Weather apps don’t cancel trips, accountable people do

Karnin ‘Tino’ Petera was 15 when he died on 9 May 2023, trapped by a flash flood inside Abbey Caves near Whangārei during a school outdoor education trip, while an orange heavy rain warning was in force for Northland. Coroner Alexander Ho’s findings, released last week, criticise Whangārei Boys’ High School’s risk tolerance and question why MetService publishes hourly rainfall numbers at all.

The MetService angle will grab headlines. The lesson that matters for anyone running weather-exposed work is less comfortable. A forecast is an input. It is not a cancellation policy, and it is not a decision-maker.

The numbers softened overnight

On the Monday night, MetService’s hourly forecast showed 12.5mm of rain between 1pm and 2pm the following day. By Tuesday morning it had changed to 1-4mm an hour from 7am to 5pm, a far gentler picture that the coroner said may have influenced the decision to proceed. Those hourly figures are computer-generated and not checked by a meteorologist before publication. Most users also misread them: a number in the ‘9am’ column covers the hour leading up to 9am, not the hour after.

Ho put it bluntly. “Not receiving the expected forecast rainfall for one’s lemon tree is inconvenient. Being caught outside in a thunderstorm during a high-risk outdoor activity is dangerous,” he wrote, before asking “why MetService publishes hourly numbers at all”. He recommended MetService clarify its website and align inconsistent disclaimers across its app and web platforms.

MetService is right, and it doesn’t help

Chief meteorologist Chris Noble said severe weather watches and warnings “should always take precedence over other forecast information” and that MetService had already been improving how it communicates forecasts before the findings landed.

He is correct. On that Tuesday morning the most authoritative product available, the orange warning, was telling everyone to be careful. But a hierarchy of information only protects people if an organisation has written it into its procedures before the bus leaves. Otherwise staff under time pressure will gravitate to whichever number looks most reassuring. That is human nature, and good risk systems are designed around it.

The policy existed, the trigger didn’t

This is the crux. Back in 2024, WorkSafe found the tragedy was preventable. The school’s own risk assessment said trips would be cancelled for heavy rain warnings. Yet there was no shared understanding among organisers of what ‘heavy rain’ meant or when cancellation kicked in. WorkSafe also found the board’s oversight of high-risk activities was ineffective, and that the emergency plan did not anticipate rising water trapping students. Sixteen other students and two supervisors survived.

The coroner has now added that there was uncertainty over who actually held authority to cancel the trip, and described the decision to go ahead as reflecting an “overly optimistic” tolerance for risk. A policy that nobody can interpret, enforced by nobody in particular, is not a control. It is paperwork.

Principals want thresholds, not better apps

Principals’ Federation president Jason Miles said the sector “definitely” agrees with the coroner. Tellingly, his asks are not about forecast accuracy. He wants help monitoring “wind speeds, rainfall thresholds and river flow rates” and “clearer guidelines on who had the authority to cancel activities”, backed by professional development with “really specific guidance”.

That is the right diagnosis, and it applies well beyond schools.

What every weather-exposed operator should rewrite

Adventure tourism operators, builders working at height, event organisers, forestry crews, transport and logistics firms all make go or no-go calls on weather. Under the Health and Safety at Work Act 2015, officers carry a personal due diligence duty, and governance commentators including the Institute of Directors have framed this case as a board-level failure where risk should have been reassessed continuously, not once days earlier.

A defensible protocol needs four things. First, name the authoritative product, so a warning beats a worded forecast and both beat hourly numbers. Second, set hard, measurable triggers in millimetres, kilometres per hour or river levels, not adjectives like ‘heavy’. Third, name the person who can cancel, and give them authority to do so on the spot without escalation or second-guessing. Fourth, require a fresh assessment on the day, and again when conditions change, including when swapping to an alternative activity.

None of this costs much. A cancelled day costs a refund or a rescheduled crew. Getting it wrong costs a life, a prosecution and a reputation.

The question that outlasts the forecast

MetService will tidy up its disclaimers and its hourly displays, and it should. But Karnin’s parents, who told the inquest in April that they “trusted our son would be safe at school”, have kept asking who was accountable for the call to proceed. Every business that sends people into the weather should be able to answer that question about itself, in writing, before the next orange warning arrives.

Sources

• 1News: MetService criticised by coroner – questions the point of their hourly forecasts (2026-11-09)
• RNZ: Karnin Petera died during a school trip to Abbey Caves – a coroner reveals what went wrong (2026-11-05)
• Institute of Directors: A board’s worst nightmare, and the ramifications
• 1News: Teen killed in Abbey Caves tragedy had concerns before trip, parents say (2026-04-30)
• NZ Herald: Karnin Petera’s parents still question accountability of decision-makers