Temporal Surface Ozone Patterns in Urban Manitoba, Canada

This study examined temporal surface ozone patternsfor two urban centres in Manitoba,Canada by analyzing hourly concentrations at theWinnipeg downtown (1995–1999), Winnipegresidential (1995–1999) and Brandon industrial(1998–1999) monitoring sites. The characteristicannual ozone cycle and diurnalcycles for June and December were attributable to: (1) theannual and diurnal solar radiation cycles,(2) temporal variations in the emissions of precursorchemical compounds, in particular the source strength ofnitrogen oxides, (3) temporal variationsin the height of the mixed layer, which determinethe degree of dilution of these emissions byatmospheric dispersion, and (4) an in situvolatile organic compound sensitive photochemicalregime, which resulted in decreased concentrations ofozone in response to increasedconcentrations of nitrogen oxides. Onlyone exceedance of the maximum acceptable level of82 ppb was recorded in the study period; itoccurred at the Brandon monitoring site on June 6, 1999.The sequence of weather and the hourly concentrationsof ozone and nitrogen oxides indicatedthat: (1) ozone fumigation, with the transition fromthe nocturnal boundary layer to the daytime mixed layer, mayhave supplemented photochemical ozone formation duringthe morning hours, and (2) during theevening hours, the post cold-frontal downward flux ofozone rich air, which was in the region dueto atmospheric transport, stratosphere-troposphere-exchangeor, possibly, due to the multiplicityof thunderstorms in the area in lateafternoon – early evening, may have been the main cause ofthis rare exceedance event.     

Ozone and PAN Formation Inside and Outside of the Berlin Plume – Process Analysis and Numerical Process Simulation

During the BERLIOZ field phase on 20 July 1998 a 40 km wide ozone-plume 30 to 70 km north of Berlin in the lee of the city was detected. The ozone mixing ratio inside the plume was app. 15 ppb higher than outside, mainly caused by high ozone precursor emissions in Berlin, resulting in a net chemical ozone production of 6.5 ppb h^–1, which overcompensates ozone advection of –3.6 ppb h^–1 andturbulent diffusion of –1.1 ppb h^–1. That means, although moreozone leaves the control volume far in the lee of Berlin than enters it at the leeside cityborder and although turbulent diffusion causes a loss of ozone in the leeside control volume the chemical production inside the volume leads to a net ozone increase. Using a semi-Lagrangian mass budget method to estimate the net ozone production, 5.0 ppb h^–1 are calculated for theplume. This means a fraction of about 20% of ozone in the plume is producedby local emissions, therefore called `home made' by the Berlin emissions. For the same area KAMM/DRAIS simulations using an observation based initialisation, results in a net production rate between 4.0 and 6.5 ppbh^–1, while the threefold nested EURAD model gives 6.0 ppbh^–1. The process analysis indicates in many cases goodagreement (10% or better) between measurements and simulations not only in the ozone concentrations but also with respect to the physical and chemical processes governing the total change. Remaining differences are caused by different resolution in time and space of the models and measurements as well as by errors in the emission calculation.The upwind-downwind differences in PAN concentrations are partly similar to those of ozone, because in the BERLIOZ case they are governed mainly by photochemical production. While in the stable boundary layer at night and windward of Berlin 0.1 to 0.3 ppb are detected, in the centre of the plume at noon concentrations between 0.75 ppb and 1.0 ppb are measured. The O₃/PAN ratio is about 80 to 120 and thus due to the relatively lowPAN concentrations significantly higher than found in previous studies. The low PAN formation on 20 July, was mainly restricted by the moderate nonmethane hydrocarbon levels, whereas high PAN concentrations of 3.0 ppb on 21 July, are caused by local production in the boundary layer and by large scale advection aloft.     

Boundary-layer ozone depletion as seen in the Norwegian Arctic in spring

Several years of measurements of ozone, hydrocarbons, sulphate and meteorological parameters from Spitsbergen in the Norwegian Arctic are presented. Most of the measurements were taken on the Zeppelin Mountain at an altitude of 474 m a.s.l. The focus is the episodes of ozone depletion in the lower troposphere in spring, which are studied in a climatological way. Episodes of very low ozone concentrations are a common feature on the Zeppelin Mountain in spring. The low ozone episodes were observed from late March to the beginning of June. When the effect of transport direction was subtracted, the frequenty of the low ozone episodes was found to peak in the beginning of May, possibly reflecting the seasonal cycle in the actual depletion process. Analyses based on trajectory calculations show that most of the episodes occurred when the air masses were transported from W-N. Ozone soundings show that the ozone depletion may extend from the surface and up to 3–4 km altitude. The episodes were associated with a cold boundary layer beneath a thermally stable layer, suppressing mixing with the free troposphere. The concentration of several individual hydrocarbons was much lower during episodes of low ozone than for the average conditions. The change in concentration ratio between the hydrocarbons was in qualitative agreement with oxidation of hydrocarbons by Br and Cl atoms rather than by OH radicals.     

