Worldwide Greenhouse Gas Reduction Potentials in Transportation by 2050

Reductions in the greenhouse gas (GHG) intensity of passenger and freight transportation are possible through adoption of fuel‐saving technologies, demand switching between modes, and large‐scale electrification of fleets, in addition to other actions. In this study, future scenarios to 2030 and 2050 are the basis for assessment of GHG reduction potentials for major passenger and freight modes (automobiles, buses, trains, aircraft, and oceangoing vessels) across eight regions of the world. New fuel‐saving technologies can significantly reduce the life‐cycle GHG footprint of both passenger and freight vehicles, but not uniformly worldwide. Countries outside of the Organization for Economic Cooperation and Development (OECD) lag behind OECD countries in GHG reduction potentials for all modes but oceangoing vessels owing to a combination of slower adoption of fuel‐saving technologies and a slower decarbonization of electricity generation and other processes. The reduction of GHG intensity will occur more slowly for freight modes than for passenger modes. However, improved fuel efficiency has negative feedbacks to the effectiveness of mode‐switching and alternative fuel adoption policies through 2050 because improvements in the fuel efficiency of vehicles alone may cause the marginal benefits of GHG abatement policies to diminish over time. This trend may be reversed if alternative fuel pathways decarbonize at faster rates than conventional transportation fuels. The largest opportunities for GHG reductions occur in non‐OECD countries. Given the many factors that distinguish transportation systems between developed and developing nations (e.g., availability of new technologies, the financial ability to acquire them, and policies to incentivize their adoption), many benefits could be gained through interregional cooperation.


Introduction
Reducing greenhouse gas (GHG) emissions from the global transportation sector is a critical step needed to slow the worsening effects of climate change. In 2010, passenger and freight vehicles emitted 7.0 gigatonnes of carbon dioxide equivalents (CO 2 -eq) of GHGs into the atmosphere or 23% of the world's anthropogenic GHGs (IPCC 2014). The relative contributions of transportation-related GHGs between countries depend upon many regional factors (population size, [An acknowledgment was added after initial online publication to correctly acknowledge important assistance with this work.] Around the world, reductions in the GHG emissions intensity of moving people and goods are driven by both technological advancements as well as targeted governmental policies. Vehicle fuel efficiency which influences the majority share of vehicle life-cycle GHG emissions (Facanha and Horvath 2007;Winebrake et al. 2007;Horvath 2009, 2012;Hawkins et al. 2013), depends heavily on both vehicle design and key powertrain characteristics. Across passenger and freight modes, lightweighting and downsizing, thermodynamic cycle improvements, hybridization, and aerodynamic improvements (ICAO 2010;IEA 2012a;ICCT 2013;NRC 2014) represent proven technological strategies for reducing fuel usage and thereby the GHG intensity of vehicles. However, the adoption of fuel-saving technologies within vehicle fleets depends on many regional factors, including income, geography, climate, and culture. To spur implementation, the world's largest GHG emitters have set incremental fuel economy and CO 2 emission standards for new light-duty passenger vehicles (ICCT 2014a), heavy-duty vehicles (ICCT 2012a), and ocean-going vessels (OGVs) (IMO 2013), modes that represent 85% of all transportation-related GHG emissions (ICCT 2012b). Private industries have also set fuel economy improvement goals for both rail (UIC 2012) and aircraft (ICAO 2012). However, because of differences in fleet characteristics (e.g., vehicle age and stock turnover rates), the benefits accrued from these policies, which target new vehicles, will likely occur at different rates around the world.
Whereas fuel efficiency improvements are critical steps toward curbing the GHG intensity of personal mobility and goods movement, there are other pathways to GHG reductions in the transport sector. Importantly, studies suggest that mode-switching and alternative fuel (e.g., electricity and biofuels) adoption strategies offer many potential climate benefits (Williams et al. 2012;Chester and Horvath 2012;Scown et al. 2012Scown et al. , 2013Craig et al. 2013;IPCC 2014;Nahlik et al. 2014), though the success of these policies is often location dependent. Structural policies require governmental intervention owing to economic, technical, and preferential barriers resisting their mass adoption (Chester et al. 2014). Nevertheless, for profound changes, many GHG reduction scenarios require mode-switching and alternative fuel adoption strategies in order to meet future climate mitigation goals (IEA 2009;Yang et al. 2009;Hill et al. 2012).
