The GAINS model Introdution The Greenhouse gas - Air pollution Interations and Synergies (GAINS) model (http://gains.iiasa.a.at/) has been developed at the International Institute for Applied Systems Analysis (IIASA) and provides an integrated assessment framework desribing the pathways of atmospheri pollution from anthropogeni driving fores to relevant health and environmental impats. It brings together information on future eonomi, energy and agriultural development, emission abatement potentials and osts, atmospheri dispersion and environmental sensitivities towards air pollution. The model addresses threats to human health posed by fine partiles and ground-level ozone, and the risk of eosystems damage from aidifiation, exess nitrogen deposition (eutrophiation) and exposure to elevated levels of ozone. These impats are onsidered in a multi-pollutant ontext, quantifying the ontributions of all major air pollutants as well as the six greenhouse gases onsidered in the Kyoto protool (Figure ). PM SO 2 NO x VOC NH 3 CO 2 CH 4 N 2 O HFCs PFCs Health impats: PM O 3 Vegetation damage: O 3 SF 6 Aidifiation Eutrophiation Radiative foring: - diret - via aerosols - via OH Figure : The GAINS multi-pollutant/multi-effet framework. The model an explore ost-effetive strategies to redue emissions of air pollutants in order to meet speified environmental targets. It also assesses how speifi ontrol measures simultaneously influene different pollutants, permitting a ombined analysis of air pollution and limate hange mitigation strategies, whih an reveal important synergies and trade-offs between these poliy areas. A omprehensive desription of the European version of the GAINS model is given in (Amann et al., 20), whih also illustrates the appliation of the model using a reent poliy analysis. Brief desriptions of those aspets of the model framework of most relevane to the EUCLIMIT projet are desribed in the subsequent setions.
2 Methodology 2. Emission estimates For eah of the pollutants shown in Figure GAINS estimates urrent and future emissions based on ativity data, unontrolled emission fators, the removal effiieny of emission ontrol measures and the extent to whih suh measures are applied. In the general GAINS methodology, emissions from soure s in region i and year t are alulated as the ativity data A its times an emission fator ef ism. If emissions are ontrolled through implementation of tehnology m, the fration of the ativity ontrolled is speified by Appl itsm, i.e., [ A ef Appl ] E its = its * ism * itsm, () m NOC where efism= efis *( remeffsm) and Appl its =, (2) m and where A its is the ativity (e.g., number of animals, amounts of fuel or waste), ef ism is the emission fator for the fration of the ativity subjet to ontrol by tehnology m, Appl itsm is the appliation rate of tehnology m to ativity s, NOC ef is is the no ontrol emission fator for ativity s, and remeff sm is the removal effiieny of tehnology m when applied to ativity s. This approah takes aount of ritial differenes aross eonomi setors and ountries that ould justify differentiated emission redution requirements in a ost-effetive strategy. Strutural differenes in emission soures are refleted through ountry-speifi ativity levels. Major differenes in emission harateristis of speifi soures and fuels are represented through soure-speifi emission fators, whih aount for the extent to whih emission ontrol measures are applied. Future emissions are estimated by varying the ativity levels along external projetions of anthropogeni driving fores and by adjusting the implementation rates of emission ontrol measures. The GAINS model holds relevant data for all European ountries, employing international energy and agriultural statistis and appropriate emission fators. 2
