Heat kills more than almost any disaster — so why can’t we define it?

Article 1 of the Heatwaves series | HansOnResilience.com — Hans Guttman | July 2026

As I write this, Europe and parts of the US are in the grip of major heatwaves. France has placed nearly half of its mainland departments under a red alert — the highest warning level. Spain has seen temperatures exceed 45°C in the south. The UK broke its all-time June temperature record. In one week alone, by 26 June 2026, 327 heat-related deaths had been registered across the continent. Trains have derailed from heat-warped rails in Sweden. Power grids have strained to breaking point. Children have died in hot cars in France. Elderly people, who had nowhere cool to go, have died at home alone.

And yet: ask any of the meteorologists issuing those red alerts what exactly a heatwave is — what precise combination of temperature, humidity, duration, and deviation from normal constitutes the event they are warning about — and you will get a different answer in most countries. Ask them when a heatwave officially begins, or ends, or crosses from dangerous to severe, and the answers diverge further still. This is not a minor technical quibble. It lies at the heart of why heat, despite killing more people than almost any other natural hazard, continues to be treated as a weather inconvenience rather than a disaster.

Europe is not alone in this, and 2026 is not unique. In April and May 2024, schools across the Philippines shut down as a heatwave overwhelmed the country — the government recognised the strain, but response was reactive, triggered by what was already happening rather than by any pre-agreed threshold. In late June 2021, a heat dome descended on the Pacific Northwest of North America that still reads like a fever dream. The small village of Lytton in British Columbia, Canada, recorded 49.6°C on 29 June — Canada’s all-time temperature record — and burned to the ground in a wildfire the very next day. British Columbia alone recorded 619 heat-related deaths in a fortnight. Back in August 2003, a prolonged heat event killed an estimated 70,000 people across Europe — an event that now serves as the baseline reference for the entire heat-health field.

The pattern across all of these events is consistent: reaction after the fact, improvised response, and a death toll that surprises people who had not realised heat was this dangerous. The reason for that surprise, I would argue, runs deeper than a lack of public awareness. It begins with a definition problem that the professional community has not fully resolved.

Heat as a killer — numbers we do not talk about enough

Let’s start with the mortality picture, because it tends to shock those who have not encountered it before. It is estimated that almost half a million people die each year from heat exposure globally, based on the World Health Organisation’s (WHO) estimates (covering the period 2000 to 2019). That is almost certainly an underestimate, for reasons I will come to in a moment. But even at face value, it places heat alongside — and often above — the annual global toll from many of the disasters that dominate news coverage and humanitarian budgets: floods, storms, earthquakes.

The comparison with seismic events is instructive. Earthquakes and tsunamis produce catastrophic, concentrated death tolls — the 2004 Indian Ocean tsunami, the 2010 Haiti earthquake — which are seared into global memory precisely because of their scale and suddenness. Hydrometeorological events, including heatwaves, drought, storms and floods, tend to kill, maim and destroy livelihoods more diffusely, across more time and a broader geography, which makes the aggregate harder to perceive. Heat, in particular, does not produce much visible wreckage. It leaves no collapsed buildings, no flooded streets, no dramatic aerial photographs. It kills quietly, mostly indoors, mostly among people who were already vulnerable.

The undercounting problem compounds this invisibility. The WHO’s figure of 489,000 deaths, significant as it is, rests on data that is patchy, methodologically inconsistent, and essentially absent in many of the most heat-exposed regions of South and Southeast Asia and Africa. When you look at how heat mortality is actually measured — and there are three quite different methods in use globally, described in Box 1 below — you begin to understand why the true number may be substantially higher. What we can say with some confidence is that WHO’s own regional office in Europe reported in June 2026 that heat had contributed to more than 200,000 premature deaths across Europe alone in just four years. That is a figure that deserves to be repeated slowly.

