Three Months of Unbroken Heat: When East Asia's Summers Never Cool Down

Three months. That is roughly how long heat could press down on East Asia in a single warm season by the end of this century if greenhouse gas emissions keep climbing. Not scattered hot spells punctuated by cooler nights, but something far more brutal: heat that does not give you a break when the sun goes down.

Almost all the heatwaves you have ever survived were daytime events. The sun beats down, the pavement radiates, you retreat inside, and at night the temperature finally drops enough for your body to recover. But there is a nastier cousin of the heatwave — the "concurrent daytime and nighttime heatwave," or CDNHW — in which both the daily maximum and the daily minimum temperatures stay above their historical thresholds for days on end. Your body never gets that nightly reprieve. Recovery time from heat exposure, the mechanism that lets cells repair and organs cool, effectively disappears.

A team of South Korean climatologists has now mapped how often, how intensely, and how widely these unrelenting heat events could strike East Asia by 2071–2100 (Kim et al., 2023). Their verdict, based on a dozen coupled regional climate models, is staggering. Under the most extreme emissions pathway, the average East Asian location could experience CDNHWs for three months out of every year — roughly a third of the entire calendar — with intense, multi-week stretches that are essentially unprecedented in the historical record.

The Science

The study comes from researchers at Pukyong National University, Kongju National University, and Jeju National University in South Korea, led by Young-Hyun Kim. Their task was to answer a question that single climate models cannot: what will future heat stress across East Asia actually look like, at a resolution fine enough to matter for real cities and real people?

Global climate models (GCMs) simulate the whole planet, but their grids are far too coarse to capture regional detail — a mountain range, a coastline, a megacity. Regional climate models (RCMs) zoom in, using the global model's output as boundary conditions to simulate weather at a much finer scale. Combining several global models with several regional models produces an "ensemble" of GCM–RCM chains, each a slightly different plausible version of the future.

The team harnessed 12 such chains from CORDEX-East Asia Phase 2, a coordinated project that runs regional climate simulations over East Asia at a 25-kilometer resolution — fine enough to represent the region's complex terrain and the sharp temperature contrasts between coast and interior. From three-hourly temperature output, they derived daily maximum and daily minimum temperatures for every grid cell.

They then applied a standard climate stress test. A day counts as a "hot day" or "hot night" when its temperature exceeds the 90th percentile of historical values for that calendar day, calculated over a 1981–2005 baseline. A CDNHW occurs when both conditions hold on the same day. The team tracked how often these events happened, how long they lasted, how hot they got, and how much area they covered — first in the historical period, then in projections for 2071–2100 under four emissions scenarios.

The four scenarios are the workhorses of climate science. Two are older "Representative Concentration Pathways" (RCPs): RCP2.6, an aggressive mitigation pathway that roughly keeps warming near 1.5–2°C, and RCP8.5, a high-emissions "business as usual" extreme. Two are newer "Shared Socioeconomic Pathways" (SSPs): SSP1-2.6, the sustainability path, and SSP5-8.5, the fossil-fueled development path. Running both old and new scenario families lets the results speak to two generations of climate modeling while keeping the emissions endpoints comparable.

The core idea is simple but medically profound. A daytime heatwave is stressful; a nighttime heatwave prevents recovery; but a heatwave that is both, simultaneously, stacks the two harms. By measuring concurrent events rather than either kind alone, the study captures the heat that actually kills.

What They Found

In the historical period (1981–2005), CDNHWs were a rare event across East Asia, appearing on average for about 10.9 days per warm season (April through September) and making up just 0.9 percent of all days in that window. In other words, the kind of heat that removes the body's nightly recovery window was, for most of the past half-century, a once-a-summer curiosity.

That changes dramatically in every future scenario, but the differences between pathways are enormous. Under RCP2.6 — the strong-mitigation world — the average CDNHW season grows to about three weeks and 3.7 percent of warm-season days. Painful, but survivable. Under RCP8.5 (high emissions), CDNHWs would occupy two months and 20.5 percent of the warm season. Under SSP1-2.6, two months and 15.6 percent. And under SSP5-8.5 — the fossil-fuel-heavy extreme — three months and a staggering 45.7 percent of the April-to-September season.

The share of summer spent in unrelenting heat

Label	Value
Historical (1981–2005)	0.9
RCP2.6 (2071–2100)	3.7
SSP1-2.6 (2071–2100)	15.6
RCP8.5 (2071–2100)	20.5
SSP5-8.5 (2071–2100)	45.7

East Asian mean CDNHW occurrence rate (% of warm-season days). Source: Kim et al. (2023), Int. J. Climatol. 43(10).

Those headline averages hide an even more disturbing shift in intensity and persistence. Historical CDNHWs were brief affairs; a handful of consecutive nights of unrelenting heat was the exception. The researchers found that only 0.2 percent of historical CDNHWs lasted longer than ten days. Under SSP5-8.5, that share jumps to 17.3 percent — nearly a fifth of all concurrent heat events would stretch past ten days, with many lasting far longer.

How heatwaves become heat regimes

Label	Value
Historical (1981–2005)	0.2
RCP2.6 (2071–2100)	1.2
SSP1-2.6 (2071–2100)	6.1
RCP8.5 (2071–2100)	7.2
SSP5-8.5 (2071–2100)	17.3

Share of CDNHWs persisting >10 days (%). Source: Kim et al. (2023).

A ten-day CDNHW is not a heatwave; it is a weather regime. It is heat that occupies the atmosphere the way a monsoon occupies a season.

