Pollution influences on atmospheric composition and chemistry at high northern latitudes: Boreal and California forest fire emissions

We analyze detailed atmospheric gas/aerosol composition data acquired during the 2008 NASA ARCTAS (Arctic Research of the Composition of the Troposphere from Aircraft and Satellites) airborne campaign performed at high northern latitudes in spring (ARCTAS-A) and summer (ARCTAS-B) and in California in summer (ARCTAS-CARB). Biomass burning influences were widespread throughout the ARCTAS campaign. MODIS data from 2000 to 2009 indicated that 2008 had the second largest fire counts over Siberia and a more normal Canadian boreal forest fire season. Near surface arctic air in spring contained strong anthropogenic signatures indicated by high sulfate. In both spring and summer most of the pollution plumes transported to the Arctic region were from Europe and Asia and were present in the mid to upper troposphere and contained a mix of forest fire and urban influences. The gas/aerosol composition of the high latitude troposphere was strongly perturbed at all altitudes in both spring and summer. The reactive nitrogen budget was balanced with PAN as the dominant component. Mean ozone concentrations in the high latitude troposphere were only minimally perturbed (<5 ppb), although many individual pollution plumes sampled in the mid to upper troposphere, and mixed with urban influences, contained elevated ozone (ΔO₃/ΔCO = 0.11 ± 0.09 v/v). Emission and optical characteristics of boreal and California wild fires were quantified and found to be broadly comparable. Greenhouse gas emission estimates derived from ARCTAS-CARB data for the South Coast Air Basin of California show good agreement with state inventories for CO₂ and N₂O but indicate substantially larger emissions of CH₄. Simulations by multiple models of transport and chemistry were found to be broadly consistent with observations with a tendency towards under prediction at high latitudes.     

A tropical rainforest clearing experiment by biomass burning in the state of Pará, Brazil

Results are described of a forest clearing experiment conducted in Tomé Acu, located approximately 250 km south of Belém, the capital of the Brazilian northern state of Pará. An area of 3 ha of virgin forest was cut in July 1994 and left to dry until October of the same year, when fire was set. Post burning was also performed 30 days after the main fire. The test location biomass content per hectare was measured by indirect methods using formulas with parameters of forest inventories. The carbon content of the several biomass compartments was determined in a CHN analyzer. The combustion completeness was estimated by selecting ten 2+2 m² areas and 24 large trunks and examining their consumption rates by fire. The 2+2 m² areas were used to determine the completeness of small parts of biomass (those whose characteristic diameters were lower than 10 cm) and the trunks to determine the efficiency of the larger parts (characteristic diameters larger than 10 cm). The overall process combustion completeness was estimated to be 20.1%. Considering that the combustion gases of carbon in open fires contain approximately 90% of CO₂ and 10% of CO in volumetric basis, the emission rates of these gases by the burning process were estimated as 70.2 and 5.0 t ha^-1, respectively.     

Validation of ADMS against wind tunnel data of dispersion from chemical warehouse fires

Comparisons are presented of the predictions of the atmospheric dispersion modelling system (ADMS) and wind tunnel data for plume dispersion from chemical warehouse fires. The focus of the comparisons is dispersion from structurally intact buildings with open roofs and dispersion of plumes flush with the ground without obstacles, however, dispersion from building shells and doors is also considered. Both buoyancy driven and momentum driven flows are treated, although emphasis is on buoyancy driven flows as these are generally more likely to occur in warehouse fires. The study shows that the ADMS building module is able to reproduce many of the features of dispersion observed in the wind tunnel. These include a recirculating region behind the building in which material may be trapped, a main wake which brings material down towards the surface, and appropriate sensitivity to the buoyancy and momentum of the emitted material, and the location of sources on the building roof. The comparisons suggest that the ADMS building model can be used to predict dispersion from the stages of fire development studied. The precise level of agreement depends (but not in a systematic way) on the buoyancy flux parameter F_B, the momentum flux parameter F_M and the number of roof lights. There are some significant differences between the wind tunnel boundary layer and the simulated atmospheric boundary layer in ADMS which have to be considered when making wind tunnel model comparisons. These relate mainly to the near surface where the wind tunnel underestimates turbulent velocities, the boundary layer height which in the wind tunnel corresponds to an atmospheric boundary layer depth of 82.5 m (atmospheric boundary layers are frequently an order of magnitude deeper), and the boundary layer top where the ADMS boundary layer is capped by an inversion and has low turbulence levels whereas the wind tunnel boundary layer has higher levels of turbulence and no capping inversion.     