What makes heat deaths particularly difficult to count is that heat is, in most cases, not a direct cause of death in the medical sense. It is a risk multiplier. When someone with cardiovascular disease dies of a heart attack during a heatwave, their death certificate states “cardiac arrest.” When a person with chronic kidney disease deteriorates on the third day of a heat event, the record shows “renal failure.” Heat is the context and trigger, but it rarely appears in the paperwork. This is not unique to heat — flood mortality statistics, for example, also struggle with the distinction between direct deaths (drowning) and indirect deaths (disease outbreaks, infrastructure failure, displacement-related mortality) — but heat’s invisibility as a primary cause makes the undercounting particularly severe.

The response gap mirrors the counting gap. After the 2003 European heatwave, much of the action was driven by the health sector: hospitals were required to maintain registers of vulnerable patients and, when a heatwave was forecast, to contact those patients, check whether they had cooling capacity, and ensure that relatives or carers were alerted. It was a relatively straightforward, low-cost intervention — and it worked. But it was a health-sector response. The meteorological community and the disaster risk reduction community were much slower to build equivalent frameworks. That gap — between what health practitioners did and what the broader disaster management community developed — has never been fully closed.

Most major hazards have a number — heat does not

Here is what I find striking about the current state of the field. We have spent decades building precise, operational, globally agreed definitions for most major natural hazards. Those definitions are not academic exercises. They are the machinery by which warnings are issued, protocols are triggered, and resources are mobilised. I am using hydrometeorological examples since heatwaves fall into that category, but seismic hazards have these “numbers” as well.

Consider El Niño. Scientists isolate a specific grid in the east-central tropical Pacific — the Niño-3.4 region — and define an El Niño event when the sea surface temperature in that specific grid exceeds 0.5°C above the long-term baseline for a sustained period. One number, one location (albeit a large one), one threshold. And once it is crossed, the consequences are well-understood and expected: greater typhoon intensity in the western Pacific, a shift in rainfall from Southeast Asia to the eastern Pacific, elevated drought risk across parts of Africa and South Asia. The definition unlocks a whole cascade of preparedness action.

Tropical storms work in a similar way, where the principles are the same with some global variation in details. The storms are categorised based on wind speed — the agreed single measure of a storm’s power — and are followed (the speed and direction of the storm system) to provide the basis of advisories on when and where the “landfall” is expected. The thresholds, naming conventions, and averaging periods differ by region (see Box 2 below), but the principle is consistent everywhere: a number triggers an action. The category and expected landfall trigger the action on the ground — a specific set of protocols, be it school closures, evacuation orders, maritime restrictions, emergency service mobilisation, or others.

Even at a regional level, impact-based forecasting for storms allows response to be targeted where it is most needed and most effective — distinguishing, for instance, between the risk of storm surge in coastal areas, the risk of inland flooding from concentrated rainfall, or infrastructure damage due to high wind speeds.

Cold spells in Northern Europe and Canada offer an instructive parallel for heat, because the mechanism is similar: a physiological threshold being crossed by ambient conditions. The wind chill index combines temperature and wind speed into a single “felt temperature” figure. At specific values, outdoor work protocols kick in: mandatory warming breaks, coverage requirements, and ultimately work stoppages. The threshold itself is locally calibrated — Canada has work-stoppage thresholds for outdoor workers at wind chill equivalents which differ between colder Saskatchewan and milder southern Ontario. Finland does the same across its own latitudinal range. Same wind chill index, same underlying formula, locally calibrated thresholds — and a downstream chain of action that is standardised, though the trigger point differs by location. The process of establishing it, and the protocols that follow from it, are standardised. The number triggers the action.

Now look at heat. The World Meteorological Organisation (WMO) and WHO’s joint 2015 guidance defines a heatwave as “periods of unusually hot and dry or hot and humid weather that have a subtle onset and cessation, a duration of at least two to three days and a discernible impact on human activities.” The United Nations Office for Disaster Risk Reduction (UNDRR )’s own terminology page notes that this definition “is not sufficient to guide National Meteorological and Hydrological Services in developing practical methods and tools for a heatwave monitoring system that would allow comparisons across regional or international” scales.

