, respectively, in 2007 than in 2004.According to the EPA 2008 Toxics Release Inventory (www.epa.gov/tri),C 2 HCl 3 emission reductions were about a factor of two larger than C 2 Cl 4 over the same time period.
An estimate of the rate of decrease in atmospheric mixing ratios of both halocarbons was made using the annual statistics in order to minimize the influence of the seasonal variation in C 2 HCl 3 and in background C 2 Cl 4 .The linear regression through the annual background mixing ratios gave decrease rates (± standard deviation) of 0.73±0.24and 0.27±0.05pptv/year for C 2 Cl 4 and C 2 HCl 3 , respectively (Table 2).In comparison, C 2 Cl 4 decrease rates (pptv/year) were estimated to be 1.0 (C 2 HCl 3 =0.1) in 1991-1996 at Alert, Canada (Gautrois et al., 2003), 0.6-1.2during 1994-1997 in the continental US (Hurst et al., 1998), 0.1-0.4throughout 1989-2002 along the North American west coast (Simpson et al., 2004), 0.18 (C 2 HCl 3 =0.01)during July 2000-December 2004 at Mace Head, Ireland (Simmonds et al., 2006), and 5% per year in 1995-2003 based on analysis of EPA and NOAA CMDL data at remote North American sites (McCarthy et al., 2006).It must be kept in mind that the trends observed at TF and other locations represent different time periods, emission rates, and industrial regulations.Nonetheless, our results indicate that the amounts of C 2 Cl 4 and C 2 HCl 3 being transported from the northeastern US to the North Atlantic are decreasing.

Comparison with tracers and source signatures
The major sources of ethyne, benzene, carbon monoxide (CO), and alkenes are incomplete combustion of fossil fuels, biomass burning, and vehicle exhaust emissions (e.g., Harley et al., 1992Harley et al., , 2001;;McLaren et al., 1996;Choi and Ehrman, 2004).The oxidation of VOCs is also a potential secondary source of CO, particularly in summer when biogenic NMHC mixing ratios are high (Sect.3.4) (Hudman et al., 2008).Studies focused on the northeast US during summer 2004 found that observed CO mixing ratios could largely be explained by primary urban/industrial and biomass burning sources (Warneke et al., 2006;Griffin et al., 2007).Thus, in our analysis, we neglect the possible secondary contribution to CO. Ethyne and CO were fairly well correlated at TF (r 2 =0.5-0.9)demonstrating a year-round impact from combustion emissions (Fig. 6a).The correlations between alkenes and tracers for combustion (ethyne), liquefied petroleum gas (LPG) (propane), and fuel evaporation (i-pentane) were strongest in winter and showed considerable scatter in the other seasons.The short-lived alkenes had presumably undergone mixing and oxidative removal during transport resulting in weaker correlations in spring, summer, and fall.Based on the winter measurements, the ethene and propene correlation slopes with ethyne (1.2 and 0.21, respectively) were similar to light duty gasoline and vehicle exhaust emission ratios (0.9-1.7 and 0.1-0.5, respectively) (Conner et al., 1995;Watson et al., 2001;Choi and Ehrman, 2004;McGaughey et al., 2004) suggesting that vehicular emissions were the dominant source of these alkenes.Propene and 1butene were well correlated throughout the majority of the study period reflecting a common source (Fig. 6b).The correlation coefficients between ethyne and the alkanes, benzene, and toluene were fairly variable within each season and year (r 2 =0.4-0.9),but overall suggest that combustion sources were collocated with or were the same as the alkane and aromatic sources.(2004,2005,2006,2007,2008).

