Port closure probability: the answer the rule doesn't give and the AI climate agent does

The container is on board, the ship is anchored and the customer is asking when it arrives. The harbor master's office rule says the port closes with waves above 2 meters. The city forecast says "rain in the afternoon." Neither answers the only question that matters: will the port operate tomorrow?

It doesn't answer because the port doesn't close when the threshold is hit. It closes when the person in the decision-making seat understands there is operational risk for that maneuver, that berthing or that open hold. This risk often appears before the threshold. Other times, only after it.

This text explains how i4sea turns this human decision into probability, and how the AI climate agent delivers that probability, the ship queue and the operational impact to whoever depends on the port from the cargo side.

In this article

• Why the rule's yardstick doesn't predict closure
• Multifactor analysis: how to find the pattern that closes the port
• What the AI climate agent answers in natural language
• What changes for whoever depends on the port
• Proof: ship waiting time in Santos dropped from 7 to 3 days
• Frequently asked questions
• Test the Port Logistics AI Climate Agent

Why the rule's yardstick doesn't predict closure

The closure rule is a simple piece of paper. It says the port closes with waves above 1.8 meters, or 2 meters, and stops there. Nature doesn't stop there.

There isn't a single wave; there's a wave train. Within it there are 3-meter waves and 1-meter waves, with different periods, different directions and different energies. And energy is what matters, because it defines how much force the sea transfers to the hull, the cable and the berth.

On top of that come wind, visibility and rain, all at the same time. It's this combination that the person responsible for the maneuver, for the berth operation or for the open hold evaluates when deciding to stop before the event arrives.

By analyzing all of a port's maneuvers, without excluding the good ones or the bad ones, i4sea finds both sides of this story. Ships that operated above the threshold with a well-executed maneuver. And ships that didn't operate, even with the sea below the closure rule. The threshold describes the rule. It doesn't describe the decision.

Multifactor analysis: how to find the pattern that closes the port

The right question isn't "will the wave exceed 2 meters?" It's "what combination of conditions leads the decision-maker to stop this port, in this operation?" Answering that requires looking at many variables together, and looking at what actually happened.

i4sea's multifactor analysis cross-references each port's maneuver history with the meteo-oceanographic conditions measured in each one: wave height, period, direction and energy, gust and sustained wind, visibility, current, tide and rain. From this cross-referencing comes that port's insecurity pattern, not the number on a sheet of paper.

The result is probability. Instead of "closes or doesn't close," the method delivers the chance of restriction or closure by port and by operation type, hour by hour, with hours or days of lead time. It's the difference between reacting to the bulletin and operating the window.

This probability is the same for every link in the chain. The terminal uses it to schedule berths. The shipowner, to decide whether to wait or skip the call. The importer, the freight forwarder and the trucking company, to hold or release the truck, the dock and the promise to the customer.

What the AI climate agent answers in natural language

The multifactor analysis sits inside the Port Logistics AI Climate Agent. The user doesn't read a technical bulletin: they ask in natural language, as they would ask an operations colleague, and receive an answer calibrated to the port and the operation they're tracking.

The questions users ask in practice are few and concrete. Is the port of Itajaí operating now? What is the probability of the bar closing in the next 72 hours? Which ships are in the queue ahead of mine? What is the maneuvering window for my ship's draft on tomorrow's tide?

For each one, the Agent delivers three layers. The probability of port restriction or closure, by time slot. The ship queue and your position in it. And the likely operational impact: lost window, berthing delay, demurrage risk and land-side rescheduling.

On top of these layers come critical recommendations, with financial data. How much it costs to wait at anchor versus reducing speed en route. How much is saved by bringing forward or postponing arrival. Where the possible operational optimization lies, with the number up front, so the decision comes before the suspension, not after it.

And each answer shows where the data comes from. In beta testing, the most recurring question from users was "where did you get this from?" The answer to that question is part of the product, because it's what allows the information to be passed on to the customer or shipowner with confidence.

