28 August 2025
Narmada S
[FYI request #31864 email]
Reference: OIA-2025/26-0087
Dear Narmada S
Official Information Act request relating to success criteria, evaluation and approach to
nationwide alerts
Thank you for your Official Information Act 1982 (the Act) request received on 2 August 2025.
You requested:
Under the Official Information Act 1982, I request the following information relating to
the Emergency Mobile Alerts (EMAs) issued on 30–31 July. Please interpret “records”
to include emails, messages, briefings, minutes/notes, incident logs, templates, and
any attachments.
A. Mandatory/opt-out/quiet-hours settings
1. Opt-out/quiet mode policy: Has NEMA/DPMC considered providing user-level opt-
out (globally or by hazard type) or a “quiet mode” for advisory-level events (e.g., “strong
and unusual currents” advisories)? Please provide evaluations, options analyses,
minutes, recommendations, and the reasons for any decision to decline.
B. Proportionality during sleeping hours
2. Proportionality test: What internal threshold or public-interest test must be satisfied
before sending an EMA during typical sleeping hours where the advice is not materially
new (e.g., a reminder)? Please provide any checklists/templates used on 31 July 2025
(06:30).
C. Novelty/duplication
3. Message novelty: Drafts/versions or records showing how NEMA determined the
06:30, 31 July alert added new value versus duplicated the prior alert. If no new value
was identified, the rationale for sending regardless.
4. Severity–template binding: Are message templates bound to severity levels
(advisory vs watch vs warning) to prevent advisory-level messaging being framed with
unnecessarily high urgency? Please provide the template library in force at the time
and governance/approval controls.
D. Risk created by the alert itself
5. Risk registers: All risk assessments/risk-register entries identifying potential harms
caused by EMA use, including sleep disruption, panic/mass movement, road traffic
incidents, 111 call surges, and alert fatigue reducing future compliance; plus any
mitigations noted.
6. Incident/near-miss reports: Any incident or near-miss reports attributable to the 30–
Level 7, TSB Building, 147 Lambton Quay | PO Box 5010 | Wellington 6140 | New Zealand
Tel: +64 4 830 5100 | [email address] | www.civildefence.govt.nz
31 July EMAs (e.g., collisions, coastal self-evacuations, medical device interference,
workplace safety events), including any preliminary tallies or summaries if final reports
are pending.
E. Channel alternatives and evaluation
7. Channel alternatives: The evaluation comparing EMA versus media/social/web push
for advisory-level scenarios (reach, timeliness, expected benefits, and cost/harm trade-
offs), including any cost–benefit or harm–benefit analysis used for 30–31 July.
8. Performance KPIs: The KPIs or success metrics used to judge whether the 06:30,
31 July alert achieved objectives (e.g., reduced coastal water use, complaints/false-
alarm cost), the KPI framework, and any preliminary evaluation against those KPIs.
A. Mandatory/opt-out/quiet-hours settings
1. Opt-out/quiet mode policy: Has NEMA/DPMC considered providing user-level opt-out
(globally or by hazard type) or a “quiet mode” for advisory-level events (e.g., “strong and
unusual currents” advisories)? Please provide evaluations, options analyses, minutes,
recommendations, and the reasons for any decision to decline.
Please find enclosed a document titled Decision for Emergency Mobile Alert to remain a non-
opt-out channel dated 2 July 2018 which summarises the considerations and decision by the
Cell Broadcast Alerting Steering Committee to continue broadcasting Emergency Mobile
Alerts solely on a non-opt-out channel.
This document is released to you in full. There is no further material in scope of this part of
your request.
B. Proportionality during sleeping hours
2. Proportionality test: What internal threshold or public-interest test must be satisfied before
sending an EMA during typical sleeping hours where the advice is not materially new (e.g., a
reminder)? Please provide any checklists/templates used on 31 July 2025 (06:30).
The Emergency Mobile Alert (EMA) system is governed by strict protocol on its use. The
Emergency Mobile Alert Protocol for User Agencies document covers the criteria for sending
Emergency Mobile Alerts to New Zealanders. The National Emergency Management Agency
(NEMA) will only send alerts when there is a severe, urgent and likely threat to life, health or
property. Dangerous tsunami activity generated by the Kamchatka earthquake affecting New
Zealand’s coastline clearly met these criteria, and the decision to use it was informed by
scientific advice and consultation with Civil Defence Emergency Management Groups.
The EMA system, and the noise it generates, is based on a global standard, and is also used
in other countries. In nearly eight years of use in New Zealand, which has seen tens of
millions of instances of alerts being received by people, NEMA is not aware of any confirmed
severe health event or accident that has been directly attributed to the use of EMA.
Conversely, every time the system is used, it potentially prevents significant impacts to life,
health or property.
