Methodology for Inventorying GHG Emissions from Global Cities

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy ARTICLE IN PRESS

Energy Policy 38 (2010) 4828–4837

Contents lists available at ScienceDirect

Energy Policy

journal homepage: www.elsevier.com/locate/enpol

Methodology for inventorying greenhouse gas emissions from global cities

Christopher Kennedy a,Ã, Julia Steinberger b,1, Barrie Gasson c,2, Yvonne Hansen d,3, Timothy Hillman e,4, Miroslav Havranek f,5, Diane Pataki g,6, Aumnad Phdungsilp h,7, Anu Ramaswami e,8, Gara Villalba Mendez i,9 a Department of Civil Engineering, University of Toronto, 35 St. George Street, Toronto, Canada M5S 1A4 b Vienna Institute of Social Ecology, IFF, University of Klagenfurt, Schottenfeldgasse 29, 1070 Vienna, Austria c School of Architecture and Planning, University of Cape Town, 7701 Rondebosch, Cape Town, South Africa d African Centre for Cities, University of Cape Town, 7701 Rondebosch, Cape Town, South Africa e Department of Civil Engineering, University of Colorado Denver, Campus Box 113, P.O. Box 173364, Denver, CO 80217-3364, USA f Charles University Environment Center, U Krize 8, 15000 Prague, Czech Republic g Department of Earth System Science and Dept. of Ecology & Evol. Biology, University of California, Irvine, Irvine, CA 92697-3100, USA h Department of Energy Management, Dhurakij Pundit University, 110/1-4 Prachachuen Rd., Laksi, Bangkok 10210, Thailand i Institut de Ciencia i Tecnologia Ambientals, Autonomous University of Barcelona, Ediﬁci Cn 08193, Bellaterra article info abstract

Article history: This paper describes the methodology and data used to determine greenhouse gas (GHG) emissions Received 21 January 2009 attributable to ten cities or city-regions: Los Angeles County, Denver City and County, Greater Toronto, Accepted 21 August 2009 New York City, Greater London, Geneva Canton, Greater Prague, Barcelona, Cape Town and Bangkok. Available online 12 September 2009 Equations for determining emissions are developed for contributions from: electricity; heating and Keywords: industrial fuels; ground transportation fuels; air and marine fuels; industrial processes; and waste. Urban metabolism Gasoline consumption is estimated using three approaches: from local fuel sales; by scaling from Life-cycle analysis regional fuel sales; and from counts of vehicle kilometres travelled. A simpliﬁed version of an Climate change intergovernmental panel on climate change (IPCC) method for estimating the GHG emissions from landﬁll waste is applied. Three measures of overall emissions are suggested: (i) actual emissions within the boundary of the city; (ii) single process emissions (from a life-cycle perspective) associated with the city’s metabolism; and (iii) life-cycle emissions associated with the city’s metabolism. The results and analysis of the study will be published in a second paper. & 2009 Elsevier Ltd. All rights reserved.

1. Introduction as centers of wealth and creativity, with high population densities and economies of scale, cities must play a significant role in With the global urban population now exceeding 50 percent, tackling global climate change. This is particularly clear in a the inhabitants of cities are recognized as a major driver of global political context where goals and actions of groups such as the greenhouse gas (GHG) emissions (Satterthwaite, 2008). Moreover, C40 mayors exceed those of many national governments. As a first step to addressing climate change many cities have established inventories of GHG emissions, often using the simple pragmatic Ã Corresponding author. Tel.: +1416 978 5978. approach of the International Council for Local Environmental E-mail addresses: Initiatives (ICLEI, 2008). Such inventories of urban GHG emissions [email protected] (C. Kennedy), [email protected] (J. Steinber ger), [email protected] (B. Gasson), [email protected] (Y. Hansen), provide a basis for action on climate change. In particular, [email protected] (T. Hillman), [email protected] (M. Havranek), identifying cities with better per capita performance than others [email protected] (D. Pataki), [email protected] (A. Phdungsilp), Anu.Ramaswa may help to guide climate change policies for urban settings, but [email protected] (A. Ramaswami), [email protected] (G.V. Mendez). before this can be done we need comparable accounting systems. 1 Tel.: ++4315224000 411. 2 Tel.: +27 21650 2366. Closer inspection shows that there are several technical issues 3 Tel.: +27 21650 5903. with GHG inventories for cities: lack of full life-cycle perspective; 4 Tel.: +1303 556 4734. problems with defining spatial and temporal context; and issues 5 Tel.: +420251080351. of assigning emissions by political jurisdiction (Ramaswami et al., 6 Tel.: +1949 824 9411. 2008; Kennedy and Mohareb, 2009). It is also questionable 7 Tel.: +662 9547300x587. 8 Tel.: +1303 556 4734. whether existing inventory schemes are being used consistently 9 Tel.: +93 58124 88. between cities. This is particularly the case for transportation

C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837 4829

Table 1 Deﬁnition of cities or city-regions used in the study, with their populations and year of GHG inventory.

City or city-region Deﬁnition Total land area (km2) Population Year

Bangkok City 1569 5,658,953 2005 Barcelona City 100 1,605,602 2006 Cape Town City 2454 3,497,097 2005 Denver City and County 397 579,744 2005 Geneva Canton 282 432,058 2005 London Greater London Area (GLA) 1579 7,364,100 2003 Los Angeles County (88 cities) 10,518 9,519,338 2000 New York City City 789 8,170,000 2005 Prague Greater Prague Area (GPA) 496 1,181,610 2005 Toronto Greater Toronto Area (GTA) 7195 5,555,912 2005

