Verifying emission factors and national POPs emission inventories for the UK using measurements and modelling at two rural locations

Abstract

Different approaches are used to verify the adequacy of emission factors (EFs) and their use in emission inventories of persistent organic pollutants (POPs). The applicability of EFs was tested using atmospheric dispersion modelling to predict atmospheric concentrations of polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs) and resulting toxic equivalents (ΣTEQ) and particulate matter <10 μm (PM₁₀) in two rural locations in northern England (UK). The modelling was based on general assumptions of fuel composition, consumption and heating needs to simulate emissions of POPs from the domestic burning of coal and wood where ambient measurements were made in the winter of 1998. The model was used to derive the local contribution to ambient air concentrations, which were estimated independently based on comparative air measurements. The results support the hypothesis that in both villages, the majority of PAHs and the lower chlorinated PCDFs were locally released. The situation for PCBs and polychlorinated naphthalenes (PCNs) was different. While the EFs show the release of both compound groups from the domestic burning of coal and wood, the ambient levels of these “legacy POPs” in the villages were still clearly dominated by other sources. Rural areas relying mainly on fossil fuels can exceed the proposed UK ambient air quality standard for benzo[a]pyrene during winter. The measured EFs were then used to estimate the importance of the domestic burning of coal and wood to national emission inventories for these compound classes. Extrapolations to the UK suggest that the domestic burning of pure wood and coal were minor emitters for chlorinated POPs but contributed strongly to PAH and PM₁₀ levels in 2000. Finally, the UK’s national POPs emission inventories based on source inventories and EF, as used here, were compared to estimates derived using the increase in atmospheric concentration of selected POPs.

---

Rainer LohmannPresent address: Graduate School of Oceanography, University of Rhode Island, Narragansett, RI, 02882, USA. E-mail: lohmann@gso.uri.edu; Fax: 001 401 874 6811; Tel: 001 401 874 6612.

Rainer Lohmann was born in Germany, in 1971. He received his PhD in Environmental Chemistry from Lancaster University, England (UK) in 2000, and a degree in Chemical Engineering from the European Higher Institute of Chemistry at Strasbourg (EHICS, France) in 1996. In 2000 he moved to the Massachusetts Institute of Technology as a postdoctoral fellow, and joined the Research Center for Ocean Margins (Bremen, Germany) as a fellow for most of 2004. Since November 2004 he has been Assistant Professor in Oceanography at the University of Rhode Island’s Graduate School of Oceanography. His current research interests are: the black carbon cycle and its effects on the (bio)availability of organic pollutants, the global fate of POPs and the use of passive samplers to understand the activities of organic compounds.

Introduction

Cities and industrialised areas are generally responsible for major emissions of a range of persistent organic pollutants (POPs), such as polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), naphthalenes (PCNs), dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs). For example, buildings continue to act as sources of PCBs to the environment.¹ The high concentration of combustion engines, cookers, heating systems, power stations and industrial plants are responsible for emissions of combustion-derived compounds such as PAHs and PCDD/Fs. It is no surprise that cities like London and Manchester in the UK routinely display much higher concentrations of PAHs, PCBs, PCDD/Fs and PCNs than rural/background sites in the UK.²

Abatement measures in the UK have resulted in reduced emissions of PCDD/F toxic equivalents (ΣTEQ) from waste incinerators and emissions of PAHs and PCDD/Fs from cars.³ On the other hand, diffusive emissions from the domestic burning of solid fuels are much more widespread and could be of major importance for POP concentrations in rural areas. Hence Lee et al. determined emission factors (EFs) for a range of POPs during the controlled burning of house coal and seasoned hardwood under controlled test conditions.⁴ Fuels and burning conditions were designed to simulate domestic burning for space heating as typically used in northern England.

In this paper, an attempt is made to verify the applicability of EFs using atmospheric dispersion modelling. The modelling is used to predict atmospheric concentrations of PCDD/Fs, ΣTEQ, PCBs, PCNs, PAHs, and particulate matter <10 μm (PM₁₀) for two rural locations, based on general assumptions of fuel composition, consumption and heating needs. The modelling was designed to simulate locations where ambient measurements were made in the winter of 1998. The model is used to derive the local contribution to ambient air concentrations, which were estimated independently based on comparative air measurements. The measured EFs are then used to estimate the importance of the domestic burning of coal and wood to national emission inventories for these compound classes. Finally, the UK’s national POPs emission inventories based on source inventories and EF, as used here, are compared to estimates using the increase in atmospheric concentration of selected POPs.

