Mapping climate hazards

Aerial view of air conditional array in the middle of a building complex.

Explore the full report

Advancing adaptation: Mapping costs from cooling to coastal defenses estimates the costs of climate adaptation using a granular, geospatial analysis and examining 20 proven, cost-effective adaptation measures.

Read the report

The McKinsey Global Institute analyzed how nine climate hazards manifest today and how their patterns might shift as the world warms, in an effort to understand the costs and benefits of adaptation.

Technical appendix (15 pages)

We find that today, about 40 percent of the Earth’s landmass experiences hazards including severe heat, destructive wildfires, prolonged drought, and severe flooding, which are rooted in current local patterns of weather extremes.¹Observations of global temperatures indicate that Earth warmed by 1.1°C relative to preindustrial levels in the decade prior to 2020, which is when our analysis starts and what we refer to as “current” or “today’s” climate conditions. (The average of the most recent estimates for the 2015–24 decade is 1.24°C of warming relative to the global temperature between 1850 and 1900.) In this report, warming levels refer to long-run average temperatures rather than temperatures in any given year, in line with typical climate analysis. See “Summary for policymakers,” in Climate Change 2021: The Physical Science Basis, Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, 2021; and “Indicators of global climate change 2024: Annual update of key indicators of the state of the climate system and human influence,” in Earth System Science Data, Copernicus Publications, 2025. Our analysis excludes Antarctica because it is not populated. Heat, wildfires, and drought affect wide swaths of land, while flooding hazards are often more localized, shaped by geography, coastlines, and local rainfall. In addition, some 35 percent of the world’s terrain experiences freezing days (see sidebar, “The hazards we examine”). In response, many places around the world have implemented measures to adapt to these hazards.

Yet today’s view offers only a snapshot of what is actually a moving picture. Global temperature increases of 1.5°C and 2°C, anticipated by roughly 2030 and 2050, respectively, under the current trajectory of global emissions, will cause kaleidoscopic local shifts.²

We assessed 25 current-trajectory emissions scenarios, based on policies announced or implemented, from the IPCC, NGFS, and IEA to reflect the wide range of climate estimates. These scenarios suggest that 1.5°C of warming relative to preindustrial levels would be reached somewhere between 2025 and 2035, when warming is measured as a 20-year average centered on a given year and relative to a preindustrial average. In 2030, temperatures across the 25 scenarios range from 1.45°C to 1.55°C. In the IPCC scenarios, the median level of warming by 2050 is 2.0°C, and in the more recent NGFS and IEA scenarios, it is 2°C. By 2100, the temperature estimates from the same set of models are 3.0°C (average across IPCC scenarios, with a maximum of 3.7°C), 2.9°C (NGFS and IEA Current Policies), and 2.5°C (IEA Stated Policies). The IPCC and NGFS scenarios are based on the assumption that currently announced policies are preserved. For NGFS, this covers national climate policies that were legislated and supported by instruments as of March 2024. For IPCC scenarios, the specific policies reflected vary by model. The IEA Current Policies scenario (2025), by contrast, considers only policies already adopted in legislation and regulation, assuming no future changes, even where governments have indicated their intention to do so, and takes a cautious perspective on the pace at which new energy technologies are deployed and integrated into the energy system. The IEA Stated Policies scenario draws on a wider interpretation of the policy environment, incorporating not only enacted measures but also formally proposed ones and other official strategy documents that signal the intended policy direction. For details, see IEA, World Energy Outlook 2025, November 2025.

On these trajectories, warming is projected to rise beyond 2°C in the second half of this century. Because some adaptation measures take more than a decade to implement, anticipating needs at least for 2050 and a 2°C world now is prudent, even as the world works to reduce emissions. Moreover, some measures implemented today—for example, stormwater networks—can have multiple-decade lifetimes, making it prudent to consider how hazards could evolve a few decades out, to avoid costly retrofits down the road.

The hazards we examine

We analyze nine types of hazards across categories—heat, wildfires, drought, flooding, and cold—to understand how people and places are exposed to them today and going forward, and analyze the costs to adapt.