emissions, where the ideal approach is based on quantification of WRI / Spatial Life-cycle Components Measure fuel consumed in a regional urban area, i.e., a commutershed. WBCSD boundary perspective definition Many estimates of transportation emissions from cities are based on computer simulated measures of vehicle kilometres travelled Embodied emissions from (VKT) within politically defined city boundaries. How well do food and materials consumed in cities these approaches compare? To develop and test a more robust and transparent inventory Out of Production Emissions upstream of boundary chain electric powerplants procedure, this research has undertaken a comparative study of energy use emissions (iii) GHG emissions, including life-cycle impacts, from a small group of Scope 3 (and further global cities. These include Los Angeles (County), Denver (City & out of Upstream emissions from boundary fossil fuel use County), Greater Toronto, New York City, Greater London, Geneva emissions, (ii) (Canton), Barcelona, Greater Prague, Cape Town and Bangkok not included in Scope 2) Combustion of aviation and (Table 1). These cities were chosen because they have been the marine fuels subject of urban metabolism studies (Gasson, 2002; Sahely et al., ICLEI 2003; Emmenegger et al., 2003; Ngo and Pataki, 2008; CUEC, Out of boundary waste 2008), urban energy or GHG studies (Baldasano et al., 1999; (landfill) emissions Phdungsilp, 2006; Ramaswami et al., 2008), or are prominent Out of boundary district global cities with detailed publicly accessible GHG inventories Single heating emissions (Mayor of London, 2006; New York City, 2007). process Note that these existing studies were useful as background emissions Scope 2 In boundary Out of boundary electricity material, but not all could be used directly for this work; indeed, electricity emissions at powerplant most of the data which are presented in this study is compiled use from original sources and has never been published before. (i) Results for the determination of GHG emissions will be published In boundary fossil fuel elsewhere (Kennedy et al., 2009). This paper describes the combustion methodology that was used in the study, and presents a In boundary waste (landfill) substantial new data set for the energy and waste components Scope 1 In boundary emissions of these cities urban metabolism. emissions In boundary industrial The global warming potential of GHGs attributable to cities, processes and product use including carbon dioxide, methane, nitrous oxide and several In boundary agriculture, other gases, is expressed in terms of carbon dioxide equivalents, forestry and other land use eCO2. From a practical perspective, emissions of CO2 itself dominate the urban inventory, with methane of significance for Fig. 1. Comparison of measures for attributing greenhouse gas emission to cities landfilled waste, and other gases mainly of significance for considering spatial boundaries and life-cycle perspective. industrial emissions where they occur. The methodology and data used to quantify the emissions for components of: electricity; heating and industrial fuels; ground transportation fuels; aviation forestry and other land use (AFOLU), which are determined as per and marine fuels; industrial processes; and waste, follows in IPCC guidelines. (Note that AFOLU emissions are typically small Section 2 below. for cities and have been excluded in this paper.) Scope 2 emissions Different metrics for expressing the total GHG emissions include out-of-boundary emissions due to electricity used in attributable to cities can be developed depending on definition cities. Several further out-of-boundary emissions attributable to of spatial boundary and life-cycle perspective. Fig. 1 shows the cities are included in Scope 3. three measures that we use in this work (with the total emissions Here we distinguish between Scope 3 emissions that are single attributable to cities increasing from the bottom of the diagram to process emissions and those that are production chain emissions the top). The figure uses the terminology of the World Resources from a life-cycle perspective. Measure (ii) that we use includes Institute/World Business Council for Sustainable Development single process emissions from out-of-boundary landfills, district (WRI/WBSCD) to express emissions in relation to the spatial heating, aviation and marine transportation that are attributable boundary of a city. Scope 1 emissions (our measure (i)) only to cities. Our measure (iii) would ideally include all the embodied include those that are produced within city boundaries. These emissions associated with producing the food and materials include in-boundary components from: fossil fuel combustion; consumed in cities, and emissions that are upstream of electric waste; industrial processes and product use; and agriculture, powerplants. Due to data limitations, however, it only considers Author's personal copy ARTICLE IN PRESS

4830 C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837 upstream emissions from fuel combustion. Further development intensity factors by the respective agencies. Since we aimed to of these three measures of total GHG emissions is given in collect data on electricity consumption by final users (i.e., from Section 3. the aggregation of consumer bills provided by retailers), the line As well as contributing to the development of methodology for losses include both those in transmission and distribution. If attributing GHG emissions to cities, this paper also presents a consumption had been based on meter readings at sub-stations substantial new data set on energy use and wastes produced in located in the cities, then only transmission losses would need to cities. Our comprehensive data set on energy use in ten cities fills be accounted for. a gap in the literature identified by previous researchers (Decker The GHG emissions factors, Ielectricity (t eCO2/GWh), are et al., 2000; Kennedy et al., 2007). typically those for the mix of powerplants in the city’s state, province or region (Table 2). In the cases of New York City, Denver, Geneva and Bangkok, the emissions factors reflect a mix of 2. Components of GHG inventory supplies from different scales, e.g., local and imported. In these cases, a weighted average of the emissions intensities from the 2.1. Electricity applicable sources is used. Where residents or business indepen- dently purchase electricity from a certified renewable energy supplier, e.g., Xcel Energy’s Windsource program in Denver, this GHG emissions (t eCO2) attributable to electricity consumption are determined by: too is reflected in the mix of supplies. The emission intensities vary widely depending on the source fuels for power generation. GHGelectricity ¼ Celectricity L Ielectricity ð1Þ The examples in Table 3 show how Cape Town has high emission intensity due to the predominant use of coal in South Africa, while The electricity consumption, Celectricty (GWh), excludes that from combined heat and power (CHP) plants within the urban Toronto’s intensity is relatively low due to substantial use of region and electricity derived from the combustion of waste. From nuclear and hydro power in Ontario. a data collection perspective it was easier to use a convention of including emissions from CHP and waste combustion under 2.2. Heating and industrial fuels heating fuels and waste, respectively. Geneva’s 2793 GWh of electricity consumption in Table 2 does not include the 670 GWh Emissions in this category are primarily due to fossil fuels used of electricity used at the CERN particle accelerator facility in 2005. for heating in buildings, e.g., space heating, water heating and CERN contracts half of its electricity from Switzerland and half cooking. Also included are fossil fuels used by CHP facilities within from France. cities (mainly natural gas and oil) and, where data are available, Electrical line losses range from 5% to 12% as shown by the loss fossil fuels combusted by industry. factor, L,inTable 2. Line loss factors for Barcelona, Geneva, London GHG emissions (t eCO2) are determined by: and Prague are estimates from national level data (IEA, 2008). The X loss factors for Denver and Los Angeles are unknown (to the GHGheating ¼ Cfuel Ifuel ð2Þ authors), but were included in the computation of the GHG heating fuel

Table 2 Electricity use, line losses and GHG intensity for the ten cities.