Methods

The sites

Clapham (187 households, 1991 census data) and Austwick (217 households) are ∼50 km east of Lancaster in north-west England (see Fig. 1). They are located on the western slopes of Ingleborough (720 m high), a peak in the Yorkshire Dales National Park. In both villages, domestic burning of coal and wood is thought to be high during winter months because there is no natural gas supply to the villages. Samplers were installed at ground-level within the two villages. Regional background concentrations were sampled at Hazelrigg, Lancaster University’s meteorological field station, located ca. 5 km outside the town, on the north-west coast of England, ca. 5 km from the Irish Sea (see Fig. 1). It is predominantly influenced by ‘clean air’ masses coming in from the Irish Sea to the west.⁵ Samples were taken concurrently using Andersen GPS-1 samplers equipped with a glass fibre filter (GFF) and two polyurethane foam plugs (PUF). Details of the sampling, analytical procedures and a summary of atmospheric concentrations are given in Lohmann et al.⁶

Fig. 1 Map of the sampling sites in north-west England in the UK. Image produced from the Ordnance Survey Get-a-map service. Image reproduced with kind permission from Ordnance Survey and MultiMedia Mapping.

Fuel use estimate

20% of households in both villages did not have central heating (1991 census data). In the absence of a local fuel use survey, as a first approximation, 50% of those houses were assumed to use wood for space heating, the remaining 50% coal. For the houses with central heating, 80% were considered to use oil or bottled liquid petroleum gas (LPG) for space heating, the remaining 20% coal. We therefore assumed the following fuel usage: 10% wood, 26% coal, 64% oil/LPG. Space heating was also assumed to be restricted to 7–9 am and to 5–11 pm. Additionally we assumed a proportion of ‘comfort heating’, i.e. the burning of 10 kg wood, at 25% of the houses every evening.

Energy consumption

The calculation of the energy consumption was based on the Building Research Establishment Domestic Energy Model.⁷ The amount of energy required for heating a ‘unit house’ was computed with fixed temperatures for the living room (21 °C) and adjacent, secondary, rooms (18 °C). The energy required for keeping these temperatures is a function of insulative losses, heating efficiencies and minimum and maximum ambient temperatures. Typical values were used for the calculation of a ‘unit house’ and measured temperatures from the Malham Tarn Field Centre were used to obtain the daily energy requirement for the ‘unit house’. The daily energy requirement was multiplied by a count of the number of houses to give the total energy usage of each village. Applying the above detailed proportion of solid fuel use by their respective energy contents gave the mass of each fuel burnt.

Emission factors

No EFs for oil or LPG were available and so these were assumed to have negligible emissions of PCDD/Fs, PCBs and PAHs. Indeed, results by Bröker et al.⁸ show emissions of PCDD/Fs from the burning of oil to be lower by factors of 5–10 compared to those for wood and coal. Otherwise, EFs determined in the recent accompanying paper⁴ were used to estimate emissions of organic pollutants and PM₁₀. The EFs for PCDD/Fs were found to be comparable to others, while EF for PCBs were lower than previously reported.

Dispersion modelling

All houses in the 2 villages were grouped into 5 blocks, of which 2 were in Clapham and 3 in Austwick (see Fig. 1 for the positioning of the housing blocks and the sampling sites, respectively). The housing blocks served as area sources and combined the majority of houses in both villages. The local topography was not included in the model runs. Atmospheric dispersion modelling system version 3 (ADMS 3)⁹ was used for the dispersion modelling. ADMS 3 describes the boundary layer by the boundary layer depth and the Monin–Obukhov length. Convective dispersion is described by a skewed Gaussian distribution. The time step was one hour. Hourly meteorological data (temperature, wind speed, wind direction, rainfall, cloud cover and relative humidity) from the Eskmeal Meteorological Office station in the nearby Lake District was taken for the dispersion modelling as these were the most suitable available data. The dispersion model used is linear with respect to emission rate. Hence concentration maps were generated for one compound, benzo[a]pyrene (BaP) and the concentrations of the other pollutants were obtained by using the ratio of the EFs relative to that of BaP.

The model assessed only the impact of the 5 housing blocks on the 2 sampling locations. Two scenarios were considered. One where background concentrations were not taken into account in the modelling runs; the other where ambient measurements made at the Hazelrigg site were assumed to represent background advected air. Three month averages of the model results were taken and compared to measured concentrations at the two sites. Simultaneous measurements made at the Lancaster site were taken as background concentrations.

To simplify the assumptions made, the pollutants were assumed to behave conservatively (i.e., with no loss mechanisms taken into account). Even though higher molecular weight PAHs and PCDD/Fs are up to ∼100% in the particulate phase, this simplification is unlikely to substantially affect the results, because: (i) the samplers were in the centres of the villages, in close proximity to the housing blocks; (ii) the greatest distance a pollutant emitted at any given point had to travel was ∼5000 m (the distance between the two villages); (iii) the travel time for the lowest wind speed (4 km h⁻¹) is ∼1 hour; (iv) reported atmospheric half-lives are in the order of a day for predominantly particle-bound species:¹⁰ and (v) measured and laboratory-based half-lives for PCDD/Fs and PAHs with a significant gaseous fraction are in the order of several days.^11,12 It is therefore hypothesised that particulate deposition or atmospheric degradation of the gaseous species was of minor significance for any pollutant, given travel times of less than an hour.

