How to build a climate risk score in logistics, link by link: The end of climate guesswork

Climate guesswork already has a price in your supply chain

A climate risk score in logistics is a rating that combines the probability of an event with what it causes in a specific link of the chain, within a defined time window, and that points to the recommended action. Without that combination, the rating becomes opinion with a number attached.

Climate risk in the transportation sector already has a number in Brazil. Seven out of ten transportation companies suffered financial losses from extreme weather events in the last five years. Another 74.6% suffered operational impact, such as interrupted flow, altered routes and lack of supplies.

The data comes from the CNT Climate Resilience Survey of the Transportation Sector, released in November 2025, which surveyed 317 business owners across different modes of transportation and all regions of the country.

These numbers count what already happened. What hurts day to day is the decision that comes after the event: the truck already inside the flood, the train already held up on the mountain pass, the customer already asking for the date. At that point, weather enters the conversation as an excuse, not as a management variable.

A climate risk score serves to reverse that order. Instead of a bulletin by region, it delivers a score by chain link and by time window, tied to an action: operate, adjust or anticipate the stoppage.

What is a climate risk score in logistics

Searching for "climate risk score" mixes two products that answer different questions.

• Strategic score. Question: where is my network exposed over the coming decades? Unit: facility or supplier, by location, with ratings from 1 to 100. Who uses it: ESG, procurement, insurance.
• Operational score. Question: what do I do tomorrow, in which link and at what time? Unit: segment, route, yard or dock, by time window. Who uses it: logistics, control tower, operations.

The strategic score decides where to invest, where to hold inventory and how to price insurance. That is the level where climate risk in supply chain sits: suppliers, factories and distribution centers seen as a whole.

A study by KPMG with BVL, with more than 90 decision makers in the German logistics sector, shows that 67% of companies were affected by resource shortages linked to climate risks. Only 37% address floods and storms comprehensively in risk management.

The two layers complement each other: the strategic layer shows where the network is fragile, the operational layer decides what to do days before the event. No strategic score tells you what to do on Tuesday.

Every rating is ordinal, not monetary: an 80 does not state the expected financial loss. That is why the operational score only works when tied to an action and to a stoppage cost that the operation already knows.

A good operational score tells you where, when, with what probability and what to do.

A score for each link of the chain

The same storm generates different decisions at each link, so a single score for the entire chain hides more than it shows. In Brazilian freight, the greater weight is on land.

The CNT survey shows that road transport accounts for 84.5% of the companies surveyed. Rail, yard, distribution center and last mile complete the picture. Ports and waterways are another link, not the center of the equation.

• Road. Accumulated rainfall and soil saturation matter most. The decision is to reroute, hold back cargo or adjust the departure window. Typical lead time of 1 to 3 days for accumulated rainfall.
• Rail. Landslides, track flooding and fallen trees matter most. The decision is to hold a train or open a maintenance window, with 1 to 3 days of lead time.
• Yard and distribution center. Heavy rain and flooding at access points, lightning, wind, temperature and humidity matter most. The decision is to reschedule docks, bring forward dispatch and protect sensitive inventory, from minutes to 2 days.
• Last mile. Heavy rain, gusts and urban flooding matter most. The decision is to cut a route and open a new delivery window, from hours to 2 days.
• Port, river and cabotage. Swell, wind, fog at the bar and draft matter most. The decision is to bring forward docking or departure, with 3 to 10 days for cyclones and cold fronts. River level has a horizon of up to 90 days.

The lead time ranges follow climate risk in logistics, link by link, and vary by region and by phenomenon. Imminent lightning calls for an alert from minutes to 6 hours.

Resolution also matters. Public global models operate at around 25 km, while hyperlocal readings work at 1 to 3 km, according to the free climate risk diagnostic from i4sea. At 25 km, two mountain segments and two neighboring yards look like the same thing.

Format matters as much as the rating. Instead of "high risk in the Southeast," the score says: "segment X of the highway, high risk of flooding between 4am and 10am, activate detour." That is the difference between information and decision.