All told, previous research has advanced our understanding of the climate impact of transportation systems on a global scale while offering possible pathways for reducing the sector's net GHG emissions. However, relying on existing comparative analyses to gauge the potential effectiveness of climate-change mitigation policies as they relate to current and future transportation systems is still challenging, given that differences in regionally and temporally specific model assumptions, emissions inventory data, and system boundaries induce many uncertainties (Hertwich et al. 2015;Reyna et al. 2015). Overall, there is limited information on how the unit GHG benefits of climateoriented transportation policies will change over time as vehicle modes and energy systems decarbonize. Addressing these chal-lenges is important, especially at a regional level (Cicas et al. 2007;, because it is unlikely that there exists a single, long-term strategy for reducing GHG emissions within the transportation sector (e.g., one policy best for all locations). Moreover, given the differences in regionalized characterization factors, such as vehicle fuel efficiency, load factors, fuel pathways, and typical trip distances, it is also unlikely that the opportunities to decarbonize the transportation sector are equivalent across global regions over time.
This article is one in a series of technology assessments initiated through the International Resource Panel (IRP) of the United Nations Environmental Program (UNEP). The collaboration seeks to quantify the environmental and natural resource benefits and trade-offs associated with wide-scale adoption of advanced energy technologies, with each assessment sharing consistent methods, system boundaries, and background lifecycle inventory (LCI) data. The goal of this study is to evaluate how an extensive, but feasible, adoption of low-carbon energy and efficient demand-side technologies would affect the incremental GHG benefits (e.g., on functional unit and trip basis) of modal shift and vehicle electrification within and across OECD countries (i.e., mostly wealthy economies) and non-OECD countries (mostly developing economies). Given a wide variety of transportation modes across the globe, we limit the scope of this assessment to the world's most common vehicle types: automobiles, buses, trains, aircraft, and OGVs. We also set our system boundaries to include only globally prominent transportation fuels (e.g., gasoline, diesel, bunker fuel, and jet fuel) as a basis for comparison. Though other alternative fuelssuch as biofuels, biogas, and natural gas-are also used in some countries, they are supplied at relatively low volumes and/or to niche applications.
We provide estimates for the life-cycle cradle-to-grave GHG footprint of both passenger and freight transportation following a prominent global energy and material production scenario established by the International Energy Agency (IEA) (IEA 2010) for reference years 2010, 2030, and 2050. Results show how the incremental GHG benefits of modal shift and vehicle electrification policies vary across all global regions through 2050 and where the largest opportunities for GHG reductions occur. We assess the sensitivity of GHG abatement policies within the transportation sector to key assumptions regarding the levels of services these modes provide (i.e., ridership and total payload), the evolution of electricity generation mixes through time and space, as well as offer a discussion on the relevance and uncertainty of the results.

Methods
The goal of this study is to assess the GHG reduction potentials associated with mode-switching and vehicle electrification policies across the world in 2030 and 2050. The scenarios that are presented in this study have been guided by previous global assessments of mobility and goods movement (IEA 2009;IPCC 2014) as well as literature focusing on transportation systems in specific regions of the world. The scope of the study includes the production, distribution, storage, and in-vehicle use of transportation fuels (i.e., well-to-wheel [W2W] processes), vehicle, vessel, and aircraft manufacturing and maintenance, including battery manufacturing for battery-electric (BEVs) and plug-in hybrid electric vehicles (PHEVs) (Majeau-Bettez et al. 2011;Hawkins et al. 2013), as well as end-of-life (EOL) processes. A complete mapping of the system boundaries is provided in the Supporting Information available on the Journal's website. Emissions associated with the construction, maintenance, and operation of supporting infrastructure are not considered because of global data unavailability, though we recognize that these emissions are important for certain transportation modes Horvath 2009, 2010).
