RAPID forward-extrapolation block

Source: RAPID-MOCHA-WBTS, 2004–2024 monthly (n = 240). Φ range [6.64, 24.99] Sv (mean 16.97, std 3.20). log²(\(\Phi_{\max}/\Phi_{\min}\)) = 1.755 — Theorem 3 floor binds.

5 rate-perturbation scenarios over 50-yr forward extrapolation:

scenarioκ_perturbed (Sv/yr)Φ@50yr median (Sv)β GLS68% CIPELT cps
−50% rate−0.04616.66−2.138[−3.29, −0.98]13
−25% rate−0.06915.49+0.158[−0.98, +1.29]16
0 (RAPID drift)−0.09213.72+1.177[+0.22, +2.13]17
+25% rate−0.11513.18+2.468[+1.62, +3.31]23
+50% rate−0.13912.04+2.349[+1.64, +3.06]24

CMIP6 SSP fan (msftmz; Pangeo catalogue)

SSPn_modelsβ GLS medianσ_crossweakening rate (Sv/decade median)
SSP1-1.90— (no models in cache)
SSP1-2.613+1.2572.526−0.519
SSP2-4.514+0.5531.730
SSP3-7.00— (no models in cache)
SSP5-8.519+0.3091.545

Same σ_cross paradox as sea-ice

σ_cross is widest under SSP1-2.6 (2.526) and tightest under SSP5-8.5 (1.545). σ_cross-best is SSP5-8.5 but is meaningless physically — it is the σ_cross paradox repeated. Read the paradox treatment.

Source: scenario_fan_amoc_v2.py + amoc_observational_rapid.py