Results and discussion

General comments

The agreement between the modelled and measured concentrations of the various pollutants at the three to four day timescale was rather poor, as would be expected given the assumptions used to calculate the solid fuel use. Dispersion modelling will give more reliable estimates of long-term model runs. The ensuing discussion therefore focuses on the average concentrations over the measurement period (i.e., 3 months). In each case, measured concentrations refer to the sum of gas phase and particle-bound pollutants.

The comparison of measured and modelled concentrations is shown in Figs. 2–5, where ‘perfect’ agreement would place the modelled values on the 1–1 line. If the modelled values exceed those measured, they will be placed above and to the left of the 1–1 line; under-predictions will fall to the right and below the 1–1 line. In each case, modelled values are shown for: (i) local emissions in the two villages (i.e., without background); and (ii) adjusted values to which background concentrations were added. The further the modelled ambient concentrations appear below the 1–1 line, the less important the contribution of domestic burning was to ambient concentrations (note the logarithmic scales on both axes).

Fig. 2 (a) Measured versus modelled ambient PCDD/F concentrations with and without background values; (b) profile of the 2,3,7,8-substituted PCDD/Fs from the ambient air and the burning of coal and wood.

Fig. 3 (a) Measured versus modelled ambient PCB concentrations with and without background values; (b) selected PCB profile in ambient air and the burning of coal and wood.

Fig. 4 (a) Measured versus modelled ambient PAH concentrations with and without background values with EF from Lee et al.;⁴ (b) selected PAH profile in ambient air and the burning of coal and wood.

Fig. 5 (a) Measured versus modelled ambient PAH concentrations with and without background values with EF from Wenborn et al.;¹⁵ (b) selected PAH profile in ambient air and the burning of coal and wood.

PCDD/Fs. Predicted and measured PCDD/Fs values are in very good agreement (see Fig. 2), with the exception of the Cl₂DFs (top right corner). Concentrations of the other lower chlorinated homologue groups (Cl₁DFs, Cl₂DDs, Cl₃DFs) can almost entirely be attributed to the domestic burning of coal and wood, as was proposed previously.⁶ On the contrary, the local contribution of coal/wood burning to the ΣCl₄₋₈DD/Fs was minor, while background-adjusted values gave a close fit to measured values. On average, background-adjusted model results accounted for 112% (Clapham) and 87% (Austwick) of measured concentrations.

In the earlier study, principal component analysis (PCA) further implied that 2,3,7,8-Cl₄DF and -Cl₄DD were both characteristic of the ambient air in the villages. However, the modelling results suggest that domestic burning of coal and wood contributed only moderately to their ambient concentrations in the villages.

Cl₂DFs. For Cl₂DFs, it was shown earlier that ambient concentrations were dominated by a near coastal emission source.¹³ As a result, ambient concentrations of Cl₂DFs at the background site in Lancaster exceeded those in the villages during air mass movements from a westerly direction.⁶ These air mass transport episodes (i.e., advection) dominated over any emissions in the villages. This is reflected in Fig. 2, where local emissions do not account for the observed values, whereas those with background values from Lancaster far exceed the measured ones.

ΣTEQ. Domestic burning produces predominantly lower chlorinated congeners, notably PCDFs. The background corrected concentrations for 2,3,7,8-Cl₄DF and 2,3,4,7,8-Cl₅DF are well predicted, but 2,3,7,8-Cl₄DD was the only analyte for which the background corrected values fell short of the measured mean values (Fig. 2). Since 2,3,7,8-Cl₄DD was a major contributor to the ΣTEQ, this explains the under-prediction of the ΣTEQ.

The individual contributions to the PCDD/F-ΣTEQ differed strongly between the ambient measurements and coal and wood emissions (Fig. 2b). In the ‘pure’ coal and wood emissions, 2,3,7,8-Cl₄DF and 2,3,4,7,8-Cl₅DF dominated the ΣTEQ, with strong contributions from 1,2,3,7,8-Cl₅DD and 2,3,7,8-Cl₄DD. However, in the ambient air, the major contributors were 2,3,4,7,8-Cl₅DF and 1,2,3,7,8-Cl₅DD. This supports the conclusion that emissions from (pure) coal and wood were not sufficient to account for the ambient concentrations, nor the profiles in the villages.

PCBs. Two major differences between the situation for PCBs and PCDD/Fs were apparent. Firstly, local emissions were minor for all PCB congeners, according to the modelling results. Secondly, ambient PCB concentrations were, with few exceptions, rather underpredicted. However, background-adjusted PCB concentrations were close to the measured values in both villages, on average 94% in Clapham and 80% in Austwick (Fig. 3). There was no discernible difference evident between co-planar PCBs (#77, 126 and 169) and ortho-substituted PCBs. In fact, adjusting PCBs #126 and #77 for their background concentrations at Hazelrigg resulted in their over-prediction in both villages. This could be the result of either their elevated concentrations at our background site, or a weaker emission strength at the villages than predicted by the modelling or a combination of both. As a result, this comparison provides little verification of the EF for PCBs.

