Despite non-linear safety level reductions when planetary boundaries are combined, this study shows that there is a low probability of meeting climate-related planetary boundaries without CDR even if CO2 emissions peak before 2025. Significant CDR implementation may allow for a later emissions peak but the trade-off is limited and peaking later than 2030 seriously jeopardizes the chances to remain within planetary boundaries. Other relevant results suggest that ambitious CDR need not be deployed before the late 21st century and that early net-zero emissions (between 2050 and 2075) can offset later peaks. In addition, the inclusion of SRM can improve the safety level, but the impact on acidification is limited due to the dominant role of atmospheric CO2 concentration on this specific impact.
Our results show that assumptions about the level of future CDR and SRM have a significant impact on the space of anthropogenic activities compatible with planetary boundaries, in line with previous results15,25-27. Our backward approach only infers compatible global CO2 emissions or non-CO2 radiative forcing. That is why we define CDR and SRM as additional mitigation solution when a floor of positive emissions or radiative forcing is reached (see Method). This floor being the lower bound of the AR6 scenarios envelope14 without these technologies, we assume that all possible mitigation efforts have been made before resorting to CDR or SRM.
With our notion of safety levels, we assess the existence of at least one emissions pathway compatible with one or more planetary boundaries in a space of emissions bounded by constraints on their key characteristics. Importantly, in a given physical configuration of the model, not all pathways with a key characteristic equal to the limit value of the compatible space adhere to the planetary boundary. Indeed, the chosen pathway characteristics are relevant proxies, but they must be combined together to comprehensively map the limits of the actual compatible space. In other words, key characteristics limits quantified in this work and necessary but not sufficient conditions to stay within planetary boundaries, unless enough of them are combined so as to become comprehensive. Furthermore, we investigated alternative definitions of the compatible space, using safer limits requiring more than one emissions pathway to be compatible with the boundaries. Safer limits naturally lead to more stringent limits of the compatible space, but we did not identify any qualitative change in our results beyond this (Methods).
It is generally accepted that global warming primarily depends on the cumulative CO2 emissions budget, rather than the magnitude or timing of emissions peaks28. As a result, long-term mitigation strategies often prioritize achieving a net-zero CO2 emissions target29. However, the diversification of boundaries diminishes the relevance of carbon budget as a unique indicator to define compatible spaces of emissions7. Our study highlights the importance of considering additional key mitigation characteristics to comprehensively map the space of compatible future CO2 emissions. In particular, the timing of peak CO2 emissions emerges as a key factor in trade-off mechanisms. We have quantified how an earlier emissions peak can enable a shift in the net-zero CO2 date, reduce the need for CDR measures, and even eliminate the need for SRM techniques. The concept of compatible space is in the continuity of the tolerable window approach introduced in early 2000’s19,30-32 and extends the notion of a “closing door” introduced by Stocker33, which examined the relationship between the starting date of mitigation efforts, the mitigation rate, and the attainment of a specific temperature target. In addition, considering different planetary boundaries independently changes the compatible space, and combining them further narrows it. The four planetary boundaries could be supplemented by additional boundaries arising from physical or anthropogenic impacts of climate change. The selected thresholds are also subject to debate due to the inherent uncertainties and judgement values associated with quantifying planetary boundaries2,34.The trade-off between peak emissions and negative emissions has already been highlighted in the results of projections based on integrated assessment models (IAM), but the current AR6 scenario’s framework14 explored a narrower range of pathways than our study, often with strong negative emissions in the second half of the 21st century25,35, without inclusion of SRM, and limited to the 2100 horizon. Here, we produced ex-ante a large ensemble of pathways, ignoring socioeconomic and technological constraints, which allows us to search for original scenarios consistent with the Paris Agreement that were not analyzed by IAMs. Figure S1 illustrates that Pathfinder covers a wider range of possible pathways than was explored by the IAMs. Although the IAMs propose scenarios that respect all planetary boundaries, they are based primarily on the global warming boundary and do not explore all alternatives.
The non-linearity of boundary combinations argues for diversification of climate impact indicators to be considered ex-ante by scenario developers. It is instructive and complementary to IAMs to expand the scope of emissions compatible with these boundaries. Even if these pathways turn out not to be economically optimal, it is critical to first draw what is physically possible and then decide what is socioeconomically feasible and acceptable. To assess the relevance of a pathway in the socioeconomic dimension, coupling of Pathfinder with an IAM or simple impact models seems a natural next step in our work, as it would allow the exploration of compatible spaces that are also constrained in the socioeconomic and technological dimensions.
In summary, the framework developed in this study can contribute to enhancing dialogue among the Earth system, impacts, and IAM communities and encourage scenario narratives that explore a wider range of possible futures in an integrated manner.
Box 1: Method to determine compatible spaces
Figure 1 illustrates, step by step, our backward approach. First, we generate around 15,000 trajectories of global mean surface temperature and atmospheric CO2 concentration that follow historical data and asymptotically reach 1.5°C with an overshoot up to the 2°C warming level (see Methods). Those trajectories are used as inputs to Pathfinder (step 1). To account for physical uncertainties, we run the model (step 2) with 1500 sets of parameters that can be seen as 1500 possible states of the world with their own physics, all independent and calibrated on historical observations.
All pathways stay below the +2°C global warming boundary by construction, and we define permissive conditions on CO2 anthropogenic emissions and non-CO2 radiative forcing to remain within reasonable ranges: in step 3, we exclude pathways rising too quickly (quicker than the most pessimistic pathway from AR6 scenarios) or using too much CDR (more than 10 PgC yr−1). We use the envelope of all the pathways that stay within a given boundary to illustrate the compatible space. We compare the envelope of the default space compatible with the global warming boundary (black envelope in Figure 1) with envelopes obtained for other planetary boundaries, either independently (green envelope for ocean acidification) or combined (purple envelope). As explained in main text, respecting more boundaries leads to narrower envelopes.
The next steps are illustrated with the ocean acidification boundary only. To express the compatible space in terms of key characteristics, we search for limit values of the characteristics for which there remains one last pathway in the compatible space. In Step 4, we extract three of those final pathways that show that delaying the CO2 emissions peak requires to deploy more CDR.
In step 5, we represent the compatible space of CDR deployment and CO2 emission peak associated with the ocean acidification boundary. Inside this compatible space, the model always finds at least one pathway compatible with the chosen boundary. The limit of the domain gives the trade-off between the two pathway characteristics that are represented, assuming conditions on other key characteristics remain unchanged.
These successive steps are illustrated for one Pathfinder configuration only, but they were repeated for all 1500 configurations. Therefore, for a given set of conditions on pathway characteristics and planetary boundaries, we assess the existence of at least one compatible pathway in all individual physical states of world (i.e. configurations). We define the safety level (in %) as the percentage of configurations that find at least one emissions pathway compatible with one or more planetary boundaries. This pathway may well be different, however, across configurations.