Trace gas variations at Cape Point,South Africa,during May 1997 following a regional biomass burning episode

During the continuous monitoring of atmospheric parameters at the station Cape Point (34°S, 18°E), a smoke plume originating from a controlled fire of 30-yr-old fynbos was observed on 6 May 1997. For this episode, which was associated with a nocturnal inversion and offshore airflow, atmospheric parameters (solar radiation and meteorological data) were considered and the levels of various trace gases compared with those measured at Cape Point in maritime air. Concentration maxima in the morning of 6 May for CO₂, CO, CH₄ and O₃ amounted to 370.3 ppm, 491 ppb, 1730 ppb and 47 ppb, respectively, whilst the mixing ratios of several halocarbons (F-11, F-12, F-113, CCl₄ and CH₃CCl₃) remained at background levels. In the case of CO, the maritime background level for this period was exceeded by a factor of 9.8. Differences in ozone levels of up to 5 ppb between air intakes at 4 and 30 m above the station (located at 230 m above sea level) indicated stratification of the air advected to Cape Point during the plume event. Aerosols within the smoke plume caused the signal of global solar radiation and UV–A to be attenuated from 52.4 to 13.0 mW cm⁻² and from 2.3 to 1.3 mW cm⁻², respectively, 5 h after the trace gases had reached their maxima. Emission ratios (ERs) calculated for CO and CH₄ relative to CO₂ mixing ratios amounted to 0.042 and 0.0040, respectively, representing one of the first results for fires involving fynbos. The CO ER is somewhat lower than those given in the literature for African savanna fires (average ER=0.048), whilst for CH₄ the ER falls within the range of ERs reported for the flaming (0.0030) and smouldering phases (0.0055) of savanna fires. Non-methane hydrocarbon (NMHC) data obtained from a grab sample collected during the plume event were compared to background levels. The highest ERs (ΔNMHC/ΔCH₄) have been obtained for the C₂–C₃ hydrocarbons (e.g. ethene at 229.3 ppt ppb⁻¹), whilst the C₄–C₇ hydrocarbons were characterised by the lowest ERs (e.g. n-hexane at 1.0 and n-pentane at 0.8 ppt ppb⁻¹).     

Biomass and combustion characteristics of secondary mixed deciduous forests in Eastern Ghats of India

Biomass quantities at three different sites in tropical moist mixed secondary deciduous forests before and after burning have been quantified in the forest patches cleared for shifting cultivation purposes. The main objective of the experiments was to study the spatial variability in the amount of biomass burnt and the contribution of different components viz., bole, branch, and mixed leaf litter in the burning process affecting the combustion factors. Species composition at the three sites varied with respect to each other. The total density of species before burning at the three sites for more than 10 cm diameter were found to be 3192 (site 1), 1194 (site 2) and 1444 (site 3) stems/area, respectively. Analysis of the results from girth-class and density relationships suggests that nearly 80% of the stems occurred in the range from 10–40 cm girth for site one, 64.2% in 10–55 cm girth class for the site two and more than 80% of stems in 10–40 cm girth class for the site three, indicating very poor and secondary nature of the forest. The fire intensity is found to be high for site one with 60 847 kJ s⁻¹ m⁻¹, when compared to 31 086 and 42 789 kJ s⁻¹ m⁻¹ for second and third sites, respectively. The values are comparatively higher than the mean value of 2566 kJ s⁻¹ m⁻¹ reported for savanna fires. The individual combustion completeness suggested that among the different components of biomass, branch material with less than 10 cm and upto 5 cm dbh contributed to more than 60% of combustion. Mixed leaf litter contributed to about more than 50% of individual combustion completeness at all the sites, with third site having the highest (84%). The contribution of dry biomass material having more than 70 cm diameter is found to be very low indicating that most of the trunks were burnt superficially. The overall combustion completeness suggested that mixed leaf litter and branch material contributed to most of the combustion. Of all the three sites, site three had been found to be having highest combustion completeness of about 30.04% when compared to 20.18% and 16.1% for first and second sites, respectively. In the study, comparison of combustion factors for different vegetation types has also been made.     