I have decided to release internal communications relating to the decision to issue, and the
timing of, the alert on the morning of 31 July 2025. Some information is withheld under the
following sections of the Act:
• s9(2)(g)(i), to maintain the effective conduct of public affairs through the free and
frank expression of opinion
• s9(2)(g)(ii), to prevent improper pressure or harassment.
A small amount of information is not in scope of your request and is marked accordingly.
5101367
2
link to page 3
C. Novelty/duplication
3. Message novelty: Drafts/versions or records showing how NEMA determined the 06:30,
31 July alert added new value versus duplicated the prior alert. If no new value was
identified, the rationale for sending regardless.
Overnight, we received confirmation that tsunami activity was impacting New Zealand’s
coastline and would continue to do so well into Thursday (31 July). We weighed up the risks
and concluded that we needed to inform the public that the threat was present and ongoing.
To ensure we did not wake large numbers of people up in the middle of the night, we sent
the alert at 6.30am, while informing those planning morning activities in beach or marine
areas while dangerous tsunami activity was happening.
I have decided to release to you a timeline of key actions/advice relating to this event and
internal communications relating to the decision to issue the EMA at 6.30am on 31 July.
4. Severity–template binding: Are message templates bound to severity levels (advisory vs
watch vs warning) to prevent advisory-level messaging being framed with unnecessarily high
urgency? Please provide the template library in force at the time and governance/approval
controls.
Templates are not bound to severity levels – the EMA only has one severity level and that is
for High Priority Alerts (as defined by the current
protocol).
There are more than 200 templates loaded into the EMA system by NEMA and the wider EM
Community. These templates are designed to be edited to reflect the situation when issued. I
have decided to refuse your request under section 18(f) of the Act as it would require
substantial collation to extract these templates and bring them together to provide to you.
Before reaching the decision to refuse this part of your request, I considered consulting with
you to refine this, charging and or extending the time for response. I did not identify a way to
meaningfully refine the request while maintaining the intention. Nor do I believe that charging
or extending the time for response would alleviate the administrative burden.
D. Risk created by the alert itself
5. Risk registers: All risk assessments/risk-register entries identifying potential harms caused
by EMA use, including sleep disruption, panic/mass movement, road traffic incidents, 111
call surges, and alert fatigue reducing future compliance; plus any mitigations noted.
I have decided to release, as an excerpt, the relevant parts of an internal advice document
titled “HRM Briefing Note – Early Warning Best Practice May 2025” which relates to alert
fatigue:
There are several key factors to consider when determining how, and when to issue
warnings. One of these is the risk of warning fatigue, often called the ‘cry-wolf effect’.
Warning fatigue is the loss in credibility of a warning system and subsequently, the warning
agency, due to false alarms. The risk of warning fatigue may need to be considered
alongside the consequences of perceived ‘missed alerts’, which occur when the public
believe they should have been warned for an event and did not receive a warning.
Recommended mitigation techniques for decreasing the risk of warning fatigue include
raising the trigger level for issuing warnings to decrease the number issued, transparent
communication of any uncertainty relating to the event and, if required, communicating
clearly why a false alarm occurred[1].
[1] Canary Innovation (2025). What do we know about warning fatigue?
5101367
3
link to page 4 link to page 4
Position
Aotearoa New Zealand is continuously working to improve its early warning systems, and
public education arrangements that maximise the effectiveness of warnings. NEMA
encourages cautious and considered approaches to improving early warning systems that
will help to protect communities during events when there is a risk to life, property or
infrastructure. Aotearoa New Zealand supports international principles for EWS and NEMA
has identified areas where we can improve our EWS, such as increasing our focus on multi-
hazard warning systems and ensuring that both the messaging and dissemination methods
are useable for everyone.
NEMA’s connections with the technology sector enable us to stay informed of, and advocate
for, technological innovations that could improve our ability to disseminate warnings for
events. Additionally, NEMA’s networks with the science and research communities provide
opportunities to inform, and maintain awareness of, the projects and research programmes
that will underpin advancements in monitoring, forecasting and warning for hazard risks.
Regarding the use of EMAs, NEMA maintains the position that the EMA system should only
be used to issue High-Priority Alerts. In accordance with the New Zealand Common
Alerting Protocol (CAP-NZ), High-Priority Alerts are at level (a) or (b) within each of the
following three criterion:
1. Certainty
a. Observed: Determined to have occurred or to be ongoing
b. Likely: Probability of its occurrence greater than 50%
2. Severity
a. Extreme: Extraordinary threat to life, health or property
b. Severe: Significant threat to life, health or property
3. Urgency
a. Immediate: Responsive action should be taken immediately
b. Expected: Responsive action should be taken soon.
