Projecting long-term pathways of greenhouse gas emissions and carbon dioxide removal with an Integrated Assessment Model emulator

Integrated Assessment Models (IAMs) combine economy, energy, and sometimes land-use modeling approaches and are commonly used to evaluate climate policies under least-cost scenarios. The marginal abatement cost (MAC) curve approach has been commonly used in climate policy analyses to show the carbon price level for a given abatement level, which has also been applied as a way to parameterize the complex behavior of IAMs. Here, we propose a new methodological framework to i) emulate the IAM’s emission reductions in response to carbon price pathways through MAC curves (i.e., IAM emulator) and then ii) extend IAM’s emission pathways (usually given until 2100) to 2300 with the emulator.

As part of the Horizon Europe RESCUE and OptimESM projects, our approach is used to extend the greenhouse gas (GHG) emission pathways from different sources and carbon dioxide removal (CDR) pathways generated by the REMIND-MAgPIE model. A key feature of the approach is that we individually capture the emission reductions associated with CDRs (i.e., afforestation, bioenergy and carbon capture and storage (BECCS), direct air capture with carbon storage (DACCS), industrial CCS, and ocean alkalinity enhancement (OAE)) through MAC curves. Our approach relies on the following simplifying assumptions: i) MAC curves are assumed time-independent over periods, ii) abatement levels are assumed independent across GHGs (CO₂, CH₄, and N₂O), sectors (energy- and non-energy-related emissions), and CDR options, and iii) a uniform carbon price is used across sectors and CDRs, with the GWP100 metric used to fix the price ratios between different GHGs.

We approximated the dynamics of REMIND-MAgPIE with MAC curves, using equation log(f(x)+1)=a*x^b+c*x^d for sectoral gases and equation f(x)=a*x^b+c*x^d for CDR options, respectively. f(x) represents the corresponding carbon price level at x, while the variable x represents the abatement level relative to the assumed baseline level, expressed as a percentage for sectoral emissions or the absolute amount of CO₂ removed for CDR options. a, b, c, and d are the parameters that are optimized for each case. Additionally, we derived the maximum abatement levels of REMIND-MAgPIE from its simulation results under all carbon budgets, which reflect the limit of, for example, CCS capacity and sectoral mitigation potential. We also calculated for each gas, sector, and CDR option the maximum first and second derivatives of temporal changes in abatement levels to capture the limits of the technological change rate and the socio-economic inertia.

By combining the IAM emulator with a reduced-complexity climate model ACC2, we further derived extended emission pathways beyond 2100 using the least-cost approach for temperature trajectories (1 °C, 1.5 °C, and 2 °C) with overshoots of up to 2 °C. These pathways illustrate various CDR use cases over the coming centuries. Our extended scenarios, generated on the basis of long-term climate-economy interactions, can serve as input to Earth System Models investigating the long-term consequences of climate change mitigation strategies, particularly the implications of CDR deployment and associated Earth system dynamics over centennial timescales.