Electricity consumption GWh Line loss factor GHG intensity t eCO2/GWh (incl. line loss) Source

Bangkok 28,500 1.081 550 BMA (2008), ERM-Siam (2005) Barcelona 7479 1.11 143 ICAEN, Baldasano et al (1999) Cape Town 12,209 1.078 969 City of Cape Town Denver 6659 Unknown 792 Xcel Energy Geneva 2793 1.07 54 Ofﬁce Cantonal de la Statistique London 39,237 1.09 469 Mayor of London (2006) Los Angeles 63,919 Unknown 368 California Energy Commission New York City 49,567 1.055 497 New York City (2007) Prague 5506 1.06 710 ERO (2008), Fott et al. (2008) Toronto 55,778 1.12 246 Hydro One, Environment Canada

Table 3 Examples of electricity supply mixes for Cape Town and Toronto, and their GHG intensity factors. The supply mixes are those for South Africa and Ontario, respectively.

Source Cape Town (South Africa) Toronto (Ontario)

Supply mix (%) GHG intensity (t CO2e/GWh) Supply mix (%) GHG intensity (t CO2e /GWh)

Coal 92.0% 978 17.7 1006 Natural gas 8.8 432 Oil 0.5 832 Hydro 2.0% 0 22.3 0 Nuclear 6.0% 0 50.2 0 Other 0.6 0

Overall generation 899 220

Line loss factor 1.078 1.12 Overall consumption 969 246 Author's personal copy ARTICLE IN PRESS

C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837 4831

The energy contents of fuels used, Cfuel (TJ), is given for each cities are particularly interesting: Cape Town uses no natural gas, city in Table 4. In some cases the energy content was calculated and mainly relies on a mixture of gasoil, heavy furnace fuel, from volumes of fuels consumed. For Los Angeles, the fuel use was parafﬁn, LPG and coal; Bangkok has a large and diverse mix of estimated based on Californian data (Ngo and Pataki, 2008). fuels including substantial quantities of rice husk and bagasse. Natural gas is the predominant fuel used in the European and The default (tier 1) IPCC (2006a) GHG emissions factors

North American cities, with the exception of Geneva, which uses for fuels, Ifuel (t eCO2/TJ), were used for calculating direct substantial quantities of fuel oil. Prague is also supplied by one of emissions. Emissions factors for upstream emissions are discussed the longest heat pipelines in the world, with steam traveling up to in section 3. 60 km from a coal ﬁred power plant at Melnik. The remaining two 2.3. Ground transportation fuels

Table 4 The fuels used for ground transportation within cities are Heating and industrial fuel use (TJ) in the ten cities. primarily gasoline and diesel, with small amounts of LPG and Bangkok (Source: Phdungsilp, 2006) natural gas used in some cases. Emissions due to use of electrified Wood 8725 modes of transportation, e.g., subways and streetcars, were Fuel oil 32,217 counted in the electricity category. Natural gas 15,407 GHG emissions for ground transportation fuels (t eCO ) are LPG 12,041 2 determined by: Kerosene 322 X Gasoline 595 GHG ¼ C I ð3Þ Diesel 6033 transport fuel fuel Coal and coke 20,331 transport fuel Lignite 22,035 with default IPCC GHG emissions factors again used for direct Anthracite 632 Rice husk 8449 emissions. Bagasse 33,427 Emissions from ground transportation can contribute as much Charcoal 314 as 20–40% of a city’s GHG emissions, but are the greatest source of uncertainty in the total inventory due to the estimation Barcelona (Source: Agencia d’Energia de Barcelona, Area de Medi Ambient de l’ajuntament de Barcelona) procedures involved. Therefore, in order to test and verify Natural gas 23,252 calculations of gasoline consumption in cities, three different LPG 895 methods were applied. Cape Town (Sources: Caltex, City of Cape Town, 2003) The preferred approach, where data are available and depend- Gasoil 20,511 ing on the spatial boundaries of the city, is to use local fuel sales Heavy furnace fuel 12,341 data. This approach is appropriate where the city-region is a Illumin. paraffin 8012 commutershed, i.e., for which the number of vehicle trips over the LPG 9734 Power paraffin 23 border of the city-region is small relative to the number of trips Coal 3788 within it. For example, in 2006, the number of auto and rail trips Wood 561 within the Greater Toronto Area was 12.078 million, while the Denver (Source: Xcel Energy) number of daily commuter trips into and out of the area was just Natural gas 42,622 535,000 (based on Transportation Tomorrow Survey data). Gaso- Geneva (Source: Office Cantonal de la Statistique, Service Cantonal de line consumption for Cape Town, Geneva and Greater Toronto is l’energie) established by this approach (Table 5a), although in the case of Natural gas 8891 Geneva there is some substantial cross-border commuting from Fuel oil 12,746 France. Sales data are also established for the City of Bangkok, Coal 3 Wood 109 even though the commutershed is much larger than the city. For Waste* 404 Greater Toronto, retail sales data are adjusted upwards by 11% Solar thermal 10 (based on provincial fuel sales data) to account for non-retail, bulk London (Source: Mayor of London, 2006) purchasing of gasoline. Natural gas 299,908 The second approach is to use gasoline consumption values Oil 12,093 estimated from vehicle kilometres travelled (VKT) within cities Coal 255 (Table 5b). The VKT for cities is typically estimated using a CHP 18,392 computer model, or by a vehicle counting survey. The calculation Los Angeles (Source: scaled from Bemis 2006) of fuel use by this approach also requires estimation of fuel Natural gas 216,409 efficiency for typical vehicles. This approach is perhaps the most Petroleum 13,410 Coal 167 common, and has been applied here to Bangkok, Barcelona, Denver, London, New York City and Prague. It has some New York City (Source: New York City, 2007) Natural gas 252,707 Residual fuel oil 34,130 Table 5a Distillate fuel oil 77,656 Estimation of gasoline consumption from sales within city or city-region Kerosene 4475 Steam (CHP) 45,890 Gasoline sales Source (ML) Prague (Source: Prague City Hall, 2008) Natural gas 37,397 Bangkok 2741 Department of Energy Business, Ministry of Fuel oil 232 Energy Coal 8181 Cape 1249 City of Cape Town Steam (coal power plant) 9454 Town Toronto (Sources: Enbridge Gas and Union Gas) Geneva 260 Office Cantonal de la Statistique Natural gas 327,000 Toronto 6691 Kent Marketing & Statistics Canada Author's personal copy ARTICLE IN PRESS

4832 C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837

Table 5b Estimation of gasoline consumption based on vehicle kilometres travelled (VKT).