Sources and concentration of nanoparticles (<10 nm diameter) in the urban atmosphere

Whilst limited information on particle size distributions and number concentrations in cities is available, very few data on the very smallest of particles, nanoparticles, have been recorded. Measurements in this study show that road traffic and stationary combustion sources generate a significant number of nanoparticles of diameter <10 nm. Measurements at the roadside (4 m from the kerb) and downwind from the traffic (more than 25 m from the kerb) show that nanoparticles (<10 nm diameter) accounted for more than 36–44% of the total particle number concentrations. Measurements designed to sample the plume of individual vehicles showed that both a diesel- and a petrol-fuelled vehicle generated nanoparticles (<10 nm diameter). The fraction of nanoparticles was even greater in a plume 350 m downwind of a stationary combustion source. On a few occasions, a temporal association between nanoparticles in the size range 3–7 nm and solar radiation was observed in urban background air at times when no other local sources were influential, which suggests that homogeneous nucleation can also be an important source of particles in the urban atmosphere.     

Inventory of aerosol and sulphur dioxide emissions from India. Part II—biomass combustion

A spatially resolved biomass burning data set, and related emissions of sulphur dioxide and aerosol chemical constituents was constructed for India, for 1996–1997 and extrapolated to the INDOEX period (1998–1999). Sources include biofuels (wood, crop waste and dung-cake) and forest fires (accidental, shifting cultivation and controlled burning). Particulate matter (PM) emission factors were compiled from studies of Indian cooking stoves and from literature for open burning. Black carbon (BC) and organic matter (OM) emissions were estimated from these, accounting for combustion temperatures in cooking stoves. Sulphur dioxide emission factors were based on fuel sulphur content and reported literature measurements. Biofuels accounted 93% of total biomass consumption (577 MT yr⁻¹), with forest fires contributing only 7%. The national average biofuel mix was 56 : 21 : 23% of fuelwood, crop waste and dung-cake, respectively. Compared to fossil fuels, biomass combustion was a minor source of SO₂ (7% of total), with higher emissions from dung-cake because of its higher sulphur content. PM_2.5 emissions of 2.04 Tg yr⁻¹ with an “inorganic fraction” of 0.86 Tg yr⁻¹ were estimated. Biomass combustion was the major source of carbonaceous aerosols, accounting 0.25 Tg yr⁻¹ of BC (72% of total) and 0.94 Tg yr⁻¹ of OM (76% of total). Among biomass, fuelwood and crop waste were primary contributors to BC emissions, while dung-cake and forest fires were primary contributors to OM emissions. Northern and the east-coast India had high densities of biomass consumption and related emissions. Measurements of emission factors of SO₂, size resolved aerosols and their chemical constituents for Indian cooking stoves are needed to refine the present estimates.     

Fire and biofuel contributions to annual mean aerosol mass concentrations in the United States