Read that slowly. The UN’s own disaster risk reduction agency is telling us that the primary international definition of a heatwave is not sufficient to do the job it needs to do. And the phrase at the heart of the problem is “discernible impact on human activities.” A discernible impact is not a threshold. It is a description of what has already happened. By the time impact is discernible, people are already affected, even dying.

The contrast with how different countries currently define heatwaves in practice makes the fragmentation vivid — but looking at it more carefully, it also reveals something more interesting than simple incomparability.

India’s Meteorological Department (IMD) declares a heatwave when maximum temperature reaches 40°C or above on the plains and represents a departure of at least 4.5°C from the station’s long-term normal. The Philippines’ weather agency, PAGASA, uses the Heat Index (HI), which combines temperature and relative humidity into a “feels like” temperature, and classifies conditions as dangerous at a HI of 42°C or above.

Both are right about what they are trying to measure. IMD has correctly grasped that what matters is not the absolute temperature but the departure from local norms — that the same 42°C means something very different in Rajasthan, where it is a hot but not unusual day, than in a station where the normal maximum is 35°C. PAGASA has correctly grasped that in a tropical, high-humidity environment, the dry-bulb temperature alone is physiologically misleading — a HI of 42°C captures what the body is actually dealing with in ways that the thermometer reading does not.

Each has identified a genuine piece of the problem. What neither captures is the overnight recovery dimension — whether the night that followed yesterday’s heat allowed the body, the building, and the urban environment to discharge the accumulated thermal load before the next day’s heat arrives — nor the accumulation that builds when it does not. And because they are built on different variables through different processes, with no shared severity scale, there is no way to read from one to the other, or to compare a heatwave in Manila with one in Delhi, or either with what European services are measuring.

That is the common language that is missing — not a single universal threshold, which would be wrong for different climates, but a standardised process for deriving locally appropriate thresholds, and a shared intensity scale that makes the results comparable. Without that, there is no common learning. PAGASA itself acknowledged in 2024 that its existing system has limitations and that it is actively exploring improved methods — the instinct is there, and it is the right instinct. What is missing is the framework that would allow that improvement to happen systematically, in a way that connects Manila to Delhi to Paris to Nairobi.

Why definition matters — from description to action

The argument for a more precise, operational definition of heatwaves is not about scientific tidiness. It is about the chain from observation to action, and right now that chain has a critical link missing.

Think about what a definition does in an emergency management context. It sets a threshold. That threshold triggers a protocol. The protocol mobilises resources, issues warnings, activates health services, protects workers, and reaches vulnerable people before they are in danger rather than after. Without a definition, each of those steps becomes a judgement call, made under pressure, with inconsistent criteria. That is where people slip through.

The humidity dimension makes this more complex for heat than for cold, and it is worth dwelling on for a moment. When temperature and humidity are both high, the human body’s primary cooling mechanism — sweating — loses much of its effectiveness. Sweat can only cool you when it evaporates, and when the surrounding air is already saturated with moisture, evaporation slows dramatically. What the thermometer says and what the body experiences diverge sharply. A reading of 38°C at 80% relative humidity — common across coastal South and Southeast Asia — is physiologically far more dangerous than 38°C in a dry desert climate, with a fan or a breeze providing real cooling. Researchers have established that a wet-bulb temperature of 35°C — the combined measure of heat and humidity — is broadly the threshold beyond which even a healthy, resting adult with water and shade faces a fatal rise in core body temperature within hours.

This is precisely why a universal numerical definition of heat danger — a single global threshold — is genuinely not appropriate. Coastal Pakistan is not Finland. Manila is not Madrid. The physiological risk at any given temperature depends on acclimatisation, humidity, building stock, access to cooling, occupational exposure, and age profile. A threshold that works for one context will either under-warn or over-warn in another.

