River discharge anomalies precede coastal sea-level station readings

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 parsimonious alternative is that a shared large-scale climate forcing—specifically ENSO—independently drives both river discharge anomalies and coastal sea-level anomalies through distinct physical pathways at different response timescales, producing a spurious apparent lead-lag between the two without any direct causal link. ENSO-driven precipitation anomalies can elevate basin discharge within 1–3 months of an El Niño peak, while ENSO-driven steric expansion, wind-driven Kelvin wave propagation, and altered gyre circulation elevate coastal sea level along the U.S. Pacific and Gulf coasts 3–8 months after the same forcing event; this differential response alone spans exactly the 2–6 month window claimed. The cross-correlation between USGS streamflow and NOAA tide gauge records would then reflect ENSO's differential fingerprint on the two systems, not any hydraulic or freshwater-mass signal propagating from river mouth to tide gauge.

   Resolved

   Regress the monthly Niño 3.4 index (NOAA CPC, 1950–present) and the PDO index (NCEI) out of both the USGS discharge anomaly and the NOAA tide gauge anomaly time series at all stations using a distributed-lag OLS with lags 0–12 months before computing the river–sea-level cross-correlations on the residuals; this is a standard partial cross-correlation controlling for the common climatic driver. The decision rule is: if the partial cross-correlation peak |r| on ENSO/PDO-residualized series falls below 0.15 or the fraction of stations meeting |r| > 0.4 drops below 30%, the original signal is attributable to shared climate forcing rather than direct river-to-ocean propagation. A further discriminating test is a station-pair specificity check—genuine hydraulic propagation requires that the strongest lags appear only at geographically co-located river-mouth/tide-gauge pairs, not at spatially randomized pairs from the same climate zone, which a permutation test on pair identity (holding climate-residualized series fixed) can quantify with a false-discovery-rate threshold of q < 0.05.

2. Fact-Checker #02
   Raised

   NOAA CO-OPS tide gauges achieve ±3 mm precision on 6-minute water level data (1-sigma instrument noise), but after de-tiding and seasonal removal, monthly residual sea-level anomalies carry ≈10–15 cm RMS dominated by meteorological forcing—inverse-barometer effect (≈1 cm/hPa) and wind-driven setup (5–30 cm during synoptic events)—not instrument noise; the freshwater-driven steric or dynamic coastal signal attributable to river discharge is estimated at 2–5 cm near major river mouths, capping the physically achievable |r| at roughly 0.13–0.33, a range that simultaneously overlaps both the SUPPORTS threshold (

   Resolved

3. Researcher #03
   Raised

   The primary uncontrolled confounder is large-scale interannual climate variability—specifically ENSO—which simultaneously forces above-normal continental precipitation (elevating river discharge) and drives coastal sea-level anomalies through wind-driven Ekman transport and steric (thermosteric/halosteric) adjustments, with no causal link between the two pathways. Because ENSO teleconnections operate on 3–9 month timescales and affect both river basins and adjacent coastal oceans via independent atmospheric and oceanic channels, the cross-correlation analysis will detect a common-driver signal that masquerades as freshwater propagation, inflating |r| and biasing the apparent lag into the 2–6 month window even when river outflow has negligible influence on local tide-gauge readings.

   Resolved

   Partial cross-correlations should be computed after pre-whitening both the discharge and sea-level anomaly series by regressing out the Niño 3.4 SST index and the PDO index (both available at monthly resolution from NOAA CPC, e.g., https://www.cpc.ncep.noaa.gov/data/indices/ersst5.nino.mth.91-20.ascii), as well as the station-relevant atmospheric pressure field from ERA5 (Copernicus CDS variable "mean_sea_level_pressure," dataset ERA5 monthly averages, to capture the inverse-barometer effect); lagged versions of these covariates (0–12 months) should enter a linear filter applied to each paired series before the residual cross-correlation is computed, ensuring that a significant peak at the predicted lag reflects the freshwater-transport mechanism rather than shared climate forcing.

4. Compliance-Guard #04
   Raised

   The downstream application explicitly names mortgage and insurance pricing for coastal property, which directly implicates NFIP actuarial-soundness requirements under 44 CFR Parts 61–62 and FEMA's Risk Rating 2.0 methodology, as well as state insurance department rate-filing standards that require actuarially credible, independently validated models before they may be embedded in premium calculations or Special Flood Hazard Area (SFHA) determinations. Premature citation of this cross-correlation signal as an established predictive tool could also trigger SEC Rule 10b-5 liability if flood-risk assumptions based on this unvalidated lag structure are embedded in prospectuses or risk disclosures for coastal-property mortgage-backed securities, constituting a material misrepresentation of the underlying hazard model.