The criteria is set at a high threshold because of the risk of warning fatigue, as noted in
Background, resulting in a lower likelihood of public response, and a reduction in level of
trust in warning agencies. This risk presents a serious concern for emergency managers, as
trust in both the warning agency and the technology is an important factor in ensuring any
warning is effective[2][3]. In the case of an imminent, high impact threat where EMA is the only
effective warning mechanism available, a low response rate poses an increased risk to life.
If agencies are looking to implement significant changes to their warning systems, often
prompted by a major event and subsequent reviews, it is recommended that the warning
agency first assesses the risks associated with changing the system. Key risks include
increasing warning fatigue if the warning threshold is lowered, causing public confusion if
education on the new system is insufficient, or misaligning with national standards and
consistent approaches. Alternative solutions should be considered, such as enhancing public
awareness and preparedness by increasing hazard risk and impact messaging on
communication channels, as widely and early as possible, or investigating possible
alternative alerting methods, to complement existing systems.
You may also be interested in the following publicly available material:
Information on the M8.8 Kamchatka Russia earthquake and tsunami prepared by GeoNet is
available on the following website:
https://www.geonet.org.nz/news/4O1zMlf5rXqq8mFCJOfcEq
[2] Tan, Vinnell, Valentin et al. (2023). The public’s perception of an earthquake early warning system: A
study on factors influencing continuance intention.
[3] McBride, Bostrom, Sutton et al. (2020). Developing post-alert messaging for ShakeAlert, the earthquake
early warning system for the West Coast of the United States of America. Retrieved from
https://doi.org/10.1016/j.ijdrr.2020.101713
5101367
4
More information on civil defence emergency management alerts and warnings are available
on the following website:
Civil Defence Emergency Management Alerts and Warnings » National Emergency
Management Agency
6. Incident/near-miss reports: Any incident or near-miss reports attributable to the 30–31
July EMAs (e.g., collisions, coastal self-evacuations, medical device interference, workplace
safety events), including any preliminary tallies or summaries if final reports are pending. I can confirm that NEMA does not hold any information relating to this part of your request.
There were no reports of incidents or near-misses arising from the EMA issued in July this
year. Further, I can confirm that there have been no reported significant harms during the
lifetime of the EMA system. Accordingly, this part of your request is declined under section
18(e) of the Act as the information requested does not exist.
E. Channel alternatives and evaluation
7. Channel alternatives: The evaluation comparing EMA versus media/social/web push for
advisory-level scenarios (reach, timeliness, expected benefits, and cost/harm trade-offs),
including any cost–benefit or harm–benefit analysis used for 30–31 July.
NEMA has a statutory responsibility to alert the public to tsunami risk. Every decision to alert
the public is based on best advice available including scientific evidence, international best
practice, the observable conditions, and the life safety risks.
When NEMA received the notification that one of the most powerful earthquakes ever
recorded had struck and was sending tsunami waves across the Pacific, it was a
straightforward decision to alert people to help them stay safe. I can confirm the EMA was
issued in line with the Emergency Mobile Alert protocol which I have linked above.
Your request for the evaluation and any cost benefit analysis is refused under section 18(e)
of the Act as the information requested does not exist.
We note that EMA is one of multiple emergency information channels available. No single
channel is failsafe or able to reach all people. For this reason, we take a multi-channel
approach to communicating tsunami risk, which means complementing EMA use with radio,
television, social media, and website.
8. Performance KPIs: The KPIs or success metrics used to judge whether the 06:30, 31
July alert achieved objectives (e.g., reduced coastal water use, complaints/false-alarm cost),
the KPI framework, and any preliminary evaluation against those KPIs.
I can confirm that when sending alerts, NEMA receives confirmation from
telecommunications providers that the alert has been transmitted to all geotargeted areas.
We can confirm that the alerts issued on 30 and 31 July were successfully transmitted to all
geotargeted areas.
In terms of “thresholds”, the User Protocol stipulates that EMAs should only be issued in the
event of a severe, urgent and likely threat to life, health or property. This threshold was met
in the Kamchatka event, with the receipt of scientific advice on the afternoon of 30 July that
dangerous tsunami activity was projected to impact our shores, and the receipt of
confirmation on 31 July that tsunami activity had subsequently been observed on our shores.
You have the right to ask the Ombudsman to investigate and review my decision under section
28(3) of the Act.
5101367
5
This response wil be published on the Department of the Prime Minister and Cabinet’s website
during our regular publication cycle. Typically, information is released monthly, or as otherwise
determined. Your personal information including name and contact details will be removed for
publication.
Nāku noa, nā
Sarah Holland
Chief Advisor to the Chief Executive
National Emergency Management Agency
5101367
6