The scenarios show the life-cycle GHG emissions associated with major passenger and freight modes for eight regions: OECD Europe (EU), OECD Pacific (PAC), OECD North America (NA), China (CN), India (IN), Latin America (LA), Africa and the Middle East (AME), and economies in transition (EIT), for example, Eastern European countries and Russia. Because of insufficient information on emerging vehicle fleets, developing countries in South Asia (e.g., Thailand and Indonesia) were excluded from the assessment. The regions reflect the system boundaries established across the collaborative of UNEP IRP studies on the environmental impacts and resource utilization of low-carbon energy and efficient demand-side technologies (Hertwich et al. 2015;Gibon et al. 2014).
In each region, GHG emission factors were reported based on an optimistic, yet attainable, evolution of electricity and material production technologies according to the BLUE Map energy scenario from the IEA Energy Technology Perspectives (ETP) (IEA 2010). The IEA's 2012 ETP was available before publication, but appears similar to the earlier version (IEA 2012b). BLUE Map employs target-oriented policies that aim to halve global GHG emissions from a 2005 baseline by 2050 through the implementation of currently available technologies (e.g., renewable energy, more efficient technologies, and carbon capture and storage). Results for the IEA's baseline scenarios, which assume business-as-usual activities, are also provided in the Supporting Information on the Web. In each scenario year (2010, 2030, and 2050), GHG emissions associated with the production, distribution, and storage of transportation fuels, vehicle manufacturing and maintenance, as well as EOL processes were estimated using the THEMIS model (Technology Hybridized Environmental-economic Model with Integrated Scenarios) (Hertwich et al. [2015] and Supporting Information on the Web). It is a regionalized model for electricity generation and materials production that integrates commercial and academic life-cycle assessment (LCA) software (ecoinvent and EXIOPOL) with original data from life-cycle emissions and resource consumption inventories of energy technologies in the future (ecoinvent Center 2010; Tukker et al. 2013;Hertwich et al. 2015;Gibon et al. 2014). The THEMIS model accounts for projected advances in efficiencies in low-carbon electricity technologies, such as photovoltaic power systems, as well as energy efficiency improvements in key industrial sectors, such as raw material production and chemical manufacturing, using the IEA's underlying energy scenario data and assumptions (IEA 2010).
There are advantages of building upon previous global scenarios using a consistent methodology, system boundary, and supporting inventory data. Although there are many benefits to utilizing life cycle tools that are tailored to specific regions of the world (Cicas et al. 2007; ANL 2012; EcoTransIT 2011), these tools do not consistently account for all transportation modes or economy-wide technological changes (e.g., electricity generation, metal production, and chemical production) that influence the GHG intensity of vehicles. By estimating the environmental performance of passenger and freight transportation through a consistent approach dedicated to global assessments, our study makes possible comparisons between transportation modes within and across different world regions from 2010 through 2030 and 2050. Table 1 provides a list and brief description of the representative vehicles analyzed in this study. We compare four passenger transportation modes (automobile, bus, rail, and airplane) and three freight modes (truck, rail, and OGV) that represent the bulk of the world's vehicle kilometers traveled (VKT) and total freight turnover (World Bank 2014). Data suggest that there is more variability in demand for different fuel types within light-duty passenger vehicles globally (gasoline, diesel, electricity, biofuels, natural gas, and so on) than for any of the other modes considered (IEA 2009). We analyze only electricity as an alternative to conventional petroleum-based fuels because its supporting infrastructure is more broadly available and more is known about the evolution of its carbon intensity through 2050 (IEA 2010). Other alternative transportation fuels, such as biofuels (Scown et al. 2012(Scown et al. , 2013, could also be potential substitutes for high-carbon fuels in some areas of the world, but are not considered herein owing to uncertainties regarding system scalability (e.g., material sourcing and infrastructure expansion) and adoption (McKone et al. 2011;Strogen et al. 2012).
