Atmos.Chem.Phys.,11,1101–1116,2011 www.atmos-chem-phys/11/1101/2011/ doi:10.5194/acp-11-1101-2011
©Author(s)2011.CC Attribution3.0
Licen.Atmospheric Chemistry and Physics
Anthropogenic sulfur dioxide emissions:1850–2005
S.J.Smith1,J.van Aardenne2,*,Z.Klimont3,R.J.Andres4,A.Volke1,and S.Delgado Arias1
1Joint Global Change Rearch Institute,Pacific Northwest National Laboratory,5825University Rearch Court,Suite3500, College Park,MD20740,USA
2European Commission Joint Rearch Centre,European Commission,Via E.Fermi2749,TP290,21027Ispra(V A),Italy 3International Institute for Applied Systems Analysis,Schlossplatz1,2361,Laxenburg,Austria
4Environmental Sciences Division,Oak Ridge National Laboratory,Oak Ridge,TN37831-6335,USA
*current address:European Environment Agency,Kongens Nytorv6,1050Copenhagen K,Denmark
Received:28May2010–Published in Atmos.Chem.Phys.Discuss.:30June2010
Revid:28January2011–Accepted:1February–Published:9February2011
Abstract.Sulfur aerosols impact human health,ecosystems,
agriculture,and global and regional climate.A new annual
又是一年春天
estimate of anthropogenic global and regional sulfur dioxide
emissions has been constructed spanning the period1850–
2005using a bottom-up mass balance method,calibrated to
country-level inventory data.Global emissions peaked in
the early1970s and decread until2000,with an increa
in recent years due to incread emissions in China,inter-
national shipping,and developing countries in general.An
uncertainty analysis was conducted including both random
and systemic uncertainties.The overall global uncertainty
in sulfur dioxide emissions is relatively small,but regional
uncertainties ranged up to30%.The largest contributors to
uncertainty at prent are emissions from China and interna-
tional shipping.Emissions were distributed on a0.5◦grid by
ctor for u in coordinated climate model experiments.
1Introduction
Anthropogenic emissions have resulted in greatly incread
图书大全
sulfur deposition and atmospheric sulfate loadings near most
industrialized areas.Sulfuric acid deposition can be detri-
mental to ecosystems,harming aquatic animals and plants,
and damaging to a wide range of terrestrial plant life.Sul-
fur dioxide forms sulfate aerosols that have a significant ef-
fect on global and regional climate.Sulfate aerosols re-
flect sunlight into space and also act as condensation nu-
clei,which tend to make clouds more reflective and change
their lifetimes,causing a net cooling.The radiative
forcing
Correspondence to:S.J.Smith (v)change wrought by sulfate aerosols may be cond only to that caud by carbon dioxide,albeit in the opposite direc-tion(Forster et al.,2007).
Sulfur is ubiquitous in the biosphere and often occurs in relatively high concentrations in fossil fuels,with coal and crude oil deposits commonly containing1–2%sul-fur by weight.The widespread combustion of fossil fuels has,therefore,greatly incread sulfur emissions into the atmosphere,with the anthropogenic component now sub-stantially greater than natural emissions on a global basis (Smith et al.,2001).
Historical reconstructions of sulfur dioxide emissions are necessary to access the past influence of sulfur dioxide on the earth system and as ba-year information for future pro-jections.This paper prents a new estimate of global and country-level sulfur dioxide anthropogenic emissions over the1850–2005period.This work reprents a substantial update of previous work(Smith et al.,2001;Smith et al., 2004)with newer data and improved methodologies,and was the basis for the sulfur emissions in Lamarque et al.(2010). The emissions reconstruction prented here accounts for re-gional differences in the pace and extent of emission con-trol programs,has annual resolution,includes all anthro-pogenic sources,and provides global coverage.Fuel-bad and activity-bad(Eyring et al.,2010)estimates of shipping emissions were reconciled for recent decades and th
en ex-trapolated to1850.A global mass balance for sulfur in crude oil was calculated as an independent estimate of petroleum emissions.Finally,a regional and global uncertainty analy-sis was conducted.
