Antarctic ice sheet contribution to sea-level rise is exceeding projections

Five independent claude-sonnet-4-6 calls, one per persona — Skeptic, Fact-Checker, Researcher, Compliance-Guard, Falsification-Auditor. Each writes its hardest objection from its own seat, paired with the methodological resolution it would accept. Run on the static catalogue spec Jun 3, 2026; a live council for any topic is at /try.

1. Skeptic #01
   Raised

   The most plausible alternative is that the apparent exceedance is an attribution artifact rather than a genuine Antarctic acceleration. Joining total GMSL records from multi-station tide gauges (/api/sealevelmulti) with /api/antarctic data without applying gravitational-rotational-deformational (GRD) sea-level fingerprinting cannot isolate the Antarctic ice sheet's contribution from concurrent accelerations in Greenland mass loss, steric thermal expansion, or terrestrial water storage — all of which project distinct but overlapping spatial patterns onto tide gauge networks. Critically, NSIDC sea ice extent measures floating ice area, which by Archimedes' principle contributes zero to GMSL, making it an invalid proxy for grounded ice sheet mass balance in this attribution chain. Additionally, GRACE/GRACE-FO Antarctic mass balance estimates carry a ±100–150 Gt/yr systematic uncertainty depending on which glacial isostatic adjustment (GIA) model is applied (ICE-6G_C vs. W12 vs. IJ05_R2), translating to ±0.28–0.42 mm/yr SLR spread that alone could explain the claimed exceedance.

   Resolved

   Apply the Frederikse et al. (2020, *Nature*, doi:10.1038/s41586-020-2591-3) budget-closure fingerprinting framework: decompose /api/sealevelmulti tide gauge records into source contributions using GRACE-FO JPL RL06 mascon solutions for Antarctic and Greenland ice sheets separately, Argo-derived steric expansion from /api/ocean OHCA fields, and GRACE terrestrial water storage. Run the GIA sensitivity ensemble across ICE-6G_C(VM5a), W12, and Geruo13 models; the hypothesis requires the Antarctic mascon term to exceed 4.0 mm/yr on a 5-yr rolling basis even at the lower GIA-correction bound. If the Antarctic mascon contribution falls within 3.0–4.0 mm/yr after GIA ensemble averaging and the residual attributed to "Antarctica" in the naive join is instead absorbed by the steric or Greenland fingerprint terms, the hypothesis is falsified at its attribution step before any projection comparison is warranted.

2. Fact-Checker #02
   Raised

   The dominant measurement uncertainty comes from GRACE/GRACE-FO Glacial Isostatic Adjustment (GIA) corrections, which carry a systematic bias of ±40–80 Gt/yr (≈ ±0.11–0.22 mm/yr sea-level equivalent) for Antarctica; combined with retrieval noise, the 1-sigma budget on a 5-year Antarctic-specific trend reaches ±0.25–0.35 mm/yr. More critically, the hypothesis misattributes the total observed global mean sea-level rise rate (~3.6–3.7 mm/yr per satellite altimetry) to Antarctica alone — AR6 WGI Table 9.5 and the IMBIE ensemble assign the Antarctic ice sheet only ~0.4–0.6 mm/yr of current contribution, rendering the stated SUPPORTS threshold of >4.0 mm/yr roughly an order of magnitude above any plausible Antarctic-specific signal. Furthermore, NSIDC sea-ice extent (passive microwave, ~15 km resolution) records floating sea-ice area, not grounded ice-sheet mass, and contributes zero information to sea-level budget attribution, making its inclusion as a convergence source a categorical measurement error.

   Resolved

   The metric must be redefined strictly as grounded ice-sheet mass loss converted to GMSL contribution (mm/yr), isolated via GRACE-FO RL06 mascon solutions (JPL or CSR) with GIA uncertainty propagated across at least three models (ICE-6G_C, W12, Geruo13) to produce 5-year rolling trends with explicit ±2-sigma confidence intervals whose half-width (~0.5–0.7 mm/yr) sets the minimum detectable difference. Against the corrected AR6 SSP2-4.5 Antarctic-specific baseline of ~0.4–0.7 mm/yr (2020–2040 likely range), testable thresholds should be reset to SUPPORTS >0.9 mm/yr and FALSIFIES <0.35 mm/yr — a band exceeding the combined GRACE-FO uncertainty budget — with NSIDC sea-ice extent removed and replaced by IMBIE ensemble mass-balance estimates or CryoSat-2 ice-volume retrievals converted to mass using a firn-densification model (e.g., IMAU-FDM or MARv3.11).