On the other hand, a comparison of the relative PCB profile in the ambient air of the villages and the emissions showed remarkably similar patterns (Fig. 3b). In all cases, PCBs 28/31 dominated the profile, although wood emissions displayed lower contributions for PCBs 17/18 and 28/31 than the other profiles. Conversely, wood emissions had higher contributions of PCBs 60/56, 61/74 and 153/132. The PCB profile from coal burning was closer to the observed ambient one. Interestingly, a similarity of emission profiles and ambient measurements was also observed for the group of PCNs.¹⁴

Regardless, results from the ambient modelling study (see above) show that pyrogenic PCB-emissions make up only a tiny fraction of total emissions to the atmosphere. This finding is also consistent with national emission inventories for PCBs (see below). The observed similarity of profiles might stem from the fact of similar product distributions from industrial formulations and burning emissions.

PAHs. Measurements showed highly elevated PAH concentrations in the villages, attributable to domestic burning of coal and wood. However, use of the EFs measured in the companion paper⁴ resulted in substantial over-prediction of village air concentrations (Fig. 4). Site-specific differences were also apparent; predicted concentrations were much higher than measured ones in Clapham (by factors of 5–10), while values for Austwick were closer to the measured ones. This supports the suspicion that the PAH-EFs derived for coal were probably not representative.⁴ Therefore, the emissions of coal and wood-derived PAHs were re-calculated using EF determined for coal and wood in an earlier study for the UK’s NAEI¹⁵ (Table 1). With these EFs, PAH-concentrations were over-predicted at Clapham by a factor of around 2, while concentrations were under-predicted at Austwick (Fig. 5). Background-adjusted ambient concentrations were only slightly higher, confirming the preponderance of local PAH emissions. A few PAHs were somewhat less well predicted: dibenz[a,h]anthracene (11 times) and fluorine (3×) were overpredicted, while benzo[b]fluoranthene (0.1×) and benzo[k]fluoranthene (0.2×) were underpredicted.

Table 1 Summary table with emission factors for PAHs, PCDD/Fs, PCBs, PCNs, and PM₁₀ for the domestic burning of coal and wood

PAHs/mg kg^−1	Coal^ab	Wood^b
a EF for coal and PM10.^15 b Ref. 4. c Ref. 27. d Relative potencies.^28
Naphthalene	123.3	8.2
Acenaphthylene	7.15	6.6
Acenaphthene	5.24	0.60
Fluorene	16.21	2.82
Phenanthrene	6.56	6.8
Anthracene	1.85	1.7
Fluoranthene	2.98	3.5
Pyrene	2.98	3.2
Benz[a]anthracene	1.79	0.81
Chrysene	1.67	0.74
Benzo[b]fluoranthene	0.07	0.33
Benzo[k]fluoranthene	0.02	0.23
Benzo[a]pyrene	1.55	0.56
Indeno[123-c,d]pyrene	1.19	0.38
Dibenzo[ah.ac]anthracene	1.79	0.06
Benz[g,h,i]perylene	0.82	0.32
ΣPAHs	175.2	43
ΣCl1−8DD/Fs/ng kg^−1	16 000	2500
ΣCl4−8DD/Fs	231	50
ΣTEQ^c	3.0	0.2
ΣPCBs/ng kg^−1	8800	630
#77	23	4.1
#126	1.8	0.20
#169	0.20	0.10
ΣTEQ^c	0.20	0.02
ΣPCNs/ng kg^−1	680	120
ΣREP-TEQ^d	0.009	0.002
PM10/g kg^−1	10.4	7.9

---

A comparison of the PAH profiles for the EF determined by Lee et al. and ambient measurements (Fig. 4b) highlighted their differences, especially for phenanthrene and the higher MW PAHs. The latter contributed >5% each to benz[a]anthracene, chrysene, benzo[a]pyrene and benzo[g,h,i]perylene in the coal emission profile, but were less abundant in the ambient measurements. The ambient profiles and the wood emissions were dominated by phenanthrene.

The PAH-emission profile as measured by Wenborn et al.¹⁵ displayed lower contributions of the higher MW PAHs (Fig. 5b), but the relative importance of phenanthrene and fluorene contributions was reversed compared to the ambient profiles. Otherwise the profiles matched rather well.

Possible reasons why modelled PAH concentrations were higher than measured by a factor of ∼2 at Clapham include: (i) imprecise assumptions about the number of households burning coal/wood; (ii) less households present than assumed (Clapham is in the Yorkshire Dales National Park and a significant proportion of houses in the area could be only temporarily occupied (as holiday cottages)); (iii) non-representative sample location(s); (iv) general modelling and analytical uncertainties; (v) a possible role of soot/enhanced localised deposition.¹⁶

In light of the modelling results giving higher predictions for all compounds groups in Clapham than in Austwick, it seems that general assumptions about the source strengths and meteorological factors/ topography played a major role. Hence the overprediction for PAHs is much more prominent than for the other POPs, as most PAHs were locally released.

