/classify-anomaly - Multi-Proxy Discrimination Skill

Purpose

Apply the multi-proxy discrimination framework to classify geochemical anomalies as SEISMIC, CLIMATIC, VOLCANIC, or COMPOUND (CVSE). This is the core methodology for distinguishing earthquake signals from false positives.

Usage

/classify-anomaly <d18O_z> <d13C_z> [--MgCa Z] [--recovery YEARS] [--date YEAR]

Examples:

/classify-anomaly -2.46 +1.60                    # Basic two-proxy
/classify-anomaly -2.16 N/A --MgCa +1.60         # Missing δ13C, use Mg/Ca
/classify-anomaly -3.77 -1.18 --date 1275        # Check volcanic correlation
/classify-anomaly -3.6 -2.36 --recovery 46       # With recovery time

Discrimination Framework

Primary Decision Tree

Is there a significant anomaly (|z| ≥ 2.0)?
├─ NO → NORMAL (no event detected)
└─ YES → Check proxy coupling...
         │
         ├─ COUPLED (δ18O and δ13C both anomalous, same sign)
         │   └─ Coupling ratio < 2.0 → **SEISMIC CANDIDATE**
         │
         ├─ DECOUPLED (δ18O anomalous, δ13C normal or opposite)
         │   ├─ Volcanic correlation? → **VOLCANIC**
         │   └─ No volcanic correlation → **CLIMATIC** (drought/wet)
         │
         └─ δ13C ONLY anomalous (δ18O normal)
             └─ Check geogenic CO₂ sources → **SEISMIC CANDIDATE**

Proxy Rules

| Proxy | Seismic Signal | Climatic Signal | Volcanic Signal | |-------|----------------|-----------------|-----------------| | δ18O | Negative (deep water) | Variable | Negative (wet) | | δ13C | Positive > -8‰ (geogenic CO₂) | Negative < -10‰ (biogenic) | Variable | | Mg/Ca | Positive (old water) | Negative (dilution) | Variable | | Coupling ratio | < 2.0 (COUPLED) | > 3.0 (DECOUPLED) | > 3.0 (DECOUPLED) | | Recovery time | 5-71 years | 1-7 years | 1-3 years |

Coupling Ratio Calculation

Coupling ratio = |δ18O_z| / |δ13C_z|

< 2.0  → COUPLED (seismic)
2.0-3.0 → AMBIGUOUS
> 3.0  → DECOUPLED (climatic/volcanic)

Recovery Time Assessment

| Duration | Classification | Example | |----------|----------------|---------| | 1-3 years | VOLCANIC | Samalas 1257 recovery | | 3-7 years | CLIMATIC | Drought recovery | | 10-30 years | SEISMIC | Typical earthquake | | 30-71 years | MAJOR SEISMIC | Lapa Grande ~96 CE (71 yr) |

Temporal Shape (Mg/Ca)

| Shape | Onset Rate | Classification | |-------|------------|----------------| | SHARK FIN | > 0.5 σ/mm | SEISMIC (rapid onset) | | HUMP | < 0.5 σ/mm | CLIMATIC (gradual onset) |

Classification Output

## Anomaly Classification: [Date/Window]

### Input Metrics

| Metric | Value | Notes |
|--------|-------|-------|
| δ18O z-score | -2.46σ | Significant negative |
| δ13C z-score | +1.60σ | Marginal positive |
| Mg/Ca z-score | +2.25σ | Significant positive |
| Recovery time | 15 years | Extended |
| Coupling ratio | 1.54 | COUPLED |

### Discrimination Tests

| Test | Result | Interpretation |
|------|--------|----------------|
| δ18O significance | ✓ PASS | z ≤ -2.0 |
| δ13C coupling | ✓ PASS | Same direction, ratio < 2 |
| Mg/Ca elevation | ✓ PASS | Deep/old water signature |
| Recovery duration | ✓ PASS | > 10 years |
| Volcanic correlation | ✗ NONE | No major eruptions ±5 years |

### CLASSIFICATION: **SEISMIC CANDIDATE** (Tier 1)

**Confidence**: HIGH
**Evidence tier**: Tier 1 (multi-proxy)

**Rationale**:
- Coupled proxies (ratio 1.54) indicate Chiodini mechanism
- Elevated Mg/Ca confirms deep water mobilization
- Extended recovery (15 yr) rules out volcanic/climatic
- No volcanic forcing in time window

Evidence Tiers

| Tier | Requirements | Confidence | |------|--------------|------------| | Tier 1 | Multi-proxy (δ18O + Mg/Ca OR δ13C) + recovery > 10 yr + no volcanic | HIGH | | Tier 2 | δ18O + historical documentation OR single additional proxy | MODERATE | | Tier 3 | Single proxy OR weak correlation | LOW |

Volcanic Correlation Check

When --date is provided, check against major eruptions:

| Date Window | Major Eruptions | VSSI (Tg S) | |-------------|-----------------|-------------| | 1255-1260 CE | Samalas 1257 | 59.42 | | 1228-1235 CE | Unknown 1230 | 23.78 | | 1105-1115 CE | Unknown 1108 | 19.16 | | 1452-1458 CE | Kuwae 1452 | 31.06 | | 1808-1816 CE | Unknown 1809 + Tambora 1815 | 22.10 + 26.03 |

If date ±5 years overlaps major eruption AND proxies are DECOUPLED → VOLCANIC

Output Classifications

| Classification | Criteria | |----------------|----------| | SEISMIC | Coupled proxies, no volcanic, recovery > 10 yr | | CLIMATIC | Decoupled proxies, no volcanic | | VOLCANIC | Decoupled proxies, volcanic correlation | | CVSE | Compound event: volcanic + seismic + hydrological | | EQUIVOCAL | Insufficient evidence to classify |

CVSE (Compound Volcanic-Seismic Event) Criteria

All THREE must be present:

Volcanic forcing - Documented eruption in ice cores
Seismic signal - Mg/Ca or δ13C proxy confirmation
Hydrological expression - Documented floods or aquifer disruption

Known CVSEs:

1285 Italy (UE6 + earthquake + Genoa floods)
936 Belize (Eldgjá + seismic + global floods)
853 Pacific NW (Churchill VEI 6 + Cascadia S + drought)

Example Full Analysis

User: /classify-anomaly -3.77 -1.18 --date 1275 --recovery 3

Classification Analysis:

Input:
- δ18O z = -3.77σ (EXTREME NEGATIVE)
- δ13C z = -1.18σ (not significant)
- Date: ~1275 CE
- Recovery: 3 years

Tests:
- Coupling ratio: 3.77/1.18 = 3.19 → DECOUPLED
- Volcanic check: 1257 Samalas (59.42 Tg S) within 18 years
- Recovery: 3 years (volcanic range)

CLASSIFICATION: **VOLCANIC**

Rationale:
- Decoupled signature (δ18O >> δ13C)
- Temporal proximity to 1257 Samalas (largest eruption in 2000 years)
- Short recovery (3 yr) matches volcanic pattern
- NOT seismic despite extreme δ18O

This is the documented Samalas volcanic response at Yok Balum.

Important Notes

Missing δ13C - Use Mg/Ca as substitute coupling test
Uncertainty propagates - Single proxy = Tier 3 max
CVSE is rare - Requires ALL THREE components
Recovery time is diagnostic - "Order of magnitude gap" between volcanic (1-3 yr) and seismic (10-71 yr)
Update ANOMALY_CATALOG.md - After classification with approval