Ambient ratios: compounds with similar lifetimes
Information on the relative impact of various sources in a region can be obtained by comparing the ambient ratio of two compounds that have similar rates of reaction with OH but different sources (e.g., Klemp et al., 1997;Jobson et al., 1999;Goldan et al., 2000).The ratio should reflect the integration of several factors, such as air mass mixing and dilution, new emission inputs, and oxidative removal, because neither compound will be removed preferentially during transport.Thus, on average, the ratio can be assumed to remain fairly constant and approximately equal to the emission ratio (Parrish et al., 1998).For example, propane, ethyne, and benzene have similar lifetimes (<30% difference in k OH ) (Atkinson et al., 2006), but these NMHCs are tracers of different sources.The propane/ethyne, propane/benzene, and benzene/ethyne vehicular exhaust and whole gasoline emission ratios are typically <1 while ratios from natural gas, LPG, or gasoline vapor are ≥∼1 (e.g., Conner et al., 1995;Fujita et al., 1995;Fujita, 2001;Mukund et al., 1996;Lawrimore et al., 1997;Watson et al., 2001;Choi and Ehrman, 2004;Mukerjee et al., 2004;White et al., 2008).Throughout 2004-2008, the propane/ethyne (Fig. 9a) and propane/benzene ratios ranged from 1-5 and 3-25, respectively, demonstrating the stronger influence of natural gas or LPG relative to incomplete combustion as a source of propane throughout the entire year.This corroborates previous work at TF and Appledore Island, Maine (10 km off the NH coast) during summer 2004 which concluded that LPG was the dominant source of propane throughout the entire day in southern NH (White et al., 2008).Furthermore, Goldan et al. (2004) illustrated that the relationship between propane and benzene was unique to the northeast US based on measurements made off the New England coast during the NEAQS 2002 campaign and suggested that this is indicative of a non-vehicular source of propane.This supports our observations of a significant influence from LPG use and/or leakage at TF.
The slope of the benzene vs. ethyne correlation was the same in each season of every year (slope of all data=0.21,r 2 =0.91) (Fig. 9b).This ratio value is indicative of a vehicular source and is consistent with observations of ambient benzene/ethyne ratios measured during several spring-summer field campaigns conducted throughout the US (Fortin et al., 2005;Harley et al., 2006;Parrish, 2006;Sistla and Aleksic, 2007;Warneke et al., 2007) and in major cities (Parrish et al., 2009).The strong correlations between propane and ethyne (Fig. 9a), propane and benzene (r 2 =0.71-0.76,not shown), and benzene and ethyne (Fig. 9b) suggest that emissions from natural gas, petroleum, and fossil fuel/incomplete combustion (e.g., vehicles) sources were concurrent and/or collocated.Moreover, this illustrates a pervasive and continuous influence of emissions from several anthropogenic sources on the chemical composition of air masses observed at TF.