What changes for whoever depends on the port

Brazil paid US$ 2.3 billion in demurrage at its ports in 2024, against US$ 2 billion in 2023, according to a Bain & Company survey. Part of that bill lands on those who decided nothing: the importer, the exporter and the freight forwarder who brokered the shipment.

With the closure probability in hand two or three days in advance, three decisions change. The truck and the dock are held instead of cancelled with a penalty. The end customer receives a likely arrival window instead of "no forecast." And the next shipment is rescheduled before the queue reorders itself.

When the wait happens anyway, the record is ready: time of restriction, measured condition that caused it and authority notice. It's this package, not the memory of "the port closed due to bad weather," that supports a force majeure or demurrage discussion.

Proof: ship waiting time in Santos dropped from 7 to 3 days

At Santos Brasil, the planning team used generic internet forecasts until 2021, and they failed. "Before, the ship would arrive, the team would be ready, but it couldn't enter due to bad weather," reports Evelyn Lima, Operational Planning Director at Santos Brasil.

With a forecast made for the channel, not the city, the port closure forecast reached 75% accuracy, and average ship waiting time dropped from 7 to 3 days. In 2024, the Port Authority of Santos signed a cooperation agreement to implement i4sea's platform in the estuary channel.

The AI climate agent puts this same data in the hands of those on the other side of the dock: importers, exporters, freight forwarders, trucking companies and shipowners.

Frequently asked questions

Does the AI climate agent replace the harbor master's office bulletin?

No. The harbor master's office declares restriction or impracticability; the Agent anticipates the probability of that happening, so the cargo decision comes before the official notice.

How much lead time does the closure probability provide?

It depends on the condition. Extratropical cyclone swells and strong winds announce themselves days ahead. Fog and basin currents form within hours. The Agent combines the multi-day horizon with continuous monitoring close to the event.

Do I need to be a port operator to use it?

No. The product was designed for whoever depends on the port: importers, exporters, freight forwarders, trucking companies, shipowners and terminal commercial teams. Questions are asked in natural language, with no technical bulletin reading required.

Test the Port Logistics AI Climate Agent

Is the port operating now? What's the chance it closes in the next 72 hours? How many ships are in the queue? Ask the Agent these questions about the port and cargo you track today, and get the probability, the queue and the recommendation before promising a date to your customer.

Sources

1. G1, 02/18/2026: climate intelligence at the Port of Santos, testimony from Evelyn Lima (Santos Brasil), 75% accuracy in closure forecast, wait time down from 7 to 3 days, APS 2024 agreement. https://g1.globo.com/sp/santos-regiao/porto-mar/noticia/2026/02/18/inteligencia-climatica-evita-tragedias-e-reduz-filas-no-porto-de-santos-entenda.ghtml · Republication: CBN Santos, 02/18/2026. https://www.cbnsantos.com.br/noticias/noticias/porto-de-santos-aposta-em-inteligencia-climatica-para-reduzir-tempo-de-espera-de-navios.html

2. Valor Econômico, 04/03/2025: port delays generate extra cost of US$ 2.3 billion for Brazil in 2024 (Bain & Company study). https://valor.globo.com/empresas/noticia/2025/04/03/atraso-em-portos-gera-custo-extra-de-us-23-bi-ao-brasil-em-2024.ghtml · Folha de S.Paulo, 04/19/2025: US$ 2.3 billion in 2024 vs US$ 2 billion in 2023. https://www1.folha.uol.com.br/mercado/2025/04/atraso-em-portos-brasileiros-atrapalha-logistica-e-gera-custo-bilionario-ao-setor.shtml

3. i4sea, statement from CEO Mateus Lima (09/04/2026): port closes due to human decision facing operational risk, before or after the threshold; analysis of all maneuvers (good and cancelled); wave train, period, direction and energy.

4. i4sea, analysis of questions from beta users of the Logistics Climate Agent (05/28 to 07/03/2026): question patterns (status now, 72h risk, queue/AIS, draft × tide in Itajaí, data provenance).