PM₁₀. Unlike POPs, concentrations of PM₁₀ were not routinely measured, but only for selected samples.¹⁷ These showed mean concentrations of 46 and 42 μg m⁻³ in Clapham and Austwick, respectively, and 29 μg m⁻³ at the background site. Use of the PM₁₀ EFs measured in the companion paper⁴ resulted in substantial over-prediction for background adjusted ambient concentrations (∼60–90 μg m⁻³; data not shown). Therefore, the emissions of PM₁₀ were re-calculated using EF determined for coal and wood in an earlier study for the UK’s NAEI¹⁵ (Table 1). These resulted in mean concentrations (43 and 35 μg m⁻³ for Clapham and Austwick, respectively) close to the measured values. These numbers suggest a significant local contribution (see below).

A thorough description of the uncertainties inherent in our study is limited by the fact that numerous uncertainties cannot be easily quantified. The analytical uncertainties linked to the measurements of the emission factors, and the ambient air measurements themselves are easily quantifiable; relative standard deviations are on the order of 20–30%. Less easily quantifiable are the uncertainties linked to: (i) the representativeness of the sampling locations; (ii) the assumptions about the fuel use; (iii) the errors introduced by the modelling itself and through the simplifications used, such as reduction to 5 housing blocks; (iv) the use of meteorological data from the nearest station; and (v) neglecting the local topography. Finally, there are questions regarding how representative the emission factors are for a real household situation, as they ignore the lighting of the fire and the addition of any household waste items.

Importance of local releases to the ambient concentrations in the villages. The modelling of ambient, near-ground concentrations of pollutants in two villages in NE England enabled the local contribution of domestic burning to be estimated. Two approaches are compared in Table 2; in an earlier study, the local contribution of various compounds was estimated, based on the difference between mean atmospheric concentrations at the villages and the background station (measurements at Lancaster University’s field station, Table 2).

Table 2 Estimated contribution of the local domestic burning of coal and wood on ambient concentrations of PCDD/Fs, PCBs, PCNs, PAHs and PM₁₀ in two villages in Northern England

	Comparative measurements^a	Modelling^b
a %local = [local measurement−background value (Lancaster)]/local measurement. b %local = local release/measured values. c Excluding Cl2DFs. d Ref. 27. e Ref. 4. f Ref. 15.
ΣCl1−8DD/Fs^c	53–57%	29–68%
ΣCl4−8DD/Fs	17–24%	5–11%
ΣTEQ^d	20–26%	5–9%
ΣPCBs	22–29%	2–3%
#77	(<0)–18%	4–13%
#126	(<0)	6–13%
#169	36–43%	2–4%
ΣPAHs^e	71–81%	160–310%
ΣPAHs^f	71–81%	62–120%
ΣPCNs	Not analysed	1–2%
PM10^e	30–36%	68–135%
PM10^f	30–36%	15–30%

---

PCDD/Fs. For ΣCl₁₋₈DD/Fs, local emissions contributed >50% according to comparative measurements, while the EF-based modelling suggested around 30–70%. ΣCl₁₋₈DD/F concentrations were dominated by the lower chlorinated PCDFs. This becomes evident when looking at the local releases for ΣCl₄₋₈DD/Fs, for which comparative measurements (17–24%) and EF-based modelling (5–11%) both derived minor contributions. The same holds true for the ΣTEQ, for which comparative measurements suggested 20–26% were locally released, while the EF-based modelling derived <10%. Unaccountably elevated concentrations of 2,3,7,8-Cl₄DD measured in the villages explains this major difference, which was not mirrored in the EF used for the modelling.

PCBs. Comparative measurements suggested that ca. 22–29% of PCBs were locally emitted. However, the EF-based modelling suggested just 2–3% was derived from the domestic burning of coal and wood. It is suggested that the increase in ambient PCB concentrations in the villages may be from the broad relationship between volatile emissions and populated areas, due to the range of past uses of these compounds.¹⁸

PAHs. In the comparative measurements, the local contribution to ambient PAH-levels was estimated to be around 70–80%. Use of the EFs determined by Wenborn et al.¹⁵ resulted in a comparable importance of local emissions (60–120%), depending on the village. Again, use of the EF for coal in the companion paper⁴ resulted in modelled concentrations which exceeded measured ones by up to a factor of 1.5–3.

PCNs. While PCN concentrations were not measured during the field campaign in 1998, the EF derived by Lee et al.⁴ can be used to predict the concentrations due to the domestic burning of coal and wood. The importance of these local emissions can be estimated by assuming that background levels from Lancaster were applicable, in a similar manner to the calculations for the other compound groups. Mean PCN concentrations from measurements at the same field station in Lancaster for 2001 were used.¹⁴ The predicted local contribution to ambient PCN-levels in the villages is very small (1–2%), and comparable to those estimated for PCBs.