Studies of Oxidant Production at the Weybourne Atmospheric Observatory in Summer and Winter Conditions

Detailed studies have been made of the behaviour of gases and radicals involved in the production of oxidants at the Weybourne Atmospheric Observatory in both summertime and wintertime conditions. In June 1995 the range of meteorological conditions experienced varied such that ozone destruction was observed in clean northerly air flows reaching Weybourne down the North Sea from the Arctic, and ozone production was observed in varying degrees in air with different loadings of nitrogen oxides and other precursors. The transition point for ozone destruction to ozone production occurred at a nitric oxide concentration of the order of 50 pptv. Plumes of polluted air from various urban areas in the U.K. were experienced in the June campaign at Weybourne. Quantitative studies of ozone production in a plume from the Birmingham conurbation on 18 June 1995 showed that the measurement of ozone production agreed well with calculated production rates from the product of the nitric oxide and peroxy radical concentrations (r2=0.9). In wintertime conditions (October–November 1994) evidence was also found for oxidant production, defined as the sum of O3+NO2. At this time of year the peroxy radical concentrations (RO2) were much lower than observed in the summertime and the nitric oxide (NO) was much higher. There was still sufficient RO2 during the day, however, for a slow accumulation of oxidant. Confirmatory evidence for this comes from the diurnal co-variance of (O3+NO2) with PAN, an excellent tracer of tropospheric photochemistry. The same type of covariance occurs in summer between PAN and ozone. The results obtained in these series of measurements are pertinent to understanding the measures necessary to control production of regional photochemical air pollution, and to the production of ozone throughout the northern hemisphere in winter.     

Pan over the Arctic; Observations during AGASP-2 in April 1986

During three of the flights with the NOAA P3 Orion over the Arctic icecap in April 1986, the atmospheric concentration of PAN (peroxyacetyl nitrate) was measured. Due to major experimental problems, the uncertainty in the data is large (+/–50%), but, nevertheless, some important trends can be resolved. More than 600 (+/–300) ppt(v) of PAN was present in a moderately dense arctic haze layer, confirming conclusions reached from surface observations at Alert, N.W.T., Canada, that PAN is a major odd nitrogen species in Arctic polluted air masses. In relatively clean air off Barrow, Alaska, PAN levels were well below 100 (+/–50) ppt(v), increasing with altitude, in agreement with theoretical predictions concerning the occurrence of PAN in clean air. PAN mixing ratios in the upper troposphere or lower stratosphere were variable (from ca. 30 (+/–15) ppt(v) on April 13 up to 140 (+/–70) ppt(v) on April 8), suggesting involvement in the tropospheric-stratospheric exchange of odd nitrogen. To place the PAN data in a broader context, measurements of other NO_y compounds as well as integrated SO_x data are also reported.     

European VOC Emission Estimates Evaluated by Measurements and Model Calculations

Observations and model calculations of the concentration of hydrocarbonsat five Scandinavian rural sites during March–June 1993are reported.Decreasing concentrations from March to June are observedat all sites. The highest concentrations of hydrocarbons were found in air massescoming in from the southwest to southeast, indicating that long rangetransport fromcontinental Europe and the U.K. is important in pollution episodes. An episode of elevated concentrations of hydrocarbons observed at three of the sites in the middle of Marchis described and discussed in relation to the synoptic situation and thepresenceof other chemical compounds (NO₂, PAN, total nitrate andozone).A Lagrangian numerical model is used to calculate the concentrations of theindividual hydrocarbons at the fivesites and comparison with observations is made.The calculated concentrations for nonmethane hydrocarbons with quite longchemicallifetimes agree well with the observations.For the sum of observed and calculated hydrocarbons the correlationcoefficientsare in the range of 0.65–0.88 for the five sitesand the ratio between calculated and measured concentrations was0.72–0.97, indicating thatthe European VOC emission inventory is quite well estimated.     