While hazards are often described as “extreme weather events,” in climate science the term refers to broadly both chronic and acute hazards. Three hazards we examine, heat stress, wildfire weather, and freezing days are chronic and happen every year in the places that are exposed to them. The remaining six hazards we analyze—coastal flooding, riverine (fluvial) flooding, flooding caused by excessive rainfall (pluvial), heat waves, nonsurvivable heat, and drought in agricultural areas—are acute, meaning they are intense but rare events. For acute events, places may be exposed to them, but the actual event may not manifest every year.

These hazards can occur along a continuum of intensity, frequency, or duration, requiring a threshold for each in order to define the hazard. Thresholds can be established in various ways. Since our focus is on understanding and quantifying adaptation costs, we define them by drawing on protection standards commonly established in developed economies. Of course, even though the hazards we examine are considered meaningful enough to protect against, not all hazards are created equal and their impacts could be very different. Heat stress, defined as more than a month of very hot and humid conditions each year, and heat waves, rare, multiday occurrences of local temperature extremes, can differ markedly in the nature and magnitude of their impacts, but both are considered important enough to warrant protection. Similarly, coastal flood protection standards in developed economies typically aim to withstand a one-in-100-year flood event, while pluvial flood protection standards often correspond to a one-in-20-year flood event.

Because protection standards are not always strictly defined, we exercised some judgment in our choice of thresholds and further validated them using literature from the Intergovernmental Panel on Climate Change (IPCC) on typical hazard definitions and their impacts (exhibit).¹We validated hazard thresholds with IPCC-referenced literature including, for heat stress, Camilo Mora et al., “Global risk of deadly heat,” Nature Climate Change, July 2017, Volume 7; for heat waves, Alessandro Dosio et al., “Extreme heat waves under 1.5 °C and 2 °C global warming,” Environmental Research Letters, May 2018, Volume 13, Number 5; and Cong Yin et al., “Changes in global heat waves and its socioeconomic exposure in a warmer future,” Climate Risk Management, 2022, Volume 38; for nonsurvivable heat, Eun-Soon Im, Jeremy S. Pal, and Elfatih A. B. Eltahir, “Deadly heat waves projected in the densely populated agricultural regions of South Asia,” Science Advances, August 2017, Volume 3, Number 8; for wildfires, John T. Abatzoglou, A. Park Williams, and Renaud Barbero, “Global emergence of anthropogenic climate change in fire weather indices,” Geophysical Research Letters, January 2019, Volume 46, Number 1; for drought, Yadu Pokhrel et al., “Global terrestrial water storage and drought severity under climate change,” Nature Climate Change, March 2021, Volume 11; for coastal flooding, Robert J. Nicholls et al., “A global analysis of subsidence, relative sea-level change and coastal flood exposure,” Nature Climate Change, April 2021, Volume 11, Number 4; and for riverine and excessive rainfall flooding, Jun Rentschler, Melda Salhab, and Bramka Arga Jafino, “Flood exposure and poverty in 188 countries,” Nature Communications, June 2022, Volume 13. Continued efforts are needed to develop robust and comparable ways to quantify the impact of hazards and their link to adaptation metrics.

Naturally, impacts do not begin to appear only when a particular threshold is crossed. Rather, our thresholds are a way to dimensionalize exposure and protection levels and apply a consistent approach in line with typical design standards for protection.²In some cases, thresholds are based on local parameters, namely the intensity of locally extreme events. For example, we have defined exposure to drought in line with literature from the IPCC reflecting a local minimum in moisture in the soil, rather than an absolute measure of soil water content. This approach reflects the fact that agricultural activity depends on local climatic conditions and that changes relative to these conditions are most relevant for guiding adaptation responses. Exposure to heat stress, by contrast, is based on absolute physiological thresholds related to human productivity loss; this is because here, too, it is these thresholds that determine when an adaptation response becomes necessary. We focus on a subset of climate hazards that meet three criteria: (1) they have a well-established and direct link to climate change as the primary driver; (2) they are not slow-onset events, the impacts of which emerge over time; and (3) they have sufficient data to enable spatial analysis.³Because this work excludes slow-onset risks such as saline intrusion, we acknowledge that this excludes the coverage of key risks in some geographies, including risks for which there may be hard limits around 2°C, such as communities on small islands at risk of freshwater shortages. See “Key risks across sectors and regions,” Table 16.3, in Climate Change 2022: Impacts, Adaptation and Vulnerability, Working Group II Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, 2022.