PM₁₀. For selected samples, PM₁₀-concentrations were determined at the different sites.¹⁷ By comparing these measurements, local emissions were found to potentially account for one third of PM₁₀ concentrations in the villages. The emission modelling came to a similar order of magnitude using the EF from Wenborn et al.,¹⁵ with local contributions of 15–30%. However, use of the EF determined in Lee et al.⁴ for coal resulted in much higher predictions (70–140%).

Benzo[a]pyrene. The proposed UK national air quality standard for BaP is an annual average of 0.25 ng m⁻³.¹⁹ Measured concentrations of BaP were often in breach of that value during the sampling period (see Fig. 2). Mean concentrations of BaP were 1.1 and 1.5 ng m⁻³ over the entire sampling periods in Clapham and Austwick, respectively. If average BaP background concentrations were on the order of ∼0.15 ng m⁻³ over the remaining 10 months of the year, the annual average would not be exceeded. It is likely that concentrations will be lower during the summer when less solid fuel is used. Concentrations of BaP at Hazelrigg, Lancaster, in the summer months are in the order of ∼0.04–0.1 ng m⁻³.²

The good agreement between measured and modelled local emissions of PAHs and PCDD/Fs provides a rough field validation of the quality of the EF used here. This field validation was not possible for PCBs and PCNs, presumably due to the dominance of secondary sources. From this we are encouraged to apply the EF to the UK as whole to re-visit estimations of the national contributions of the domestic burning of coal and wood.

National importance of domestic burning

Lee et al.⁴ used their determined EFs to calculate the national emissions from domestic coal and wood burning, assuming they are representative for the country as a whole. These emissions are based on national domestic coal and wood consumptions of ca. 2.4 and 0.7 million tonnes per year in the late 1990s, respectively.²⁰ These domestic burning emissions were compared to national atmospheric emissions inventories (NAEI) as compiled by the UK’s government for PAHs, PCBs and PCDD/Fs²¹ (Table 3). A crude relationship between global PCB and PCN production was used as a scaling factor for PCN emissions.⁴ Their results are briefly summarised here.

Table 3 Estimated inputs from the domestic burning of coal and wood to the UK atmosphere and potential contribution to the national inventory with emission factors from Lee et al.

	Wood + coal	UK NAEI^e for 2000	Ambient estimation^f	% due to domestic burning
a EF for coal and PM10.^15 b ΣPAHs: acenapthene, fluorene, phenanthrene, anthracene, pyrene, benz[a]anthracene, chrysene, benzo[a]pyrene, benzo[g,h,i]perylene, Σbenzo[b,j,k]fluoranthenes. c ΣPCBs: PCBs #28, #52, #101, #118, #138, #153, #180. d Ref. 27. e Ref. 21. f Based on measurements at High Muffles, see text.
ΣPAHs/tonnes year^−1^ab	106	2200	940	11%
BaP/tonnes year^−1^a	4.6	10	3.7	124%
ΣPCBs/kg year^−1^c	2.6	1600	4000	0.1%
ΣCl1−8DD/Fs / kg year^−1	40	320	400	10%
ΣTEQ/g year^−1^d	7.3	320	255	3%
ΣPCNs/kg year^−1	1.7	na	9000	0.0%

---

For PAHs and PM₁₀, domestic burning of coal and wood accounted for ca. 15–20% of nationwide emissions, which was close to that estimated by the NAEI. For PCBs and PCNs, emissions from the domestic burning made up ca. 1–2% of annual emissions. It was concluded that for these industrially produced high volume chemicals, volatilisation of the industrial formulations dominated ambient air concentrations. For PCDD/Fs -ΣTEQ, NAEI estimated a 20% contribution from domestic burning, whereas the use of the EF determined by Lee et al. put ΣTEQ emission from domestic burning at just 2% of total emissions.

Estimating national emissions from ambient measurements

Another way of estimating annual emissions for the UK is to extrapolate ambient measurements from representative locations. Thus Lohmann et al.¹¹ estimated the annual PCDD/F-TEQ emissions based on measurements at the Eastern coast of the UK (High Muffles) and the west coast (Lancaster) for favourable weather transport periods. The results (ca. 700–1300 g TEQ year⁻¹) were found to be in good agreement with other estimates of ca. 560–1100 g TEQ year⁻¹ by Eduljee and Dyke²² and 220–660 g TEQ year⁻¹ by Alcock et al.³

In general, the UK receives westerly winds from the Atlantic. These air masses will receive emissions from the densely populated west coast conurbations (such as Liverpool, Manchester etc.) while moving across the UK. The air will become well mixed when crossing the upland Pennines, which run north-south through the middle of northern England, before being sampled at the North Yorkshire Moors station (High Muffles, see Fig. 1).