We estimate the contributions from biomass burning (summer wildfires, other fires, residential biofuel, and industrial biofuel) to seasonal and annual aerosol concentrations in the United States. Our approach is to use total carbonaceous (TC) and non-soil potassium (ns-K) aerosol mass concentrations for 2001–2004 from the nationwide IMPROVE network of surface sites, together with satellite fire data. We find that summer wildfires largely drive the observed interannual variability of TC aerosol concentrations in the United States. TC/ns-K mass enhancement ratios from fires range from 10 for grassland and shrub fires in the south to 130 for forest fires in the north. The resulting summer wildfire contributions to annual TC aerosol concentrations for 2001–2004 are 0.26 μg C m⁻³ in the west and 0.14 μg C m⁻³ in the east; Canadian fires are a major contributor in the east. Non-summer wildfires and prescribed burns contribute on an annual mean basis 0.27 and 0.31 μg C m⁻³ in the west and the east, highest in the southeast because of prescribed burning. Residential biofuel is a large contributor in the northeast with annual mean concentration of up to 2.2 μg C m⁻³ in Maine. Industrial biofuel (mainly paper and pulp mills) contributes up to 0.3 μg C m⁻³ in the southeast. Total annual mean fine aerosol concentrations from biomass burning average 1.2 and 1.6 μg m⁻³ in the west and east, respectively, contributing about 50% of observed annual mean TC concentrations in both regions and accounting for 30% (west) and 20% (east) of total observed fine aerosol concentrations. Our analysis supports bottom-up source estimates for the contiguous United States of 0.7–0.9 Tg C yr⁻¹ from open fires (climatological) and 0.4 Tg C yr⁻¹ from biofuel use. Biomass burning is thus an important contributor to US air quality degradation, which is likely to grow in the future.     

The evolution of photochemical smog in a power plant plume

The evolution of photochemical smog in a plant plume was investigated with the aid of an instrumented helicopter. Air samples were taken in the plume of the Cumberland Power Plant, located in central Tennessee, during the afternoon of 16 July 1995 as part of the Southern Oxidants Study – Nashville Middle Tennessee Ozone Study. Twelve cross-wind air sampling traverses were made at six distance groups from 35 to 116 km from the source. During the sampling period the winds were from the west–northwest and the plume drifted towards the city of Nashville TN. Ten of the traverses were made upwind of the city, where the power plant plume was isolated, and two traverses downwind of the city when the plumes were possibly mixed. The results revealed that even six hours after the release, excess ozone production was limited to the edges of the plume. Only when the plume was sufficiently dispersed, but still upwind of Nashville, was excess ozone (up to 109 ppbv, 50–60 ppbv above background levels) produced in the center of the plume. The concentrations image of the plume and a Lagrangian particle model suggests that portions of the power plant plume mixed with the urban plume. The mixed urban power plant plume began to regenerate O₃ that peaked at 120 ppbv at a short distance (15–25 km) downwind of Nashville. Ozone productivity (the ratio of excess O₃ to NO_y and NO_z) in the isolated plume was significantly lower compared with that found in the city plume. The production of nitrate, a chain termination product, was significantly higher in the power plant plume compared to the mixed plume, indicating shorter chain length of the photochemical smog chain reaction mechanism.     

Identification and assessment of ship emissions and their effects in the harbour of Göteborg,Sweden

The effect of ship emissions in the urban environment of Göteborg has been studied by multivariate analysis. The simultaneous measurements of relevant gases and sub-micron particles make identification of ship plumes possible. Increased concentrations of these species due to ship emissions are quantified for ships entering the inner part of the harbour. Annual depositions of SO₂ and NO₂ are estimated to be 220 and 115 kg km⁻² yr⁻¹, respectively. Exposure of transient particles (less than 0.1 μm in diameter) to this part of the harbour increased by a factor of 3 in number concentration when a ship plume was recorded. Ni, Pb, V and Zn are shown to have positive correlation with NO emissions from ships.     

A comparison of operational Lagrangian particle and adaptive puff models for plume dispersion forecasting

Transport and dispersion of pollutants in the lower atmosphere are predicted by using both a Lagrangian particle model (LPM) and an adaptive puff model (APM2) coupled to the same mesoscale meteorological prediction model PMETEO. LPM and APM2 apply the same numerical solutions for plume rise; but, for advection and plume growth, LPM uses a stochastic surrogate to the pollutant conservation equation, and APM2 applies interpolated winds and standard deviations from the meteorological model, using a step-wise Gaussian approach. The results of both models in forecasting the SO₂ ground level concentration (glc) around the 1400 MWe coal-fired As Pontes Power Plant are compared under unstable conditions. In addition, meteorological and SO₂ glc numerical results are compared to field measurements provided by 17 fully automated SO₂ glc remote stations, nine meteorological towers and one Remtech PA-3 SODAR, from a meteorological and air quality monitoring network located 30 km around the power plant.     