3. Researcher #03
   Raised

   The critical uncontrolled confounder is steric sea-level rise — thermosteric and halosteric ocean volume expansion driven by non-Antarctic forcing (ENSO-phase heat redistribution, Atlantic Multidecadal Oscillation, and secular anthropogenic ocean warming) — embedded unresolved within the tide gauge records used to estimate Antarctic contribution. Tide gauges measure total relative sea level and cannot distinguish barystatic (mass-addition from ice melt) from steric (density-change) components; because global ocean heat content has been rising at ~10 ZJ/yr independently of Antarctic ice dynamics, attributing the total tide gauge trend acceleration to Antarctic mass loss without first removing the steric signal will systematically overstate the ice-sheet contribution. This aliasing channel is large enough — steric anomalies can add 1–2 mm/yr regionally during positive PDO or El Niño episodes — to plausibly explain the entire claimed 0.9–1.4 mm/yr excess above the AR6 SSP2-4.5 central estimate.

   Resolved

   The steric component must be subtracted from each tide gauge record prior to Antarctic attribution using gridded thermosteric and halosteric anomaly fields derived from Argo profiling floats — specifically the NOAA/NCEI Global Ocean Heat Content product (WOA23, 0–2000 m thermosteric sea-level anomaly) or the Met Office EN4.2.2 dataset. Residual barystatic trends should then be cross-validated against GRACE/GRACE-FO JPL RL06M mascon solutions restricted to Antarctic ice-sheet drainage catchments, with the glacial isostatic adjustment correction uncertainty explicitly bounded across the Caron et al. (2018) probabilistic GIA ensemble (rather than a single Earth model such as ICE-6G_C) to confirm that the claimed mass-loss acceleration survives both steric removal and GIA model sensitivity analysis before any comparison to AR6 projections is declared significant.

4. Compliance-Guard #04
   Raised

   The downstream prediction that "coastal-property tenure exposures revise upward by decades, not years" directly implicates SEC Rule 10b-5 (17 CFR § 240.10b-5) and the SEC's March 2024 climate-risk disclosure framework if this pre-threshold rate estimate is embedded in material representations about asset valuations, REIT portfolio risk, or municipal bond prospectuses for coastal issuers. Simultaneously, IFRS S2 and EU CSRD require that physical-risk disclosures rest on "reliable and verifiable" data rather than preliminary experimental estimates; citing a 4.5–5.0 mm/yr figure as observed fact — when the 5-year sustain window required by the SUPPORTS threshold has not yet elapsed and the Antarctic-specific contribution has not been formally disaggregated from the multi-source composite — would constitute a material misstatement under both regimes. The additional channel of risk is NFIP/FEMA actuarial reliance: if this hypothesis is cited in support of revised Flood Insurance Rate Map elevations or Advisory Base Flood Elevation changes before peer review, affected property owners could face premium and title-insurance consequences anchored to a statistically unconfirmed rate.

   Resolved

   No downstream regulatory or fiduciary reliance on the 4.5–5.0 mm/yr figure is permissible until the following gating conditions are satisfied: (1) the Antarctic-specific contribution is formally disaggregated using an independent altimetry or gravimetry ground-truth dataset (e.g., GRACE-FO mascon solutions independently cross-validated against ICESat-2 elevation change) and confirmed to exceed 4.0 mm/yr across a completed, non-overlapping 5-year rolling window; (2) the methodology is documented in a peer-reviewed or equivalent technically-reviewed publication that explicitly attributes the excess rate to Antarctic mass balance rather than thermal expansion or other contributors conflated by the multi-source join; and (3) any disclosure using this estimate must carry the explicit qualifier: "This rate reflects a preliminary experimental estimate that has not yet crossed the pre-registered SUPPORTS threshold; it must not be used as a basis for property valuation, insurance underwriting, securities disclosure, or regulatory compliance until independent validation is complete."