Climate risk in distribution centers

The distribution center feels the weather through two doors: the one facing outward, through which cargo enters and leaves, and the one facing inward, where inventory and teams are located.

• Access and docks. Heavy rain and flooding in the surrounding area trap trucks, halt transshipment and disrupt the dock schedule.
• Storage. Temperature, humidity and rain affect sensitive products. Each product's limit becomes a rule in the DC's score.
• Equipment and teams. Lightning strikes call for attention to electronic equipment maintenance and outdoor work. In extreme heat, the report Extreme heat poses challenges to pharmaceutical transport, from Mundo Logística, points to growing demand for mapping critical routes and upgrading packaging.

A DC's score works with short windows: rescheduling docks, bringing forward dispatch, protecting sensitive inventory and notifying the customer. The limits matrix applies to a DC dock just as much as to a road segment.

Operational limits: where the rule comes from and why it fails

Every link already operates with climate limits, written or not. They come from equipment manuals, standards, contracts or accumulated practice, and vary by facility.

The material from UNCTAD on operational thresholds shows how the limit changes from component to component. For one type of crane, 25 m/s of wind. In the access channel, winds above 18 m/s force a stoppage. Between 12.8 and 18 m/s, the decision is discretionary.

The example is port related, and the mechanics repeat on land: on a mountain road, the hard limit is closure due to landslide, but allowing the descent with high accumulated rainfall already falls into the discretionary range.

That range is where guesswork lives: the rule does not decide, the person on duty decides, with the information they have at that moment.

The fixed limit fails for three reasons:

1. Gray zone. Between "operate" and "stop" there is a range in which the outcome depends on a combination of factors, not a single number.

2. Isolated variable. A review in the Journal of Marine Science and Engineering (2023, DOI 10.3390/jmse11030536) notes that downtime is usually measured by a single variable, even though it is multidimensional.

3. Limit without cost. The rule says when to close, not how much it costs to close too early or too late.

How to establish the limits for your link

1. List critical assets and segments: road, rail, yard, distribution center.

2. Record where each limit comes from: equipment, standard, contract or practice.

3. Separate the hard limit (safety) from the decision limit (the gray zone).

4. Cross-check each limit against the history of stoppages and delays. What does not appear there is a hypothesis, not a limit.

5. Review it each season, because the asset and the operation change.

The matrix organizes the survey, one row per asset: link (DC dock 2), variable (sustained wind), hard limit, decision limit, rule origin and review date. Without it, any rating becomes opinion with a number attached.

Anticipating the stoppage: from the yard to the last mile

When a link stops, the cost does not stay there. A closed mountain segment pushes the truck onto a longer route, the DC dock window is lost and the delivery is delayed.

The port is just the most visible case, because the queue sits anchored for everyone to see. The math is the same at any link: the wait becomes a queue, the queue becomes a lost window, and the lost window becomes demurrage, detention or an SLA penalty.

We have already written about probability of port closure and about demurrage as a predictable cost.

What the score needs to deliver

• probability of restriction or stoppage by time window and by asset, not just by region;
• sufficient lead time to act: 3 to 10 days for cold fronts and cyclones, hours to 2 days for wind and heavy rain;
• expected impact on the queue and on gate or dock windows, with the suggested action and automatic notification to the following links, all the way to the customer.

How a stoppage at one link travels through the chain

The diagram shows the path: a delayed link pushes cost to the next one.

The score needs to act in the first box: the sooner the decision about the delayed link reaches the following links, the less cost accumulates at each one.

From score to decision and evidence

A score that nobody knows what to do with becomes just another dashboard. What turns it into a tool is the rule that ties each range to an action and an owner, defined before the event.

• Low. Window within the link's limits. Action: operate normally. Decides: operations.
• Attention. The decision limit may be crossed within the window. Action: adjust schedule, prepare plan B, notify the next link. Decides: planner.
• High. Hard limit likely within the window. Action: anticipate the stoppage, change route, notify the customer. Decides: operations manager.

