Delayed emergence of a global temperature response after emission mitigation

The study asks when measurable benefits in global mean surface temperature (GMST/GSAT) can be detected following strong mitigation of specific climate forcers, given that internal climate variability and system inertia obscure near-term trends. Although anthropogenic forcing has driven ~0.2 °C per decade warming since the 1970s and will likely continue for decades, documenting near-term benefits is crucial for sustaining public support for mitigation. Prior work shows delayed detectability of mitigation benefits due to variability and inertia, with global temperature signals emerging decades after scenario divergence and even later for regional scales. Marotzke (2018) also found that, despite strong mitigation (RCP2.6), many realizations could still exhibit near-term warming rates equal to or exceeding recent decades, risking a “hiatus debate in reverse.” This paper extends previous scenario-based analyses by isolating individual forcers (e.g., CO₂, CH₄, N₂O, aerosols) to assess when statistically significant temperature differences emerge under idealized single-component mitigation relative to a baseline (RCP4.5). It evaluates cumulative temperature impacts and near-term warming rates, highlighting that even very strong mitigation of most forcers will not yield discernible surface temperature changes for years to decades and that fully removing some short-lived forcers (e.g., BC) would not be statistically discernible for about a decade.

Tebaldi and Friedlingstein (2013) quantified delayed detection of mitigation benefits, finding ~25–30 years to detect differences in GSAT between strong mitigation (RCP2.6) and higher-emission scenarios (RCP8.5/RCP4.5), with delays longer for smaller regions due to higher variability. Marotzke (2018) showed that under RCP2.6, more than a third of initial-condition ensemble members continue to warm faster until 2035 than in 2006–2020, and used Bayesian methods to estimate that only a modest fraction of trend reductions could be attributed necessarily and sufficiently to emission changes. Other studies on time-of-emergence (e.g., Hawkins & Sutton 2012; Frame et al. 2017, 2019) contextualize detectability thresholds. Research on linear additivity of forcing-response (e.g., Shiogama et al. 2013) supports analyzing individual forcers, though complexities remain for aerosols and interactions. Multi-model work on BC and aerosol impacts (e.g., Samset et al. 2016, 2018; Stjern et al. 2017) informs expectations for short-lived forcers and aerosol-cloud interactions.

The authors simulate future global mean surface air temperature (GSAT) using the reduced-complexity model MAGICC6 (default configuration; ECS = 3 °C; Bern carbon cycle; aerosol radiation interaction efficacy 0.9). Emissions follow RCP pathways with RCP4.5 as the baseline. Idealized single-forcer mitigation scenarios begin in 2020: (1) Zero emissions of the targeted forcer, (2) −5% per year reduction, and (3) switching that forcer to follow RCP2.6 (all other components remain on RCP4.5). For forcers considered (CO₂, CH₄, N₂O, SO₂, CO, NMVOC, NOₓ, BC, OC, CF₄, SF₆), MAGICC6 converts emissions to concentrations via parameterized lifetimes/carbon cycle, estimates radiative forcing, and computes temperature response. To incorporate internal variability, the study uses 32 members from CESM1 Large Ensemble (CESM1 LENS) forced with historical+RCP8.5, detrends each member by removing the ensemble mean to obtain variability-only series, and adds these anomalies to MAGICC6 outputs to create 32 synthetic realizations per scenario. To avoid identical variability alignment between compared scenarios, one series is shifted by five years before pairwise comparison. Emergence is defined via a Student’s t-test (p < 0.05) applied to the cumulative annual GSAT time series from 2020 up to year y for each baseline–scenario pair; the time of emergence is the first year when at least 66% (21/32) of pairs show significant difference. Consistency checks with trend-based methods from TF13 and Bayesian analysis per M18 are provided. Additional analyses include: (i) emergence times for RCP pathway differences (RCP8.5 vs RCP4.5 vs RCP2.6), (ii) avoided warming in 2100 for each single-forcer scenario, (iii) decadal warming rates (10- and 15-year trends) under mitigation, and (iv) a mitigation potential versus cumulative mass reduced comparison. Sensitivity tests examine alternative ECS settings for CH₄ and additivity by summing multi-component RCP2.6 perturbations.