The results of the LCI for each vehicle are normalized by either passenger-kilometer (pkm) or metric ton-kilometer (tkm) to obtain the life cycle GHG emission factor (equation 1): where, e f : life-cycle emission factor (personal travel: grams [g] CO 2 -eq/pkm; freight: g CO 2-eq / tkm), I : total number of vehicle life-cycle phases i considered, T i : total GHG emissions for each life-cycle phase i (g CO 2-eq : weighted CO 2 , methane, nitrous oxide emissions), VKT: lifetime vehicle kilometers traveled (km), E i : emissions rate for each life phase i (g CO 2 -eq/km), A: activity or level of service (personal travel: passengers; freight: metric tons, t) The process of normalizing GHG emissions from the vehicle manufacturing, maintenance, and EOL stages slightly differs from W2W emissions because of the way in which emissions are reported. The former are estimated based on a lump-sum Note: 1 The terms "ton," "tons," and "ton-km" found in this document refer to the metric ton, which is equivalent to 1,000 kg.
value (e.g., the manufacturing of one vehicle results in X g CO 2 -eq), whereas the latter is estimated over a continuum (e.g., g CO 2 -eq/km). These differences can have an impact on the formation of the life cycle emission factor (Taptich and Horvath 2014) and are important for understanding the relative role each life cycle phase has on influencing the vehicle's total GHG footprint. These factors are discussed in the following sections. For a summary of model inputs used for each of the modes, see the Supporting Information on the Web. Fuel consumption estimates (g fuel/km) for each mode were derived based on current and projected fuel economy standards (IEA 2012a; ICCT 2012a; US EIA 2014; ICCT 2014a), fleet-specific inventory data (OAG 2008;KPMG 2011;Lissys Ltd 2010;Clarkson Research 2014), and industry-specific fuel consumption forecasts (ICAO 2009;ICAO 2012;UIC 2012) (table 1). Overall, information detailing the fuel efficiency of specific vehicles was more readily available for the OECD regions (Europe, North America, and the Pacific Rim) than for the non-OECD countries and regions. In cases where little data were available, we assumed fuel economies based on regional fleet compositions, known fleet turnover rates, as well as information provided in the literature. For instance, the world's high-speed rail (HSR) infrastructure currently only exists in three study regions (OECD Europe, OECD Pacific, and China), but plans are in place to expand this network to other regions by 2030 (IEA 2009). Information regarding HSR fuel consumption is available over a range of values for older systems, and we rely on fuel consumption estimates from Chester and Horvath (2012) for new HSR technologies to represent theoretical fleets (20 kilowatt-hours [kWh]/VKT) because these technologies would reasonably be adopted upon initial implementation. A full list of fuel consumption estimates used in this study and their respective references are provided in the Supporting Information on the Web. Table 2 provides an overview of future trends within each mode considered.
Methods for estimating operational GHG emissions vary by mode. For on-road vehicles and rail, operational GHG emission rates, E op (g CO 2 -eq/km), are calculated from fuel consumption rates, which are already reported over distances traveled. However, the standard for aircraft and OGVs is to report total fuel usage per trip, which is a function of time spent at varying operational modes and loading conditions (ICCT 2013(ICCT , 2014b. For aircraft, we estimate E op for intraregional trips using a linear regression model of fuel consumption simulations of an Airbus 320, a commonly used, average-size aircraft under varying loadings and trip distances (Lissys Ltd 2010) and using historical flight data (OAG 2008) such that (equation 2): where β 0 : fuel usage associated with ground-based operations (idling, takeoff, and landing) (g of fuel), β 1 : fuel usage rate associated with flight operations (climbing and cruising) (g fuel/km), d: distance traveled per trip (km), γ : carbon dioxide emissions to fuel mass ratio (g CO 2 -eq/g fuel).
For OGVs (containerships and crude tankers), fuel consumption rates are reported in terms of engine braking power (g fuel/kWh) for both primary and auxiliary vessel engines. We rely upon vessel composition data obtained from the World Fleet Register (Clarkson Research 2014) and vessel-specific engine loading estimates from the literature (ICCT 2013) to Rail For diesel-powered rail, we assume a 10% improvement in fuel efficiency (l/100 km) or 0.5% per year by 2030 and 20% improvement by 2050. We assume that electric rail, which is currently only available in select regions, achieves a 25% efficiency improvement (kWh/km) by 2030 and an additional 10% improvement (0.5% per year) by 2050.