The ranges and the owners are a model. The right cutoffs come from calibration against each asset's history.

The trail that becomes evidence

The second gain comes after the event. With each alert logged with time, source, recommendation and decision owner, the operation has in hand what it predicted, what happened and what was decided.

That trail supports discussions about demurrage and detention, disputes over SLA penalties, incident reports to the insurer and internal alignment about what was a decision and what was a surprise.

The score does not suspend charges by itself. The effect depends on the contract's laytime and force majeure clauses. What the trail does is take the conversation away from "we think it was the weather" and move it toward "here is the data."

How to choose: five questions and a test

Before comparing vendors, compare the questions. A score is defended with verifiable answers, not with a nice screen. And run the test before the critical window: according to Revista Apólice, January through March is the most critical period for cargo transportation, with flooding, landslides and closures on strategic highways.

1. Is it by link and by window, or by region? A regional rating does not distinguish between two segments or two yards.

2. Was it tested against my own history? A good test uses the events you have already recorded, including stoppages that were not climate related, so as not to mix up causes.

3. What is the false alarm rate and the unpredicted event rate? Every score makes mistakes. The question is at what cost: how many false alerts per year and how much it costs to act too early.

4. What is the lead time with measured accuracy? Lead time without accuracy measurement is just a promise.

5. Does the alert reach where the decision happens, and is it logged? TMS, control tower, Teams or WhatsApp. If it is just another portal, nobody opens it. Without a trail with time, source and owner, it does not become evidence.

The history test

Ask the vendor to run the score against your history of stoppages and delays, without seeing the result beforehand. Compare how many stoppages would have been flagged with useful lead time, how many alerts would have been false and how much it would cost to act too early. If they refuse the test, you already have your answer.

What changes in practice: a real case

Santos Brasil: from 7 to 3 days of average ship waiting time

At Tecon Santos, i4sea's climate intelligence reduced the average ship waiting time from 7 to 3 days, according to a G1 report from 02/18/2026.

The change is procedural: the operation started deciding the window before the event, instead of reacting to a queue that had already formed.

The case is from a port because that is where the gain became public. The logic travels through the chain: in the yard, in the DC and in the last mile, the question is the same.

Start with your history, not with the promise

Climate guesswork ends when every link has logged limits, every window has a probability and every range has an action with an owner. The score brings these three things together into a decision that fits the daily routine of those who operate.

The first step does not require hiring anything. Choose an asset (a highway segment, a DC, a yard or a route) and find out how many times extreme weather has hit it in recent years.

The free climate risk diagnostic for your asset does this survey with data calibrated to 1 to 3 km resolution, for a single asset, without you having to send any data from your operation.

Frequently asked questions

Is a climate risk score the same as a weather forecast?

No. A forecast tells you what the weather will be. The score tells you what that weather will likely do to a specific link in the chain, in a defined window, and what action it recommends. The forecast is the raw material, and the score is the risk reading on top of it.

What is the difference between a strategic and an operational climate score?

The strategic score assesses the exposure of facilities and suppliers over long horizons and serves investment, procurement and insurance decisions. The operational score assesses risk by link and by time window, from hours to a few days, and serves next day decisions. Many companies need both.

How do you define operational limits for a link?

Start with the source: equipment, standard, contract or practice. Separate the hard limit from the decision limit, where someone weighs in. Cross-check each one against the asset's history of stoppages and delays, and review it each season.

What lead time is useful for each type of event?

Imminent lightning calls for an alert from minutes to 6 hours. Wind and heavy rain, from hours to 2 days. Accumulated rainfall and cabotage windows, from 1 to 3 days. Cold fronts, extratropical cyclones and swell, from 3 to 10 days. River level and draft, up to 90 days.

How does climate risk affect a distribution center?

Through two doors. Outside, heavy rain and flooding trap trucks and delay docks. Inside, heat, humidity and lightning affect inventory and equipment. The DC's score works with short windows: rescheduling docks, bringing forward dispatch.

References