Chokepoint Methodology

What we track

WorldMonitor has 13 monitored waterways in the canonical chokepoint registry and the GetChokepointStatus RPC. The status RPC returns one row for each canonical id: suez, malacca_strait, hormuz_strait, bab_el_mandeb, panama, taiwan_strait, cape_of_good_hope, gibraltar, bosphorus, korea_strait, dover_strait, kerch_strait, and lombok_strait.

Only seven of those 13 currently carry a live oil/gas flow estimate from energy:chokepoint-flows:v1, because only those seven have an EIA baseline id in the current seeder:

The other six canonical waterways still receive transit summaries, warnings, AIS-disruption matching, threat classification, disruption score, and war-risk tier, but flowEstimate is absent until a baseline-backed flow model is added.

Live Flow

Live flow is computed by scripts/seed-chokepoint-flows.mjs on the seven-item subset above:

1. Read PortWatch history from supply_chain:portwatch:v1.
2. Choose the signal: use capTanker deadweight tonnage when at least half of the prior-window days have DWT coverage; otherwise use tanker counts.
3. Compute current7d as the average of the latest seven days.
4. Compute baseline90d as the average of the previous window, up to 90 days (history.slice(-97, -7)), requiring at least 20 baseline days.
5. Compute flowRatio = clamp(current7d / baseline90d, 0, 1.5).
6. Compute currentMbd = baselineMbd * flowRatio, rounded to one decimal.

The published ratio is therefore a recent 7-day average against a prior up-to-90-day rolling baseline. The baselineMbd field is the annual EIA 2023 reference level; it is used to convert the observed ratio into mb/d. Flow ratios can publish from 0% to 150% of baseline after the clamp.

Live flow is eligible only when the PortWatch series has at least 40 total days of history. After that gate, the seeder still requires at least three recent days, at least 20 prior-window baseline days, and a non-thin baseline: at least 1 DWT-day when using tanker deadweight tonnage, or at least 0.5 tanker-count when using vessel counts. Rows below those floors are omitted rather than reported as zero flow.

The disrupted boolean is separate from the color badge: it is true when each of the latest three individual days is below 85% of the same baseline90d.

Each published flow estimate may also include live hazard context from portwatch:disruptions:active:v1. The seeder looks for the nearest active GDACS RED or ORANGE alert within 500 km of the supported chokepoint and surfaces it as hazardAlertLevel and hazardAlertName. This is annotation only: hazard enrichment does not change currentMbd, flowRatio, or the disrupted calculation.

Score Badge

The public status field on ChokepointInfo is a traffic-light score badge: green, yellow, or red. It is not an operational closure-state label.

The score is:

disruptionScore = min(
  100,
  threatLevelWeight + warningComponent + aisComponent + anomalyBonus
)

Where:

• threatLevelWeight is the geopolitical baseline weight in the taxonomy below.
• warningComponent = min(15, activeWarnings * 5).
• aisComponent = min(15, maxCongestionSeverity * 5), where low/elevated/high AIS congestion severity maps to 1/2/3.
• anomalyBonus = 10 only when transit history shows a 50% or larger traffic drop against the prior 30-day transit-count baseline and the threat level is war_zone or critical.

The score badge thresholds are:

Threat Taxonomy

The geopolitical threat baseline is assigned per chokepoint in server/worldmonitor/supply-chain/v1/get-chokepoint-status.ts and weighted in server/worldmonitor/supply-chain/v1/_scoring.mjs.

Current assignments:

Transit Anomaly

Transit anomaly detection for anomalyBonus is based on compact transit-count history, not the live mb/d flow model. It compares the latest seven transit counts with the preceding 30 days converted to a seven-day baseline:

baselineAvg7 = (sum(days -37..-8) / availableBaselineDays) * 7
dropPct = round((baselineAvg7 - recent7) / baselineAvg7 * 100)
signal = dropPct >= 50 AND threatLevel in {war_zone, critical}

If signalled, the handler appends a description such as Traffic down 55% vs 30-day baseline, vessels may be transiting dark (AIS off). The anomaly path is gated before that comparison: it requires at least 37 days of compact transit-count history, and the normalized seven-day baseline must be at least 14 transits. Thin baselines return no anomaly signal.

Country Exposure

GetCountryChokepointIndex returns a separate exposure model for a country and HS2 chapter. It is PRO-gated and cached at supply-chain:exposure:{ISO2}:{HS2}:v1.

When bilateral HS4 import products are available, exposure is computed as:

productWeight = product.totalValue / totalSectorValue
routeCoverage = overlappingRouteIds / chokepoint.routeIds.length
contribution = routeCoverage * exporter.share * productWeight * 100

Contributions are summed per chokepoint across matching products and top exporters. For HS2 27, chokepoints with shockModelSupported = true receive a 1.5x energy-model boost capped at 100.

When bilateral product data is unavailable, the fallback score is:

exposureScore = overlappingCountryRouteIds / chokepoint.routeIds.length * 100

with the same HS2 27 supported-energy boost and 100 cap. The vulnerabilityIndex is the weighted top-three score:

top1 * 0.5 + top2 * 0.3 + top3 * 0.2

Provenance

Chokepoint status combines Redis-backed transit summaries, flow estimates, navigational warnings, AIS disruption matching, and the static threat taxonomy. If a canonical chokepoint lacks upstream transit coverage for the current cycle, transitSummary.dataAvailable is false and the response-level upstreamUnavailable flag is set so clients can render partial-coverage UI.

Refresh cadence:

• energy:chokepoint-flows:v1: Railway seed loop every 6 hours.
• supply_chain:chokepoints:v4: status warm-ping every 30 minutes.
• supply_chain:transit-summaries:v1: AIS relay compact transit summary.

Known Limits

• The seven live-flow rows are energy-baseline-backed; the other six monitored waterways do not yet publish mb/d estimates.
• The Dover Strait live-flow row currently maps to the EIA danish baseline id. Treat baselineMbd as the source reference and id as the canonical WorldMonitor row id.
• AIS and PortWatch coverage can degrade near heavily jammed or conflict-heavy regions. Missing upstream coverage is surfaced as unavailable data, not synthesized as zero traffic.
• Baselines are annual EIA reference levels used for mb/d conversion. The live ratio itself uses recent PortWatch observations against the prior rolling baseline window described above.

Corrections

See /corrections for the planned revision-log shape. If you spot a wrong number today, open a GitHub issue at the public repository.