There is no single best mitigation pathway. The right choice depends on what you care about most.
If aggressive mitigation were uniformly best, every row of the table below would say SSP1-1.9. It doesn't. The pattern of where the framework's per-observable best-pathway calls land — and where they refuse to land — is itself the headline.
The seven calls
| Observable | Best pathway | Why |
|---|---|---|
| CO₂ trajectory (#23) | SSP1-1.9 / SSP1-2.6 | Deepest sub-rate β post-2024; latest β=1 forward crossing. |
| Sea-level rise (aggregate) (#24) | SSP1-1.9 | Wins three orthogonal criteria with rank-sum 7 of 24. |
| Permafrost onset (#26) | SSP1-2.6 | Delays first regime-change to median 2033.5 vs 2027.5 under SSP5-8.5 — about 5 years gained. |
| Glaciers (#24) | SSP1-1.9 | Sub-rate everywhere, but rate scales with forcing — mitigation buys decades of glacial retention. |
| Sea-ice (#16/17) | No SSP qualifies | The σ_cross paradox — aggressive mitigation widens cross-model disagreement. |
| AMOC (#16/29) | No SSP qualifies | Same σ_cross paradox; the σ_cross-tightest SSP is SSP5-8.5 but is meaningless physically. |
| Amazon NPP (#16) | Inconclusive | 5/26 models project decline under SSP1-2.6, similar under SSP5-8.5 — genuine model disagreement. |
Climate Mitigation Atlas — \(\beta\) by observable × emissions pathway
Each cell's colour is one number — the rate exponent \(\beta\). Green = stoppable (returns to rest). Orange = super-rate. Red = locked-in.
Click any cell for the full reading: \(\beta\), cross-model dispersion, theorem anchor, and the source code on GitHub.
How to read this chart · what the SSPs mean
The axes
- Rows: 8 climate observables — what is changing.
- Columns: 5 emissions pathways from "very aggressive mitigation" (left) to "no mitigation" (right).
- Cell colour: the rate exponent \(\beta\). Below 1 is stoppable; above 1 is locked-in.
- Cell label: the actual \(\beta\) value cross-model median.
The five emissions pathways (SSPs)
- SSP1-1.9 — ~1.5°C. Aggressive net-zero by mid-century. Paris lower bound.
- SSP1-2.6 — ~2°C. Moderate mitigation. Net-zero by ~2070.
- SSP2-4.5 — ~2.7°C. Current policies, middle-of-the-road.
- SSP3-7.0 — ~3.6°C. Regional rivalry, fragmented action.
- SSP5-8.5 — ~5°C. Continued fossil-fuel growth, no mitigation.
SSPs (Shared Socioeconomic Pathways) are the IPCC's standard set of emissions futures. The number after the dash is the radiative forcing in 2100 (W/m²).
What this table tells you
Aggressive mitigation works for the forcing-driven cascades
CO₂ trajectory, aggregate sea-level rise, permafrost onset, glaciers — all four have aggressive-mitigation pathways as the unambiguous best call. Mitigation works on these; the framework's call agrees with the IPCC literature on the direction of effect.
It backfires on robustness for sea-ice and AMOC
Two observables have no SSP that clears the framework's three criteria — and in both cases the reason is the same: cross-model disagreement widens under aggressive mitigation. The σ_cross paradox. Read the paradox →
Some risks are simply locked in
Antarctic ice-sheet, Greenland ice-sheet, permafrost regime change (already crossed at +0.36 K) — none of these have a "best pathway" because no pathway exits the locked-in regime. Mitigation reduces magnitude and timing, not regime. See locked-in risks →
Amazon NPP is genuinely inconclusive
The framework declines to call this one. 5/26 models project decline under aggressive mitigation; 5/25 models project decline under no mitigation. The difference is not the SSP — it is which model you ask. Honest reporting under Law V.
How "best" is decided
For each observable, the framework tests three orthogonal criteria:
- Tightness of cross-model agreement (smallest σ_cross — Law III).
- Smallest projected rate (smallest mean ρ).
- Furthest below the rest threshold (β farthest below 1 — Theorem 1).
For aggregate sea-level rise the criteria converge — SSP1-1.9 wins rank-sum 7 of 24 with clear separation from SSP5-8.5 at 17 of 24. For sea-ice the criteria diverge: σ_cross is best under high forcing (paradox), β CI clears 1 under mid–high forcing, and the no-PELT-pre-2050 criterion fails everywhere. "No SSP qualifies" is itself the framework's reading, not a methodological failure.
Source: scenario_fan_synthesis.md