Ambient ratios: compounds with different lifetimes
Ambient ratios between compounds with different rates of reaction with OH and with well characterized sources and sinks are frequently used to estimate the relative photochemical age of air masses, transport times and distances, or OH concentrations (e.g., Jobson et al., 1994;McKeen et al., 1996;Parrish et al., 1998;Smyth et al., 1999;Dimmer et al., 2001;Kleinman et al., 2003;Russo et al., 2003).A fundamental drawback to estimating air mass processing times using ambient ratios is a lack of information on seasonal variations in sources, especially when analyzing data from short-term field campaigns.Analysis of the long-term measurements from TF provided a unique perspective on the interrelationships between seasonal variations in sources and chemical processing in this region.
Four common ratios with the shorter-lived compound in the numerator are ethyne/CO, propane/ethane, toluene/benzene, and C 2 HCl 3 /C 2 Cl 4 (Fig. 10).The general behavior of these ratios can be predicted based on the differential removal of the compounds in each ratio.For example, if reaction with OH was the only factor influencing the seasonality in mixing ratios, a decrease in ratio values from winter to summer would be expected to occur concurrently with the increase in atmospheric OH concentrations because the shorter-lived compound is removed preferentially.The ethyne/CO ratio trend reflects the seasonal variation in OH concentrations with higher winter ratios (4-5 pptv/ppbv) indicating less processed emissions and low summer ratios (1-2 pptv/ppbv) reflecting more processed air masses (Fig. 10a).The propane/ethane and C 2 HCl 3 /C 2 Cl 4 ratios tracked each other very well, and the temporal variation of both ratios resembled the ethyne/CO ratio with maximum values in winter and minimum values in late spring-summer (Fig. 10b).However, the propane/ethane and C 2 HCl 3 /C 2 Cl 4 ratios increased throughout summer and fall which likely reflects the similar lifetimes and seasonal variation of propane and C 2 HCl 3 combined with the mid to late summer minimum ethane and C 2 Cl 4 mixing ratios.
The seasonal variation of the toluene/benzene ratio was opposite of the ethyne/CO, propane/ethane, and C 2 HCl 3 /C 2 Cl 4 ratio behavior, and thus contrary to the expected photochemical trend (Fig. 10a).The toluene/benzene ratio was lowest in winter-spring (∼0.5-1.5) and highest in summer-fall (0.5-7).These ratio values are comparable to ambient ratios (1-5) observed in numerous continental/urban areas (e.g., Parrish et al., 1998;Monod et al., 2001).However, the fact that the ratio increases in the summer is indicative of an additional source or enhanced emissions of toluene in this region.White et al. (2009) illustrated that the anomalous toluene behavior at TF could not be fully explained by fuel evaporation and industrial emissions and provided evidence for a biogenic contribution to the summer toluene enhancements.It was also noted that the toluene enhancements were larger with each successive year (White et al., 2009).This trend continued into 2007 when the highest toluene/benzene ratios (mean=2.1)were observed (Fig. 10a).
This analysis provides a clear illustration for the necessity of characterizing VOC sources in individual regions.As recently discussed by Schnitzhofer et al. (2008), these results have important implications because numerous studies have used the toluene/benzene ratio to estimate photochemical air mass ages (e.g., Roberts et al., 1984;Gong and Demerjian, 1997;Kang et al., 2001;de Gouw et al., 2005;Warneke et al., 2007) or to distinguish between industrial, evaporative, and exhaust emission sources (e.g., Barletta et al., 2008).The TF measurements demonstrate that the toluene/benzene ratio may not be appropriate for estimating relative air mass ages in this region because variations in the ratio values reflect both chemical processing and additional toluene emissions.An additional source of toluene in this region causes the initial toluene/benzene emission ratio used in the processing time calculations to be erroneously high leading to overestimated photochemical air mass ages and transport distances.