There is also a time dimension that temperature-at-a-moment cannot capture. Heatwaves that persist through the night are more dangerous than day-only events, because the body uses cooler nights to recover from daytime heat stress. When nights stay above 25°C or 30°C, that recovery does not happen, and the cumulative physiological burden builds day by day. The 2003 European heatwave was lethal partly because of its duration and the persistent overnight warmth — not simply because daytime temperatures were high.

None of this means, however, that we must tolerate the current definitional vacuum. The cold-spell analogy is instructive here too. Canada and Finland both use wind chill protocols, but their precise thresholds differ because their climates, worker populations, and baseline conditions differ. What is standardised is not the number itself, but the process by which the number is derived and the framework of action that flows from it, much like tropical storms. That is what is needed for heatwaves.

What the institutions have — and have not — done

It would be unfair to suggest that nothing has happened at the international level. In the past two years in particular, there has been a genuine acceleration of institutional attention to extreme heat, and it is worth acknowledging that before identifying what is still missing.

At COP30 in Belém in November 2025, UNDRR, WMO, the Global Heat Health Information Network, and Duke University jointly launched an Extreme Heat Risk Governance Framework and Toolkit — the direct product of the UN Secretary-General’s Call to Action on Extreme Heat issued in July 2024. The framework is intended to help countries assess their governance maturity, identify gaps, and build toward integrated, systematic heat-risk reduction across short, medium, and long time horizons. Pilot implementation is under way in 2026 in Barbados, Senegal, and Cambodia.

In June 2026, WHO/Europe launched a second edition of its Heat–Health Action Plans Guidance, framing heat as “no longer just an exceptional seasonal phenomenon” but a “critical public health issue amplified by climate change.” The guidance covers eight core elements of a heat-health action plan and explicitly targets policymakers and healthcare professionals, pushing them to shift from emergency response toward long-term prevention.

These are genuinely significant developments, and they should not be undervalued. The fact that the UN Secretary-General is personally championing extreme heat, that a governance framework has been produced at global scale, and that WHO is updating its operational guidance in real time during a continent-wide emergency — all of this represents a step change from where the field was a decade ago.

And yet. A governance framework is not an operational definition. Knowing who is responsible for coordinating the heat response — which ministry leads, which agency issues the warning, which budget line covers the cooling centres — is important. But it only works if those responsible parties have a clear, agreed threshold at which their responsibilities are activated. Without that threshold, the framework has no ignition point. Right now, that alarm is not standardised, not agreed, and in many countries, not even designed.

The case for localised, operational definitions — and a standardised process to derive them

I want to be precise about what I am and am not arguing here, because the logical tension is real and it matters.

I began by noting that most major hazards have a number — a precise, agreed, operational definition. I then said that a single global temperature threshold for heatwaves is not appropriate, because the danger depends heavily on local conditions. These might seem contradictory. They are not.

The reason why is this: what needs to be standardised is not the number itself, but the process by which locally-appropriate numbers are derived, and the framework of action that flows from those numbers once established. El Niño has one number because it is measured in one specific ocean grid where global conditions are directly relevant. Tropical storms have some different scales, but they are calculated in the same way — they measure the sustained wind speed.

A heatwave in Dhaka and a heatwave in Helsinki are both heatwaves — but the physiological risk, the acclimatisation baseline, the building stock, and the vulnerable populations are so different that the same temperature means different things in each city. That is not a failure of the definition; it is a feature of the hazard.

What is urgently needed is a standardised methodology — a clear, reproducible process — that national meteorological and hydrological services, working alongside public health authorities and disaster risk reduction practitioners, can apply in their own context to derive a locally-valid heat threshold. The work of researchers like John Nairn and colleagues in Australia on the Excess Heat Factor (EHF) offers one serious approach: rather than comparing temperature against a fixed absolute value, it measures the departure from local climatological norms combined with the cumulative heat stress built up over several consecutive days. The resulting metric can be applied in Manila or in Melbourne, with locally calibrated baselines, and produce a result that is both meaningful and comparable within each context.