For example, the link between water stress and climate change is not direct, since other factors, particularly population growth and rising demand for water, are at play.⁴ While these and other hazards may be highly relevant for adaptation planning in any given location, they fall outside the scope of our global analysis. See the technical appendix for more details.

We examine chronic and acute hazards the world experiences today.

Importantly, change will occur, but not everywhere, not in the same way, and not all at once. While many hazards will spread across more parts of the world, some will not. Freezing is expected to decline, and in some small areas, drought or flood risks may also decrease—even while increasing in others. Hazards may not shift in the same way, either. Many will intensify, lengthen, or happen more frequently as the planet warms, but the scale of the change and where it occurs will vary. For example, regions like Latin America and parts of emerging Asia could face much longer periods of heat stress at 2°C, whereas India, already subject to prolonged heat conditions, is likely to see a smaller relative increase. Finally, exposure to hazards won’t increase all at once; more places are expected to become exposed between now and 1.5°C, and still more by 2°C.

The following hazard spotlights explain where, when, and how each type of hazard could change as global temperatures rise from current levels to 1.5°C and 2°C above preindustrial levels. For each hazard, we highlight the change, whether in intensity, frequency, or duration, that matters most for guiding adaptation planning.³These hazards occur along a continuum of intensity, frequency, or duration, requiring a defined threshold for understanding the hazard. Thresholds can be defined in various ways. Since our focus is on understanding and quantifying adaptation costs, we draw on protection standards commonly established in developed economies. Of course, the hazards we examine may not manifest in the same way or have comparable impacts. For instance, coastal flood protection standards in developed economies typically aim to withstand a one-in-100-year flood event, while pluvial flood protection standards often correspond to a one-in-20-year flood event. We further validated these thresholds using literature from the IPCC on typical hazard definitions and their impacts. For more details on our approach to understanding climate hazards and their behavior with warming, refer to the technical appendix. The findings presented apply only to the specific hazard metrics and definitions used in this research. The view of how hazards evolve over time and adaptation needs could look different for other thresholds and hazard definitions. Naturally, the prevalence of a hazard does not necessarily mean places need to or will adapt to it, but understanding today’s patterns of extreme weather and their evolution is the first step in adaptation planning.

In doing this work, we could define hazards in many ways, influencing the results shown here. Since our focus is on understanding and quantifying adaptation costs, we define hazards using protection standards typically established in developed economies. And, of course, even though all the hazards examined here are considered meaningful enough to protect against, they are not all created equal, and their impact can be very different. For example, heat stress events are prolonged periods of high heat or humidity, lasting more than a month each year, that broadly affect labor productivity and health. By contrast, heat waves are rare, short-lived, localized extreme heat events that particularly put vulnerable populations at risk.

Acknowledgments

This report builds on a larger body of work from the McKinsey Global Institute (MGI) that explores responses to a shifting climate, including an assessment of the costs of mitigation and the tools needed to achieve it. The research was led by Mekala Krishnan, an MGI partner; Olivia White and Sylvain Johansson, MGI directors; Sven Smit, MGI’s chair; and Annabel Farr and Kanmani Chockalingam, MGI senior fellows.

Many McKinsey colleagues and alumni also participated in this research. During the early stage of research, Gualtiero Jaeger led the working team, which included Riley Brady, Enrico Calvanese, Clark Doman, Jamie Dunn, Meredith Fish, Marie Houdou, Caroline Jupe, Sonya Kalara, Francois Klein, Dhiraj Kumar, Jake Lieberfarb, Sophia Lien, Trevor Liu, Shilpita Mathews, Reid McLaughlin, Adam Nowakowski, Brandon Nguyen, Shreyangi Prasad, Khalyani Sankar, Aidan Sarazen, and Emily Spittle. Expert members of the team included Jean-François Lamarque and Jan Lenaerts.