Emission rates estimated from observations at TF
Emission rates of speciated VOCs based on ambient measurements are needed for developing regional budgets, implementing effective control strategies for reducing emissions of photochemical smog precursors and toxic compounds, and evaluating emission inventories and air quality models.Estimates of emission rates are limited and are primarily reported on global scales (e.g., Boissard et al., 1996;Gupta et al., 1998) or in urban areas during specific campaigns (e.g., Blake and Rowland, 1995;Chen et al., 2001, Velasco et al., 2005).Additionally, emissions for individual VOCs are usually lumped into specific classes (i.e., alkane, alkene, aromatic, biogenic).Major reasons for the lack of information on regional VOC emission rates include the complications involved with differentiating between local, regional, and distant sources and the scarcity of long-term continuous measurements.In order to minimize the complications associated with air mass transport, we focus on (1) winter measurements because the C 2 -C 8 NMHC lifetimes are longer than regional transport and mixing timescales at this time of year and (2) nighttime data when it has been determined that mixing is minimal.Under these conditions, we can assume that a change in NMHC mixing ratios reflects local sources or sinks.In contrast, the daily canister samples are representative of daytime conditions when mixing and transport may be occurring and can be assumed to reflect sources from a larger area.Thus, emission rates of NMHCs were calculated in two steps.
First, we estimated emission rates of NMHCs using hourly measurements from the automated in situ GC system at TF during December 2005-January 2006.We followed a simple box model approach which has been effectively used in previous studies to calculate emission and removal rates of trace gases in this region (i.e., Talbot et al., 2005;Zhou et al., 2005;Sive et al., 2007;White et al., 2008;Russo et al., 2010).This method uses measurements made on nights with low wind speeds and when a stable inversion layer has developed because under these conditions, the exchange of air between the nocturnal boundary layer (NBL) and the residual layer above is limited (e.g., Hastie et al., 1993;Gusten et al., 1998;Talbot et al., 2005).Therefore, advection and vertical mixing of air masses can be neglected.Two criteria were used for identifying nights when a stable inversion layer developed: (1) wind speeds <1 m/s and (2) O 3 ≤5 ppbv.The two criteria conditions were met on several nights between approximately midnight and 05:00 with concurrent increases in NMHC mixing ratios.Emission rates (ER GC ) were calculated by multiplying the slope of the linear regression between the change in hourly average concentrations (dC in molecules cm −3 ) per unit time (dt=5 h) by the boundary layer height: Stable nocturnal boundary layer heights typically range from ∼50-200 m at midlatitude continental locations (e.g., Hastie et al., 1993;Gusten et al., 1998;McKendry and Lundgren, 2000;White et al., 2003); thus we chose 125 m as a representative value for the TF site (e.g., Talbot et al., 2005;Sive et al., 2007;Mao et al., 2008;White et al., 2008;Zhou et al., 2005).If we use H =50 m or 200 m, our emission rate estimates discussed below vary by ±60%.It must be kept in mind that the resulting emission rate estimates are directly proportional to the boundary layer height which varies with meteorological conditions, time of day, and season (Talbot et al., 2005).
In the second step, emission rates (ER DC ) were calculated by multiplying the emission ratio for each compound from the daily canister samples by the emission rate of a reference compound (Eq.2): Ethyne was used as the reference compound because of its relatively long lifetime and because its major source (combustion) is well known (e.g., Conner et al., 1995;Goldstein et al., 1995;de Gouw et al., 2005;Lee et al., 2006).Emission ratios using the winter 2006 daily canister data were determined from the slope of the correlation between a specific NMHC and ethyne (NMHC/ethyne).A comparison between the winter 2006 data from the automated in situ GC and the canister samples demonstrated that the mixing ratios agreed well (±10%, r 2 =0.82-0.94;orthogonal distance regression) and that the results from the two instruments can be meaningfully compared (Fig. 11).The correlations are based on comparing the TF GC sample corresponding most closely to the time the canister was collected.These times agreed within less than one hour.The canister and GC data tracked each other illustrating that the daily canister samples www.atmos-chem-phys.net/10/4909/2010/Atmos.Chem.Phys., 10, 4909-4929, 2010 captured a wide range of air mass types and compositions including background air masses and significant winter pollution events with enhanced NMHC mixing ratios.The percent difference between the emission rates calculated from the TF GC nighttime data (Eq. 1) and the daily canister emission ratios (Eq.2) ranged between ±5-80% (Table 3).The emission rate of propane (∼2 to 4×10 10 molecules cm −2 s −1 ) was 1-2 orders of magnitude larger than the other NMHCs (range 0.2-7×10 9 molecules cm −2 s −1 ) (Table 3).The high propane emission rate is another indication of the persistent impact of leakage from LPG tanks or refilling stations throughout the region (Sect.4.2).Additionally, assuming the TF results are representative of the state and region, the NMHC emission rates extrapolated to the state of New Hampshire and New England are ∼2-61 Mg/day and ∼12-430 Mg/day, respectively (Table 3).The emission rates of propane, i-butane, and n-butane from New Hampshire were 20-90% lower than observed from LPG leakage in Mexico City, Mexico (Elliott et al., 1997) and in Santiago, Chile (Chen et al., 2001).However, taking into consideration the larger land area for New Hampshire and New England, our results suggest that potential emissions of C 3 -C 4 alkanes from the northeast US are comparable to emission rates that have been observed in densely populated urban areas (see also White et al., 2008).Furthermore, these high levels of precursor compounds in air masses transported to and across the North Atlantic may contribute to O 3 and aerosol production, thus influencing the air quality of downwind regions.
Emission rates determined using the daily canister emission ratios from winters 2004, 2005, 2007, or 2008 were within ∼10-50% of the winter 2006 values.In addition, the winter 2006 emission rates of propane, i-butane, n-butane, ipentane, and propene agreed (within factors of ∼0.7-6) with estimates made using nighttime measurements from the automated TF GC during summers 2003 and 2004 (White et al., 2008).The consistency between the estimates for different winters and for winter and summer suggests that emission rates do not appear to be varying detectably with season or year.We consider the fact that the winter 2006 TF GC and daily canister emission rates and the summer 2003 and 2004 emission rates from White et al. (2008) agree within an order of magnitude (with the exception of m+p-xylene) to be good agreement.Furthermore, despite the different areas (i.e., source footprints) potentially represented by the TF GC and daily canister emission rates, both approaches yield similar results.Possible explanations for this similarity include: (1) the two estimates are not completely independent because the daily canister emissions are calculated using the TF GC ethyne emission rate; (2) local NMHC emission rates are similar to rates throughout New England; (3) the midday daily canister emission rates retain a signature from local nighttime emissions following the breakup of the NBL and subsequent mixing with remnant air from the previous day's mixed layer in the morning.Overall, this analysis provides promising results that the daytime canister samples provide representative information on regional emission rates.