There is important recent precedent here too. The UNDRR terminology page on heatwaves notes that temperature thresholds are often defined using the 90th or 95th percentile of local temperature distributions — in other words, local conditions relative to what that specific population is acclimatised to. That approach, consistently applied through a standardised process, is the direction of travel.

The second article in this series will explore what that process might look like in practice — drawing on the EHF framework, the HI approach used in Southeast Asia, and the emerging work on impact-based forecasting to show how a locally-calibrated, operationally useful heatwave definition can be built from available data and methods. But that article rests on an argument that this one needs to make clearly: the definitional work is the foundation. Without it, the governance frameworks, the action plans, the cooling centres, and the public warnings are all built on sand.

The urgency is right now

Returning to where I started. Europe is in a heatwave right now, in July 2026. The warnings are real. The deaths are real. The frameworks developed since 2003 have demonstrably saved lives — France’s post-2003 national heatwave plan is estimated to have saved around 4,400 lives in the 2006 heatwave alone, and mortality in the 2022 European heatwaves was substantially lower than it would have been under the pre-2003 baseline. Progress has been made.

But the foundation — a clear, locally-calibrated, operationally useful definition of what a heatwave actually is, built through a standardised process that any country can apply — is still missing. The institutions know it. The UNDRR says its own definition “is not sufficient.” WHO is updating guidance in real time. WMO is mobilising early warning systems. And yet the alarm that triggers all of it — the agreed, precise, locally-derived threshold — remains, in most countries and most regions of the world, absent.

National meteorological and hydrological services, working with public health authorities and disaster risk reduction practitioners, need to agree on a standardised process for deriving local heat thresholds. The threshold itself will vary by location. The process for deriving it should not. Until that happens, the heat emergency response is often improvised. And people are dying who should not be.

References section

• DOST-PAGASA (2024) Press Statement: Official Statement on Creating Extreme Heat Protocol, 12 April 2024. Quezon City: PAGASA.
• Gasparrini, A., Guo, Y., Sera, F. et al. (2017) ‘Projections of temperature-related excess mortality under climate change scenarios’, The Lancet Planetary Health, 1(9), pp. e360–e367.
• Nairn, J.R. and Fawcett, R.J.B. (2015) ‘The Excess Heat Factor: A Metric for Heatwave Intensity and Its Use in Classifying Heatwave Severity’, International Journal of Environmental Research and Public Health, 12(1), pp. 227–253.
• UNDRR (2023) Heatwave (MH0501). Sendai Framework Terminology.
• UNDRR, WMO, Global Heat Health Information Network and Duke University (2025) Extreme Heat Risk Governance Framework and Toolkit. Launched at COP30, Belém, Brazil, 11 November 2025.
• World Health Organization Regional Office for Europe (2026) Heat–Health Action Plans Guidance, 2nd edn. Copenhagen: WHO Europe.
• World Health Organization (2024) Climate Change and Health: Heat and Health Fact Sheet.
• World Health Organization Regional Office for Europe (2026) Statement — Europe lost 200,000 people to heat in 4 years yet nearly all of them were preventable, 11 June 2026.
• World Meteorological Organization (2026) Records fall as extreme heat grips Europe, WMO News, 26 June 2026.
• World Meteorological Organization and World Health Organization (WMO/WHO) (2015) Heatwaves and Health: Guidance on Warning-System Development. Geneva: WMO.
• British Columbia Coroners Service (2021) Chief Coroner’s Statement on Public Safety During High Temperatures. Victoria: BC Government.
• India Meteorological Department (IMD) (n.d.) FAQ on Heat Wave. New Delhi: IMD. World Weather Attribution (2026) Fossil fuel emissions have rapidly worsened European heatwaves in just a few decades, 26 June 2026.
• DOST-PAGASA (2024) Press Statement: Official Statement on Creating Extreme Heat Protocol, 12 April 2024. Quezon City: PAGASA.