Measurements from High Muffles, a rural location, were performed there as part of the toxic organic micropollutants (TOMPs) network and could be used for estimating annual emissions for the UK in 2000. In this simplified approach, the UK was considered a box with a width of 150 km, an area of 200 000 km² and an average mixed boundary layer height of 750 m. Based on an average wind speed of 3 m s⁻¹, the UK atmosphere would be replaced almost twice per day (630 year⁻¹). Background, input values were taken from measurements at Mace Head, at the western coast of Ireland (see Fig. 1).

TOMPs reported the following concentrations as an annual average for High Muffles in 2000: 10 ng m⁻³ for ΣPAHs (0.04 for BaP), 54 pg m⁻³ for ΣPCBs and 6.3 fg TEQ m⁻³ for PCDD/Fs;²³ 110 pg m⁻³ for ΣPCNs, albeit measurements from Lancaster in 2001.¹⁴ For the background air concentration, the following concentrations were determined at Mace Head, Ireland: 99 pg m⁻³ for ΣPAHs (1.1 for BaP) and 12 pg m⁻³ for ΣPCBs, all in 1998;²⁴ 3.6 fg TEQ m⁻³ for PCDD/Fs in 1997;¹¹ 15 pg m⁻³ for ΣPCNs in 2000.¹⁴

The results of these ‘ambient estimations’ suggest lower emissions for ΣPAHs (factor 2 lower) and BaP (factor 3) but a good agreement for PCDD/Fs and ΣTEQ (Table 3). ΣPCB emissions are higher by a factor of 2–3, while there is no NAEI estimation for ΣPCNs. An important result from this comparison is that PCBs and PCNs appear to have comparable annual emissions in the UK, despite a much lower global production volume of the latter compounds.²⁵ In all cases, similar conclusions can be drawn regarding the importance of domestic burning of coal and wood. It seems negligible for ΣPCBs and ΣPCNs, of minor importance for ΣTEQ and most important for PAHs (and ΣPM₁₀).

Outlook and discussion

There was broad agreement for the importance of local emissions, based on comparative measurements and the EF-based village modelling. These results support the hypothesis that in both villages, the majority of PAHs and the lower chlorinated PCDFs were locally released. These organic pollutants are hence good indicators of the domestic burning of coal and wood. Domestic burning of solid fuels was calculated to account for ∼15–20% of the ΣTEQ at the villages, in line with estimates based on differences in ΣTEQ concentrations from Lancaster to Austwick/Clapham.⁶ The situation for PCBs and PCNs was different. The EFs determined in the companion paper by Lee et al.⁴ demonstrate that both compound groups are emitted from the domestic burning of coal and wood. However, ambient levels in the villages were still clearly dominated by other sources. Comparison of model results and ambient measurements suggest that PAH-EF values determined by Wenbornn et al. for coal and PM₁₀ are likely to be more representative of the ambient conditions considered here.¹⁵

The study confirms and validates the use of EFs, as derived from controlled burning experiments, in the UK’s national emission inventory. In the national context, the contribution of the burning of coal and wood is of minor importance for PCBs and PCNs. Conversely, it is one of the main emission sources for PAHs and the lower chlorinated PCDD/Fs. It should be noted in this context, that the EFs used here are only valid for the controlled burning of pure materials. The burning of household waste or treated wood can increase emissions of PCDD/Fs, for example.²⁶

Acknowledgements

We are grateful to the UK Department of the Environment, Food and Rural Affairs for financial support. This is publication number 0342 from the Research Center for Ocean Margins (RCOM).