- Scenario detectability (multi-component): With internal variability included, temperature differences between RCP8.5 and RCP2.6 emerge around 2035; RCP8.5 vs RCP4.5 around 2037; RCP4.5 vs RCP2.6 around 2046. - Avoided warming in 2100 (relative to RCP4.5; MAGICC6, ECS=3 °C): - CO₂: Zero emissions 1.05 °C; −5%/yr 0.74 °C; RCP2.6 0.74 °C (all bold/significant before 2100). - CH₄: 0.19, 0.16, 0.17 °C. - N₂O: 0.18, 0.14, 0.06 °C. - SO₂: −0.13, −0.09, −0.09 °C (mitigation increases warming due to loss of cooling). - CO: 0.02, 0.01, 0.00 °C; NMVOC: 0.04, 0.04, 0.01 °C; NOₓ: −0.08, −0.06, 0.00 °C. - BC: 0.09, 0.07, 0.04 °C; OC: −0.12, −0.10, 0.07 °C. - CF₄: 0.01, 0.01, 0.01 °C; SF₆: 0.01, 0.01, 0.01 °C. - Combined zeroing of BC+OC+SO₂ would yield a net +0.16 °C warming in 2100 due to removal of aerosol cooling. - Emergence times for single-forcer mitigation (year first 66% significant; from 2020 start): - CO₂: Zero 2033; −5%/yr 2044; RCP2.6 2047. - CH₄: 2039; 2055; 2050. Sensitivity to ECS: with ECS=2, 2042/2060/2057; ECS=4, 2038/2052/2048. - N₂O: 2061; 2079; no emergence for RCP2.6 following. - SO₂: 2026; 2050; 2052 (note: signal is additional warming). - NMVOC: 2081; no emergence; no emergence. - NOₓ: 2039; 2075; no emergence. - BC: 2028; 2048; 2083. - OC: 2036; 2064; 2081. - Near-term warming rates (10-year trends): MAGICC6 baseline rates ~0.25 °C/decade. Single-forcer mitigation generally does not push decadal rates outside variability envelopes before mid-century. Exception: complete removal of anthropogenic CO₂ rapidly lowers the warming rate outside 1 SD in 2021–2030 and keeps it low. Strong BC or CH₄ mitigation can reduce the mean decadal rate in the first decade, but effects wane as CO₂ dominates later. SOx mitigation increases near-term warming rates. - Mitigation potential vs effort: BC reductions yield the most rapid detectable temperature response with small ultimate payoff; CO₂ offers the largest long-term avoided warming but requires orders-of-magnitude greater cumulative mitigation mass before emergence; CH₄ provides intermediate benefits, combining relatively rapid emergence with several tenths of a degree potential avoided warming by 2100. - Consistency checks: Trend-based probabilities for lower 2021–2035 GSAT trends vs 2006–2020 in CESM1 are 59% (RCP4.5) and 80% (RCP2.6), with ~20% necessary-and-sufficient causation probability, broadly consistent with M18.

Findings demonstrate that internal variability substantially delays the detectability of temperature responses to even strong mitigation, addressing the core question of when benefits become observable. While simplified modeling indicates clear avoided warming by 2100 for key forcers (CO₂, CH₄) and rapid responses for short-lived forcers (BC), emergence against internal variability typically occurs years to decades after mitigation begins. This underscores the challenge of communicating progress if GMST continues rising despite mitigation. Aerosol mitigation, especially SO₂, yields early detectable signals but increases temperatures due to loss of cooling, revealing policy trade-offs if air-quality and climate goals are pursued jointly. The rate-of-warming analysis indicates that only drastic CO₂ elimination markedly lowers near-term decadal warming rates beyond variability; for other single-forcer actions, rates can remain at or above present levels for decades, supporting expectations of limited near-term trend changes. The study highlights additive potential of multi-component mitigation resembling low-RCP pathways, which could yield detectable signals around ~2040. Caveats include dependence on CESM1 LENS variability under RCP8.5, MAGICC6’s higher near-term warming rates and parameter choices (ECS, aerosol efficacies), simplified carbon cycle and linear additivity assumptions, and the 2020 starting point embedded in RCP4.5 emissions. Nevertheless, the overall message is robust: rigorous detection and attribution of mitigation impacts on global mean temperature will remain difficult for years to decades, necessitating complementary indicators (e.g., GHG concentrations, emissions, carbon intensity) for tracking progress.

The paper extends emergence analyses to individual climate forcers, combining MAGICC6 with CESM1 LENS variability to quantify when statistically significant GSAT differences from mitigation become detectable. It shows that strong single-forcer mitigation generally yields delayed temperature emergence, with faster signals for SO₂ (warming), BC (small avoided warming), CO₂ (largest long-term benefit), and CH₄ (intermediate, combining near- and long-term benefits). Multi-component mitigation akin to low-RCP pathways is detectable around ~2040. The work emphasizes the need to manage expectations: even effective mitigation may not promptly manifest in GMST trends due to internal variability and system inertia. Future research should: employ comprehensive ESM ensembles to capture non-linear interactions and regional patterns; refine representation of aerosol-cloud processes and variability modes; harmonize starting conditions with observed emissions; and develop robust, policy-relevant detection frameworks and auxiliary indicators to track mitigation progress.

- Internal variability source: reliance on CESM1 LENS (RCP8.5-forced) may not capture the full model spread of variability characteristics; variability representation differs among CMIP5/CMIP6 models. - Model structure and parameters: MAGICC6 near-term warming rates are higher than some comparable models (e.g., FaIR); results depend on assumed ECS (default 3 °C) and parameter tunings (e.g., aerosol efficacies, Bern carbon cycle). - Aerosol processes: simplified and globally averaged treatment may under-represent regional aerosol-cloud interactions and teleconnections; MAGICC6 tuning based on CMIP3 may limit realism relative to modern ESMs. - Scenario initialization: mitigation perturbations start in 2020 on RCP4.5, which already assumes some short-lived forcer declines; no harmonization with actual 2020 emissions (e.g., rapid SO₂ reductions in China) potentially affects magnitudes and timing. - Linearity/additivity assumption: real-world co-emissions, interactions, and non-linear responses may reduce additivity; comprehensive ESMs would be needed for sectoral/combined policies and regional impacts. - Global focus: analysis centers on GSAT; regional emergence and other variables (e.g., precipitation extremes) are not assessed.