VKT (billion km) Fuel efﬁciency (km/L) Gasoline consumption (ML)

Bangkok (Source: BMA, 2008; Tanatvanit et al., 2003) Passenger car 23.272 11.67 1994 Microbus & pickup 0.665 12.32 54 Van & pickup 0.873 12.42 70 Motortricycle 0.005 12.00 0.4 Urban taxi 1.102 11.63 95 Fixed route taxi 0.090 13.00 7 Motortricycle taxi 0.005 12.50 0.4 Motorcycle 10.790 24.57 439 Business taxi 0.015 11.63 1 2662

Barcelona (Source: Servei d’Ordenacio & i Seguretat Vial de la Generalitat de Catalunya: Road Planning and Safety Service of the Barcelona City Council)- Cars 1.675 10.15 165 Motorcycles 0.600 35.29 17 Trucks and vans 0.168 6.21 27 209

Denver (Source: Ramaswami et al., 2008) Gasoline vehicles 8.045 6.38 1234

London (Source: Mayor of London, 2006) Cars Unknown Unknown 1711 Motorcycles Unknown Unknown 32 Goods vehicles Unknown Unknown 54 1797

New York City (Source: New York City, 2007) Passenger vehicles 18.167 7.48 2428 Light trucks 10.274 5.87 1751 4179

Prague (Source: Transport yearbook, CDV, 2007 estimates) Gasoline cars Unknown Unknown 357

Table 5c Estimation of gasoline consumption by scaling from sales within a wider region.

Wider region Gasoline consumption of wider region (ML) Scaling factor Estimated gasoline consumption (ML)

Los Angeles California 52,700 0.28 14,751 New York City New York State 21,448 0.192 4107 Toronto Ontario 15,857 0.441 6988 advantages over the fuel sales approach when applied to central approaches can be gained. For Toronto, the estimate of 6988 cities that experience large cross-boundary commuter trafﬁc million litres scaled from provincial consumption is 4.4% higher (sales may occur outside boundaries). A modelling approach is than the 6691 million litres determined from fuel sales. While for also preferred by the IPCC for determining CH4 and N2O New York City, the scaled estimate of 4107 million litres is 1.7% emissions, which depend on trafﬁc conditions and cold-starts. lower than the 4179 million litres determined from VKT. Finally,

For CO2 emissions, however, the IPCC prefers the use of fuel sales for Bangkok, the estimate of 2662 million litres based on VKT, is data. One of the drawbacks of modelling VKT, or estimating it 2.9% lower than the estimate of 2741 million litres from fuel sales. from vehicle counts, is that techniques may differ between cities; Overall, the comparisons show that the three approaches can each city has its own unique model. produce reasonably close estimates, provided that the VKT The third approach is to estimate a city’s gasoline consumption calculations are conducted appropriately. by scaling from state, provincial, or regional data (Table 5c). Motor Consumption of other ground transportation fuels is estimated vehicle registrations, R, were used to determine a scaling factor for using the same approaches as used for gasoline (Table 6). Toronto and New York City: 2.4. Aviation and marine transportation Rcity Ccity ¼ Cregion ð4Þ Rregion There are differing approaches to the quantiﬁcation of aviation The scaling factor for Los Angeles County is based on and marine emissions that occur with the transportation of population, which corresponded closely to aggregate travel people and goods to and from cities. In some studies, aviation and commute times from the US census (Ngo and Pataki, 2008). This marine emissions are excluded. This is because they are mainly approach is the least desirable of the three, as is assumes that (but not entirely) cross-boundary emissions; and, moreover, may average annual vehicle kilometres in the city are the same as for be considered as beyond the control of local governments. Other the wider region. Nevertheless, scaling from regional data may be studies, such as that by Carney et al. (2009), include emissions necessary when reliable data at the city level is unavailable. from domestic aviation, but only include take-off and landings for As gasoline use in three of the cities has been determined by international aviation. Such an approach is consistent with the more than one method, a sense of the accuracy of the three United Nations Framework Convention on Climate Change, which Author's personal copy ARTICLE IN PRESS

C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837 4833

Table 6 Consumption of other ground transportation fuels.

Diesel (ML) LPG (TJ) Natural gas (TJ) Source

Bangkok 2094 0 0 BMA (2008), Tanatvanit et al. (2003) Barcelona 266 0.054 131 Barcelona City Council Cape Town 724 0 0 City of Cape Town Denver 197 0 0 Ramaswami et al. (2008) Geneva 51 0 0 Ofﬁce Cantonal de la Statistique London 1238 0 0 Mayor of London (2006) Los Angeles 3212 1699 4296 scaled from Bemis (2006) New York City 657 0 0 New York City (2007) Prague 281 0 0 Transport yearbook, est. (2008) Toronto 2011 0 0 Kent Marketing, Statistics Canada