   Resolved

   This hypothesis must not be represented as validated predictive infrastructure for any actuarial, underwriting, or securities-disclosure purpose until it formally crosses the SUPPORTS threshold—cross-correlation peak |r| > 0.4 at lag 60–180 days in ≥60% of N≥30 independent USGS–NOAA station pairs, with phase-randomization null rejected at p<0.05 across rolling 5-year windows. All interim outputs must carry an explicit disclaimer that the signal is a research-stage hypothesis under active evaluation, that no flood-risk premium, SFHA boundary revision, mortgage underwriting criterion, or securities offering document may cite it as a validated lead indicator, and that formal operational adoption requires independent peer review, FEMA/NFIP actuarial audit, and state insurance department approval of any resulting rate-change methodology before deployment.

5. Falsification-Auditor #05
   Raised

   Both river discharge anomalies and coastal sea-level anomalies are jointly driven by large-scale atmospheric forcing — principally ENSO, NAO, and regional precipitation teleconnections — so even a null world with zero direct hydrological causal pathway would still produce spurious cross-correlations of |r| ≈ 0.20–0.40 at seasonal lags, well above the stated FALSIFIES ceiling of 0.15. The phase-randomization null used in the SUPPORTS arm destroys all temporal autocorrelation structure and therefore dramatically underestimates this atmospheric co-forcing noise floor, meaning the FALSIFIES band (|r| < 0.15) is practically unreachable under any realistic null. The experiment is consequently one-sided: outcomes in the range |r| = 0.15–0.40 — the most likely result — neither support nor falsify, and the falsification arm provides no actual discriminating power against a confounded null.

   Resolved

   Build a physically-grounded null ensemble by pairing each discharge record with 500 surrogate sea-level series drawn from tide gauges that share the same ENSO/NAO forcing regime but whose drainage basins are geographically swapped (e.g., Pacific-basin river paired with Atlantic tide gauge and vice versa), preserving atmospheric co-variance while eliminating direct hydrological coupling; the empirical 95th-percentile |r| of this surrogate distribution — expected near 0.25–0.35 — then defines the operational FALSIFIES ceiling, replacing the arbitrary 0.15 with a threshold that is actually reachable under the null. Supplement this with a sensitivity sweep across five reanalysis-forced discharge reconstructions (ERA5, MERRA-2, JRA-55, CFSR, 20CR) to quantify model-spread uncertainty on the lag peak, and add a direct-validation arm comparing correctly-paired versus drainage-swapped station pairs as an internal control; if the correctly-paired correlations cannot be statistically separated from the swapped-pair null at p < 0.05, the FALSIFIES condition is triggered.

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 (forming) — 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. forming

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

• NOAA National Water Center, upon operational integration of USGS discharge-anomaly signals into coastal flood guidance products, would extend probabilistic inundation advisory lead times from the current 3-5 day tide-gauge window to 60-180 days, enabling FEMA regional offices and state emergency managers to trigger pre-event resource mobilization 15-30× earlier within 18 months of a validated operational deployment.
• Swiss Re Capital Markets and Aon Securities, structuring next-generation parametric catastrophe bonds for Atlantic and Gulf Coast river-estuary exposures, could reduce trigger-mismatch basis risk by 15-25 bps per transaction by incorporating USGS discharge exceedance thresholds as co-triggers alongside tide-gauge readings, with first issuances achievable within two annual ILS renewal cycles.
• Fannie Mae and Freddie Mac, updating climate-risk underwriting frameworks under FHFA direction, could impose conforming-loan spread widening of 20-50 bps on coastal properties at correlated river-estuary junctions within FEMA Special Flood Hazard Areas, affecting an estimated $340 billion in outstanding mortgage collateral within 24 months of FHFA guidance issuance.
• FEMA Risk Rating 2.0 actuarial teams, recalibrating NFIP premiums using discharge-precursor flood-frequency data, could increase annual policy costs by 8-15% for approximately 1.2 million river-adjacent coastal properties, shifting an estimated $2-4 billion in annual flood exposure from insured to uninsured status within the next biennial NFIP actuarial review cycle.
• Moody's Investors Service and S&P Global Ratings, incorporating 2-6 month discharge lead-time signals into climate credit overlays, could widen general obligation and revenue bond spreads by 10-30 bps for coastal municipal issuers with material river-correlated flood exposure, affecting an estimated $180 billion in outstanding coastal-jurisdiction paper by the next annual credit surveillance cycle.