Peroxyacetyl Nitrate (PAN) Measurements During the POPCORN Campaign

During the POPCORN campaign between 3 and 24 August 1994 we measured peroxyacetyl nitrate (PAN) in a rural area of Mecklenburg-Vorpommern (North-Eastern Germany) above a corn field. A total of about 5000 PAN measurements were carried out within the three weeks of the campaign. Measured PAN mixing ratios ranged from below the detection limit of 10 ppt up to an afternoon maximum of 1 ppb. The mean value of all data was 140 ppt. The daily mean PAN mixing ratios were typically in the range of 50 to 250 ppt, but during a clean air episode PAN mixing ratios of well below 40 ppt were observed. The characteristic relative diurnal variation of the PAN mixing ratios with a late night/early morning minimum and an afternoon maximum persisted during these episodes. The daily averages of the PAN mixing ratios showed clear episodic variations which coincided with the duration of typical synoptic episodes of two to six days duration. Based on the measurements of the various parameters determining the PAN formation and destruction rates, the local budget for PAN was calculated. During daytime the calculated net photochemical formation rate of PAN was nearly always significantly higher than the observed change of the PAN concentration. This demonstrates that substantial amounts of PAN (often in the range of several hundred ppt/h) were exported from the corn field. The resulting removal of NOx to some extent effects the budget of nitrogen oxides (NOx), but the export of odd oxygen radicals in the form of PAN during daytime often amounted up to 30–50% of the OH-radical formation by ozone photolysis. Thus the importance of PAN as reservoir and transport medium for odd oxygen radicals can be very substantial and may have a significant impact on the budget and distribution of odd oxygen radicals.     

High PAN concentrations during nonsummer periods: A study of two episodes in Creteil (Paris), France

The effect of sunlight on air containing hydrocarbons and nitrogen oxides leads to a very complex series of reactions which form, among other secondary pollutants, peroxyacetylnitrate (PAN). This compound is believed to be the most useful indicator of photochemical pollution, and has been monitored for several months at the University of Creteil (Paris, France). Although it is commonly accepted that high PAN levels occur more frequently in summer, our results show elevated PAN concentrations during nonsummer periods, precisely during two episodes in October 1985 and February 1986. A close examination of these episodes indicates that, in addition to local photochemical production, long-range transport of polluted air masses can occur in winter, revealing the influence of emissions from central Europe.     

The partitioning of nitrogen oxides in the lower Arctic troposphere during spring 1988

Surface observations of several nitrogen oxides in the Canadian high Arctic during the period March-April 1988 are reported. These include data on NO₂, the inorganic nitrates HNO₃ and particulate nitrate, and the organic nitrates PAN and C₃–C₇ alkyl-nitrates. It is found that the organic nitrates make up 70–80% of the sum of the measured nitrogen oxides. Based on concurrently measured sulphur oxides, the period of observation was divided into two halves with the first half representing less polluted, more aged air than the second. The preponderance of the organic nitrates was less in the first period than the second. In contrast, there was little difference in the inorganic nitrates and NO₂ concentrations. The dominant inorganic nitrate shifted from particulate nitrate in the first period towards gaseous HNO₃ in the second. No correlation between the nitrates (inorganic or organic) and O₃ was observed; although some indication of a positive correlation between NO₂ and O₃ has been reported earlier (Bottenheimet al., 1990). Possible explanations for these observations are proposed. A survey of other potential nitrogen oxides that may be present in the Arctic air but not measured in these experiments suggests that the nitrogen oxides not measured here constitute a minor fraction of the total reactive nitrogen (NO _y ).Paper submitted to the 7th International Symposium of the Commission for Atmospheric Chemistry and Global Pollution on the Chemistry of the Global Atmosphere held in Chamrousse, France, from 5 to 11 September 1990.     

Anthropogenic aerosols and gases in the lower troposphere at Alert Canada in April 1986