Published by Copernicus Publications on behalf of the European Geosciences Union.
最成功的一件事2Methodology
Sulfur emissions from combustion and metal smelting can,in principle,be estimated using a bottom-up mass balance ap-proach where emissions are equal to the sulfur content of the fuel(or ore)minus the amount of sulfur removed or retained in bottom ash or in products.Data limitations,however, make the bottom-up approach uncertain since sulfur contents vary and information on sulfur removals is not always re-ported.Countries with air pollution policies in place gener-ally require detailed reporting of emissions,including direct measurement of emissions for many large sources.The data are likely to be more accurate than bottom-up estimates; therefore,we constrain our calculations to match country-level inventories where the data are available and judged to be reliable.This method produces an emissions estimate that is consistent with the available country inventory data, contains complete coverage of all relevant emissions sources (assuming the country inventory data are complete),and is consistent across years.
Emissions were estimated annually by country for the fol-lowing sources:coal combustion,petroleum combustion, natural gas processing and combustion,petroleum process-ing,biomass combustion,shipping bunker fuels,metal smelt-ing,pulp and paper processing,other industrial process, and agricultural waste burning(AWB).The approach is sum-marized in Fig.1,and further detailed in the Supplement (S.2,S.3).Thefirst step was to develop a detailed inventory estimate by ctor and country for three key years:1990, 2000,and2005.The years were chon due to data avail-ability and becau the years span a time period of sig-nificant change in emissions controls.Initial estimates for other years were constructed by interpolating emissions fac-tors,withfinal estimates by source and country found af-ter calibrating to country-level emissions inventories where tho were available(S.3).This methodology accounts for changing patterns of fuel consumption and emission controls in a context of limited detail for earlier years.
Emissions by end-u ctor for decadal years were esti-mated for a t of standard reporting ctors(energy,indus-try,transportation,domestic,AWB)before downscaling to a 0.5◦spatial grid.Descriptions of each element in this calcu-lation are given below.
2.1Fossil fuel combustion
Emissions from coal and petroleum combustion were es-timated starting with the regional emissions factors from Smith et al.(2001).A composite fossil fuel consumption time ries was constructed(e Supplement S.2)using data from IEA/OECD(2006),UN energy statistics(1996),and Andres et al.(1999).Country-level emissions estimates for Europe,North America,Japan,Australia,and New Zealand were compiled from:UNFCCC(2009)for1990–2005;En-vironment Canada(2008),the EEA(2002),Fujita(1993), Atmos.Chem.Phys.,11,1101–1116,2011
time ries developed here with IEA datafinds that the dif-ference is often within the IEA“statistical error”category. There is generally no other consumption category in the IEA data that is large enough to include the difference between our regional fuel consumption estimate and the IEA reported bunker fuel u.We presume,therefore,that the difference is unreported consumption and no adjustment to the IEA con-sumption data has been made.
让座看图写话二年级In order to estimate sulfur dioxide emissions,we cali-brate to the year2000value from Eyring et al.(2010)of 11080Gg SO2/year,which is the mean of estimates from Corbett and K¨o hler(2003),Eyring et al.(2005)and Endren et al.(2007).Using the global division between fuels derived above,we match this value using the following sulfur con-tents:residual:2.9%and distillate/other:1.3%.The average value for sulfur in bunker coal is assumed to be1.1%.For simplicity the values are kept co
冰鉴全文及翻译
nstant over time.The values,and their time trends,are not known with precision. The most authoritative data on marine fuel sulfur contents is from the IMO(2007),however the uncertainty in the values difficult to estimate(e Supplement S.9).While the estimate here is similar to other estimates in the literature (Corbett and K¨o hler,2003;Eyring et al.,2005,2010;En-dren et al.,2007),as prented in the Supplement,the emissions are particularly uncertain.The estimate here falls within the2001uncertainty of8400–13100Gg as estimated by Corbett and K¨o hler(2003,2004).
