Summary
Eight methodological recommendations distilled from the climate arc, anchored to specific theorems. Source: domains/climate/docs/methodological-recommendations.md.
1. Choosing observables (Theorem 4)
Prefer absolute Φ over anomaly-form. Anomalies cross zero; \(\log|\Phi|\) diverges. Atmospheric and land reservoirs in the energy-budget analysis fail brake-p Layer-A specifically because the reservoirs are anomaly-form. Strictly monotone cumulative quantities (ice mass, integrated OHC) give cleaner brake fits.
Be explicit about reparametrisation choice. By Theorem 4(b), \(\Phi \mapsto \Phi^a\) shifts \(\beta\) according to \(\beta \mapsto (\beta + a - 1)/a\). When comparing β values across studies, verify the same reparametrisation. The canonical presentation is the natural Φ — neither logged nor squared.
2. Constructing shadow ensembles (Theorem 6)
Verify shadow admissibility (S1)–(S3). Two observables that measure different aspects of a complex system are not shadows of one signal — they are observables of two different cascades. Stratospheric-ozone (instance #25): hole-minimum responds to polar-vortex chlorine chemistry; zonal-mean responds to global photochemical balance. \(\mathfrak{A} = 22.4 \gg \tau_{T3}\) — honest physical-shadow-mismatch, not poor methodology.
SOI is a true ENSO shadow; Niño regions are not. The four Niño sub-region indices (1+2, 3, 4, 3.4) are spatial sub-regions of one SST field with correlated noise (S2 fails). SOI (Southern Oscillation Index, SLP-based) is independent.
3. Interpreting σ_cross (Theorems 6 + 8)
σ_cross must be paired with forcing context. Without forcing-context, σ_cross is uninterpretable. The σ_cross paradox is the cleanest illustration: aggressive mitigation produces wider σ_cross for sea-ice and AMOC because internal variability dominates external forcing under low-emission scenarios. Read the full paradox treatment.
Cross-component σ_cross is meaningful when components are framework-admissible shadows of a parent cascade with linear composition (Theorem 5, ρ-additivity).
4. Reading β values (Theorems 1 + 8)
β = 1 is the only universal threshold. No imported climate constant — no °C threshold, no ppm target.
Report β with R² and Theorem-3 precision floor. A β of 0.131 with \(R^2 = 0.05\) and log-range 0.7 (instance #24, GMSL) is precision-floor-bound. The framework is honest about its own resolution.
β = 0 means forced-rate degeneracy, not "regular cascade". By Theorem 8, when \(\rho\) is determined by external forcing rather than by \(\Phi\), the brake form admits no unique \(\beta\). Forced trajectories give \(\beta \approx 0\) by construction; that is the framework's "I cannot answer this question" signal.
5. Component decomposition (Theorem 5)
Decompose linear-composed observables. GMSL = steric + glaciers + GrIS + AIS + TWS. The aggregate hides the component story: aggregate β = -1.12 (SSP1-1.9, sub-rate) but AIS β = +3.76 (super-rate, locked-in). Components are framework-admissible shadows by Theorem 5.
Identify the dominant brake-rate driver. AIS dominates GMSL post-tipping cascade rate (β = 3.21–3.76 across all SSPs). Mitigation on AIS-specific feedbacks matters more than aggregate sea-level metrics.
6. Cadence and class (Theorems 7 + 9)
Report β with explicit cadence. Theorem 7 cadence-stamps every brake reading. The same cascade can be brake-class at one cadence and resonance-class at another (Mid-Pleistocene Transition: brake-rate stable, but the dominant frequency shifts 41-kyr → 100-kyr at the kyr cadence).
Classify cascade-class vs resonance-class first. ENSO is oscillatory, not brake-cascade. The framework correctly reports that ENSO doesn't have rate-vs-magnitude scaling brake-p was built to detect.
7. Honest scope (Law V)
State framework limitations explicitly. Every published finding carries an \(\mathscr{A}\) scope-reporter (Theorem 12). The arc's principal contribution to climate is itself a Law-V finding: a unified cross-observable diagnostic in domain-internal units, paired with the explicit identification of where the diagnostic does and does not apply.
8. Reproducibility (Pattern-4)
Open code, open data, framework-native units. Every climate experiment writes a TXT human summary + JSONL payload. All code lives under domains/climate/experiments/; all results under domains/climate/results/.