During April 1986, as part of an international arctic air chemistry study (AGASP-2), ground level observations of aerosol trace elements, oxides of sulphur and nitrogen and particle number size distribution were made at Alert Canada (82.5N, 62.3W). Pollution haze was evident as indicated by daily aerosol number (size > 0.15 m diameter) and SO₄ ⁼ concentrations in the range 125 – 260 cm^–3 and 1.6 – 4.5 g m^–3, respectively. Haze and associated acidic gases tended to increase throughout the period. SO₂ and peroxyacetylnitrate (PAN) mixing ratios were in the range 140 – 480 and 370 – 590 ppt(v), respectively. About 88% of the total end-product nitrogen was in the form of PAN. In air dried to 2% relative humidity by warming to room temperature, the aerosol mass size distribution had a major mode at 0.3 m diameter and a minor one at 2.5 m. Aerosol mass below 1.5 m was well correlated with SO₄ ⁼, K⁺ and PAN. There was a steady increase in the oxidized fraction of total airborne sulphur and nitrogen oxide throughout April as the sun rose above the horizon and remained above. The mean oxidation rate of SO₂ between Eurasia and Alert was estimated as 0.25 – 0.5% h^–1. The molar ratio of total nitrogen oxide to total sulphur oxide in the arctic atmosphere (0.67±0.17) was comparable to that in European emissions. A remarkably strong inverse correlation of filterable Br and O₃ led to the conclusion that O₃ destruction and filterable Br production below the Arctic surface radiation inversion is associated with tropospheric photochemical reactions involving naturally occurring gaseous bromine compounds.     

Fluxes of gases and particles above a deciduous forest in wintertime

Eddy-correlation measurements of the vertical fluxes of ozone, carbon dioxide, fine particles with diameter near 0.1 m, and particulate sulfur, as well as of momentum, heat and water vapor, have been taken above a tall leafless deciduous forest in wintertime. During the experimental period of one week, ozone deposition velocities varied from about 0.1 cm s^–1 at night to more than 0.4 cm s^-1 during the daytime, with the largest variations associated primarily with changes in solar irradiation. Most of the ozone removal took place in the upper canopy. Carbon dioxide fluxes were directed upward due to respiration and exhibited a strong dependence on air temperature and solar heating. The fluxes were approximately zero at air temperatures less than 5 °C and approached 0.8 mg m^–2 s^–1 when temperatures exceeded 15 °C during the daytime. Fine-particle deposition rates were large at times, with deposition velocities near 0.8 cm s^–1 when turbulence levels were high, but fluxes directed upward were found above the canopy when the surface beneath was covered with snow. Diffusional processes seemed to dominate fine-particle transfer across quasilaminar layers and subsequent deposition to the upper canopy. Deposition velocities for particulate sulfur were highly variable and averaged to a value small in magnitude as compared to similar measurements taken previously over a pine forest in summer.     

2017年8—9月湖州市臭氧污染特征及其生成机制研究

近年来近地面臭氧问题日益凸显,成为影响空气质量持续改善的瓶颈.本研究基于2017年8—9月在湖州市城区开展的为期1个月的臭氧及其前体物挥发性有机物（VOCs）和氮氧化物（NO_x）在线观测数据,分析了臭氧及其前体物污染特征,利用正矩阵因子分析（PMF）解析了VOCs来源,并采用基于观测的模型（OBM）对臭氧生成机制进行研究.研究结果表明：1）观测期间湖州市VOCs平均体积分数为（24.78±9.10）×10^-9,其中占比最高的组成为烷烃、含氧VOCs （OVOCs）和卤代烃;2）在臭氧非超标时段,湖州市臭氧生成处于VOCs控制区,而在臭氧重污染期间湖州市处于以VOCs控制为主的过渡区;3）在臭氧超标时段,对臭氧生成潜势（OFP）贡献最大的是芳香烃（39.6%）,其次是烯烃（21.5%）和OVOCs （19.4%）,排名前三的关键组分为甲苯、乙烯和间/对二甲苯;4）源解析结果显示观测期间湖州市VOCs的主要来源是溶剂使用（27.0%）、交通排放（22.7%）、背景+传输（19.3%）、工业排放（16.9%）、汽油挥发（7.7%）和植物排放（6.4%）,重污染过程期间对OFP贡献最大的两类源是交通排放源和溶剂使用源,贡献百分比分别为35.1%和30.5%.因此,对交通排放和溶剂使用方面进行控制管理对湖州市大气臭氧污染防控有重要意义.     

Long-term measurements of surface ozone in the German Democratic Republic

The ozone concentration near the earth's surface has been measured at some stations in the GDR for more than 30 yr using the wet chemical method. Even at rural stations the ozone data show a significant linear increase by about 1–3% yr^–1. The ozone increase being stronger in summer than in winter is assumed to be due to photochemical ozone production from increasing anthropogenic emissions of trace gases that are transported over long distances. A weaker ozone increase by only about 0.2% per year was observed in the free troposphere (5.5 km) from balloon-soundings at Lindenberg within the period 1975–1984. If the ozone trends continue, the ozone concentration near the surface and its seasonal amplitude will have doubled around the turn of the century as compared to the mid-fifties.     

