a, CH4 has 102 times the radiative forcing per gram of CO2 but decays more quickly, with the gases having equal radiative forcing (RF) 67 years after emission4. b, As a result, the impact of using technologies decays over time at different rates, as the comparison of gasoline and CNG illustrates. c,d, These dynamics explain why the impacts of technologies, notably algae biodiesel with a biogas co-product, change when evaluated over a 100-year (c) versus a 20-year (d) time horizon. (BD, biodiesel; CNG, compressed natural gas.)
In this section we describe the approach to generating the reference scenarios used to calculate the range of CCI and ICI values and to test metric performance. We also describe the technology emissions data used in the research.
Reference scenarios.
The reference scenarios are CO2 emissions, multi-gas concentration scenarios: all emissions in the simulation are composed entirely of CO2, but previous emissions of non-CO2 greenhouse gases are also modelled (Supplementary Section 1.1). These reference scenarios are used to calculate the CCI and ICI and to test all metrics, by allocating a portion of CO2-equivalent emissions to non-CO2 gases using the metrics.
Emissions scenarios are constructed30, where initial emissions e0 change over time according to
where g(t′′) is an evolving, exponential growth rate (t′′ is a dummy, integration variable). Emissions grow at a constant rate g0, based on present growth rates, until the mitigation onset time (t1), after which g(t′′) is reduced by a fixed annual amount until it reaches the final growth rate gf.
Concentrations ci(t) of each gas i are a function of pre-industrial concentrations ci(t0), historical emissions (t0 < t′ ≤ 0) and new emissions (0 < t′ ≤ t),
where fi(t, t′) gives the fraction of a gas emitted at t′ remaining at time t (ref. 4),
and ai and τi are constants (see Supplementary Table 2 for CO2, CH4 and N2O values). (Equation (5) is also used in the CCI and ICI formulations (equations (1) and (2)), where t is replaced by t′′ for the CCI, which ranges from t′ to tS, and is replaced by tS for the ICI.) As n = 1 and a0 = 0 for non-CO2 greenhouse gases, no information about the emissions timing is needed to calculate concentrations from historical emissions. For CO2, where n = 3 and a0 ≠ 0, the rate of removal must be approximated using historical emissions data (Supplementary Section 1.1.2).
Radiative efficiency (radiative forcing per unit concentration) values are used to determine radiative forcing from concentrations4.
where RFA(t) refers to all radiative forcing not due to the presence of modelled gases i, and Ai is the radiative efficiency of gas i (Supplementary Section 1.1.3).
A scenario family is a set of pathways RF(t) that stabilize at the same radiative forcing threshold but approach it at different rates. To generate stabilization scenarios, emissions are adjusted after the threshold is reached such that radiative forcing equals the threshold value in all subsequent years. Emissions scenarios within a scenario family are defined based on their values of t1, which is varied to the greatest extent possible. Earlier values of t1 result in gradual emissions reductions, whereas later values of t1 result in delayed emissions reductions followed by rapid reductions. The scenario family for 3 W m−2 stabilization defines the range of stabilization times (tS) for the analysis presented in the paper.
Metric testing.
The performance of emissions metrics is tested by budgeting a trajectory e(t′) for total CO2-equivalent emissions, and allocating a fraction q of these emissions to a non-CO2 greenhouse gas. Consider the case of two gases, CO2 and CH4. Given a metric μ(t′), the sum of CO2 emissions eK(t′) and CO2-equivalent CH4 emissions μ(t′)eM(t′) must equal e(t′). The radiative forcing scenario can be derived from equation (6),
where K and M subscripts refer to CO2 and CH4 respectively, all other contributions to radiative forcing are encompassed in the term RFA(t), and concentrations are disaggregated into the three contributions given in equation (4): pre-industrial concentrations, concentrations from historical emissions (abbreviated ciL(t)), and concentrations from new emissions. The difference between the actual radiative forcing in the mixed gas case (where q ≠ 0) and the budgeted, CO2 emissions case (where q = 0) is
A similar approach is used to test metric performance for technology evaluation, given a budgeted emissions trajectory e(t′) and using historical data to allocate a fraction p of these emissions to the sector of interest (Supplementary Section 1.2.2).
Data.
Global emissions, concentration and radiative forcing data are published by the International Institute for Applied Systems Analysis. Technology life-cycle emissions are taken from the Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Version 2012, published by Argonne National Laboratory. In GREET, natural gas CH4 emissions that arise from liquid unloading in conventional production offset increased leakage during unconventional well completion, with conventional gas having 27% higher emissions than unconventional gas. Present US breakdowns of conventional and unconventional gas are used22. In GREET, corn co-products are used as animal feed and algae co-products are used to create biogas in a state-of-the-art facility with CH4 leakage rates of 2% (ref. 2). Emissions for electricity generation technologies are taken from a recent study1. Low-CH4 emissions scenarios for natural gas are based on updated EPA estimates of natural gas system leakage and an alternative catalytic hydrothermal gasification scenario for algae biodiesel2 (Supplementary Section 4).