Comparison with the 2002 and 2005 EPA National Emissions Inventory
The emission rates of benzene, toluene, ethylbenzene, xylenes, and ethyne in the two most recent versions (2002 and 2005) of the EPA National Emissions Inventory (NEI) (www.epa.gov/ttn/chief/net) were compared with each other and with the emission rates estimated at TF (Sect.5.1).These five NMHCs were chosen because the NEI provides speciated emission rates for some toxic compounds (including aromatics) and because ethyne or VOC/ethyne ratios are commonly used as tracers of specific sources.The emission rates of benzene, toluene, ethylbenzene, and xylenes for NH from the major source categories (onroad, nonroad, nonpoint, point) were obtained directly from the NEI.Emission rates of ethyne were estimated using the total VOC emissions from its major sources (gasoline and diesel exhaust, recreational equipment, lawn and garden equipment, and stationary source fuel combustion) and the EPA recommended The emission rates of benzene, toluene, ethylbenzene, xylenes, and ethyne in both the 2002 and 2005 NEI agreed within the error limits and were the same order of magnitude as the TF emission rates (Table 4).The emission rates of benzene, ethylbenzene, and xylenes were fairly similar (within ±15%) in the two versions of the NEI.The relative distribution of emissions in the NEI and at TF was the same in 2002 with toluene > xylenes ∼ ethyne > benzene > ethylbenzene.Toluene and ethylbenzene had the highest and lowest emission rates, respectively, in the 2005 NEI as well.Overall, it appears that the 2002 NEI emission estimates were in better agreement with the TF emission rates than the 2005 NEI.
A notable difference between the two versions of the NEI is the higher toluene emissions from nonroad sources (specifically recreational, lawn, and garden equipment) in 2005 than in 2002 (Table 4).While still within the error limits, the largest difference between the central TF emission rate and the NEI was for toluene (out of the five compounds shown in Table 4).The 2002 and 2005 total NEI toluene emissions were 30% and 50%, respectively, higher than the TF estimates.Toluene has many potential sources in this region (e.g., vehicular exhaust, fuel evaporation, solvent utilization, biogenic, White et al., 2009) which is a likely fac-tor contributing to the disagreement between the inventory and our estimates.Previous studies focused on the eastern US have reported that toluene emissions were overestimated in earlier versions of the NEI.For example, Choi et al. (2006) found that toluene emissions were overestimated in a modified version of the 1996 NEI compared to PAMS measurements made in Mid-Atlantic states.In contrast to this work in which the higher toluene in 2005 was associated with larger nonroad emissions, the overestimate reported by Choi et al. (2006) was related to solvent utilization.Furthermore, WRF-CHEM model predictions based on the 1999 NEI of the vertical profile of toluene over the northeast US during the NEAQS-ITCT 2004 campaign were approximately a factor of 3 larger than observations (Warneke et al., 2007).Toluene is an important precursor of O 3 and SOA which highlights the critical need for updated source profiles, improved inventory emission estimates, and additional observational constraints on toluene emissions.
The total 2002 NEI and TF ethyne emission rates agreed remarkably well (∼2300 Mg/year) (Table 4).However, the total 2005 NEI ethyne emission rate was ∼40% lower than the central TF value, but was still within the error limits of the TF emission rate.Ethyne emissions in the nonpoint source category reflect residential stationary source fuel combustion with a minor contribution from open burning (yard and household waste).Residential wood and fireplace Table 4. Emission rates (Mg/year) of benzene, toluene, ethylbenzene, xylenes (m+p+o), and ethyne and the toluene/benzene and benzene/ethyne ratios from the 2002 and 2005 EPA National Emissions Inventory (NEI) for New Hampshire.The TF column is the emission rate (Mg/year) estimated using the winter 2006 TF GC nighttime measurements from Table 3.The toluene/benzene and benzene/ethyne ratios were calculated by converting the emission rates for TF and for the respective source category listed in this table to molar concentrations.The TF toluene/benzene and benzene/ethyne ratio errors were propagated from the error in the TF emission rates.
The toluene/benzene and benzene/ethyne ratios based on the NEI emissions for New Hampshire were also examined and compared to the ambient ratios (Sects.4.2 and 4.3) (Table 4).Both the 2002 and 2005 NEI toluene/benzene and benzene/ethyne ratios agreed within the error limits of the TF ratio values.The total benzene/ethyne ratios in the 2002 (0.26) and 2005 (0.35) NEIs were lower than the ratios given in Parrish (2006) for the 1996 and 1999 NEIs (0.9 and 0.7, respectively).The benzene/ethyne ratio in the onroad exhaust source category (∼0.5) was higher than ambient ratios (∼0.2).Our results are consistent with Parrish (2006) who reported that there was a downward trend in NEI benzene/ethyne ratio values but that there are still problems with the apportioning of benzene and ethyne emissions in the onroad source category.Overall, the range of NEI toluene/benzene and benzene/ethyne ratio values in the different source categories are within the range of the ambient values observed at TF (Table 4,Figs. 9,10).The general agreement between our results and the inventory may be fortuitous or it may indicate that VOC emissions are more accurately represented in the 2002 and 2005 NEIs than in earlier versions.Furthermore, these results suggest that the ethyne emission rate from the NEI may be useful for estimating emission rates of other VOCs by using their emission ratios relative to ethyne from the daily canister samples.This may be a valuable tool for modeling or predicting ambient VOC concentrations.