5. Falsification-Auditor #05
   Raised

   The stated FALSIFIES band (3.0–4.0 mm/yr) is drawn from total global mean sea-level rise, not from the Antarctic-specific contribution the metric claims to measure. Current GRACE/GRACE-FO estimates put Antarctica's isolated contribution at roughly 0.4–0.8 mm/yr — roughly one-tenth of the falsification floor — so no 5-year rolling average of an Antarctic-only signal can naturally enter the 3.0–4.0 mm/yr band under any plausible null. Compounding this, inter-annual snowfall variability over the Antarctic ice sheet introduces ±100–200 Gt/yr of mass-balance noise (≈ ±0.3–0.6 mm/yr GMSL-equivalent), and reanalysis-ensemble spread in surface mass balance adds another ±30%, meaning even a correctly scoped Antarctic-only metric carries sufficient variance to blur any threshold set near its actual central value but cannot possibly reach 3.0 mm/yr.

   Resolved

   Redefine the metric explicitly as the Antarctic ice-sheet-only contribution (GRACE mascon solutions from JPL, CSR, and GSFC processed identically), set the FALSIFIES threshold at < 0.3 mm/yr sustained over 5 years (i.e., a deceleration below the pre-2010 mean) and the SUPPORTS threshold at > 1.2 mm/yr (materially above the AR6 SSP2-4.5 Antarctic-only central projection of ~0.7 mm/yr for the current decade), then validate reachability by running 10,000 Monte Carlo draws from the AR6 Antarctic contribution posterior to confirm both thresholds lie outside the null's 95% envelope; any residual ambiguity from GIA-model spread should be addressed via a sensitivity sweep across the three leading GIA reconstructions (ICE-6G, W12, Bedmap) to bound the irreducible attribution floor.

Unlike the static stress tests above (synthesised against the frozen catalogue spec), this is what a 3-voice council found in the most recent biweekly review. Refreshed on the 1st and 15th of each month at 09:00 UTC. Each voice runs one bounded web search via Anthropic's web_search_20260209 tool, cites what it finds, and recommends a verdict. The verdict diverges from the curated catalogue status (monitoring) — the synthesis below explains why.

Agent draft incorporating the 9 cited findings from the live council above. Not auto-merged — surfaces here for human review. To accept, open a PR editing site/src/_data/hypotheses.json with the revised fields below. To reject, ignore and the proposal will refresh on the next council run.

1. monitoring

   May 30, 2026 · added to catalogue at status "monitoring"

• FHFA, acting on revised Antarctic SLR trajectories, would require Fannie Mae and Freddie Mac to apply a 15–20% sea-level stress haircut to appraised collateral values for conforming loans on properties below 2-meter NAVD88 elevation along the Gulf and Atlantic seaboards — affecting an estimated $900B in outstanding GSE-backed collateral — within 24 months of the projection revision entering regulatory guidance.
• Guy Carpenter and Munich Re NatCatSERVICE would revise Atlantic and Gulf Coast expected-annual-loss models upward 18–25%, widening catastrophe reinsurance rate-on-line 50–90 bps for affected coastal treaties at the next two January/June renewal cycles.
• FEMA's National Flood Insurance Program actuarial shortfall, currently ~$20.5B, would widen by an additional $9–14B under a confirmed 4.5–5.0 mm/yr Antarctic contribution scenario, forcing Congress to accelerate NFIP premiums 20–30% above the existing Risk Rating 2.0 glide path or seek supplemental appropriation at the next statutory reauthorization window, expected by 2027.
• Moody's Investors Service would place general-obligation bonds for high-exposure low-elevation coastal municipalities — including Miami Beach FL, Norfolk VA, and Atlantic City NJ — on negative-outlook watch within 18 months of peer-reviewed confirmation, initiating a 1–2-notch downgrade path that would widen spreads 40–70 bps on their combined ~$18B in outstanding GO debt.
• Plaintiff-side economic experts in pending U.S. state climate-attribution suits — including Hawaii v. Sunoco LP and City of Annapolis v. BP — would revise present-value coastal-damage liability estimates upward by $30–80B within 18 months of peer-reviewed confirmation, materially strengthening causation arguments before trial courts that have already cleared Daubert and 12(b)(6) thresholds.