2.2 Mitigation potentials and osts A wide range of tehnial measures has been developed to apture emissions at their soures before they enter the atmosphere. GAINS onsiders about 3500 end-of-pipe measures for reduing emissions of SO 2, NOx, VOC, NH 3, PM, CH 4, N 2 O and F-gases, as well as 350 options to redue CO 2 through strutural hanges. In order to assess emission ontrol osts aurately, it is important to identify the fators leading to variations in osts between ountries, eonomi setors and pollutants. Diversity is aused by differenes in the strutural omposition of existing emission soures (e.g., fuel use pattern, fleet omposition, et.), the state of tehnologial development, and the extent to whih emission ontrol measures are already applied. Assuming no trade barriers in the market for emission ontrol tehnologies, the same tehnology is assumed available to all ountries at the same investment osts. However, ountry- and setor-speifi irumstanes (e.g., size distributions of plants, plant utilisation, fuel quality, energy and labour osts, et.) lead to justifiable differenes in the atual osts at whih a given tehnology removes pollution at different soures. For eah of the 3500 emission ontrol options, GAINS estimates the osts of loal appliation onsidering annualised investments, fixed and variable operating osts, and how they depend on tehnology, ountry and ativity type. Mitigation osts per unit of ativity are alulated in GAINS as the sum of investment osts, labour osts, fuel osts (or ost-savings), and operation and maintenane osts (or ost-savings) unrelated to labour and fuel osts. The unit ost of tehnology m in ountry i and year t is defined as: ( + r) ( + r) T r C itm = I im M T + im + im it is im where ( + r) ( + r) fuel ( L W w ) + ( F p ) T r I im T is the annualized investment ost for tehnology m in ountry i and with interest rate r and tehnology lifetime of T years, it (3) M im is the sum of annual operation and maintenane osts (or ost-savings) unrelated to labour or fuel osts, L im is the fration of annual work hours for operating tehnology m, W it is the annual average wage in ountry i in year t, w is is a ountry-speifi wage adjustment fator for type of setor s (agriulture or manufaturing industry), F im is the additional amount of energy used or reovered when applying tehnology m, and fuel p it is the fuel prie in ountry i in year t for the energy used or reovered under tehnology m. 3
3 Modelling non-co 2 greenhouse gas emissions and mitigation Non-CO 2 greenhouse gases (GHGs) in the GAINS model inlude methane (CH 4 ), nitrous oxide (N 2 O), hydrofluoroarbons (HFCs), perfluoroarbons (PFCs) and sulphur hexafluoride (SF 6 ), whih are all addressed in the Kyoto protool. The GAINS model has been used on several oasions to estimate urrent and future emissions of non-co 2 GHGs in the European Union in support of the EU limate strategy (Amann et al., 2008; Höglund-Isaksson et al. 200; Höglund-Isaksson et al., 202). A detailed desription by setor of the methodology applied to the estimation of non-co 2 greenhouse gases in the EU was ompiled in a report by Höglund-Isaksson, Winiwarter and Tohka (200). An updated version of the report is sheduled for April 202. Baseline emissions of greenhouse gases in the EU and poliy senarios for mitigation are estimated in a joint effort by the PRIMES, CAPRI, GAINS and GLOBIOM models. All models use eonomi foreasts from the European Commission (DG-ECFIN) as starting point for model senarios. While CO 2 emissions are modelled within the PRIMES energy systems model, emissions of non-co 2 GHGs and air pollutants onsistent with the estimated CO 2 emissions, are modelled using the GAINS model. The onsisteny between the models is maintained in GAINS through the use of energy ativity data on e.g., fuel prodution and onsumption from the PRIMES model and agriultural ativity data on e.g., livestok numbers and fertilizer appliation from the CAPRI agriultural setor model. The ativity data generated externally for energy and agriultural ativities are omplemented with data generated internally for other setors e.g., waste, wastewater, air onditioning and refrigeration. Consisteny in these setors is maintained by using the same maroeonomi foreast from DG-ECFIN as in the PRIMES and CAPRI models. The proedure is illustrated in Figure 2. The ativity data are ombined with emission fators to produe estimates of unontrolled emissions. To the extent possible given the available information, ountry-speifi emission fators are derived following the reommendations in the IPCC guidelines (997 and 2006). Deriving emission fators from ountry-speifi information on fators important for the emission generation in a setor makes it possible to produe emission estimates that are onsistent and omparable aross ountries. When ountry-speifi information is unavailable or insuffiient, default emission fators from the IPCC guidelines or