Dispersion modelling of a wintertime particulate pollution episode in Christchurch,New Zealand

This paper examines the inter-suburb dispersion of particulate air pollution in Christchurch, New Zealand, during a wintertime particulate pollution episode. The dispersion is simulated using the RAMS/CALMET/CALPUFF modelling system, with data from a detailed emissions inventory of home heating, motor vehicles and industry. During the period 27 July–1 August 1995, peak 1 h and 24 h PM₁₀ concentrations of 368 and 107 μg m⁻³, respectively, were observed. Peak concentrations occurred at night, when particulate emissions from wood- and coal-burning domestic heating appliances were at a maximum and emitted into a stable boundary layer. The model is generally able to reproduce the observed PM₁₀ time series recorded at surface monitors located throughout the urban area. For this simulation, the fractional gross error ranges between 0.69 and 0.99, and the fractional bias ranges between −0.17 and 0.30. Strong horizontal concentration gradients of 100 μg m⁻³ km⁻¹, both in the observational record and model predictions, are apparent. Three emission reduction options, designed to reduce the severity of particulate pollution episodes in Christchurch, are simulated. When both domestic open-hearth fires and all coal burning are removed, the 24 h average peak concentration is reduced by 55%. The number of guideline exceedences of PM₁₀ in the modelled period is reduced from five to one. Removing open-hearth fires results in 42% reduction in PM₁₀ concentration, resulting in three exceedences of the guideline, and removing coal-burning fires yields a 32% reduction in PM₁₀, resulting in four exceedences of the guideline.     

On the prediction of concentration variations in a dispersing heavy-duty truck exhaust plume using k–ε turbulent closure

This work presents the computational fluid dynamic modeling of an exhaust plume dispersed from the exhaust pipe of a class-8 tractor truck powered by 330 hp Cummins M11 electronically controlled diesel engine. This effort utilizes an advanced CFD technique to accurately predict the variation of carbon dioxide concentration inside a turbulent plume using a k–ε eddy dissipation model. The simulation includes the “real-world” operation of a truck and its exhaust plume in a NASA, Langley aircraft testing wind tunnel, that had an effective volume of 226, 535 m³ (8,000,000 ft³). The predicted results show an excellent agreement with the experimentally measured values of CO₂ concentrations, dilution ratios, and the temperature variations inside the plume. A specific goal of this effort was to study the effect of recirculation region near the truck walls on dispersion of the plume. For this purpose, growth of the plume from the center of the exhaust pipe is also presented and discussed. This work also shows the benefits of CFD modeling in applications where dispersion correlations are not required a priori, instead the dispersion coefficients are calculated precisely by solving the turbulent kinetic energy and dissipation equations.     

Emission strengths for primary pollutants as estimated from an aircraft study of Hong Kong air quality

An aircraft study of air quality in the Hong Kong region during the fall of 1994 has allowed for an estimation of the daytime source strengths for CO and NO_y from the Hong Kong metropolitan center. Emission rate estimates for the Hong Kong urban plume for NO_y and CO were 5.4×10e(25) molecules s^-1 and 1.8×10e(26) molecules s^-1 as determined for the case study of 18 October. All emission rate estimates have uncertainties of a factor of 2. On one occasion a distinct plume emanating from Shenzhen in the People’s Republic of China was encountered. While plume delimitation was insufficient for source strength calculations, transect integrals did allow for a CO/NO_y ratio of about 16 to be determined. The CO/NO_y ratio for the Hong Kong urban plume was about 3.3. The difference in these ratios indicates differences in the overall combustion processes and efficiencies taking place within Hong Kong and the PRC.     