Chester and Horvath (2009) calculate E op (g CO 2-eq /km) for fleet-average vessels such that (equation 3): where sfc: specific fuel consumption of the primary engine (g fuel/kWh), P : power capacity of the primary engine (kW), L : loading of the primary engine (%), sfc aux : specific fuel consumption of the auxiliary engine (g fuel/kWh), P aux : power capacity of the auxiliary engine (kilowatts; kW), L aux : loading of the auxiliary engine (%), V : average speed of the vessel (km per hour [km/hr]). GHG reduction potentials associated with new technology adoption, mode-switching, and electrification polices are assessed based on both a unit as well as a trip basis. These metrics provide two major points of insight with regard to identifying opportunities to reduce GHG emissions within fleets. The former allows us to compare the environmental efficiency within and across transportation modes on common terms. This is important for monitoring the performance of these modes from an environmental viewpoint and can offer guidance on how to improve underproductive vehicles. The latter measures the marginal life-cycle GHG impacts scaled by a unit demand. Thus, we gain an understanding of the magnitude of GHG reduction policies as they will occur on a unit basis within fleets. This is important when considering modes such as rail and OGVs, which, as we will show, are very efficient but have large associated trip distances (OAG 2008; World Bank 2011).

Results
The share of GHG emissions from W2W life-cycle stages relative to a vehicle's total life-cycle footprint is a function of vehicle fuel efficiency and lifetime levels of service (i.e., Figure 1 Results from year 2010, 2030, and 2050 greenhouse gas emission scenarios for major passenger and freight modes in each global region. Values adjacent to each circle represent the average carbon intensity of the mode (g CO 2 -eq per functional unit) in a respective policy year. Percent reduction from the 2010 regional baseline is reported. Within each mode class, we highlight the lowest carbon intensity region in 2050 with a darker background. g CO 2 -eq = grams carbon dioxide equivalents. ridership or freight turnover). Results from the 2010 scenario show that high-capacity, long-distance modes have significant shares of W2W emissions (93% to 100%). These modes include heavy heavy-duty (HHD) trucks, rail, aircraft and OGVs. This finding has previously been reported across the literature for LCAs of vehicles (Winebrake et al. 2007;Facanha and Horvath 2007;Horvath 2009, 2012) and provides direction for developing policies that aim to reduce GHG emissions from these modes. For instance, focusing on policies that only improve the W2W GHG intensity (e.g., fuel-saving technologies or low-carbon fuels) of light-duty vehicles ignores a significant portion (17% to 33%) of their full GHG impact (Majeau-Bettez et al. 2011;Hawkins et al. 2013). For modes such as buses, heavy-duty trucks, rail, and OGV, lowering the GHG intensity of vehicle manufacturing, maintenance, and EOL stages will result in small savings over the lifetime of these vehicles. In summary, the sensitivity of life cycle GHG emission factors to the adoption of new fuel-saving technologies or alternative fuels varies across modes. Small improvements to the W2W GHG intensity of high-capacity, long-distance modes will have greater marginal improvements to emission factors. Figure 1 summarizes our findings for autos, buses, heavy-duty trucks, trains, and OGVs for the 2030 and 2050 scenario years. Overall, non-OECD countries lag behind OECD countries in GHG reduction for all modes, excluding OGVs, owing to a combination of slower adoption of fuel-saving technologies and a slower decarbonization of electricity generation and other industrial processes. The largest differences between these two economic classifications can be observed for vehicles powered entirely or in part by electricity (BEV, PHEV, and HSR). For instance, the life-cycle GHG footprint of a BEV operated in India could be over 2.5 times larger than the footprint of the same vehicle operating in North America as a result of higher GHG emissions from electricity generation (using mostly coal).
In fact, we find that there would be no difference between BEV and gasoline-powered cars in India for the 2030 scenario year. This highlights an important insight derived from this study: there is no one strategy for all modes and all regions. Each region has its own unique levels of ridership, freight turnover, and other technological characteristics that influence its respective GHG footprint.