The fossil fuel consumption data from IEA ud to es-timate combustion emissions(Sect.2.1)were procesd to exclude fuels reported as international bunkers but included fuels ud for domestic shipping andfishing,using ctoral definitions as discusd further in Sect.2.5.In parallel with this assumption,domestic shipping andfishing emissions in the inventory data were included in the surface transportation ctor(e Supplement S.2,S.9).Emissions due to domes-tic shipping andfishing emissions from inventory data were subtracted from the total shipping estimate in the tables and figures reported below to avoid double counting.
2.3Metal amelting
Smelting emissions were estimated using a mass balance ap-proach where emissions are equal to the gross sulfur con-tent of ore minus reported smelter sulfuric acid production, with estimates adjusted to match inventory data where avail-able.Smelter production of copper,zinc,lead,and nickel was tabulated by country,primarily using USGS mineral yearbooks and predecessor publications for1930–2005and a variety of supplemental sources,particularly for earlier years.Some mineral sources and production technologies emit minimal sulfur and the were accounted for where they could be identified.Further details are available in the Supplement(S.10).
A number of data bias can affect the estimate of sul-fur emissions from metal smelting.To evaluate this possibil-ity,we can compare emissions from inventories to emissions estimated purely using the mass balance approach.For the period1990–2005,emissions by ctor were available for a number of countries.When compared to emissions from the mass balance approach,the inventory values forfifteen countries in1990appear to be significantly lower than tho estimated by the mass balance approach,although the oppo-site was the ca in the United States.This indicates that ei-ther the sulfur content of ore was overestimated or that the amount of sulfur removal was underestimated.The latter may be the more likely possibility,given that the sulfur re-covery tabulations from USGS are not necessarily complete, since data will be more readily available for co
吃鸡蛋上火吗mmodities, such as ores,which are internationally traded than for sulfu-ric acid,which might be ud locally.While the changes impact estimates of emissions from smelting,this has only a small impact on total emissions for recent years since to-tal emissions for most of the countries are constrained by inventory values.
The comparisons indicate that emissions from areas without inventory data could be overestimated if sulfur re-moval data are underreported.In contrast,there are veral regions with inventory data where the sulfur content of ore appears to be larger than default values,which could also be the ca in regions without inventory data,potentially lead-ing to an underestimate.The comparisons emphasize the importance of site-specific information in order to accurately estimate metal smelting emissions.
2.4Other emissions
Natural gas deposits often contain significant amounts of sul-fur compounds,particularly hydrogen sulfide,that are either flared,thus producing sulfur dioxide,or removed and con-verted to a salable product.Natural gas production emissions were estimated over time for the United States,the Former Soviet Union,and other regions without inventory data(e Supplement S.11).
While natural gas distributed for general u has mini-mal sulfur,natural gas containing larger amounts of sulfur, known as sour gas,can be combusted for industrial applica-tions,resulting in sulfur emissions.The US EPA inventory contains an estimate for emissions of376Gg SO2in2000, and the US inventory estimates were ud for US emissions from this source.This is the only explicit inclusion of this source in our estimate.Any sour gas emissions in industri-alized countries were presumed to be included in country in-ventories,but no data were available to explicitly account for the emissions,which would be included,indirectly,though our calibration procedure.
Petroleum production emissions were taken either from country level inventories or,where tho were not available, from the EDGAR3.2(Olivier and Berdowski,2001)and3.2
Atmos.Chem.Phys.,11,1101–1116,2011www.atmos-chem-phys/11/1101/2011/
FT inventories(Olivier et al.,2005).Emissions were scaled with petroleum production over time where inventory data were not available.
Emissions from pulp and paper operations were estimated using emissions factors from Mylona(1996)and inventory data combined with wood pulp production statistics(e Supplement S.11).
Remaining process emissions originate from a variety of sources,with sulfuric acid production one of the largest sources,particularly in earlier years.Process emissions were taken from the above sources and scaled over time prior to 1990where inventory data were not available by the regional HYDE estimate(van Aardenne et al.,2001).Where updated 2005data were not available,year2000values were ud. Emissions from biomass combustion(exclusive of open burning)were estimated using a historical reconstruction of biomass consumption bad on the estimate of Fernandes et al.(2007)combined with IEA data(e Supplement S.2). Emissions from agricultural waste burning onfields were from EDGAR v4.0(JRC/PBL,2009).Production statistics for24crop types from FAO(FAOSTAT Crop Production) were combined with information on the fraction burned on thefields(Yevich and Logan,2003;Eggleston et al.,2006; UNFCCC,2008)and emission factors from Andreae and Merlet(2001).Emissions from waste burning were calcu-lated in a similar manner,although the are small and the data for this ctor are incomplete.EDGARv4.0emissions from agricultural waste burning and waste were included in the Reprentative Concentration Pathway(RCP)emissions relea described in Lamarque et al.(2010).