excludes GHG emissions from cruising on international flights, Table 7 and international marine activity. Other studies, such as those for Jet fuel loaded onto planes at major airports in the ten cities. London (GLA, 2007), New York City (2007) and Denver (Ramas- Jet fuel Airports (Source) wami et al., 2008), have reported GHG emissions based on fuels (ML) loaded at airports and marine ports within their boundaries. This approach, including all emissions from domestic and international Bangkok 4293 Donmaung Airport (Dept of aviation/marine transportation, provides a more complete mea- Energy Business, Ministry of Energy) sure of the carbon dependence of a city’s economy. It recognizes Barcelona 1058 Barcelona Airport- El Prat cities as being global service centres—the headquarters, financial (scaled from national jet fuel centers, and key gateways between national/regional economies stats.) (Sassen, 1991; Taylor, 2004). Overall, the choice of approach Cape Town 1211 Cape Town International perhaps depends on the user of the inventory; there is currently Airport (Caltex) Denver 326 Denver International Airport no standard approach for quantifying aviation and marine (Ramaswami et al., 2008) emissions for cities. Geneva 294 Geneva International Airport In our study, GHG emissions from international and domestic (Office Cantonal de la aviation and marine transportation were sought, in order to Statistique) London 9100 Heathrow and City Airports include impacts from the movement of people and goods to and (calculated from GLA, 2007) from the ten cities. (Technically, it is movement away from cities Los Angeles 4202 LAX, BUR, and LGB Airports that is quantified, since calculations are based on fuel loaded in (scaled from statewide CO2 cities.) There are still challenges to assigning emissions to cities stats.) under this wider approach. Perhaps it might be appropriate to New York City 5615 JFK and LaGuardia Airports (New York City, 2007)) include only those emissions associated with travel by residents Prague 420 Ruzyn Airport (Fott et al, 2008 of a city, or movements of goods consumed only by residents of a Toronto 1830 Toronto Pearson Airport city. On the other hand, if a city is a major gateway, or an attractor (GTAA) for visitors and conventions, as is the case for some of the ten, then that gateway function is part of what the city does—and contributes to its economy. There are further challenges with passengers transferring between planes at airports; and airports Table 8 serving wider regions than just cities. Ideally, the GHG emissions GHG emissions from marine fuels loaded onto ships at port cities. from passenger aviation attributed to a city might be determined GHG emissions Source from Eq. 5 below, as has been demonstrated for Denver from marine (Ramaswami et al., 2008): fuels (kt eCO2) X ncity Bangkok Unknown Fuel data from Caltex GHGaviation ¼ Ifuel Cfuel ð5Þ Barcelona Unknown nregion regional airports Cape Town 10,100 London Small Scaled from statewide stats, where Ifuel is the emissions factor for jet kerosene (other aviation using cargo tonnage Los Angeles 16,096 New York City (2007) fuels could also be included); Cfuel the energy content of jet kerosene loaded onto planes; n the annual number of surface New York City 6200 city Toronto Small passenger transportation trips from the city to the airport; and nregion the annual number of surface passenger transportation trips from the wider region to the airport. In the case of Denver, Ramaswami et al. (2008) were able establish that 22% of fuel loaded at Denver International Airport within city boundaries has been allocated to the city, despite the was attributable to activities of the City and County of Denver, potential for airports to serve a larger regions and transferring which is one of nine counties and 25 cities in the region. passengers. Based on availability of data, however, the only consistent It might also be noted that passengers and freight can enter means to determine GHG emissions from air and marine and leave cities by ground transportation modes such as railways, transportation for the other nine cities was from the full volumes buses and trucks. Emissions from these modes were not of fuels loaded onto planes and ships at the cities’ respective considered in this work, since these emissions would be counted airports and marine transport terminals (Tables 7 and 8). For the in the ground transportation of neighbouring and nearby regions. nine cities other than Denver, 100% of fuel loaded at airports (i.e., double-counting is avoided). Author's personal copy ARTICLE IN PRESS

4834 C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837

2.5. Emissions from industrial processes and product use are given by:

GHG^ ¼ 21 M L ð1 f Þð1 OXÞð6Þ Industrial process emissions include those from activities such landfill landfill 0 rec as cement manufacturing and limestone consumption. (They where Mlandfill is the mass of urban waste sent to landfill in the exclude emissions from combustion for industrial energy supply.) inventory year (Table 10); L0 the methane generation potential; Data for this category of emissions was only available for three the value 21 is the 100-year global warming potential of methane cities: Los Angeles, Prague and Toronto (Table 9); in other cases (IPCC, 2007); and frec the fraction of methane emissions that are industrial process emissions may or may not have been negligible. recovered at the landfill. The oxidation factor, OX, in Eq. (6) is For Los Angeles, industrial emissions were estimated from a state- typically no higher than 0.1; it was excluded from our approach wide inventory, weighted by industry employment. Prague has due to lack of supporting data. one cement plant reporting emissions for a national allocation The methane recovery factor, frec, is a significant source of plan for emissions trading. The Toronto data was that from uncertainty. Several studies have tried to measure recovery industrial facilities with large emissions (4100,000 t eCO2) that system efficiency. Oonk and Boom (1995) reports efficiency report to Environment Canada, excluding energy supply facilities. ranging between 10% and 85% (average 37%). More recently, Scharff et al. (2003) measured recovery efficiencies on four sites to be 9, 50, 53 and 33 percent, whereas Spokas et al. (2006) and Diot 2.6. Waste et al. (2001) measured efficiency above 90 percent. Willumsen (2003) gives maximum recovery efficiency in OECD countries of We used a simplified version of the IPCC recommended 80% and in non-OECD countries of 50%. Methane recovery approach for estimating the GHG emissions from landfill waste. efficiencies for the ten cities in this study have been determined The ideal aim would be to calculate the methane emissions for a at very few landfills. The general assumption made by most of the given year due to the decay of waste placed in the landfill during cities is that the methane recovery factor is about 0.75. This value that year and previous years. The IPCC (2006b) recommends an has been used for eight of the ten cities, the exceptions being approach called First Order Decay for estimating such emissions Bangkok and Cape Town, which do not currently capture methane. based on the Scholl Canyon model. The data requirements are, The methane generation potential, L (t CH4/t waste), is however, cumbersome, requiring ideally 20 or more years of data 0 determined using the IPCC (2006b) approach as follows: for each facility within each city, and good estimates of decay coefficients. We undertook a pragmatic adaptation of the IPCC 16 L ¼ MCF DOC DOC F ð7Þ (1996) approach called Total Yield Gas. 0 12 F This alternative approach estimates the long-term GHG where: emissions for the waste landfilled in the inventory year, recognizing that the emissions would be released over several MCF CH correction factor (equal to 1.0 for managed land- years. The long-term GHG emissions from landfill waste (t eCO ) 4 2 fills); DOC degradable organic carbon (t C/t waste); Table 9 DOC fraction DOC dissimilated (default range 0.5–0.6; as- Industrial process emissions determined for Los Angeles, Prague and Toronto. F sumed equal to 0.6); Industrial process GHG F fraction of methane in landfill gas (range: 0.4–0.6; emissions (t eCO2) assumed equal to 0.5); 16/12 stoichiometric ratio between methane and carbon. Los Angeles (Source: scaled from statewide GHG inventory using employment statistics in the industrial sector) Cement production 1,700,000 The degradable organic carbon, DOC, is estimated from waste Limestone & dolomite consumption 100,000 fractions, fi, as follows: Soda ash consumption 100,000 X Semiconductor manufacture 200,000 DOC ¼ Wi fi ð8Þ i 2,100,000 Prague (Source: National Allocation Plan for emission trading) where the weightings Wi are as shown in Table 11. Waste fractions Cement production (one plant) 505,211 in categories of food, garden, paper, wood, textiles and industrial Toronto (Source: Environment Canada) are given for the study cities in Table 12. Two cement plants 2,755,000 Two points of clarification are required for Table 12. First, in Petro-Canada Lubricant Centre 430,370 3,185,370 the cases of Barcelona, Los Angeles and Prague, the waste fractions used are regional default values from the IPCC (2006b). Second,

Table 10 Incinerated and landﬁll waste.