Evaluation of the reactive and optics model of emissions (ROME)

The reactive and optics model of emissions (ROME) is a reactive plume visibility model that simulates the potential atmospheric impacts of stack emissions. We present here an evaluation of the ability of ROME to simulate several plume physical and chemical variables, using an experimental data base that consists of a total of 40 case studies from four field programs. The evaluation variables include plume height, horizontal width, NO_x and SO₂ maximum concentrations, NO₂/NO_x concentration ratio at the plume centerline, and plume-to-sky radiance ratios. Three algorithms used to simulate plume dispersion in ROME were compared: (1) the empirical Pasquill–Gifford–Turner (PGT) scheme, (2) a first-order closure (FOC) algorithm and (3) a second-order closure (SOC) algorithm that simulates the instantaneous plume dimensions.The plume height results show a correlation of 0.82 between simulated and measured values and a gross error that is 13% of the mean measured value. For plume horizontal dispersion, the second-order closure algorithm produces a moderate correlation (0.54) and a small bias (5% of the mean measured value) in comparison with the field data. Although the PGT scheme also demonstrates moderate correlation with the measurements, it produces a negative bias by significantly underestimating plume horizontal dispersion. The first-order closure algorithm overestimates plume width and shows the least correlation (with the measurements) of the three dispersion algorithms.For the NYSEG data set where coordinated measurements of stack emissions, meteorology at plume height and plume characteristics were available, the SOC algorithm provides better correlations for NO_x concentrations, NO₂/NO_x ratios and plume visibility than the FOC and PGT algorithms. For plume visibility, the SOC algorithm shows a correlation of 0.96 at 405 nm, the wavelength where the plume was visible, and it simulates no visible plume at the other wavelengths (550 and 700 nm).A comparison of ROME simulations with those of the plume visibility model PLUVUE II shows that ROME, with the SOC algorithm, performs better for all variables.     

Photoacoustic spectrometer for measuring light absorption by aerosol: instrument description

A photoacoustic spectrometer has been developed to measure in situ light absorption by aerosol. The measured quantity is the sound pressure produced in an acoustic resonator caused by light absorption. The current lower detection limit for light absorption is 0.4 Mm^-1 which corresponds to an elemental carbon mass density of ≈40 ng m^-3 assuming an efficiency for light absorption of 10 m² g^-1. Calibration is performed using simple theory for the instrument along with use of a calibrated microphone and laser. The acoustic resonator is operated in the plane wave mode, which has a quality factor of ≈80, a resonance frequency of ≈500 Hz, and a photoacoustic coefficient of 12.8 Pa (W m^-1)^-1. The equivalent noise bandwidth of the resonator is ≈5 Hz. Coherent acoustic noise was supressed through the use of acoustic notch filters and laser beam ports at pressure nodes of the resonator. The relatively low-quality factor made it possible to use phase-sensitive detection having an equivalent noise bandwidth of ≈7.5 mHz. This was achieved by vector time averaging the microphone signal for ≈8 min. Two compact, efficient lasers were used during instrument evaluation performed in the Northern Front Range Air Quality Study (Colorado, 1996/97). One was a laser diode pumped, frequency doubled, solid state laser, and the other was a laser diode. Laser wavelengths were 532 nm and 685 nm, and corresponding average powers were 60 and 87 nW. Some examples are provided for light absorption measurements using the photoacoustic instrument and a nearby aethalometer.     

Very high weight ratios of S/K in individual haze particles over Kalimantan during the 1997 Indonesian forest fires

The elemental composition of individual aerosol particles of 0.15–3 μm radius, collected over Kalimantan during the 1997 Indonesian forest fire event, was analyzed using a transmission electron microscope equipped with an energy-dispersive X-ray analyzer (EDX). Although 60–90% of the particles collected at altitudes of 1–5 km contained K, they exhibited high weight ratios of S/K with median values of 9–18 independent of particle size. These were much larger than those (median values of 2–4) obtained from the forest fires in northern Australia. The high weight ratios over Kalimantan are considered to be due to the heterogeneous growth of particles through the oxidation of SO₂. In addition to SO₂ from the combustion of forest biomass, SO₂ originating from the combustion of peat below the ground is believed to have been important in producing the high S/K ratios.     