Summary
This work characterized the mixing ratios, seasonal to interannual variability, and sources of C 2 -C 8 NMHCs, C 2 HCl 3 , and C 2 Cl 4 from samples collected during January 2004-February 2008 at Thompson Farm in Durham, NH.The midday canister samples provided a comprehensive and representative picture of the day-to-day and interannual VOC trends and captured a wide range of mixing ratios and various sources (fossil fuel combustion, gasoline, LPG, fuel or solvent evaporation, industry, biogenic).Estimates of NMHC emission rates using the daily canister sample and automated in situ TF GC measurements ranged from 10 9 -10 10 molecules cm −2 s −1 .Additionally, benzene, toluene, ethylbenzene, xylene, and ethyne emission rates from the 2002 and 2005 EPA National Emissions Inventory were within ±50% of the TF emission rates.The alkanes, ethyne, benzene, toluene, and halocarbons exhibited consistent and reproducible seasonal trends each year, whereas the alkenes and C 8 aromatics illustrated greater interannual variability reflecting their shorter lifetimes and/or varying sources or emission rates.Furthermore, this multi-year data set illustrated that the mixing ratios of reactive NMHCs (specifically C 5 -C 6 alkanes, propene, toluene), and C 2 HCl 3 were enhanced in the summer relative to their springtime minimums indicating a persistent influence from evaporative emissions each year.Previous fuel emission studies have observed increased evaporative emissions from gasoline containing ethanol (e.g., Durbin et al., 2007;Graham et al., 2008).The results from this study will be valuable for evaluating and documenting the impact of new federal regulations enacted in 2005-2007 for motor vehicle emissions (http: //www.epa.gov/otaq/gasoline)(i.e., MTBE ban and switch to E10) and dry cleaning solvents on the ambient levels of VOCs in this region.