Incinerated waste (t) Landﬁll waste (t) Source

Bangkok 0 4,562,500 BMA Dept. of Public Cleaning Barcelona 296,600 265,900 ICAEN (2008) Cape Town 0 2,626,320 City of Cape Town Denver 0 725,000 Denver Arapahoe Disposal Site Geneva 321,400 149,460 Rapport d’exploitation des Cheneviers London 0 10,320,000 Chartered Institute of Wastes Management (2002) Los Angeles 0 11,379,617 LA County, Public Works New York City 0 7,931,868 New York City (2007) Prague 201,200 54,000 Prague City Hall (2008) Toronto 65,700 4,091,500 Waste Diversion Ontario Author's personal copy ARTICLE IN PRESS

C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837 4835

Table 11 The second measure is similar to that currently reported by Weight fraction of DOC (degradable organic carbon) of particular waste streams most municipalities in that it includes all emissions from (Source: IPCC waste model spreadsheet, available at http://www.ipcc-nggip.iges.or. electricity use and waste disposal (see for example, ICLEI’s jp/public/2006gl/vol5.html). measure in Fig. 1). Only, our measure also includes the direct

Waste fraction Wi emissions from fuel loaded at airports and harbours, which result from the city’s participation in a global trading network. Food 0.15 The third measure has the same components as the second Garden 0.2 measure, but uses GHG emission factors for the lifecycle of fuels, Paper 0.4 Wood 0.43 rather than for direct emissions. Ideally, this measure would Textiles 0.24 include GHG emissions associated with energy embodied in food, Industrial 0.15 construction materials and other critical products imported by cities to meet basic human needs. Some input–output studies

have established the upstream CO2 emissions or indirect energy Table 12 associated with household consumption (Weber and Matthews, Components of landfill waste, used in calculation of degradable organic content. 2007; Cohen et al., 2005; Pachauri and Spreng, 2002; Reinders et al., 2003). Specifically at an urban level, Ramaswami et al. (2008) Waste fraction, f , of landfill waste i found that upstream emissions for food and cement contributed

Food Garden Paper Wood Textiles Industrial Other 1.7 Mt eCO2 to Denver’s GHG inventory, equivalent to 12% of the city’s emissions. The data for including similar contributions for Bangkok 0.44 – 0.10 0.05 0.05 0.30 0.06 the other nine global cities has not been established here. (A data Barcelona 0.37 – 0.17 0.11 – – 0.35 set for 8 US cities is currently being determined by Hillman and Cape Town 0.10 0.05 0.60 0.025 0.025 – 0.20 Denver 0.34 0.12 0.13 0.065 0.065 0.28 0.28 Ramaswami.) For eight of the ten global cities, however, it has Geneva – – – – – 0.07 0.93 been possible to determine the upstream emissions associated London 0.04 – 0.08 0.05 – 0.08 0.75 with fuels. There are still challenges though with selecting Los Angeles 0.34 – 0.23 0.06 0.04 0.10 0.21 appropriate life-cycle emissions factors, as these have been found New York City 0.22 0.05 0.10 0.14 – – 0.50 Prague 0.30 – 0.22 0.075 0.05 0.08 0.28 to vary significantly between studies (MacLean and Lave, 2003). Toronto 0.14 0.07 0.33 0.06 – 0.12 0.28 Selection of upstream emissions factors for fuels was largely based on review of two European and two North American datasets (Table 13). Appropriate life-cycle data for fuels could not there are significant differences between the waste fractions for be found for South Africa and Thailand, so Cape Town and various cities in part because each city has its own unique way of Bangkok were not included in this part of the analysis. The first of reporting waste outputs. In particular, different cities may or may the European datasets, MEET, was useful in that it included not report construction wastes (under ‘other’ in Table 12). Thus, separate upstream factors for individual nations (Lewis, 1997). differences in the reported waste fractions have to be interpreted Differences between factors for Spain, Switzerland and the UK with an understanding of what is included in the ‘whole.’ were found to be relatively small; only for jet kerosene did the

For four cities, Barcelona, Geneva, Prague and Toronto, emissions factors differ by more than 1 t eCO2/TJ. Thus, the use of emissions from waste incineration were included, based on local generalized European factors seems reasonable. The upstream studies in those cities. emissions reported by MEET were, however, consistently smaller than the European values in the Ecoinvent database (Frischknecht and Rebitzer, 2005). Possible explanations for the discrepancy 3. Overall measures of emissions may include: the age of the data, changes in crude oil supply mix, selection of heating values, and exclusion of infrastructure in the To accommodate the effects of spatial and life-cycle bound- life-cycle analysis. The Ecoinvent database was hence chosen for aries, three different measures of overall emissions are used in all four European cities. this study (Fig. 1). These are: The two North American datasets were generally in close agreement. Natural Resource’s Canada’s GHGenius (2004) database has emissions factors for both Canada and the United States, (i) physical emissions within the boundary of the city (WRI/ which only differ by more than 1 t eCO /TJ in the case of diesel. WBCSD Scope 1); 2 The US GHGenius upstream values for gasoline and diesel fuels (ii) single process emissions from end-use activities associated were also in good agreement with values in the US DOE’s GREET with the city’s metabolism; database; although there was a difference of 3 t eCO /TJ for LPG. (iii) extended life-cycle emissions (i.e., including upstream pro- 2 Neither GREET nor GHGenius had values for jet fuel, but the duction chain processes) associated with the city’s metabo- upstream processing of jet fuel is similar to that of diesel, so an lism. approximate life-cycle intensity of 92.5 t eCO2/TJ was established. Also, while GHGenius has upstream emissions factors for natural

The first measure is predominantly composed of emissions from gas, a slightly higher value of 67.9 t eCO2/TJ was used; this was fossil fuel combustion within the city boundaries. Emissions from based on Canadian specific processing data from GaBi4 (2002) and electricity production typically occur outside of the physical distribution losses reported by TransCanada pipelines. With these boundaries of most of the cities; where this occurs they are two exceptions above, GHGenius was used to determine Toronto’s excluded in measure (i). Only Barcelona, New York City and upstream emissions, while a combination of GREET and US Geneva have significant quantities of electricity generated within GHGenius data were used for Denver, Los Angeles and New York their boundaries (Geneva’s domestic electricity, however, is City. mostly hydraulic which has negligible emissions). Some cities Perhaps the most interesting finding from this comparison of also export their solid waste outside of their boundaries. So where life-cycle databases was that North American upstream emission this occurs, emissions from landfill methane are not included in factors are typically 5–10 t eCO2/TJ higher than those in the this measure. European databases. Possible explanations for the lower European Author's personal copy ARTICLE IN PRESS

4836 C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837

Table 13

Direct and lifecycle GHG emissions intensity for fuels (t eCO2/TJ).