The results also show that reduction of the GHG intensity of freight modes occurs more slowly than of passenger modes around the world by 2050. The reason is twofold. First, for the most part, fleet turnover rates occur more slowly within freight modes than within passenger modes. The longer vehicle lifetimes (e.g., 30 years for rail, aircraft, OGV, and HHD truck) (IEA 2009) and infrastructure design choices (e.g., nonelectrified rail) lock in the technologies available for long periods of time. Hence, if there are large improvements to new vehicle stock relative to older generations of vehicles, these savings may not be realized until years later as fleets gradually adopt these technologies. Second, freight vehicles face performance conditions that are unlike those for most passenger modes. Higher levels of service (e.g., tonnage) come at a cost of higher required power rating for engines and therefore fuel usage. Reducing fuel usage given these constraints is challenging even for an economic sector with incentives to reduce fuel costs. Meanwhile, passenger modes (e.g., cars) can feasibly roll back the power delivered by engines or the mass of a vehicle, reducing total fuel usage, without affecting levels of services. For the same reasons stated, passenger vehicles have a greater variety of alternative fuels available, which may reduce the GHG footprint of these modes. Because of data availability and large supply potential, we only consider electricity as an alternative fuel in this study, but other lower-GHG fuels should also be considered in future work, such as biofuels, hydrogen, and natural gas.
The unit GHG reduction potentials associated with modeswitching and vehicle electrification policies across all eight regions through 2050 is evaluated in figure 2. The results of the 2010, 2030, and 2050 global scenarios show both positive (net savings) and negative (net gains) reduction potentials. This variability in benefits again reinforces the importance of regional considerations when comparing the effectiveness of GHG abatement policies on a global scale. Though the efficiency of these policies varies between regions, the overall trend is that GHG reductions converge on a similar range of numbers over time, and the range of unit GHG reduction potentials is smaller in 2050 than it is in 2010 and 2030. In terms of meeting future climate mitigation goals (IEA 2009;Yang et al. 2009;Hill et al. 2012), this finding is both positive and negative. We see that the effectiveness of some policies to reduce GHGs increases over time and, in some cases, switches from negative to positive reduction potentials, whereas the benefits of other policies diminish over time. Overall, the long-term effectiveness of mode-switching and vehicle electrification policies depends on the policy type and region of interest.
The results of the trip-based assessment (kg CO 2 -eq saved per passenger trip or metric ton trip) also indicate that mode-switching policies will be the most effective in the short term, whereas vehicle electrification policies achieve the largest savings by 2050. The differences between these trends can be attributed to differences in improvements in fuel efficiency between vehicles and the rate at which well-to-pump processes decarbonize for petroleum-based fuels relative to electricity generation. Vehicle electrification coupled with both fuel efficiency improvements and large reductions in the carbon intensity of electricity generation under the IEA BLUE Map energy scenarios result in a net improvement over time. It is important to note, however, that this scenario does not consider the role of advanced biofuels as a liquid fuel additive or substutitue (Scown et al. 2013), which could lower the incremental GHG benefits of eletrification policies. In contrast, mode-switching policies for vehicles powered by petroleum-based fuels do not receive the added benefit of well-to-pump decarbonization. Thus, as the fuel efficiency for each vehicle within a respective scenario improves over time, the difference between modes in terms of GHG intensity diminishes. These findings strengthen the position that environmental assessments of transportation modes should extend beyond characterizations of tailpipe emissions alone. Here, the key factors influencing the climate benefits of these policies occur upstream of the vehicle's operation. Figure 2 also shows that the largest unit savings per policy occur in non-OECD countries. Under the same reasoning from the preceeding policy assessments, larger benefits occur in these regions as a result of their currently higher W2W GHG intensities (i.e., large emissions factors tend to have larger relative GHG reduction potentials). Because GHGs are global pollutants, we could achieve the greatest GHG reductions if the more affluent OECD countries facilitated mode switching and alternative fuel adoption polices in less affluent countries. For example, the GHG savings resulting from switching freight deliveries by HHD truck to freight rail in OECD Europe is nearly 6 times less effective than the same policiy in EIT on a unit-trip basis.

