The 2015 Paris Agreement's 1.5°C target necessitates deep and sustained emission reductions. However, detecting the impact of mitigation on surface climate variables like global surface air temperature (GSAT) is challenging due to substantial internal variability and the lagged response of the climate system. This paper explores the use of stratospheric temperature trends as a more sensitive indicator of mitigation's effects. The globally averaged stratosphere is close to radiative equilibrium, resulting in significantly lower internal variability than the troposphere. This higher signal-to-noise ratio in stratospheric temperatures makes them potentially more effective for detecting externally forced trends, as noted in the IPCC First Assessment Report. Observed stratospheric cooling over the past 40 years has primarily been attributed to increasing greenhouse gases and ozone depletion. Future stratospheric cooling will be partly offset by ozone recovery; however, the signal-to-noise ratio is still expected to remain high, making it a promising avenue for detecting mitigation's impacts. The study uses large initial condition ensemble projections from three climate models to assess the externally forced climate change signal relative to internal variability. By comparing scenarios representing stringent mitigation (SSP1-1.9), current commitments (SSP2-4.5), and a lack of mitigation (SSP3-7.0), the study aims to demonstrate the potential of stratospheric temperature trends as an early indicator of successful climate mitigation.

The paper reviews existing literature highlighting the challenges of detecting the effects of climate mitigation on surface climate variables due to large internal variability and system inertia. It cites previous research emphasizing the stratosphere's proximity to radiative equilibrium and its consequently lower internal variability compared to the troposphere. The authors cite prior work on the canonical pattern of atmospheric temperature change (tropospheric warming and stratospheric cooling) as a key indicator of human influence. Existing studies on observed stratospheric cooling, primarily attributed to greenhouse gases and ozone depletion, are reviewed. The paper also mentions research on the use of large ensembles in climate modelling to disentangle forced changes from internal variability. Finally, past literature highlighting the importance of stratospheric temperature as a detection variable for climate change is referenced, including its mention in the IPCC's First Assessment Report and studies quantifying the anthropogenic signal in atmospheric temperature trends.

The study utilizes data from three large initial condition ensembles (≥30 members) from the CMIP6 ScenarioMIP project: CanESM5, EC-Earth3, and MIROC6. These models represent different climate sensitivities. Four global average temperature indicators were examined: GSAT, temperature of the lower stratosphere (TLS), temperature of the middle stratosphere (TMS), and temperature of the upper stratosphere (TUS). Satellite observable atmospheric layer temperatures were created by applying weighting functions based on Remote Sensing Systems and NOAA STAR SSU data to the model output. Three future emission scenarios (SSP1-1.9, SSP2-4.5, and SSP3-7.0) were used, representing stringent mitigation, current commitments, and a lack of mitigation, respectively. Data quality control was implemented in CanESM5 to remove outlier spikes. Least squares linear trends (5, 10, 15, and 20 years, starting in 2023) were calculated for each ensemble member. The overlap of trend distributions between pairs of scenarios was quantified using percentile analysis to assess the statistical significance of differences in trends between scenarios. The sensitivity of the results to the inclusion of additional layers below 1 hPa was tested using CanESM5.

The analysis reveals that under the stringent mitigation scenario (SSP1-1.9), the trends in middle and upper stratospheric temperatures (TMS and TUS) show a detectable weakening within 5-10 years compared to the scenario representing current commitments (SSP2-4.5). The separation of TMS and TUS trends between scenarios is more rapid for TUS due to the altitude-dependent increase in CO2-driven cooling. The ensemble spread in TMS and TUS trends is smaller than that observed for GSAT and TLS across all trend lengths. The overlap in temperature trends between SSP1-1.9 and SSP2-4.5 diminishes to near zero for TMS and TUS within 5-10 years. For GSAT, achieving a similar level of statistical confidence requires at least 20 years. Figure 1 shows projections of global temperature indicators for the three models, and Figure 2 displays near-term global temperature trends for the three scenarios. Figure 3 quantifies the overlap of trend distributions between pairs of scenarios for different trend lengths. The study concludes that the reduction in TUS cooling under SSP1-1.9 would be clearly detectable within 5 years, compared to the counterfactual path of SSP3-7.0. Comparing SSP1-1.9 and SSP2-4.5, a clear separation of trends is achieved within 5-10 years for TMS and TUS, even accounting for internal variability.

The key finding—that stratospheric temperature trends could offer a considerably faster signal of effective climate mitigation than surface temperature trends—is significant. This has important implications for policymakers and the public, providing early, tangible evidence of the impact of climate action. While GSAT remains a crucial indicator due to its direct connection to climate risks and impacts, this study suggests that a multivariate assessment of the climate system, incorporating indicators with high signal-to-noise ratios like stratospheric temperature, could enhance the early detection of mitigation's effectiveness. This would strengthen public and political support for sustained long-term mitigation efforts.

This study demonstrates the potential of using stratospheric temperature trends as an early indicator of successful climate mitigation. The rapid response of the stratosphere to emission reductions, as opposed to the lagged response of surface temperatures, offers valuable feedback for policymakers and the public. Future research could expand the analysis to include other high signal-to-noise indicators and refine the methodology for robust detection and attribution of mitigation effects in a multivariate context. This could include exploring different weighting functions and incorporating other climate variables in a multivariate framework for enhanced early detection of climate change impacts.

The study relies on climate model projections, which are subject to uncertainties in model representation of climate processes and future emission scenarios. The choice of weighting functions to simulate satellite-observable atmospheric layers introduces some simplification. While the sensitivity to the inclusion of additional layers below 1 hPa was tested, other potential limitations of the model output data itself cannot be entirely ruled out, necessitating caution in the interpretation of results. Finally, the analysis is focused on global average temperatures and may not capture regional variations in temperature response to mitigation.