RCP4.5: a pathway for stabilization of radiative forcing by 2100
- First online:
Representative Concentration Pathway (RCP) 4.5 is a scenario that stabilizes radiative forcing at 4.5 W m−2 in the year 2100 without ever exceeding that value. Simulated with the Global Change Assessment Model (GCAM), RCP4.5 includes long-term, global emissions of greenhouse gases, short-lived species, and land-use-land-cover in a global economic framework. RCP4.5 was updated from earlier GCAM scenarios to incorporate historical emissions and land cover information common to the RCP process and follows a cost-minimizing pathway to reach the target radiative forcing. The imperative to limit emissions in order to reach this target drives changes in the energy system, including shifts to electricity, to lower emissions energy technologies and to the deployment of carbon capture and geologic storage technology. In addition, the RCP4.5 emissions price also applies to land use emissions; as a result, forest lands expand from their present day extent. The simulated future emissions and land use were downscaled from the regional simulation to a grid to facilitate transfer to climate models. While there are many alternative pathways to achieve a radiative forcing level of 4.5 W m−2, the application of the RCP4.5 provides a common platform for climate models to explore the climate system response to stabilizing the anthropogenic components of radiative forcing.
Representative Concentration Pathway (RCP) 4.5 is a scenario of long-term, global emissions of greenhouse gases, short-lived species, and land-use-land-cover which stabilizes radiative forcing at 4.5 W m−2 (approximately 650 ppm CO2-equivalent) in the year 2100 without ever exceeding that value. The defining characteristics of this scenario are enumerated in Moss et al. (2008, 2010). RCP4.5 is based on the MiniCAM Level 2 stabilization scenario reported in Clarke et al. (2007) with additional detail on the non-CO2 and pollution control assumptions documented by Smith and Wigley (2006), and incorporating updated land use modeling and terrestrial carbon emissions pricing assumptions as reported in Wise et al. (2009a, b).
Unlike the scenarios developed by the IPCC and reported in Nakicenovic et al. (2000), which examined possible global futures and associated greenhouse-related emissions in the absence of measures designed to limit anthropogenic climate change, RCP4.5 is a stabilization scenario and assumes that climate policies, in this instance the introduction of a set of global greenhouse gas emissions prices, are invoked to achieve the goal of limiting emissions and radiative forcing.
While RCP4.5 is based on the MiniCAM Level 2 scenario reported in Clarke et al. (2007), it differs in several important regards. First, the Clarke et al. (2007) scenario considered a slightly different definition of radiative forcing than RCP4.5. In Clarke et al. (2007), radiative forcing is defined in terms of a suite of six greenhouse gases, carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs) and sulfur hexafluoride (SF6). RCP4.5 considered the influences of a broader set of anthropogenic emissions including CO2, CH4, N2O, HFCs, PFCs, and SF6, but also chemically active gases such as carbon monoxide (CO) and volatile organic compounds (VOCs). Importantly, RCP4.5 considers the influence of sulfur aerosols, as well as black and organic carbon. The Clarke et al. (2007) scenario stabilized radiative forcing at approximately 4.7 W m−2, slightly higher than RCP4.5. For CO2, the most important anthropogenically released greenhouse gas, year 2100 concentrations are somewhat higher in the Clarke et al. (2007) scenario, approximately 550 ppm CO2, than in RCP4.5, approximately 525 ppm CO2. RCP4.5 also employed updated historical data series by calibrating to the year 2000 consensus emissions inventories from Lamarque et al. (2010), and the HYDE crop and pasture land use history (Klein Goldewijk et al. 2010) as well as implementing a new representation of residue biomass supply (Gregg and Smith 2010). RCP4.5 also employed a more sophisticated land-use and land-cover model (Wise et al. 2009a, b) than was available for use in Clarke et al. (2007). Perhaps the most important difference is the downscaling of emissions and land-use-land-cover from the 14 geopolitical GCAM regions to a 0.5° grid for the RCPs in order to enable use of the scenario in atmospheric chemistry models and global climate models. Finally, RCP4.5 takes advantage of one technology that was not modeled in Clarke et al., namely combining bioenergy production with CO2 capture and geologic storage (CCS). This technology combination is capable of producing final energy such as electricity with net-negative carbon emissions (Luckow et al. 2010).