Convective Redistribution of Ozone and Oxides of Nitrogen in the Troposphere over Europe in Summer and Fall

The fluxes of ozone and NO_x out of the atmospheric boundary layer (ABL) over Europe are calculated in a mesoscale chemical transport model (MCT) and compared with the net chemical production or destruction of ozone and the emissions of precursors within the ABL for two 10 days' periods which had quite different synoptic situations and levels of photochemical activity (1–10 July 1991 (JUL91) and 26 October–4 November 1994 (ON94)). Over the European continent, about 8% of the NO_x emissions were brought from the ABL to the free troposphere as NO_x, while about 15% of the NO_x emissions were brought to the free troposphere as NO_y–NO_x, i.e. as PAN or HNO₃. The convection dominates over the synoptic scale vertical advection as a transport mechanism both for NO_x and NO_y out of the boundary layer in the summertime high pressure situation (JUL91), while in the fall situation (ON94) the convective part was calculated to be the smallest. NO_x was almost completely transformed to NO_y–NO_x or removed within the ABL. Also for NO_y the major part of the atmospheric cycle is confined to the ABL both for JUL91 and ON94. The vertical transport time out of the ABL is of the order of 100h both for the total model domain and over the European continent. The net convective exchange of ozone from the ABL is not a dominant process for the amount of ozone in the ABL averaged over 10 days and the whole domain, but convection reduces the maximum ozone concentration in episodes significantly. The ozone producing efficiency of NO_x is calculated to increase with height to typically 15–20 in the upper half of the troposphere from around 5 in the ABL, but in the middle free troposphere the concentration of NO_x is often too low to cause net chemical formation of ozone there.     

Surface Trace Gases at a Rural Site between the Megacities of Beijing and Tianjin

The North China Plain （NCP） has recently faced serious air quality problems as a result of enhanced gas pollutant emissions due to the process of urbanization and rapid economic growth. To explore regional air pollu- tion in the NCP, measurements of surface ozone （O3）, nitrogen oxides （NOx）, and sulfur dioxide （SO2） were car- ried out from May to November 2013 at a rural site （Xianghe） between the twin megacities of Beijing and Tianjin. The highest hourly ozone average was close to 240 ppbv in May, followed by around 160 ppbv in June and July. High ozone episodes were more notable than in 2005 and were mainly associated with air parcels from the city cluster in the hinterland of the polluted NCP to the southwest of the site. For NOx, an important ozone precur- sor, the concentrations ranged from several ppbv to nearly 180 ppbv in the summer and over 400 ppbv in the fall. The occurrence of high NOx concentrations under calm condi- tions indicated that local emissions were dominant in Xianghe. The double-peak diurnal pattern found in NOx concentrations and NO/NOx ratios was probably shaped by local emissions, photochemical removal, and dilution re- sulting from diurnal variations of surface wind speed and the boundary layer height. A pronounced SO2 daytime peak was noted and attributed to downward mixing from an SO2-rich layer above, while the SO2-polluted air mass transported from possible emission sources, which differed between the non-heating （September and October） and heating （November） periods, was thought to be responsible for night-time high concentrations.     

Subsonic aircraft and ozone trends

Growth in subsonic air traffic over the past 20 years has been dramatic, with an annual increase of }6.1% over the decade between 1978 and 1988. Furthermore, aircraft activities in the year 2000 are predicted to be double those of 1990, with a shift towards more high-flying, longhaul subsonics. Aircraft exhaust gases increase the amount of nitrogen oxides (NO _x ) in the upper troposphere/lower stratosphere through injection at cruise altitudes. Given that NO _x is instrumental in tropospheric ozone production and stratospheric ozone destruction, it is important to determine the influence of subsonic aircraft NO _x emissions on levels of atmospheric ozone. This paper describes calculations designed to investigate the impact that subsonic aircraft may already have had on the atmosphere during the 1980s, run in a 2-D chemical-radiative-transport model. The results indicate a significant increase in upper tropospheric ozone over the decade arising from aircraft emissions. However, when comparing model results with observational data, certain discrepancies appear. Lower stratospheric ozone loss over the 1980s does not appear to be greatly altered by the inclusion of aircraft emissions in the model. However, given the trend in greater numbers of long-haul subsonic aircraft, this factor must be considered in any further calculations.     