We are grateful to the academic advisers who provided guidance, challenged our thinking, and sharpened our insights: Simon Dietz, professor of environmental policy in the Department of Geography and Environment at the London School of Economics; Sam Fankhauser, professor of climate change economics and policy at the Smith School and the School of Geography and the Environment at Oxford University; David Stainforth, fellow of the Royal Meteorological Society and professorial research fellow at the London School of Economics; Wim Thiery, associate professor at Vrije Universiteit Brussels, who additionally reviewed the methodological design and the results on the evolution of climate hazards; and John Ward, visiting senior fellow at the Grantham Research Institute on Climate Change and the Environment at the London School of Economics.

This research benefited greatly from the expertise of academics who shared their insights and helped shape analysis: Sara Boettiger, senior fellow at Harvard University; Matthew Huber, professor in the Department of Earth, Atmospheric, and Planetary Sciences at Purdue University; Karen Aline McKinnon, associate professor at the Institute of the Environment and Sustainability in the Department of Statistics at the University of California, Los Angeles; Robert Nicholls, professor of climate adaptation at the University of East Anglia; Lorenzo Rosa, principal investigator at Carnegie Science and assistant professor in the Doerr School of Sustainability at Stanford University; and Xianli Wang, research scientist at the Canadian Forest Service of Natural Resources Canada. We also thank our external data partners at Fathom.

Many McKinsey colleagues and alumni contributed their analysis and perspectives, including Antara Banerjee, Chris Bradley, Zach Bruick, Dave Eickholt, Kandice Harper, Nick Kingsmill, Gillian Pais, Adam Rubin, Alexis Trittipo, Sean Turner, and Lola Woetzel.

This report was edited and produced by executive editor Stephanie Strom, together with senior data visualization editor Chuck Burke, data visualization editor Laura M. Mandujano, assistant managing editor Rishabh Chaturvedi, editor Mary Gayen, and lead designer Nathan R. Wilson. We also thank Suzanne Albert, Ankush Arora, David Batcheck, Nienke Beuwer, Ashley Grant, Cathy Gui, Stephen Landau, Janet Michaud, Marie Morris, Rebeca Robboy, Rachel Robinson, and Julie Schwade for their contributions and support.

This research contributes to our mission to help business and policy leaders understand the forces transforming the global economy. As with all MGI research, it is independent and has not been commissioned or sponsored in any way by any business, government, or other institution. While we benefited greatly from the variety of perspectives we gathered, we have independently formed and articulated our views in this report. Any errors are our own.

Since we aren’t climate scientists, we rely on external climate models, primarily those used in the sixth assessment report by the Intergovernmental Panel on Climate Change (IPCC).⁴ In using these models, we have employed standard techniques to manage uncertainties inherent in climate modeling, though we recognize that important residual uncertainties remain (see the technical appendix for more details). Thus, the results presented here are not high-resolution predictions of climate hazards in specific places around the world. Rather, they are directional assessments of how hazards could evolve. This serves as the basis for our order-of-magnitude assessment of adaptation costs and benefits across the world.

Heat stress: Present across the tropics today, becoming more widespread and prolonged at 2°C

Heat stress refers to sustained periods in an average year of more than four weeks of high temperatures, often combined with humidity, that can significantly reduce labor productivity.¹ Prolonged exposure slows work, increases the need for rest, and raises health risks.²

Heat stress exposure is highest today across South Asia, North Africa and the Arabian Peninsula, and parts of Southeast Asia, where some places could have more than 14 weeks a year of heat stress conditions. At 1.5°C of warming relative to preindustrial levels, landmass exposed to heat stress could increase by roughly 25 percent globally compared with today.

At 2°C, global land exposure to heat stress could increase by more than 50 percent compared with today. Areas with little or no exposure now may experience meaningful periods of heat stress. And for regions already exposed today, conditions could intensify as heat stress lasts longer, from roughly 12 weeks today to more than 16 weeks, or an extra month.

Places exposed to heat stress by developed-economy standards, by warning level

Heat waves: Clustered today in a few places in the tropics and subtropics, spreading at 1.5°C and beyond

Heat can also manifest as heat waves, defined here as rare periods of at least seven consecutive days of unusually high temperatures for a particular place.¹ Not all heat waves across regions are the same. In Nice, France, a week or more of daily highs exceeding 35°C would be considered a heat wave, while in New Delhi, a heat wave event would occur only when temperatures top 45°C. And unlike heat stress, which is an annual occurrence, in any given year neither Nice nor New Delhi might experience such a heat wave. The impact of heat waves can be far lower than heat stress, but when they do occur, they can particularly affect vulnerable populations like elderly people, children, and pregnant women.