Because the RCPs are based on previously existing scenarios documented in the open literature, each reflects a different set of underlying socioeconomic assumptions. RCP4.5 is a stabilization scenario and thus assumes the imposition of emissions mitigation policies. RCP4.5 is derived from its own “reference”, or “no-climate-policy”, scenario. This reference scenario is unique to RCP4.5 and differs from RCP8.5 as well as from the reference scenarios associated with RCP6 and RCP2.6 (also referred to as RCP3PD)(van Vuuren et al. 2011a).
In the remainder of this paper we will discuss the modeling environment employed to develop RCP4.5 (the Global Change Assessment Model; GCAM), from the original MiniCAM Level 2 scenario (Clarke et al. 2007). We will then proceed to describe the underlying socioeconomic assumptions that shape RCP4.5 and its associated reference scenario and discuss the characteristics of RCP4.5, highlighting the global energy, economic, land use, and land cover systems, as well as the mechanisms employed to limit radiative forcing to 4.5 W m−2 and contrast RCP4.5 to its reference scenario. Next, we compare the RCP4.5 to other 4.5 Wm−2 stabilization scenarios in the literature. Finally, we will describe alternate GCAM scenarios that follow the radiative forcing pathways of the other three RCPs (RCP2.6, RCP6, and RCP8.5).
2.1 The Global Change Assessment Model
The GCAM is a global integrated assessment model and a direct descendent of the MiniCAM model (Kim et al. 2006; Clarke et al. 2007; Brenkert et al. 2003). It combines representations of the global economy, energy systems, agriculture and land use, with representation of terrestrial and ocean carbon cycles, a suite of coupled gas-cycle, climate, and ice-melt models. GCAM tracks emissions and concentrations of greenhouse gases and short-lived species including CO2, CH4, N2O, NOx, VOCs, CO, SO2, carbonaceous aerosols, HFCs, PFCs, NH3, and SF6.
GCAM is a dynamic recursive economic model driven by assumptions about population size and labor productivity that determine potential gross domestic product in each of 14 regions at 15 year time steps. GCAM establishes market-clearing prices for all energy, agriculture and land markets such that supplies and demands for all markets balance simultaneously. The GCAM energy system includes primary energy resources, production, energy transformation to final fuels, and the employment of final energy forms to deliver energy services such as passenger kilometers in transport or space conditioning for buildings. GCAM contains detailed representations of technology options in all of the economic components of the system with technology choice determined by market competition.
The agriculture and land use component is fully integrated (i.e., solved simultaneously) with the GCAM economic and energy system components. Land is allocated between alternative uses based on expected profitability, which in turn depends on the productivity of the land-based product (e.g. mass of harvestable product per ha), product price, and non-land costs of production (labor, fertilizer, etc.). The productivity of land-based products is subject to change over time based on future estimates of crop productivity change. This increase in productivity, adopted from projections by Bruinsma (2003), is not specifically attributed to individual components, which may include changes in management practices, increases in fertilizer or irrigation inputs, or development of new crop varieties. Emissions of gases related to agricultural productivity, for example N2O and CH4, are tied to the level of production. All agricultural crops, other land products and animal products are globally traded within GCAM. A full description of the agriculture and land use modeling in GCAM as used for RCP4.5 can be found in Wise et al. (2009a).
The GCAM physical atmosphere and climate are represented by the Model for the Assessment of Greenhouse-Gas Induced Climate Change (MAGICC; Wigley and Raper 1992, 2002; Raper et al. 1996). To construct the RCP4.5 scenario, we use MAGICC version 5.3, which is initialized to the IPCC 4th Assessment Report.1 The definition of total radiative forcing for the RCP4.5 does not include albedo, nitrate, and mineral dust. These three forcing agents have a fixed future forcing of −0.4 W m−2 in MAGICC version 5.3.
Pollutant gas emissions depend on modeled activity levels in each region, such as fuel consumption, and the assumed level of pollution controls. Pollution control levels increase over time in all countries as a function of income (Smith et al. 2005; Smith and Wigley 2006). Some further increases in emission controls for high-income countries are assumed, with larger increases in developing countries as incomes rise toward current OECD levels. Pollutant gas and aerosol emissions levels in the reference scenario were checked for consistency by estimating regional surface particulate and ozone levels using the MOZART atmospheric chemistry model. Two rounds of analysis were performed using reference scenario emissions, with emission control levels adjusted so that regional surface pollutant concentrations were consistent with the assumed regional income levels (Smith et al. 2011).
