Who Decides When a Disaster Is Over — and Why the Answer Is Almost Never “Nobody Is in Danger”

Direct answer: nothing announces that a disaster is over. Different actors close it at different moments for different reasons: a meteorological warning expires on a schedule, a response operation ends when its funding period does, a government lifts an emergency declaration for legal reasons, and news coverage stops when the story stops being new. The one thing none of those marks is the point at which people stop needing help — that comes much later, and it is the part almost nobody watches.

What “active” means on a screen, including ours

Live feeds carry a status, and it is far more mechanical than it looks. Most warning products have an expiry time built into them when they are issued: a warning valid until a stated hour simply stops being valid at that hour. Nobody re-inspected the flood. The timer ran out.

Satellite-derived records behave the same way. A fire detection is current until the satellite passes again and does not see heat, which can mean the fire is out or can mean cloud moved in. A record aging out of a feed is a statement about the feed, not about the ground — the same distinction as in what “no alerts” actually means, seen from the other end.

So when this site shows an event as no longer active, read it as “no current warning product covers this”, not as “this is finished”.

The four endings, and the order they actually happen in

1. The warning ends. Hours to days. A hazard-specific, technical decision: the wind dropped below a threshold, the river fell below a level, the ash plume stopped.
2. The news ends. Days to a couple of weeks, and it tracks novelty rather than need. Coverage is heaviest while the outcome is uncertain, and uncertainty resolves long before need does.
3. The response ends. Weeks to months, usually determined by the appeal period and the funding cycle rather than by an assessment that assistance is no longer required.
4. The recovery ends. Years. Housing, livelihoods, schooling, debt. This has no announcement and no feed.

The uncomfortable part is the gap between the second and the fourth. Attention peaks in the first week; need usually peaks well after that, once savings are gone, insurance has or has not paid, and temporary shelter has become long-term.

Why the counting keeps moving after everyone stops looking

Deaths attributed to a disaster continue to accumulate after the event window closes — from injuries, from interrupted medical care, from contaminated water, from heat in damaged housing. Whether those are counted at all depends on the attribution method in use, which is a subject of its own.

This produces a specific and misleading pattern: the figure that gets fixed in public memory is the one published while the story was still running, and it is almost always the earliest and lowest one. Later revisions are more accurate and far less read.

Cascading effects have their own timelines

• Water and disease follow flooding by days to weeks, after the water has visibly gone.
• Displacement can persist for years, and people in protracted displacement rarely appear in any event feed at all.
• Economic harm lands on households as debt long after the physical damage is repaired, and it does not resolve on the same schedule as buildings.
• Mental health effects are documented to persist well beyond the visible recovery, and they are absent from every impact figure discussed in what “people affected” actually counts.

Why a declaration being lifted is a legal event, not a medical one

Emergency declarations exist to unlock powers and money: procurement rules, movement of resources, access to funds. They are lifted when those powers are no longer needed by the administration, which is a different question from whether households have recovered. A lifted declaration can even make things harder for people still in need, because it can close the mechanism that was helping them.

None of this implies bad faith. Response systems are built around defined operations with start and end dates, because that is how they can be funded and staffed at all. The structural consequence is simply that the end of an operation is not the end of a problem.

How to read the end of a disaster honestly

1. Treat status as a statement about warnings, not about safety. No active warning means no current product, nothing more.
2. Assume the final death toll is not the one you remember. The memorable figure is usually the early one.
3. Do not read the end of coverage as the end of need. It is closer to the opposite: the quiet period is when the need is largest.
4. If you are giving, give late as well as early. Immediate donations are plentiful and recovery funding is not.
5. For a place you care about, follow the national agency and local sources rather than a global feed. Recovery is entirely local and does not generate international records. Our country pages link each country's own agencies.

What this site does, and what it cannot do

• We publish the source's status and do not invent our own. If an agency has not closed an event, neither have we.
• We never say a disaster is over, for the same reason we never publish an all-clear. It is not a statement we are in a position to make.
• We keep event pages after the warning expires, because the record of what happened stays useful long after the alert stops being live.
• We do not cover recovery, and we say so. Live feeds are built to detect onset. Nothing in them measures whether a family got their house back, and presenting an aggregator as though it did would be a false picture of the world. Our methodology sets out what the feeds do and do not carry.

The honest summary

A disaster has at least four endings, they arrive in the wrong order, and the only one with a timestamp is the least meaningful. The warning expiring is a scheduling fact. The cameras leaving is an editorial one. Somewhere after both, and usually years after, is the ending that actually matters, and no system on this site or anywhere else marks it.

This article describes how disaster events are closed in monitoring and response systems. This site is not a warning authority; whether it is safe to return, rebuild or drink the water is a question for your local authorities.

Frequently asked questions

What does it mean when a disaster is no longer marked active?

Usually that no current warning product covers it. Most warnings carry an expiry time set when they are issued, so the status changes on a timer rather than because somebody reassessed the situation. Satellite-based records behave similarly — a fire detection lapses when the next pass sees no heat, which can also mean cloud moved in. Read it as a statement about the feed, not about the ground.

Who officially declares a disaster over?

Nobody, globally. A meteorological warning expires, a response operation reaches the end of its funding period, a government lifts an emergency declaration for legal and administrative reasons, and news coverage stops when the story stops being new. These are four different endings decided by four different actors, and none of them marks the point at which people stop needing help.

Why do death tolls keep rising after a disaster ends?

Because deaths continue from injuries, interrupted medical care, contaminated water and unsafe housing well after the event window closes. Whether they are counted depends on the attribution method. This creates a predictable distortion: the figure fixed in public memory is the early, low one published while coverage was still running, and the later, more accurate revisions are far less read.

When is the need actually greatest after a disaster?

Usually well after the coverage ends. Attention peaks in the first week while the outcome is still uncertain, but uncertainty resolves long before need does. Savings run out, insurance pays or does not, and temporary shelter becomes long-term. The quiet period is generally when assistance matters most and is hardest to fund.

Does lifting an emergency declaration mean it is safe?

Not necessarily. Declarations exist to unlock powers, procurement and funding, and they are lifted when the administration no longer needs those powers. That is a legal and financial judgement rather than an assessment that households have recovered. In some cases lifting a declaration closes the mechanism that was still helping people.

Why does this site not cover the recovery period?

Because the feeds it aggregates are built to detect onset, not to measure whether a family got their house back. Recovery is local, slow and largely undocumented internationally, so no global feed carries it. Presenting an aggregator as though it tracked recovery would give a false picture, so we state the limit instead.

Do event pages disappear when the warning expires?

No. The live alert stops, but the page stays, because the record of what happened remains useful long after it is no longer current. What changes is the status, and that status should be read as describing the warning rather than the situation.

Sources

• UN OCHA — Humanitarian programme cycle, appeals and response timeframes
• IFRC — Recovery programming and the transition from response
• UNDRR — Sendai Framework: recovery, rehabilitation and reconstruction
• WHO — Health consequences of disasters beyond the emergency phase
• IDMC — Protracted internal displacement
• NOAA / NWS — Warning products, validity periods and expiry
• ReliefWeb (UN OCHA) — Situation reporting through the recovery period

Found an error? See our corrections policy.