Energy content (TJ/ Direct emissions (t eCO2/TJ) Lifecycle emissions (t eCO2/TJ) ML) IPCC1 Canada USA USA Europe Spain, Switz. UK (GHGenius) (GHGenius) (GREET) (Ecoinvent) (MEET)

Gasoline (low 34.8 72.2 94.9 94.6 92.8 85.9 81.6–81.9 sulphur) Diesel 37.8 75.2 92.3 93.8 94.0 85.4 82.4–82.8 Jet kerosene 35.1 72.0 Not given Not given 92.52 81.8 76.4–79.1 LPG (petroleum 26.8 66.1 81.0 80.8 76.4 76.43 71.9–72.7 based) Marine fuel Not used 78.9 94.1 94.4 Not given Not given Not given Natural gas (dry) N/A 56.1 67.94 63.8 Not given 61.6 Not given Fuel oil Varies 77.83 91.6 91.5 Not given 88.1 83.3–83.7

Coal: N/A Anthracite 98.09 100.8 100.8 Not given 101.6 Not given Coking/ 94.43 107.1 107.0 Not given 1075 Not given Bituminous

Values in bold are used in the study.

Note 1: Includes CO2,N2O and CH4 emissions

Note 2: North American emission factor for Jet Kerosene of 92.5 t eCO2/TJ was determined by assuming the same upstream contribution as diesel fuel. Note 3: Lifecycle emission factor for LPG in Europe is an estimate based on differences between Ecoinvent and MEET data for other fuels. Note 4: Canadian lifecycle emission factor for natural gas was determined from processing data from GaBi4 and distribution losses reported by TransCanada pipelines. Note 5: European emissions factor for coking coal is based on GHGenuis (USA). factors include differences in feedstocks, shorter distribution Acknowledgements distances, and, where relevant, lower GHG intensities for electricity production. Further study into the differences in Julia Steinberger was supported by the European Community’s upstream emission factors between Europe and North America – Seventh Framework Programme FP7/2007-2013 under grant and possibly other continents – is warranted. For the purpose of agreement 212034. The GHG accounting for Denver was sup- this study, it means that even if levels of consumption were equal, ported by grants from the US Department of Education GAANN larger GHG footprints are expected for Denver, Los Angeles, New Program and the City & County of Denver. York City and Toronto, than Barcelona, Geneva, London and The assistance of Julius Mattai of the Greater London Authority, Prague. Jonathan Dickinson of the City of New York and two anonymous reviewers is greatly appreciated.

4. Conclusions References

This paper has made several contributions to the quantification Baldasano, J.M., Soriano, C., Boada, L., 1999. Emission inventory for greenhouse of energy use and GHG emissions for cities. Primarily, we have gases in the City of Barcelona, 1987–1996. Atmospheric Environment 33, 3765– developed a consistent methodology for determining the GHG 3775. Bemis, 2006. Inventory of California Greenhouse Gas Emissions and Sinks: 1990– emissions attributable to cities. The method provides clarity on 2004, California Energy Commission. important details, such as explicitly identifying electricity line BMA, 2008. Bangkok Metropolitan Administration Data Centre-/www.bangkok. S losses, inclusion of emissions from aviation and marine transport, go.th/info/ . Carney, S., Green, N., Wood, R., Read, R., 2009. Greenhouse Gas Emissions and consistent determination of waste emissions. Moreover, cities Inventories for Eighteen European Regions, EU CO2 80/50 Project Stage 1: are often inconsistent in the means by which they determine the Inventory Formation. The Greenhouse Gas Regional Inventory Protocol (GRIP). quantities of fuels used for ground transportation. Our analysis Chartered Institute of Wastes Management, 2002. A Resource Flow and Ecological Footprint Analysis of Greater London, Best Foot Forward, London. has shown, however, that differences between three approaches – City of Cape Town, 2003, State of Energy Report for Cape Town, /www.capetown.gov. using fuel sales data, scaling from wider regions, and estimating za/en/EnvironmentalResourceManagement/publications/Documents/State_of_ from VKT – can be less than 5%. Thus, this methodology allows Energy_Report_2003.pdfS. Cohen, C., Lenzen, M., Schaeffer, R., 2005. Energy requirements of households in comparable emissions to be determined for a wide variety of Brazil. Energy Policy 33, 555–562. cities. CUEC, 2008. Charles University Environment Center, project 2B06183, Socio- In our methodology, we encourage the determination of full economic metabolism of urban systems and its impact on ecosystems, ongoing project. life-cycle emissions attributable to cities. Life-cycle emissions are Decker, H., Elliott, S., Smith, F.A., Blake, D.R., Sherwood Rowland, F., 2000. Energy a more comprehensive measure of a city’s contribution to global and material flow through the urban ecosystem. Annual Review of Energy and climate change than the ‘within city’ emissions and direct the Environment 25, 685–740. emissions measures, also discussed here. In focussing on the Diot, M., Bogner, J., Chanton, J., Guerbois, M., Hebe, I., Moreau le Golvan, Y., Spokas, K., Tregoures, A., 2001. LFG mass balance: a key to optimize LFG recovery, in life-cycle emissions for fuels, we found that the upstream Proceedings of the Eighth International Waste Management and Landfill emission intensity factors for North American cities are typically Symposium Sardinia 2001, S. Margherita di Pula (Cagliari, Italia), October 1–5. higher than those for European cities. Emmenegger, M.F., Frischknecht, R., Cornaglia, L., Rubli, S., 2003. Metabolisme des activites economiques du Canton de Geneve –Phase 1. Rapport final. In this work, we have also established a substantial new data ERM-Siam, 2005. Study on Electricity Sector Baselines in Thailand. Final Report set for energy use and waste disposal in ten cities over four Submitted to ECON Centre for Economic Analysis. ERM-Siam, Co. Ltd., Bangkok, continents. We hope that this contributes towards continuing Thailand. ERO, 2008, Energy regulatory office (Czech) /www.eru.czS. efforts to establish consistent and reliable statistics on the Fott, P., Pretel, J., Vacha, D., Blaha, J., Neuzˇil, V., Havranek, M., 2008. National GHG metabolism of global cities. Inventory Report of the Czech Republic (NIR), last year 2005. CHMI, Prague. Author's personal copy ARTICLE IN PRESS

