When warnings lose trust
Shaun Eaves, Cathrine Dyer, and James Renwick argue that as climate-driven hazard warnings become more frequent, building public trust through relational governance and transparent communication is essential to prevent emergency warning fatigue for The Conversation.
Image by Dominic Wajda | Unsplash
Shaun Eaves is a Senior Lecturer in Physical Geography at the Victoria University of Wellington, New Zealand; Cathrine Dyer is a Lecturer in Climate Change at the Victoria University of Wellington; and James Renwick is a Professor of Physical Geography (Climate Science) at the Victoria University of Wellington.
Earlier this month, Wellington, New Zealand, declared a local state of emergency, including evacuation orders, when forecast powerful swells threatened to inundate coastal properties. Hundreds of people evacuated, but when the damage and inundation remained limited, mainstream and social media raised questions about whether the hazard warning was an overreaction that risked creating ‘emergency warning fatigue’.
As extreme weather events are becoming more frequent as a consequence of climate change, communities will be dealing with more hazard warnings. Some of them may turn out to be too cautious. Researchers are investigating why some people already fail to follow protective orders in emergency warnings.
We argue a ‘relational governance’ approach can address this issue by building trust and shared responsibility between communities, experts and governing bodies before and after events. This concept provides a platform for effective communication, not just in the lead-up to a hazardous event but also afterwards. This reduces the risk of people taking warnings less seriously even when some turn out to be false alarms.
Probabilities, not promises
Science does not provide certainties, but rather probabilities informed by observations, models and prior understanding.
In the case of the Wellington swell, scientists could say with high confidence that waves would be large. Atmospheric conditions drove a sustained and strong southerly airflow extending some 2,000 kilometres from near the Antarctic margin all the way past the eastern coast of the North Island.
Strong winds sustained over a long, uninterrupted distance provided conditions for generating large swells with damaging effects on shore. Indeed, one wave buoy recorded a maximum wave height of around ten metres, with peak averages closer to four metres. This was without doubt an extreme event.
But, local effects, such as coastal inundation and damage, depend on many additional factors. This includes tide levels, local-scale coastal topography and bathymetry, and subtle variations in storm evolution. Models capture some of these variables, but not perfectly. Small differences in these factors can lead to large differences in outcomes from one place to the next.
Nevertheless, decision-makers are often forced to act in situations of deep uncertainty in which scenarios may not play out as expected.
Crying wolf
Based on Aesop’s Fable, the ‘crying wolf’ syndrome suggests that repeated false alarms undermine the efficacy of early warning systems.
The syndrome has been studied in psychology, economics, sociology and hazard-management disciplines. Generally, research suggests the likelihood of people responding appropriately to a hazard warning is not diminished by it, as long as the basis for the warning is well understood. Social connectedness, previous experiences of disasters and institutional trust all play a role in how people respond to alarms.
In April 2026, a widespread red warning was issued over the North Island, New Zealand, due to the uncertain track and potential severity of ex-tropical cyclone Vaianu. Some councils in the Hawkes Bay region declared local states of emergency. One local mayor refused to join them, suggesting the country was becoming ‘woke’ when it came to emergency declarations.
This narrative escalation mirrors dynamics that have been observed during other politically contested events such as pandemics, wildfires and floods. When trust in institutions is contested, precautionary warnings can become entangled with narratives about government competency, bureaucratic overreach and legitimacy. The overall effect is more akin to institutional trust fatigue than emergency warning fatigue. When that happens, hazard warnings become subject to political agendas rather than acting as technical communications.
In New Zealand, emergency declarations are designed primarily as legal and operational tools, rather than risk-communication tools. Sound precautionary emergency management can appear indistinguishable from overreaction when damage is either successfully avoided or the hazard effect is less than forecast.
One solution is to move from a model focused on issuing warnings to one that offers clear justification and publicly accessible reviews following an event. Such systems already exist elsewhere. For instance, California’s, US, emergency-management system requires timely, open post-event reports that explore what happened, what decisions were made, the effectiveness of actions, and the lessons learned.
However, such systems often focus on what an agency learned from an event rather than public legitimacy. While public accountability is required, it is an insufficient measure for rebuilding trust. Relational governance focuses attention on relationships – between communities, experts and agencies – before, during and after extreme events.
Iwi and community groups should be involved, as critical knowledge-holders, in the design of such frameworks, including input into thresholds that trigger actions. Post-event reporting can then form the basis of iterative and reflective processes of public engagement that build trust, legitimacy and a shared understanding of uncertainty.
As communities encounter climate hazards with increasing frequency, such relational work to strengthen connections between communities and agencies may prove as important as the warning systems themselves.
This piece is originally published by The Conversation and can be found here.