Sensitivity of Policies to Levels of Service and Electricity Mixes
In each of the scenarios presented, we assume that levels of service remain constant through 2050. This assumption introduces varying levels of bias into the reporting of GHG emission factors (see Taptich and Horvath [2014] for more), the largest of which occurs within smaller modes. We provide a range of values for each vehicle-region pair in the Supporting Information on the Web. However, we assess the sensitivity of our conclusions for each of the six policies analyzed under different assumptions of average ridership and freight turnover. Figure 3 shows a comparison of two policy scenarios (Auto to Bus, TOP; Diesel Rail to HSR, BOTTOM) under varying levels of feasible load factors for each region. Assumptions of levels of services for each vehicle map onto a region where the alternative vehicle is preferred (gray region). For highcapacity, long-distance modes, small variability (±20%) in the levels of ridership or freight turnover does not affect the recommendation to choose the alternative vehicle given that Figure 2 Results of a unit assessment of GHG reduction potentials for six abatement policies on a trip (average kg CO 2 -eq saved per pass-trip or metric ton-trip) basis show both diminishing and increasing benefits by 2050 under the BLUE Map energy scenario. We highlight the regional policies with the greatest 2050 GHG reduction potentials in gold; each occurs in non-OECD regions. GHG = greenhouse gas; kg CO 2 -eq = kilograms carbon dioxide equivalents; OECD = Organization for Economic Cooperation and Development. d e f d a is small at larger values of A (equation 1). The only exemptions to this finding are passenger rail (diesel) to HSR cases (in China, India, and Africa and the Middle East), where the carbon intensity of electricity is so high that it requires large average ridership to make HSR feasible as an alternative. In contrast, the conclusions stemming from policies that involve cars whose ridership ranges from one to five passengers are highly sensitive to this model parameter. This is owing to the nonlinear relationship between the life-cycle emission factor and the normalizing activity unit. For these modes, improving average ridership can be as beneficial as policies that improve the GHG intensity of these modes through the adoption of fuel-saving technologies and alternative fuels, assuming higher ridership is displacing the need for additional vehicle trips. Looking out to 2050, the sensitivity of our conclusions to average levels of service is unaffected by small variability (±20%) in this assumption as the intensity of W2W GHG emissions signifi-cantly decreases (i.e., d e f d a diminishes as E f decreases; equation 1). In summary, we find that the use of average levels of service is sufficient for providing robust GHG reduction potentials for HHD trucks, trains, aircraft, and OGVs, but could possibly lead to errors for lower-capacity and short-range modes such as automobiles, buses, and medium-heavy duty trucks.
For each policy scenario, we model GHG reduction potentials assuming an optimistic, yet attainable, decarbonization of electricity and material production technologies. To test our findings that vehicle electrification policies improve through time, we reanalyze the life-cycle GHG footprint of each mode under the IEA's baseline energy scenario (IEA 2010), which assumes modest reduction in the carbon intensity of the grid by 2050. Under these assumptions, we find that GHG reduction potentials for all policies, excluding passenger rail to HSR, result in diminishing, but still positive, benefits by 2050. This result arises because the rates at which W2W processes decarbonize Figure 3 Transportation GHG reduction policies are sensitive to load factors (ridership or payload). For each region and mitigation strategy (ex., Auto to Bus, TOP; Diesel Rail to HSR, BOTTOM), the gray domain implies that the y-axis mode is less GHG intensive than the x-axis mode, and the white domain implies that the x-axis mode is less GHG intensive than the y-axis. The overlying red region represents the feasible set of all load factor combinations between low-and high-ridership conditions for each respective vehicle. The dot in each graph represents average load factors estimated for each region. Results for the other four cases are provided in the Supporting Information on the Web. GHG = greenhouse gas; HSR = high-speed rail.
are comparable for both vehicles in mode-switching or electricity adoption partnerships. Hence, the difference between the status quo and the alternative vehicle decreases over time. This diminishing return on GHG abatement policies caused by reductions in the GHG intensity of fuel-related, life-cycle processes has policy implications, which we discuss in greater detail in the following section. To view the results of our scenarios under the baseline assumption, see the Supporting Information on the Web.