Fig. 1 .
Fig. 1.Location of the UNH AIRMAP Atmospheric Observing Station at Thompson Farm in Durham, New Hampshire.

Fig. 4 .
Fig. 4. (a) Time series of isoprene (pptv) (green line) and the monthly mean temperature ( • C) (gray line) at TF during January 2004-February 2008.(b) Correlation between log(isoprene) (in ppbv) and the hourly average temperature (corresponding to the hour the canister sample was collected) during June-September of2004, 2005, 2006, and 2007.

Fig. 5 .
Fig. 5. Time series of (a) C 2 HCl 3 and (b) C 2 Cl 4 (pptv) at TF during January 2004-February 2008.Black line and circles are the monthly mean ± standard error mixing ratios.

Fig. 7 .
Fig. 7. Correlation (m=slope ± standard error (r 2 )) of n-butane with (a) propane and (b) i-butane for each year.(c) Correlation between n-pentane and i-pentane.Gray diamonds are all of the samples collected in October-May of 2004-2008.Red squares are all of the samples collected in June-September of 2004-2007.

Fig. 11 .
Fig. 11.Comparison between (a) propane, (b) ethyne, (c) benzene, and (d) toluene data from the automated TF GC system during December 2005-January 2006 and the daily canister samples collected during the same time period.Note: the canister samples with mixing ratios >95th percentile for each month have been removed.

Table 2 .
Annual C 2 HCl 3 and C 2 Cl 4 statistics (pptv) for 2004-2008, and the rate of decrease (pptv/year) in the annual background mixing ratios ± standard deviation (SD).N is the number of samples.Background is the 10th percentile for the entire year.

Table 3 .
Emission rates (molecules cm −2 s −1 ) of C 3 -C 8 NMHCs calculated using nighttime measurements made by the automated TF GC when a stable inversion layer developed during winter 2006.The error bars were calculated by propagating the standard error of the linear regression between the change in NMHC concentration per unit time (00:00-05:00 EST) and the assumed variation in nocturnal boundary layer height (125 m±75 m).The winter 2006 daily canister emission ratio (±standard error (SE)) was calculated from the slope of the correlation between a NMHC and ethyne.The winter 2006 daily canister emission rate (molecules cm −2 s −1 ) was calculated by multiplying the daily canister emission ratios by the TF GC ethyne emission rate.The error bars were calculated by propagating the error associated with the winter 2006 TF GC ethyne emission rate and the standard error of the NMHC/ethyne emission ratio.New Hampshire (land area=2.3×10 10 m 2 ) and New England (land area=1.6×10 11m 2 ) emissions (Mg/day) were extrapolated from the winter 2006 TF GC emission rates.
was the largest nonpoint source of VOC emissions in northern New England in both versions of the NEI.According to the EPA, the emission factor used to calculate emissions for the woodstove and fireplace source category was reduced for the 2005 NEI (www.epa.gov/ttn/chief/net/2005inventory).Consequently, the VOC emission rate from the woodstove and fireplace source category was ∼70% lower in the 2005 NEI than in the 2002 NEI.This resulted in the lower ethyne emissions in 2005 and the larger discrepancy with the TF emission rates.This illustrates the necessity of additional studies to accurately quantify the VOC emissions from residential combustion in the northeast US (e.g., combustion