The impact of high altitude aircraft on the ozone layer in the stratosphere

The paper discusses the potential effects on the ozone layer of gases released by the engines of proposed high altitude supersonic aircraft. The major problem arises from the emissions of nitrogen oxides which have the potential to destroy significant quantities of ozone in the stratosphere. The magnitude of the perturbation is highly dependent on the cruise altitude of the aircraft. Furthermore, the depletion of ozone is substantially reduced when heterogeneous conversion of nitrogen oxides into nitric acid on sulfate aerosol particles is taken into account in the calculation. The sensitivity of the aerosol load on stratospheric ozone is investigated. First, the model indicates that the aerosol load induced by the SO₂ released by aircraft is increased by about 10–20% above the background aerosols at mid-high latitude of the Northern Hemisphere at 15 km for the NASA emission scenario A (the NASA emission scenarios are explained in Tables I to III). This increase in aerosol has small effects on stratospheric ozone. Second, when the aerosol load is increased following a volcanic eruption similar to the eruption of El Chichon (Mexico, April 1982), the ozone column in spring increases by as much as 9% in response to the injection of NO _x from the aircraft with the NASA emission scenario A. Finally, the modeled suggests that significant ozone depletion could result from the formation of additional polar stratospheric clouds produced by the injection of H₂O and HNO₃ by the aircraft engines.     

Surface ozone values in anomalous summer 2010 measured in Obninsk

Given and analyzed are the results of the measuring of concentration of ozone O₃, nitrogen oxides NO_x, and carbon monoxide CO at the surface as well as the sum of hydrocarbons and the aerosol optical depth in Obninsk (Kaluga oblast) during the warm period 2010 and 2011. The relationship between the air temperature and the maximum daily ozone values in May–September 2010 are characterized by the higher correlation coefficient than in May–September 2011: 0.82 ± 0.05 versus 0.64 ± 0.07. Increased concentration of surface ozone in Obninsk in July–August 2010 as compared to the similar period in 2011 were caused by the higher concentrations of the compounds-predecessors of ozone. The concentration of O₃ in August 2010 exceeded 200μg/m³ and was never registered in Obninsk during the observation period of 2004–2011. This is associated with the air masses that came to Obninsk from the areas with peat and forest fires that resulted in the dramatic increase in surface concentrations of NO_x, CO, and hydrocarbons.     

The photochemistry of synoptic-scale ozone synthesis: Implications for the global tropospheric ozone budget

The oxidation of nonmethane hydrocarbons represents a source of tropospheric ozone that is primarily confined to the boundary layers of several highly industrialized regions. (Each region has an area greater than one million km²). Using a photochemical model, the global tropospheric ozone budget is reexamined by including the in-situ production from these localized regimes. The results from these calculations suggest that the net source due to this photochemistry, which takes place on the synoptic scale, is approximately as large as the amount calculated for global scale photochemical processes which consider only the oxidation of methane and carbon monoxide. Such a finding may have a considerable impact on our understanding of the tropospheric ozone budget. The model results for ozone show reasonable agreement with the climatological summer distribution of ozone and the oxides of nitrogen at the surface and with the vertical distribution of ozone and nonmethane hydrocarbons obtained during a 1980 field program.     

Emission scenarios for a global hydrogen economy and the consequences for global air pollution

Hydrogen is named as possible energy carrier for future energy systems. However, the impact of large-scale hydrogen use on the atmosphere is uncertain. Application of hydrogen in clean fuel cells reduces emissions of air pollutants, but emissions from hydrogen production and leakages of molecular hydrogen could influence atmospheric chemistry. This paper combines a global energy system model and a global atmospheric model to explore the range of impacts of hydrogen on atmospheric chemistry. We found that emissions of molecular hydrogen may range from 0.2 up to 10% (or 25-167 Tg hydrogen/yr) for a global hydrogen energy system. The lower end of this range would in fact be equal to current emissions from fossil fuel combustion. Hydrogen energy use leads to a clear decrease in emissions of carbon monoxide, nitrogen oxides and sulphur dioxide, but large-scale hydrogen production from coal may lead to net increase in emissions of nitrous oxide and volatile organic compound. Compared to a reference scenario, this would lead to positive impacts on surface concentrations of carbon monoxide, nitrogen oxides and ozone. However, if hydrogen leakage would not be minimised it leads to an increase in methane lifetimes and a decrease in stratospheric ozone concentrations.