References

1. (a) R. E. Alcock, R. E. Alcock, C. J. Halsall, C. A. Harris, A. E. Johnston, W. A. Lead, G. Sanders and K. C. Jones, Environ. Sci. Technol., 1994, 28, 1836–1842; (b) M. Kohler, J. Tremp, M. Zennegg, C. Seiler, S. Minder-Kohler, M. Beck, P. Lienemann, L. Wegmann and P. Schmid, Environ. Sci. Technol., 2005, 39, 1967–1973.
2. TOMPs, Results from the Toxic Organic Micropollutants (TOMPS) network: 1991 to 1995, Report for the Department of Environment, AEAT-1643/16419146/Issue 1, 1997, http://www.defra.gov.uk/environment/statistics/index.htm.
3. R. E. Alcock, R. Gemmill and K. C. Jones, Chemosphere, 1999, 38, 759–770.
4. R. G. M. Lee, P. Coleman, J. L. Jones, K. C. Jones and R. Lohmann, Environ. Sci. Technol., 2005, 39, 1436–1447.
5. (a) R. G. M. Lee, H. Hung, D. Mackay and K. C. Jones, Environ. Sci. Technol., 1998, 32, 2172–2179; (b) R. G. M. Lee and K. C. Jones, Environ. Sci. Technol., 1999, 33, 705–712; (c) R. G. M. Lee, V. Burnett, T. Harner and K. C. Jones, Environ. Sci. Technol., 2000, 34, 393–398.
6. R. Lohmann, G. C. N. Northcott and K. C. Jones, Environ. Sci. Technol., 2000, 34, 2892–2899.
7. BREDEM, BRE Domestic Energy Model: background, philosophy and description, Department of the Environment, Building Research Establishment, Garston, Watford, UK, 1997, ISBN 0 85125 351 2.
8. G. Bröker, K. J. Geueke, E. Hiester and H. Niesenhaus, Emission polychlorierter Dibenzo-p-dioxine und – furane aus Hausbrandfeuerungen, LIS-Report 103, Landesanstalt für Immisionsschutz Nordrhein-Westfalen, Essen, Germany, 1992.
9. Cambridge Environmental Research Consultants, Industrial Air Pollution Model ADMS 3, Cambridge Environmental Research Consultants, Cambridge, UK, 1999.
10. W. A. J. van Pul, F. A. A. M. de Leeuw, J. A. van Jaarsveld, M. A. van der Gaag and C. J. Sliggers, Chemosphere, 1998, 37, 113–141.
11. R. Lohmann, N. J. L. Green and K. C. Jones, Environ. Sci. Technol., 1999, 33, 2872–2878.
12. M. F. Simcik, H. Zhang, S. J. Eisenreich and T. I. Franz, Environ. Sci. Technol., 1999, 31, 2141–2147.
13. R. Lohmann, N. J. L. Green and K. C. Jones, Environ. Sci. Technol., 1999, 33, 4440–4447.
14. R. G. M. Lee, G. O Thomas and K. C. Jones, Environ. Sci. Technol., 2005, 39, 4729–4738.
15. M. J. Wenborn, P. J. Coleman, N. R. Passant, E. Lymberidi, J. Sully and R. A. Weir, Speciated PAH inventory for the United Kingdom, Report AEAT 3512, AEA Technology, 1999, http://www.airquality.co.uk/archive/reports/list.php.
16. (a) J. Dachs and S. J. Eisenreich, Environ. Sci. Technol., 2000, 34, 3690–3697; (b) B. T. Mader and J. F. Pankow, Environ. Sci. Technol., 2002, 36, 5218–5228; (c) R. Lohmann and G. Lammel, Environ. Sci. Technol., 2004, 38, 3793–3803.
17. R. Lohmann, T. Harner, G. O. Thomas and K. C. Jones, Environ. Sci. Technol., 2000, 34, 4943–4951.
18. (a) K. Breivik, A. Sweetman, J. M. Pacyna and K. C. Jones, Sci. Tot. Environ., 2002, 290, 181–198; (b) K. Breivik, A. Sweetman, J. M. Pacyna and K. C. Jones, Sci. Tot. Environ., 2002, 290, 199–224.
19. Expert Panel on Air Quality Standards (EPAQS), Polyaromatic Hydrocarbons, DETR, London, 1999.
20. http://www.dti.gov.uk/energy/inform/energy_trends/coal_art_ sep2001.pdf .
21. National Atmospheric Emission Inventory (NAEI), http://www.naei.org.uk/.
22. G. H. Eduljee and P. Dyke, Sci. Tot. Environ., 1996, 177, 303–321.
23. http://www.defra.gov.uk/environment/statistics/index.htm .
24. A. Sweetman, Lancaster University, Lancaster, UK, personal communication.
25. J. Falandysz, Environ. Pollut., 1998, 101, 77–90.
26. (a) C. Huebner, R. Boos and T. Prey, Chemosphere, 2005, 58, 367–372; (b) P. M. Lemieux, C. C. Lutes and D. A. Santoianni, Prog. Energy Combust. Sci., 2004, 30, 1–32.
27. M. Van den Berg, L. Birnbaum, A. T. C. Bosveld, B. Brunström, P. Cook, M. Feeley, J. P. Giesy, A. Hanberg, R. Hasegawa, S. W. Kennedy, T. Kubiak, J. C. Larsen, R. F. X. van Leeuwen, A. K. D. Liem, C. Nolt, R. E. Peterson, L. Poellinger, S. Safe, D. Schrenk, D. Tillit, M. Tysklind, M. Younes, F. Waern and T. Zacharewski, Environ. Health Perspect., 1998, 106, 775–792.
28. A. L. Blankenship, K. Kannan, S. A. Villalobos, D. L. Villeneuve, J. Falandysz, T. Imagawa and J. P. Giesy, Environ. Sci. Technol., 2000, 34, 3153–3158.

---

Footnotes

† Presented at the Fifth International Symposium on Modern Principles of Air Monitoring & Biomonitoring, June 12–16 2005, Norway.

‡ Present address: Graduate School of Oceanography, University of Rhode Island, Narragansett, RI, 02882, USA. E-mail: lohmann@gso.uri.edu; Fax: 001 401 874 6811; Tel: 001 401 874 6612.

---