other soures are used. Baseline emission estimates reflet emissions inluding the effets of urrently implemented ontrol measures or future implementation of ontrol foreseen in already adopted EU-wide or national legislation. EU-wide legislation affeting emissions of non-co 2 GHGs diretly or indiretly that is onsidered in the GAINS modelling of non-co2 GHGs as of Deember 20 inlude: the Landfill Diretive (999/3/EC), the Waste Diretive (2006/2/EC), the Waste Management Framework Diretive (2008/98/EC), the Nitrate Diretive (99/676/EEC), the Common Agriultural Poliy Reform (2006/44/EC), the CAP Health hek and Set aside regulation (73/2009), the F-gas regulation (2006/842/EC), the MAC Diretive (2006/40/EC), the Biofuels Diretive (2009/28/EC), the EU Emission Trading System (2003/87/EC and its subsequent amendments) and the Effort Sharing Deision (2009/406/EC). National legislation affeting emissions of non-co 2 GHGs inludes omplete bans on deposition of biodegradable waste on landfills in Denmark, Germany and Sweden, and national legislation ontrolling emissions of nitrogen ompounds (NO X, NH 3 ), whih indiretly affet N 2 O emissions. In addition, baseline emission 4
estimates also regard the effets of a voluntary agreement to redue PFC emissions in the semiondutor industry (ESIA, 2006). As shown in Figure 2, one a first set of draft baseline emission estimates has been produed, the result for a historial base year, e.g., 2005 or 200, is ompared on a setor level with emissions reported by ountries to UNFCCC under the Kyoto protool. Reasons for disrepanies are srutinized and adjustments made when onsidered justified, i.e. to the extent that the onsistent modelling aross ountries is preserved. To math base year emission estimates with reported emissions on a ountry level, the remaining setor disrepanies in emissions are summed up on a ountry-level and enter the modelling of future emissions as a onstant fator. The resulting adjusted baseline emission estimates are provided to experts in the EU member states for review. Improved information reeived through feedbak from member state experts is inorporated into emission estimates to produe a final baseline senario for non-co 2 greenhouse gases in the EU. GDP and Population projetions External from DG-ECFIN PRIMES model Ativity data e.g., energy use, fossil fuel prod., fuel pries, ETS arbon prie CAPRI model Ativity data e.g., animal numbers, fertilizer use, milk yield GAINS model Ativity data e.g., waste and wastewater generation and F-gas applianes, use and prodution GAINS Non-CO2 GHGs emission estimation and mitigation ost urve simulation Input data: Ativity data, emission fators, urrent and max appliation of ontrol tehnology, removal effiieny, osts GAINS Draft Non-CO2 GHGs baseline emission estimation GAINS Identifiation of reasons for disrepany between GAINS BL emissions in baseyear and National reporting to UNFCCC Information inorporated in emission estimations GAINS Review of baseline assumptions by EU member state experts Information inorporated in emission estimations GAINS Final Baseline non-co2 GHG emissions and mitigation ost urves Figure 2: Work proedure for estimation of senarios for non-co2 greenhouse gases in the European Union using the GAINS model. The unertainty surrounding the adoption rate of tehnology and the effet of tehnologial development on redution potentials and osts several deades from now, is of ourse very high, in partiular for tehnologies that are urrently not ommerially available. To give an indiation of the unertainty surrounding a partiular tehnology, we speify three levels of tehnial maturity of tehnologies: fully ommerially available, starting ommerial adoption and not yet ommerially available. For the fully ommerially available tehnologies, we assume no or very limited effets of tehnologial development. For tehnologies that are just starting to be adopted ommerially a faster rate of tehnologial 5