The RCP4.5 stabilization scenario is a cost-minimizing pathway. It assumes that all nations of the world undertake emissions mitigation simultaneously and effectively, and share a common global price that all emissions to the atmosphere must pay with emissions of different gases priced according to their hundred-year global warming potentials (Schimel et al. 1996). All sectors of the economy are covered, including agriculture and land use emissions. That emissions price also rises over time so as to minimize the present discounted cost of emissions mitigation. The policy also assumes that deployment mechanisms and measurement and monitoring of both fossil fuel and terrestrial carbon are not barriers to implementation of emissions mitigation.
This cost-minimizing price path has two components. Prior to reaching the target, 4.5 W m−2, cost minimization requires that the greenhouse gas emissions price rise at the interest rate, adjusted by the rate of ocean uptake (Edmonds et al. 2008; Clarke et al. 2007; Hotelling 1931; Peck and Wan 1996). An emissions price path with this property precludes all opportunities for arbitrage because the discounted marginal cost of abatement is constant across time. The second component of the pathway occurs after the target is reached. At this time, the emissions price is adjusted to ensure that the radiative forcing level remains at its target. For the RCP4.5, stabilization occurs in 2080; prior to 2080 the emissions price rises at 5% per year, and after 2080 the emissions price is roughly constant.
2.2 Emissions downscaling
Emissions from GCAM were downscaled using a two-step method, first downscaling to the country level and then mapping to a spatial grid within each country. Detailed emissions from GCAM were aggregated into the 12 RCP reporting sectors for the 14 GCAM regions at each 15-year model time period.2 Emissions from each sector were downscaled to a country level for 231 countries using the methodology outlined in Van Vuuren et al. (2007). Input data for this step includes the country-level population projection used for the RCP4.5 scenario, a gridded base-year GDP data set from van Vuuren et al. (2007), who combined World Bank GDP information with the GPW gridded population data set (CIESIN & CIAT 2005), and year 2000 gridded emissions data from Lamarque et al. (2010). Gridded GDP and emissions data were used so that base-year GDP and emissions differences could consistently be estimated for any set of countries or regions. A convergence year of 2200 for GDP and emissions intensity calculations (slightly larger than the value used by Van Vuuren et al. 2007) was used for all anthropogenic emissions sectors. The pattern of forest and grassland emissions within each region was held constant by setting a high convergence year of 10,000.
The downscaled sectoral emissions for each country were mapped to a 0.5° grid using the base-year 2000 gridded emissions data from Lamarque et al. (2010). The relative spatial emissions distribution within each country was held constant over time for each emissions sector. For the downscaling calculations, each 0.5° grid cell was subdivided, if necessary, by country boundaries using 2.5 min boundary data (CIESIN & CIAT 2005). After downscaling on a sub-divided grid for each country and emissions sector, emissions were summed to a 0.5° resolution for the final data product.
Emissions from international shipping and aircraft primarily occur outside country boundaries. Thus, emissions from these sectors were aggregated to one global figure. Gridded shipping emissions were globally scaled from the RCP consensus year 2000 emissions grid (Lamarque et al. 2010). Aircraft emissions used a time changing pattern from the QUANTIFY B2i emissions scenario (Lee et al. 2010), as the overall pathway for this scenario closely matched the GCAM model output. The QUANTIFY three dimensional emissions pattern was collapsed to two dimensions (latitude and longitude) for 2000, 2025, 2050, and 2100. This pattern was interpolated to decadal intervals and scaled globally to match the GCAM global aviation emissions values.
2.3 Land use downscaling
The RCP scenario process is the first to explicitly provide land use projections in addition to future emissions pathways for input to global climate models. Because all four participating integrated assessment models, and all receiving climate models, use different characterizations and definitions of land use types and transitions, a harmonization step was necessary. The harmonization was designed to provide a continuous, consistent set of land use inputs for climate models from 1500 through 2100 with a smooth transition between historical data (1500–2005) and future projections (2005–2100) (see Hurtt et al. 2011).