2.5Emissions by ctor and grid
The emissions estimates developed here were mapped to a standard t of reporting ctors and th
en downscaled to a 0.5◦spatial grid as part of the production of historical data for the new RCP scenarios(Moss et al.,2010;Lamarque et al.,2010).The reporting ctors for the RCP histori-cal data were the following:energy transformation,residen-tial/commercial,industry,surface transportation,agricultural waste burning onfields,waste burning,solvent u,and agri-cultural activities(non-combustion).There are no apprecia-ble SO2emissions from the last two ctors.
Emissions from smelting and other industrial process were mapped to the industrial ctor,biomass fuel emissions to the domestic ctor,and fossil fuel extraction and process-ing emissions to the energy ctor.Emissions from coal and petroleum combustion were split into thefirst four ctors above by using inventory data,ctor-specific emissions fac-tors,and IEA fuel u data,where the data were available, and additional information from van Aardenne et al.(2001) and Bond et al.(2007),e Supplement for details(S.12). The emissions estimate was distributed onto a0.5◦reso-lution global grid for each decade from1850to2000.The sub-national split within a grid cell was estimated by using the2.5min national boundary information from the Gridded Population of the World datat(CIESIN and CIAT,2005). From1960through2000,emissions were distributed using a preliminary version of the year2000emissions distribu-tion from the EDGAR4.0project,parated into energy c-tor combustion,industrial combustion and other industrial, transportation combustion,and agricultural waste burning on fields within each country(Supplement S.12).
For1850–1900,emissions from combustion and other in-dustrial ctors for each country were distributed using the HYDE gridded population distribution(Goldewijk,2005). The emissions distribution for each country was interpolated from the“modern”grid in1960to the population-bad grid in1900by linearly increasing the weighting for the population-bad distribution in each year1950to1910and decreasing the weighting for the“modern”grid,until a pure population-bad grid is ud in1900.
The emissions grids were produced to facilitate the u of the data in global modeling experiments.In most cas, the distribution of emissions within each country is deter-mined by proxy data,not by actual emissions data.Alter-native methods of downscaling the emissions estimates to a spatial grid(van Vuuren et al.,2010),including incorpo-ration of emissions measurements,could produce improved emissions distributions.No consideration of country bound-ary changes was made during the emissions gridding proce-dure.Incorporation of the changes over time was beyond the scope of this project.
3Uncertainty
It is uful to examine uncertainty in emissions by source and region.To our knowledge,this is thefirst
consistent estimate of global and regional uncertainty in sulfur diox-ide emissions.For this estimate,we apply a relatively sim-ple approach to uncertainty analysis whereby a t of un-certainty bounds are applied to broad class of countries. This is warranted in large part since,as noted by Sch¨o pp et al.(2005),limited data are available to specify parame-ter uncertainty bounds,leading to bounds that are generally specified through expert judgment.This is particularly true for developing countries.In addition,sulfur dioxide emis-sions are principally determined by fuel sulfur content and not technology-specific emissions factors,at least in the ab-nce of emissions controls.Data on fuel sulfur content are spar in general,and tho that contain uncertainty informa-tion even rarer.The considerations make a more complex asssment of global uncertainty unwarranted at this time. We consider two sources of uncertainty,random and sys-temic uncertainties,as summarized in Eq.(1)
www.atmos-chem-phys/11/1101/2011/Atmos.Chem.Phys.,11,1101–1116,2011
Table1.Uncertainty bounds(as95%confidence interval)by country category and emissions type.The uncertainty bounds shown in the table are ud for random effects in Eq.(1).An additional systemic uncertainty was added(e text)with a magnitude of2.5%for countries with category I inventories,and5%for all other countries(and all countries prior to1970).