development is assumed. For tehnologies that are not yet ommerially available, the unertainty surrounding both removal effiienies and osts is very high. We therefore refrain from speulating about the future development in removal effiieny and osts and keep these onstant over time for this tehnology group. 4 Modelling air pollution and mitigation 4. Atmospheri dispersion An integrated assessment of air pollution needs to link hanges in preursor emissions from the various soures to responses in impat indiators at reeptors. Traditionally this task is aomplished by omprehensive atmospheri hemistry and transport models, whih simulate a omplex range of hemial and physial reations. The GAINS integrated assessment analysis relies on the Unified EMEP Eulerian model, whih desribes the fate of emissions in the atmosphere onsidering more than one hundred hemial reations involving 70 hemial speies with time steps down to 20 seonds, inluding numerous non-linear mehanisms (Simpson et al., 2003)(Fagerli and Aas, 2008). However, the joint analysis with eonomi and eologial aspets in the GAINS model requires omputationally effiient soure-reeptor relationships. For this purpose, response surfaes of the impat-relevant air quality indiators are desribed through mathematially simple formulations. Funtional relationships have been developed for hanges in annual mean PM 2.5 onentrations, deposition of sulphur and nitrogen ompounds as well as in long-term levels of ground-level ozone. The (grid- or ountry-speifi) parameters of these relationships have been derived from a sample of several hundred runs of the full EMEP Eulerian model with systematially perturbed emissions of the individual soures. Soure-reeptor relationships have been developed for hanges in emissions of SO 2, NOx, NH 3, VOC and PM 2.5 for 43 ountries in Europe and five sea areas, desribing their impats for the European territory on a 50 km 50 km grid resolution. 4.. Fine partiulate matter The health impat assessment in GAINS relies on epidemiologial studies that assoiate premature mortality with annual mean onentrations of PM 2.5 monitored at urban bakground stations. Thus, the soure-reeptor relationships developed for GAINS desribe, for a limited range around a referene emission level, the response in annual mean PM 2.5 levels to hanges in the preursor emissions of SO 2, NOx, NH 3 and primary PM 2.5. This formulation desribes the formation of partiulate matter (PM) from anthropogeni primary PM emissions and seondary inorgani aerosols only. It exludes PM from natural soures and primary and seondary organi aerosols due to insuffiient onfidene in the urrent modelling ability. Thus, the approah does not reprodue the full mass of PM 2.5 that is observed in ambient air. Consequently, results should be ompared only against observations of the individual speies that are modelled. The health impat assessment in GAINS is onsequently only onduted for hanges in the speified anthropogeni 6
preursor emissions, and exludes the largely unknown role of seondary organi aerosols and natural soures. The regional-sale assessment is performed for all of Europe with a spatial resolution of 50 km 50 km. Health impats are, however, most pertinent to urban areas where a major share of the European population lives. Any assessment with a 50 km resolution will systematially underestimate higher pollution levels in European ities. Based on the results of the City-Delta model interomparison, whih brought together the 7 major European urban and regional sale atmospheri dispersion models (Thunis et al., 2007), a generalized methodology was developed to desribe the inrements in PM 2.5 onentrations in urban bakground air that originate on top of the long-range transport omponent from loal emission soures. These relationships assoiate the differene in the annual mean PM 2.5 onentrations between an urban area and the average onentrations alulated over the 50 km 50 km grid ell surrounding the ity with spatial variations in emission densities of low-level soures and ityspeifi meteorologial and topographi fators. The GAINS/City-Delta methodology starts from the hypothesis that urban inrements in PM 2.5 onentrations originate predominantly from primary PM emissions from low-level soures within the ity. The formation of seondary inorgani aerosols, as well as the dispersion of primary PM 2.5 emissions from high staks, is refleted in the bakground omputed by the regional-sale dispersion model. 