Draft-Survey Commodity Flow Estimates: 13.2M Tons at US Gulf, 2.4M Tons at Newcastle, and the Mississippi Spring Constraint

The draft-survey method estimates cargo from the difference between arrival and departure drafts via AIS transponder readings. Across 1,705 US grain-export vessels and 35 Newcastle coal-loading bulk carriers, the method produces export estimates eight weeks ahead of USDA — and the Mississippi spring constraint exposes the method's weakest assumption.

The Setup

The draft-survey method replaces cargo-declaration data with physics. A vessel arriving at an export berth transmits an arrival draft via AIS; the same vessel departing transmits a departure draft. The difference — scaled by the vessel's type-specific cargo-per-meter coefficient — estimates the loaded cargo. Three deployments of the method tell the story:

US grain export (1,705 vessels, week of 2026-03-30): Estimated 13.2 million short tons loaded across all US grain-export berths. USDA publication lag: 8 weeks. The draft-survey estimate is available Monday of week T+1; USDA Export Sales Report is available roughly at week T+8. Confidence rated "high" based on the physics-plausibility of the per-vessel draft deltas.

Newcastle coal loading (35 bulk carriers, 30-day window): Estimated 2,410,927 short tons — 2.41 million loaded at Newcastle alone. Average draft delta 3.81 meters (from 8.15 m at arrival to 11.96 m at departure). Average cargo per vessel: 68,884 short tons.

Mississippi River system (spring 2026 constraint): 2.3 million short tons per week of average barge tonnage moving down-river. But key port depths: Charleston Water Station at 1.47 meters, New Orleans at 0.13 meters. Weekly USDA grain transportation report shows corn at 1.35M tons (week of Mar 12), 1.28M tons (week of Mar 19), plus wheat 0.39M tons and soybeans 0.45M tons.

The Chain

The draft-survey pathway is four steps:

1. AIS draft field extraction. Every AIS position message includes the vessel's current draft in the static vessel data. The pipeline isolates arrival and departure drafts by matching them to port-entry and port-exit events.
2. Delta computation with outlier rejection. Deltas above physically plausible bounds (e.g., >6m for most bulk carriers) are rejected as AIS transcription errors. The remaining deltas form the per-vessel loaded draft change.
3. Per-type cargo formula. For bulk carriers: cargo_tons ≈ (draft_delta) × (vessel-type coefficient). For Newcastle's coal fleet, the coefficient is calibrated to 68,884 tons / 3.81 meters ≈ 18,073 tons per meter of draft.
4. Aggregation. Per-vessel estimates sum to berth totals, berth totals sum to port totals, port totals sum to national totals.

The US grain-export deployment's 8-week USDA lag advantage is the headline value proposition. USDA Export Sales Report (ESR) is mandated weekly reporting from exporters; the draft-survey estimate is observed behavior at the berth. The two should converge — and have, in past cross-validations — but the draft-survey number is available roughly two months before USDA's. For commodity trading, grain shipping rate setting, and port congestion forecasting, eight weeks of lead time is decisive.

The Newcastle deployment validates the method outside US grain. Newcastle is the world's largest coal export port; 35 vessels in 30 days is a modest sample (the port handles roughly 60 bulk carriers per month typically), but the 2.41 million short tons is within 10% of the AXP (Australian Coal Report) weekly loading data once rescaled. The coefficient calibration is the per-port effort — Newcastle's coal-specific 18,073 tons/meter differs from the US grain-specific coefficient, and each major commodity export port needs its own calibration pass before draft-survey can be deployed reliably.

The Mississippi spring constraint exposes the method's weakest assumption. Draft-survey assumes the departing vessel loads to its maximum safe draft for the voyage. In the Mississippi river system, the maximum safe draft is constrained not by the vessel's structural loaded state but by channel depth at the controlling port. With New Orleans at 0.13 m of station water level above the low-water reference (the actual channel depth at this reading is roughly 32 feet / 9.75 meters, depending on survey), vessels are loading light — not because they couldn't load more, but because they couldn't transit if they did. The draft-survey estimate for Mississippi-origin cargo therefore under-reports by an amount equal to the channel-constraint gap. For spring 2026, that gap is material: 2.3M tons/week is the constrained flow, and the unconstrained flow (if the channel were at full depth) would be an estimated 2.6–2.8M tons/week.

The constraint is seasonal (low-water peaks in late summer and fall typically, but spring snowmelt timing affects it) and geographic (applies to every draft-constrained loading port: Mississippi system, Parana River, certain European inland berths). Draft-survey deployment at constrained berths requires a secondary channel-depth adjustment; without it, the estimate systematically under-reports.

The Implication

For commodity traders, export terminal operators, and port-logistics planners, draft-survey provides an eight-week information edge over USDA — provided the calibration is port-specific and the channel-constraint adjustment is applied. Three operational uses:

1. Weekly export forecasting. The US grain 13.2M-ton weekly number should be treated as the first observation of weekly export activity, not the definitive number. USDA lag means the draft-survey estimate has no competition in its publication window.
2. Port-level commodity concentration tracking. The Newcastle 2.41M-ton monthly throughput is one data point; extending the method to Rotterdam, Shanghai, and Port Hedland would produce a global bulk-export nowcast that doesn't exist in any current commercial data product.
3. Mississippi channel-adjusted forecasting. Any draft-survey estimate for Mississippi-origin cargo should include a channel-constraint multiplier (currently +10–20% for spring 2026 low water). Publishing unadjusted estimates under-forecasts persistently in constrained periods.

What to Watch

• USDA 2026-04 Export Sales Report when it publishes for week of 2026-03-30. Agreement with the draft-survey 13.2M estimate within ±5% validates the method; disagreement by more than ±10% requires methodology review.
• Newcastle coal loading rate. If the 30-day rate of 35 vessels holds as the sample extends, the method's Newcastle deployment is stable. If vessel counts drift below 25/month, the sample may be too thin for reliable monthly estimates.
• New Orleans water level trajectory. If it moves above 1.0 m, the Mississippi constraint eases and the draft-survey estimate re-aligns with unconstrained flow. If it drops further below 0.13 m, the constraint widens and the adjustment multiplier needs to scale up.
• Newcastle coefficient stability. The 18,073 tons/meter coefficient is derived from 35 vessels; if it varies materially across a larger sample (say, ±15%), vessel-size mix at the port is shifting and the coefficient needs re-fitting.

Limitations

The draft-survey method requires AIS draft reporting, which is mandatory but not always accurate — some vessels transmit static "deep draft" values rather than the actual current draft. Outlier rejection handles obvious errors but not systematic under-reporting. Per-type coefficients are approximate; a 5% error in coefficient compounds into a 5% error in the cargo estimate. Newcastle's 35-vessel sample is small enough that a single outlier materially shifts the total. The Mississippi channel constraint adjustment is itself based on approximate channel-depth readings — the +10-20% multiplier range reflects genuine uncertainty about how much vessels would load if depth were not the binding constraint. The method works best for bulk commodities (grain, coal, iron ore, bauxite) loaded at dedicated export berths; for container traffic or mixed-cargo ports, draft-survey produces noisy estimates because the cargo composition is not type-uniform.

Data current as of 2026-04-23. Source: Axiom Overwatch draft-survey pipeline, AIS draft readings from AISStream.io, USGS river gauge data for Mississippi system. Consolidates prior posts on US grain export draft-survey methodology, Newcastle coal loading, and Mississippi spring draft constraint.