The geography of the increase is telling. Not all of East Asia suffers equally. The largest jumps in both occurrence and intensity concentrate over Indochina, eastern and western China, and India — regions that are already among the most densely populated and least air-conditioned on Earth. These are places where outdoor labor is the norm, where agriculture depends on field work, and where millions of people sleep in dwellings with no mechanical cooling. The models project that these regions, which already flirt with dangerous heat, will bear the brunt of the escalation.

Spatial extent expands as well. Heat events that historically hugged certain corridors begin to sprawl across entire subregions, covering more grid cells simultaneously, so that a much larger population is exposed to the same unrelenting heat at the same moment — which simultaneously strains power grids, water supplies, and health systems across national borders.

Why This Changes Things

To grasp why concurrent heat matters so much, consider what a hot night does to the human body. During a normal hot day, the body sheds heat through sweating and peripheral blood flow; at night, when temperature falls, that load eases and tissues repair. But when nighttime minimums stay elevated, the body cannot unload the day's accumulated heat. Core temperature stays high through the night, sleep is disrupted or lost, cardiovascular strain persists, and by the next morning the body starts the day already in deficit (Kim et al., 2023). Accumulate that across consecutive days and the result is heat exhaustion, heatstroke, aggravated heart and kidney disease, and — in the most extreme events — mass mortality.

This is not a theoretical concern. The 2003 European heatwave, which killed tens of thousands, and the 2010 heatwave in Russia, which killed over 50,000, both featured unusually warm nights that denied victims any respite. Epidemiological studies consistently find that mortality risk rises sharply when nighttime temperatures stay high, because vulnerable populations — the elderly, the chronically ill, the outdoor workers, the urban poor — lose their only line of defense.

What this study shows is that East Asia, under high emissions, could face this compounding stress as the default summer state rather than an exceptional event. A seasonal average of 45.7 percent means that nearly every other day in the warm half of the year offers zero nightly cooling. For a region containing roughly two billion people, many of whom live in some of the hottest and least-adapted cities on the planet, the public-health load this implies is difficult to overstate.

Critically, the results also reveal how much hinges on which pathway the world takes. The gap between the moderate-mitigation scenarios (RCP2.6 and SSP1-2.6) and the high-emissions ones (RCP8.5 and SSP5-8.5) is not a matter of degrees; it is a matter of whole orders of magnitude in exposure. Under strong mitigation, CDNHWs remain a serious but periodic threat. Under continued high emissions, they become a defining feature of East Asian summers. In plain terms: the difference between the futures is not whether East Asia heats up, but whether its summers remain livable for those who have to work and sleep in them.

How many days of unbroken heat, per summer

Label	Value
Historical (1981–2005)	10.9
RCP2.6 (2071–2100)	21
RCP8.5 (2071–2100)	60
SSP1-2.6 (2071–2100)	60
SSP5-8.5 (2071–2100)	90

East Asian mean CDNHW occurrence period in days. Approximations of the study's reported values of 3 weeks, 2 months, and 3 months. Source: Kim et al. (2023).

The study's reliance on a multi-model ensemble is both a strength and a necessary disclaimer. No single climate model is perfectly faithful; each has biases in how it represents clouds, soil moisture, and temperature extremes. By averaging across 12 GCM–RCM chains, the researchers smooth out individual model quirks and present a range of plausible outcomes. But the results remain projections, not predictions — they describe what the physics says is likely under each emissions pathway, with real uncertainty in the exact numbers.

What's Next

The most urgent next step is adaptation that treats nighttime heat as a first-class threat, not an afterthought. Urban planning has historically focused on shading and ventilation for daytime comfort; this study argues that cooling people at night matters just as much. That means expanding access to air conditioning but also deploying cheaper, lower-carbon interventions: reflective and green roofs, urban tree cover that cools the night air, cool pavements, and public cooling shelters in neighborhoods whose residents cannot afford mechanical cooling at home.

It also raises hard equity questions, because the scenario in which CDNHWs explode — SSP5-8.5 — is precisely the world in which the least-adapted nations are left to cope alone. Indochina, India, and interior China are among the regions projected to face the largest increases, and they are also the regions with the lowest penetration of air conditioning and the highest dependence on outdoor labor. A carbon-constrained world that also invests in climate-resilient infrastructure for vulnerable regions would substantially soften the blow that the high-emissions scenarios describe.

Scientifically, the next step is to push these projections still finer and connect them directly to health outcomes. The 25-kilometer grid is a major improvement over coarser global models, but urban heat islands — cities amplified heat relative to their surroundings — operate at scales of a kilometer or less. Coupling regional climate projections with epidemiological models of heat-mortality risk could convert "days of CDNHW" into "expected excess deaths," giving planners a number they can actually plan around. Refining the definition of CDNHWs to account for humidity, which the human body experiences as heat more than raw temperature does, would likewise sharpen the picture of real-world stress.

There is also a specific modeling question the authors flag: whether current models capture the persistence of these events faithfully. The dramatic rise in multi-week CDNHWs under high emissions rests on model dynamics of soil-moisture feedback, where dry ground heats up more and reinforces heat — a feedback loop that is only as trustworthy as the models' representation of land–atmosphere coupling. Verifying this against observed extreme years, and against models with better land-surface physics, will tell us whether the most alarming projections are robust.

What this study makes unambiguous is that the choice confronting East Asia — and the world — is a choice about how much unbroken heat people must endure. The difference between 10.9 days of unrelenting day-and-night heat and three months of it is not an abstract climate statistic. It is the difference between a summer that punishes the vulnerable and one that overwhelms them. The models say both futures are physically possible. Which one arrives depends on emissions choices made now, and on whether the regions projected to suffer most are given the resources to cool their nights before those nights stop cooling on their own.