Category Coal Petroleum Smelting Other Process,
Biomass
I.Recent-Country-Inventory±11%±21%±14%±22%
合作共赢的作文
II.Older Inventory±18%±27%±25%±38%
IIa.OECD(pre inventory)±25%±43%±25%±52%
III.Other Countries±28%±45%±36%±54%
IV.Int Shipping±28%
IV.Int Shipping(earlier)±42%
uncertainty=
r
s
Emissions r s·CI random r s
2
+
r
s
Emissions r s·CI systemic r s
2
,(1)
where CI is the assumed5–95%confidence level,in percent from Table1,for a given region and category.The sums s and r are conducted over the source categories and regions listed in the Supplement(S.1,S.15).
评价总结Thefirst component of the uncertainty analysis considers errors in the individual components of the emissions calcu-lation.The t of uncertainty bounds given in Table1are applied to countries categori
zed depending on the estimated quality of the data ud to construct the inventory values(e Supplement S.15).Uncertainties are applied parately in each country to emissions from the following sources:coal, petroleum,biomass,fuel processing,smelting,and other pro-cess.Uncertainties in each of the categories are assumed to be independent and are combined in quadrature.Con-ceptually,aggregate uncertainty can be divided into uncer-tainty in driving forces,such as fuel consumption or smelter metal output,and uncertainty in sulfur content(and assump-tions such as ash retention),such as shown in the Supple-ment(S.15).Only the total values,however,are ud in this calculation.
The values in Table1are bad on the authors’judgment, informed by previous work in the literature(Sch¨o pp et al., 2005;Gregg et al.,2008;Eyring et al.,2010),comparisons with previous versions of this work,and changes over time in EPA inventories(e Supplement,S.13,S.14).The sources suggest that,overall,uncertainty is smallest where emissions are directly measured,such as in coal-fired power plants,and is relatively larger for emissions from petroleum products (except for countries with well-enforced and comprehensive sulfur standards),and process emissions.
In recent decades,sulfur emissions in most high-income countries have come under increasingly stringent control regimes.In earlier years,information on sulfur emissions was less complete,and we,t
herefore,assume that uncertain-ties are larger at the times.For similar reasons,we also as-sume that emissions are more uncertain in countries without comprehensive control regimes,or where such regimes have only been implemented recently.In addition,information on activity levels are also more uncertain in the past and in de-veloping countries generally.Becau common assumptions and data sources are ud for large portions of the world,we assume that uncertainties with each source category are per-fectly correlated within14world regions.
This procedure assumes uncertainties are symmetric.This is likely not strictly true since,for example,sulfur removal (for petroleum and metal smelting)is bounded above,sulfur retained in ash is bounded below,and some emissions drivers have potential bias in one direction–for example,underre-porting of consumption(Logan,2001).It is not clear,how-ever,if a more nuanced calculation is warranted given the number of assumptions that would need to be made.
The uncertainty estimate calculated as described above re-sults in uncertainty bounds on annual global total SO2emis-sions that are relatively small:6–10%over the20th century. This low value is due to cancellation between source cate-gories and regions.This uncertainty level would appear to be unrealistically low given that a number of previous global sulfur dioxide emissions estimates do not fall within this es-timated uncertainty bound(Smith et al.,2001;¨Om et al., 1996;van Aardenne et a
l.,2001;Lefohn et al.,1999;Spiro et al.,1992;Stern,2006).The reason is that additional,esn-tially correlated uncertainties are prent that add to the un-certainty value estimated above.Examples include reporting or other bias in global data ts for energy,sulfur removal, and other driver data,methodological assumptions,and the u of common default assumptions for sources where lit-tle data exists.Comparing the prent inventory with that of Smith et al.(2004),for example,indicates that the differ-ences between the two estimates involve veral method-ological and data changes that impacted emissions estimates over multiple world regions(e Supplement S.13).
Atmos.Chem.Phys.,11,1101–1116,2011www.atmos-chem-phys/11/1101/2011/