In the GCAM model results, land use is simulated at the 14 region level and land use changes and transitions are not spatially attributed. In this case, the land use was first downscaled to the 0.5° harmonization grid, following the algorithms of the global land-use model (GLM) (Hurtt et al. 2006), preserving GCAM regional land use area totals and generating smooth spatial patterns in the transition from historical to future states. These downscaling algorithms were developed and implemented by the land use harmonization group (Hurtt et al. 2011) and are fully described in Thomson et al. (2010).
3.1 GCAM reference scenario
3.2 RCP4.5 stabilization scenario
CO2 constitutes the largest contribution to total radiative forcing in the RCP4.5, followed by CH4, halocarbons, tropospheric ozone, and N2O (Fig. 8). The relative proportion of the non-CO2 components of positive radiative forcing remains constant over time. Increases in activity levels as income and population grow tend to increase emissions; however, since these substances are greenhouse gases, we assume that emissions controls are implemented as the carbon price rises. The net result is that emissions of these gases are roughly constant over time. Sulfate forcing is net negative throughout the century but this influence declines over time largely due to assumed increases in pollution control with income, although there are also indirect sulfur dioxide emissions reductions due to the greenhouse gas mitigation policy (Smith et al. 2005).
3.3 Comparison to the literature
Population in the RCP4.5 (Fig. 10a) is among the lowest of the scenarios considered. There are six EMF 22 scenarios and one CCSP scenario with similar population trajectories to the RCP4.5; however, it should be noted that all of these scenarios were produced by integrated assessment models from the Pacific Northwest National Laboratory.6 However, the range in population estimates in 2100 across the 28 scenarios is small; the largest population estimate is only 20% higher than the lowest.
Global GDP (Fig. 10b) varies more significantly across the 28 scenarios. The highest GDP estimate in 2100 is more than double the lowest estimate. The RCP4.5 and the CCSP RCP4.5 marker scenario on which the RCP was based fall in the middle of these estimates. Additionally, the RCP4.5 has slightly higher GDP than the CCSP, as discussed previously.
Cumulative energy and industrial CO2 emissions, as well as the time path of emissions, vary across models (Fig. 10c). Cumulative emissions range from 2043 GtCO2 to 3573 GtCO2 between 2000 and 2100. RCP4.5 falls in the middle of this range with 3010 GtCO2 emitted by the energy and industrial systems over the century. Several reasons exist for differences in cumulative emissions. First, these are only energy and industrial CO2 emissions and do not include land use and land-use change CO2 emissions. The amount of CO2 emitted from the terrestrial sphere is also likely to vary across the 28 scenarios. Second, the models use different climate models and have different characterizations of the carbon cycle. Thus, two models with the same level of cumulative total anthropogenic CO2 emissions may reach different atmospheric CO2 concentrations (see Smith and Edmonds 2006). Next, these scenarios limit radiative forcing to 4.5 Wm−2. Different models may find different contributions of the various gases to radiative forcing due to underlying pollution abatement assumptions.
The time path of emissions also varies across models for a variety of reasons, including (1) differences in reference scenario emissions, (2) differences in the cost of abatement over time, (3) differences in the speed at which capital stock can be replaced, and (4) differences in assumptions about foresight (Fawcett et al. 2009). The RCP4.5 has a slightly different emissions time path than its predecessor, the CCSP marker scenario. These two scenarios have similar cumulative CO2 emissions (RCP4.5 has 3010 GtCO2; CCSP has 3212 GtCO2) but the RCP4.5 has higher emissions in the near term and lower emissions in the long-term than the CCSP. These two scenarios use the same model, and thus the same assumptions about foresight and capital stock turnover. Additionally, they use very similar reference scenarios. One major difference between these two scenarios is the inclusion of biomass with CCS in the RCP4.5. This technology generates net negative CO2 emissions therefore making it cost effective to delay some emissions mitigation until the second half of the century. Thus, we observe higher emissions in the near term in the RCP4.5 and lower in the long term.
Finally, we compare the carbon price needed to reach 4.5 Wm−2 in the 28 scenarios (Fig. 10d). This price varies significantly across the models, ranging from $55/tCO2 in 2100 to $2141/tCO2 in 2100. RCP4.5 and the CCSP Marker Scenario both fall in the lower part of this price range. Differences in carbon prices can be attributed to differences in reference scenario emissions, and thus the level of abatement required, along with differences in the cost of abatement technologies.