4..2 Deposition of sulphur and nitrogen ompounds The ritial loads approah employed by the GAINS model for the quantifiation of eosystems risks from aidifiation and eutrophiation uses (eosystem-speifi) annual mean deposition of aidifying ompounds (i.e., sulphur, oxidized and redued nitrogen) as the impat-relevant air quality indiator. Signifiant non-linearities in the spatial soure-reeptor relationships due to o-deposition with ammonia have been found for the substantial emission redutions that have ourred over the last two deades (Fowler et al., 2005). However, the EMEP Eulerian model suggests for the tehnially feasible range of further emissions redutions beyond the baseline projetion nearly linear responses in annual mean deposition of sulphur and nitrogen ompounds towards hanges in SO 2, NOx and NH 3 emissions. 4.2 Air quality impats 4.2. Health impats from PM Based on the findings of the WHO review on health impats of air pollution (WHO, 2003)(WHO, 2007), the GAINS model quantifies premature mortality that an be attributed to long-term exposure to PM 2.5, following the outomes of the Amerian Caner Soiety ohort study (Pope et al., 2002) and its reanalysis (Pope et al., 2009). Cohort- and ountry-speifi mortality data extrated from life table statistis are used to alulate for eah ohort the baseline survival funtion over time (Mehler et al., 2002). The survival funtion l (t) indiates the perentage of a ohort alive after time t elapsed sine starting time w 0. l (t) is an exponential funtion of the sum of the mortality rates µ a,b, whih are derived from life tables with a as age and b as alendar time. As the relative risk funtion taken from (Pope et al., 2002) applies only to ohorts that are at least w 0 =30 years old, younger ohorts were exluded from this analysis. Aordingly, for a ohort aged, l (t) is: 7
t ( t ) = exp µ z, z + w. () z= l 0 The survival funtion is modified by the exposure to PM pollution, whih hanges the mortality rate and onsequently the remaining life expetany (e ). For a given exposure to PM 2.5 (PM), life expetany l is alulated as the integral over the remaining life time: w w t e = l t dt RR ( ) = exp PM µ z, z + w dt (2) 0 z= where w is the maximum age onsidered and RR PM the relative risk for a given onentration of PM 2.5. With some simplifying assumptions and approximations (Vaupel and Yashin, 985), the hange in life expetany per person ( e ) of a ohort an be expressed as: e = w β PM l ( t ) log l ( t ) dt (3) where within the studied exposure range RR PM has been approximated as RR PM = β PM+ with β = 0.006 as given in (Pope et al., 2002). For all ohorts in a ountry l the hange in life years L l is then alulated as the sum of the hange in life years for the ohorts living in the grid ells j of the ountry l: L l = w = w0 L w w Pop j, i = β PM j Pop, l l ( t) log l ( t) dt (4) Pop j l l = w0 where L,l Pop,l Change in life years lived for ohort in ountry l Population in ohort in ountry l Pop j Total population in grid ell j (at least of age w 0 =30) Pop l Total population in ountry l (at least of age w 0 =30). 4.2.2 Health impats from ozone Based on a omprehensive meta-analysis of time series studies onduted for the World Health Organization (Anderson et al., 2004) and on advie reeived from the UNECE/WHO Task Fore on Health (UNECE/WHO, 2003), the GAINS model quantifies premature mortality through an assoiation with the so-alled SOMO35 indiator for long-term ozone onentrations in ambient air. SOMO35 is alulated as the daily eight-hour maximum ozone onentrations in exess of a 35 ppb threshold, summed over the full year. In essene, the GAINS alulation estimates for the full year daily hanges in mortality as a funtion of daily eight-hour maximum ozone onentrations, employing the onentrationresponse urves derived in the meta-analysis (Anderson et al., 2004). The threshold was introdued (i) to 8
aknowledge unertainties about the validity of the linear onentration-response funtion for lower ozone onentrations, and (ii) in order not to overestimate the health effets. The annual ases of premature mortality attributable to ozone are then alulated as Mort l = 365 2 Deaths RR O3 (5) l O3 l where Mort l Deaths l RR O3 Cases of premature mortality per year in ountry l Baseline mortality (number of deaths per year) in ountry l Relative risk for one perent inrease in daily mortality per µg/m 3 eight-hour maximum ozone onentration per day. O3 l Population-weighted SOMO35 in ountry l In addition to the mortality effets, there is lear evidene of aute morbidity impats of ozone (e.g., various types of respiratory diseases). However, the GAINS