C. Kennedy et al. / Energy Policy 38 (2010) 4828–4837 4837

Frischknecht, R., Rebitzer, G., 2005. The ecoinvent database system: a comprehen- Pachauri, S., Spreng, D., 2002. Direct and indirect energy requirements of sive web-based LCA database. Journal of Cleaner Production 13, 1337–1343. households in India. Energy Policy 30, 511–523. GaBi 4, 2002. Developed by IKP, Institute for Polymer Testing and Polymer Science. Phdungsilp, A., 2006. Energy Analysis for Sustainable Megacities, Licentiate of University of Stuttgart, Germany. Engineering Thesis, Department of Energy Technology, Royal Institute of Gasson, B., 2002. The ecological footprint of Cape Town: Unsustainable resource Technology, Sweden. use and planning implications, SAPI International conference: Planning Africa, Prague City Hall, 2008, Yearbook Prague – the Environment 2007 /http://envis. Durban South Africa. praha-mesto.cz/(mxvnumegyysr2fnbrb0xzo45)/rocenky/Pr07an_htm/index_ GHGenius Version 3.12, 2004. (S&T)2 Consultants, Natural Resources Canada. en.htmS. GLA, 2007. Action Today to Protect Tomorrow: The Major’s Climate Change Action Ramaswami, A., Hillman, T., Janson, B., Reiner, M., Thomas, G., 2008. A demand- Plan. Greater London Authority. centered, hybrid life cycle methodology for city-scale greenhouse gas ICAEN, 2008. Institut Catala d’Energia de la Generalitat de Catalunya (Energy emissions. Environmental Science and Technology 42, 6455–6461. Institute of Catalonia). Reinders, A.H.M.E., Vringer, K., Blok, K., 2003. The direct and indirect energy ICLEI, 2008. Cities for Climate Protection, Toronto, Canada: International Council of requirement of households in the European Union. Energy Policy 31 (2), 139– Local Environmental Initiatives. [Online]. Available HTTP: /http://www.iclei. 153. org/index.php?id=800S (Accessed July 22, 2008). Sahely, H.R., Dudding, S., Kennedy, C.A., 2003. Estimating the urban metabolism of IEA, 2008. Electricity Information 2007. International Energy Agency, OECD/IEA, Paris. Canadian cities: GTA case study. Canadian Journal for Civil Engineering 30, IPCC, 1996. Revised 1996 IPCC Guidelines for National Greenhouse Gas Inventories, 468–483. Vol. 1–3. Sassen, S., 1991. The Global City. Princeton University Press, NJ. IPCC, 2006a. Guidelines for National Greenhouse Gas Inventories, Volume 2 Energy. Satterthwaite, D., 2008. Cities’ contribution to global warming: notes on the IPCC, 2006b. Guidelines for National Greenhouse Gas Inventories, Volume 5 Waste. allocation of greenhouse gas emissions. Environment and Urbanization 20 (2), IPCC, 2007, Fourth Assessment Report /http://www.ipcc.ch/S. 539–549. Kennedy, C.A., Cuddihy, J., Engel Yan, J., 2007. The changing metabolism of cities. Scharff, H., Martha, A., van Rijn, D., Hensen, A., v.d. Bulk, W., Flechard, C., Oonk, H., Journal of Industrial Ecology 11, 43–59. Vroon, R., de Visscher, A., Boeckx, P., 2003. A comparison of measurement Kennedy, C.A., Mohareb, E., 2009. Greenhouse Gas Emissions from Cities, Chapter 7 methods to determine landfill methane emissions. Afvalzorg Deponie B.V., in Cities in Evolution: Urbanization, Environmental Change and Sustainability, Haarlem, the Netherlands. Xuemei, Bai., Thomas, Graedel., Akio, Morishima., (Eds.), Cambridge University Spokas, K., Bogner, J., Chanton, J., Morcet, M., Aran, C., Graff, C., Moreau-le-Golvan, Press, in press. Y., Bureau, N., Hebe, I., 2006. Methane mass balance at three landfill sites: what Kennedy, C., Steinberger, J., Gasson, B., Hillman, T., Havranek, M., Pataki, D., is the efficiency of capture by gas collection systems?. Waste Management 26 Phdungsilp, A., Ramaswami, A., Villalba Mendez, G., 2009. Greenhouse Gas (5), 516–525. Emissions from Global Cities, accepted to Environmental Science and Tanatvanit, S., Limmeechokchai, B., Chungpaibulpatana, S., 2003. Sustainable Technology, doi:10.1021/es900213p. energy development strategies: implication of energy demand management Lewis, C.A., 1997. Fuel and Energy Production Factors, MEET Project (Methodol- and renewable energy in Thailand. Renewable and Sustainable Energy Reviews ogies for Estimating Air Pollutant Emissions from Transport) funded by the 7 (5), 367–395. European Commission under the Transport RTD programme of the 4th Taylor, P.J., 2004. World City Network: A Global Urban Analysis. Routledge, New framework. York. MacLean, H.L., Lave, L.B., 2003. Evaluating automobile fuel/propulsion system Weber, C.L., Matthews, H.S., 2007. Quantifying the global and distributional aspects technologies. Progress in Energy and Combustion Science 29, 1–69. of American household carbon footprint. Ecological Economics (pp. 1–13), 1–

Mayor of London, 2006. London Energy and CO2 Emissions Inventory 2003: 13, doi:10.1016/j.ecolecon.2007.09.021. Methodology Manual. Willumsen, H., 2003. Landfill gas plants: Number and types worldwide, New York City, 2007. Inventory of New York City Greenhouse Gas Emissions, Proceedings Sardinia.03, Waste Management and Landfill Symposium, CISA, Mayor’s Office of Operations, Office of Long-term Planning and Sustainability. University of Cagliari, Sardinia, Italy. Ngo, N.S., Pataki, D.E., 2008. The energy and mass balance of Los Angeles County. WRI/WBCSD (World Resources Institute and World Business Council for Sustain- Urban Ecosystems 11, 121–139. able Development) The Green house Gas Protocol: A Corporate Accounting and Oonk H., Boom T., 1995. Landfill gas formation, recovery and emission, TNO-report Reporting Standard: Revised Edition. Accessed May 2009: /http://www. 95-203, TNO, Apeldoorn, the Netherlands. ghgprotocol.org/S.