Uncertainty and Scenario Limitations
The scenario results presented are subject to many uncertainties regarding the types of energy technologies that will be adopted across transportation modes, as well as how these modes will be utilized in the future. Fuel consumption rates and other vehicle performance metrics (e.g., annual distances traveled, vehicle age, and load factors) by transportation mode are the best estimates available from the literature and various governmental agencies (table 2). However, regions and individual vehicles may experience differences in the GHG intensity reported in this study based on vehicle size and technology adoption, including material composition, model year, setting (e.g., urban vs. rural), demand level (e.g., on-vs. off-peak), and topology (Reyna et al. 2015). We also recognize that additional low-carbon fuels other than electricity could be used to reduce the GHG footprint of passenger and freight transport (Yang et al. 2009;Hill et al. 2012;Scown et al. 2013). In addition, though we offer information regarding trends in the GHG footprint of different modes within a region, considerations of costs are not within the scope of our study, but are critical for defining the optimal pathways for meeting our long-term climate-change mitigation goals in an efficient manner. On the issue of selecting a specific climate-change mitigation pathway, we chose the IEA's BLUE Map scenario to maintain a consistent analytical structure and system boundaries with other studies in the UNEP IRP collaboration (Hertwich et al. 2015;Gibon et al. 2014).

Discussion
Through a life-cycle GHG assessment at a global scale, we offer estimates on the marginal GHG benefits of new technologies, mode-switching, and electrification polices for both passenger and freight transport. We show that new technology adoption can greatly reduce the GHG intensity of passenger and freight transport by 2050. Passenger vehicles are projected to see larger fuel efficiency improvements than freight vehicles, amounting to reductions as large as 90% from the 2010 scenario baseline. The analysis also reveals that the unit GHG intensity of vehicle modes varies across the world, suggesting that effectiveness of GHG mitigation policies will be location dependent. Accordingly, accounting for region-specific technology adoption, vehicle productivity, as well as supply-chain processes occurring along fuel pathways facilitates the identification of GHG abatement policies with the greatest net benefits. We find that investments into decarbonizing fleets in non-OECD countries may amount to the greatest net reductions in the years to come. It is important to note, however, that many factors distinguish transportation systems between developed and developing nations, such as the availability and cultural appropriateness of new technologies, the financial ability to acquire them, and policies to incentivize their adoption. To mobilize the human and financial capital needed to achieve the greatest global GHG reductions, interregional cooperation should be encouraged.
There are both short-and long-term benefits to GHG abatement policies analyzed in this study. Overall, we find that the GHG reduction potentials of mode-switching policies were greatest in 2010 and see diminishing, but still positive, benefits through 2050 resulting from fuel efficiency improvements in the status quo vehicle. This effect can be considered a negative feedback caused by equal rates of improvements to W2W GHG intensity between currently utilized and alternative passenger and freight vehicles. The most apparent impact on transportation policy is that it incentivizes the early adoption of modal shift; however, it may limit the value of these policies from a GHG perspective in the future. In contrast, vehicle electrification policies become more effective over time under the IEA's BLUE Map energy scenario. In the policy cases considered, we could see an increase as high as 400% in net benefits as economies transition away from carbon-intensive electricity generation. However, the long-term success of these policies requires a progressive decarbonization of electricity. Without this, alternative fuel policies may also be subjected to diminishing marginal benefits over time.
There are many opportunities to improve the way we perform transportation-focused analyses at a global scale. Data availability, especially in the developing world, remains a key contributor to uncertainty in the studies. A first step toward reducing these uncertainties is identifying which pieces of information influence the final decision-making process the greatest. For smaller modes, understanding the extent and skew of ridership or payloads can have a large impact on the calculated GHG intensity of these modes. In contrast, modes that provide large levels of services throughout their lifetime are not sensitive to such variability. For these modes, identifying how fuel rates differ over time and space are more important for improving the certainty of life-cycle GHG estimates. Future work should consider the influence of these important factors for other important GHG abatement policies, such as coloading or ride sharing services, vehicle automation, and the adoption of alternative fuels not discussed herein. It is important to also note that supporting infrastructure systems coevolve with the deployment of low-carbon vehicle technologies and that this codependency is shaped by additional economic, political, social, and cultural factors (Chester et al. 2014). In this article, these considerations were outside the analysis boundaries, but future research should examine their influence on GHG reduction potentials.