model quantifies only mortality impats of ozone, as they are seen to be the dominant fator in any eonomi benefit assessment. Referenes M. Amann, I. Bertok, J. Borken-Kleefeld, J. Cofala, C. Heyes, L. Höglund-Isaksson, Z. Klimont, B. Nguyen, M. Posh, P. Rafaj, R. Sandler, W. Shöpp, F. Wagner, W. Winiwarter: Cost-effetive ontrol of air quality and greenhouse gases in Europe: Modelling and poliy appliations, Environmental Modelling and Software, Vol. 26, pp. 489-50, 20. M. Amann, L. Höglund-Isaksson, W. Winiwarter, A. Tohka, F. Wagner, W. Shöpp, I. Bertok and C. Heyes: Emission senarios for non-co 2 greenhouse gases in the EU-27, Final Report to DG Environment, International Institute for Applied Systems Analysis (IIASA), May 2008. http://e.europa.eu/lima/poliies/pakage/dos/ir_07_nono2_en.pdf. Anderson, H.R., R.W. Atkinson, J.L. Peaok, L. Marston, K. Konstantinou: Meta-analysis of time-series studies and panel studies of Partiulate Matter (PM) and Ozone (O3). World Health Organization, Bonn, 2004. L. Höglund-Isaksson, W. Winiwarter, F. Wagner, Z. Klimont and M. Amann: Potentials and osts for mitigation of non-co 2 greenhouse gases in the European Union until 2030 Results, Report to DG Climate Ation, International Institute for Applied Systems Analysis (IIASA), Laxenburg, May 200. http://e.europa.eu/lima/poliies/pakage/dos/non_o2emissions_may200_en.pdf. L. Höglund-Isaksson, W. Winiwarter, and A. Tohka: Potentials and osts for mitigation of non-co 2 greenhouse gases in the European Union Methodology, International Institute for Applied Systems Analysis (IIASA), Laxenburg, May 200. http://gains.iiasa.a.at/images/stories/reports/eu27-nonco2-25may200_methodology.pdf. 9
L. Höglund-Isaksson, W. Winiwarter, P. Purohit, P. Rafaj, W. Shöpp and Z. Klimont: EU Low Carbon Roadmap 2050: Potentials and osts for mitigation of non-co 2 greenhouse gas emissions, Paper submitted to Energy Strategy Reviews, January 26, 202. ESIA: Intermediate status report of the progress towards the redution of perfluoroompound (PFC) emissions from European semiondutor manufaturing, European Semiondutor Industry Assoiation (ESIA), Brussels, 2006. Fagerli, H. and W. Aas: Trends of nitrogen in air and preipitation: Model results and observations at EMEP sites in Europe, 980 2003. Environmental Pollution 54, 448-46, 2008. Fowler, D., J. Muller, R.I. Smith, J.N. Cape, J.W. Erisman: Nonlinearities in Soure Reeptor Relationships for Sulfur and Nitrogen Compounds. Ambio 34, 4-46, 2005. IPCC: IPCC Revised 996 Guidelines for National Greenhouse Gas Inventories, J.T. Houghton, L.G. Meira Filho, B. Lim, K. Treanton, I. Mamaty, Y. Bonduki, D.J. Griggs, B.A. Callender (Eds.), IPCC/OECD/IEA, UK Meteorologial Offie, Braknell, 997. IPCC: 2006 IPCC Guidelines for National Greenhouse Gas Inventories, L. B. S. Eggleston, K. Miwa, T. Ngara, K. Tanabe (Eds.), Intergovernmental Panel on Climate Change, Japan, 2006. Pope, C.A., R. Burnett, M.J. Thun, E.E. Calle, D. Krewski, K. Ito, G.D. Thurston: Lung Caner, Cardiopulmonary Mortality and Long-term Exposure to Fine Partiulate Air Pollution. Journal of the Amerian Medial Assoiation 287, 32-4, 2002. Pope, C.A., M. Ezzati, D.W. Dokery: Fine-Partiulate Air Pollution and Life Expetany in the United States. The New England Journal of Mediine 360, 376-86, 2009. Simpson, D., H. Fagerli, J. E. Jonson, S. Tsyro, P. Wind, T.P. Tuovinen: Transboundary Aidifiation, Eutrophiation and Ground-level Ozone in Europe Part I. Unified EMEP Model Desription. Norwegian Meteorologial Institute, Oslo, 2003. Thunis, P., L. Rouil, C. Cuvelier, R. Stern, A. Kershbaumer, B. Bessagnet, M. Shaap, P. Builtjes, L. Tarrason, J. Douros, N. Moussiopoulos, G. Pirovano, M. Bedogni: Analysis of model responses to emission-redution senarios within the CityDelta projet, Atmospheri Environment 4, 208-220, 2007. UNECE/WHO: Modelling and Assessment of the Health Impat of Partiulate Matter and Ozone ( No. EB.AIR/WG./2003/). United Nations Eonomi Commission for Europe, World Health Organization, 2003. WHO: Health Aspets of Air Pollution with Partiulate Matter, Ozone and Nitrogen Dioxide. World Health Organization, Bonn, 2003. WHO: Health relevane of partiulate matter from various soures. Report on a WHO Workshop Bonn, Germany, 26-27 Marh 2007. World Health Organization, Regional Offie for Europe, Copenhagen, Denmark, 2007. 0