In situ measurements of volatile aromatic hydrocarbon biodegradation rates in groundwater

Benzene and alkylbenzene biodegradation rates and patterns were measured using an in situ microcosm in a crude-oil contaminated aquifer near Bemidji, Minnesota. Benzene-D6, toluene, ethylbenzene, o-, m- and p-xylenes and four pairs of C₃- and C₄-benzenes were added to an in situ microcosm and studied over a 3-year period. The microcosm allowed for a mass-balance approach and quantification of hydrocarbon biodegradation rates within a well-defined iron-reducing zone of the anoxic plume. Among the BTEX compounds, the apparent order of persistence is ethylbenzene > benzene > m,p-xylenes > o-xylene ≥ toluene. Threshold concentrations were observed for several compounds in the in situ microcosm, below which degradation was not observed, even after hundreds of days. In addition, long lag times were observed before the onset of degradation of benzene or ethylbenzene. The isomer-specific degradation patterns were compared to observations from a multi-year study conducted using data collected from monitoring wells along a flowpath in the contaminant plume. The data were fit with both first-order and Michaelis-Menten models. First-order kinetics provided a good fit for hydrocarbons with starting concentrations below 1 mg/L and Michaelis-Menten kinetics were a better fit when starting concentrations were above 1 mg/L, as was the case for benzene. The biodegradation rate data from this study were also compared to rates from other investigations reported in the literature.     

Air pollutants in rural homes in Guizhou,China – Concentrations,speciation, and size distribution

Several types of fuels, including coal, fuel wood, and biogas, are commonly used for cooking and heating in Chinese rural households, resulting in indoor air pollution and causing severe health impacts. In this paper, we report a study monitoring multiple pollutants including PM₁₀, PM_2.5, CO, CO₂, and volatile organic compounds (VOCs) from fuel combustion at households in Guizhou province of China. The results showed that most pollutants exhibited large variability for different type of fuels except for CO₂. Among these fuels, wood combustion caused the most serious indoor air pollution, with the highest concentrations of particulate matters (218～417 μg m^?3 for PM₁₀ and 201～304 μg m^?3 for PM_2.5), and higher concentrations of CO (10.8 ± 0.8 mg m^?3) and TVOC (about 466.7 ± 337.9 μg m^?3). Coal combustion also resulted in higher concentrations of particulate matters (220～250 μg m^?3 for PM₁₀ and 170～200 μg m^?3 for PM_2.5), but different levels for CO (respectively 14.5 ± 3.7 mg m^?3 for combustion in brick stove and 5.5 ± 0.7 mg m^?3 for combustion in metal stove) and TVOC (170 mg m^?3 for combustion in brick stove and 700 mg m^?3 for combustion in metal stove). Biogas was the cleanest fuel, which brought about the similar levels of various pollutants with the indoor case of non-combustion, and worth being promoted in more areas. Analysis of the chemical profiles of PM_2.5 indicated that OC and EC were dominant components for all fuels, with the proportions of 30～48%. A high fraction of SO₄^2? (31～34%) was detected for coal combustion. The cumulative percentages of these chemical species were within the range of 0.7～1.3, which was acceptable for the assessment of mass balance.     

Validation of a hybrid forecasting system for the ozone concentrations over the Paris area

A simplified hybrid statistical-deterministic chemistry-transport model, is used in real time for the prediction of ozone in the area of Paris during Summer 1999. We present here a statistical validation of this experiment. We distinguish the forecasts in the urban area from forecasts in the pollution plume downwind of the city. The validation of model forecasts, up to 3 days ahead, is performed against ground based observations within and up to 50 km outside of Paris. In the urban area, ozone levels are fairly well forecast, with correlation coefficients between forecast and observations ranging between 0.7 and 0.8 and root mean square errors in the range 15–20 μg m⁻³ at short lead times. While the bias of urban forecast is very low, the largest peaks are somehow underestimated. The ozone plume amplitude is generally well reproduced, even at long lead times (root mean square errors of about 20–30 μg m⁻³), while the direction of the plume is only captured at short lead times (about 70% of the time). The model has difficulties in forecasting the direction of the plume under stagnant weather conditions. We estimate the model ability to forecast concentrations above 180 μg m⁻³, which are of practical relevance to air quality managers. It is found that about 60% of these events are well forecast, even at long lead times, while the exact monitoring station where the exceedance is observed can only be forecast at short lead times. Finally, we found that about half of the forecast error is due to the error in the estimation of the boundary conditions, which are forecast by a simple linear regression model here.