3.4 GCAM-simulation of the four pathways
Differences in emissions by gas between the official RCPs and the GCAM pathways that follow the RCP forcing levels can be attributed to any number of factors. First, GCAM employs different population and GDP assumptions than the other three models. GCAM has the smallest population and the second highest GDP of the four models (van Vuuren et al. 2011a). Second, the four models have different assumptions about technological change and resource availability. Third, the GCAM model uses a terrestrial carbon policy that has a significant impact on land use and land-use change emissions. While each of the models consider abatement opportunities in the terrestrial system, the method of attaining these opportunities differs across the four RCP models. The differences listed here are only a subset of differences between the four models. However, they illustrate an important point; namely, there are numerous ways that a given radiative forcing goal can be achieved. The RCP4.5 is only one possible pathway to stabilization of radiative forcing at 4.5 W/m2.
The RCP4.5 scenario is intended to inform research on the atmospheric consequences of reducing greenhouse gas emissions in order to stabilize radiative forcing in 2100. It is also a mitigation scenario – the transformations in the energy system, land use, and the global economy required to achieve this target are not possible without explicit action to mitigate greenhouse gas emissions. However, there are many possible pathways in GCAM and other integrated assessment models that would also achieve a radiative forcing level of 4.5 Wm−2. For example, simulations with GCAM can reach 4.5 Wm−2 even if some technology options, such as CCS or nuclear power, are removed from consideration or even if not all countries enter into an emissions mitigation agreement at the same time (Clarke et al. 2009). Such alternate scenarios have different characteristics – higher emissions prices and different energy system transformations, for example – than the RCP4.5. GCAM can also reach 4.5 Wm−2 under different assumptions of crop productivity growth (Thomson et al. 2010). Changing these assumptions, however, affects the amount of dedicated bioenergy crops grown and the cost of food. Additionally, we have used GCAM to stabilize at 4.5 Wm−2 without a terrestrial carbon policy. In this case, substantial deforestation occurs as land is cleared for bioenergy production. The result is significantly higher land use change emissions, with compensating reductions in energy system emissions. The pathway discussed here and released as RCP4.5 is cost-minimizing, and therefore invokes all available technology options that can cost-effectively contribute to mitigation.
RCP4.5 aims to achieve stable radiative forcing in 2100; however, this does not imply that greenhouse gas emissions, greenhouse gas concentrations, or the climate system are stable. Radiative forcing is stable from 2080–2100 in the RCP4.5, but emissions and concentrations of greenhouse gases continue to vary in the underlying scenario. The application of RCP4.5 in climate models provides a platform to explore the long-term climate system response to stabilizing the anthropogenic components of radiative forcing.
Note that the final concentration pathway values for the RCP4.5 were produced in MAGICC version 6 (see Meinhausen et al. 2011).
GCAM models emissions at the technology level.
GCAM uses the same population estimates for all scenarios. However, we include an energy price feedback effect on GDP. As a result, the introduction of a carbon policy, as in the RCP4.5, results in a slight reduction in GDP from the reference scenario values.
526 ppmv is the CO2 concentration from GCAM before harmonization in MAGICC6 (see Meinhausen et al., this issue). The harmonized RCP4.5 CO2 concentration in 2100 is 538 ppmv. However, the total radiative forcing in GCAM prior to harmonization is 4.5 Wm−2 in 2100, but the harmonized RCP4.5 total radiative forcing is 4.3 Wm−2. Thus, using MAGICC6 endogenously in GCAM would allow for slightly higher emissions and slightly lower carbon prices, while still reaching 4.5 Wm−2.
The 50 EJ/yr of bioenergy is from dedicated bioenergy crops only. An additional 120 EJ/yr of bioenergy from waste products (including crop residues, pulp and paper mills, and municipal solid waste) and 8 EJ/yr of traditional bioenergy are consumed in the RCP4.5. The amount of dedicated bioenergy crops produced is sensitive to assumptions about crop productivity improvements (see Thomson et al 2010).
The EMF 22 study included two variations of MiniCAM and the SGM model. The CCSP included MiniCAM.
Funding was provided by the US Department of Energy, Office of Science through the Integrated Assessment Research Program. The authors wish to thank the many scientists and collaborators involved in the planning and development of the RCP process, and especially the AIM, MESSAGE and IMAGE modeling teams for time devoted to scenario review and coordination. We also thank Dr. Yuyu Zhou and three anonymous reviewers for helpful improvements to earlier versions of this paper.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.