Today, parts of South Asia, Southeast Asia, the Arabian Peninsula, and northern parts of South America have a 5 percent likelihood of experiencing such heat waves. In Southeast Asia and along the Arabian Peninsula, such a heat wave can last as long as 12 days. But at 1.5°C of warming, land exposed to a likelihood of heat waves could increase fivefold, and some places could experience heat waves that persist, should they occur, for more than three weeks.

At 2°C, two-thirds of global land could be exposed to a likelihood of such heat waves. Globally, heatwaves of the duration and intensity of today’s one-in-20-year event could occur nearly once every four years. However, while large areas may be exposed to such events, protection against heat waves typically relies on relatively low-cost measures, such as early-warning systems and cooling shelters, that can, once put in place, be readily activated when a heat wave occurs.

Places exposed to heat waves by developed-economy standards, by warning level

Nonsurvivable heat: Extremely rare today, becoming slightly more widespread and likely at 2°C

Nonsurvivable heat is defined as very rare periods of extremely high, humid heat, such that the body can no longer cool itself by sweating, and during which human survival is at risk without access to cooling.¹ Isolated heat events approaching or exceeding such levels have been recorded in the Persian Gulf and the Indus Valley in recent decades.²

Today, only a handful of areas face a very small likelihood of such extreme conditions, including isolated parts of northern Australia, Central and West Africa, South Asia, mainland Southeast Asia, southern China, and around the Red Sea. At 1.5°C of warming relative to preindustrial levels, however, the chance of nonsurvivable heat increases slightly in regions currently at risk.

By 2°C, the total area exposed to this hazard could reach 2 percent of the global landmass. Overall, they would mostly remain very rare events and affect only a very small share of global land. However, in a handful of places—such as a few parts of South Asia and around the Red Sea—the annual probability of an event with such extreme heat could exceed 20 percent then.¹ Measures ranging from air conditioning to early-warning systems could protect against this hazard.

Places exposed to nonsurvivable heat, by warning level

Wildfire weather: Present today, lengthening slightly, and becoming somewhat more widespread at 2°C

Wildfire weather occurs when there are more than two weeks of hot, dry, and windy conditions that heighten the risk of wildfires in places where vegetation can sustain burning.¹ We measure the weather conditions that make wildfires more likely, though of course the occurrence of actual wildfires also requires an ignition event. The effects of these fires can be severe, destroying homes and infrastructure, damaging ecosystems, and releasing smoke that worsens air quality over vast areas. For example, the 2023 Canadian wildfires produced hazardous air quality in cities as far away as New York and Washington, DC.²

Wildfire weather is common today across forested and grassland regions worldwide. Parts of southern Africa and Australia could experience more than 180 days a year when conditions are ripe for fire spread.¹ The geographic extent of wildfires increases only modestly at 1.5°C of warming.

At 2°C, land area exposed globally to wildfires may increase by about 10 percent relative to current levels, with affected areas seeing roughly an additional week of such conditions on average.

Places exposed to wildfires by developed-economy standards, by warning level

Drought: Present today in some parts of the world, expanding and lengthening at 1.5°C and beyond

Drought, a result of low rainfall and prolonged heat, depletes water in soils, aquifers, and surface waters. We define it as a rare, unusually dry period for a given place, lasting more than six months, that affects agricultural areas.¹ Such events do not occur every year, but when they do, they can disrupt growing cycles, lower yields, and, in severe cases, cause widespread harvest failure. For example, the 2012 drought in the United States caused an estimated $30 billion in agricultural losses as corn and soybean harvests were reduced.²

Today, about 13 percent of global agricultural land has a 5 percent chance of having droughts of this kind, primarily in Brazil, the Middle East, the Sahel and southern Africa, Greater China, India, and smaller pockets in the western United States and Mexico. Exposed areas can experience six to eight dry months in a rare drought. The areas exposed to such a rare drought could increase by almost three times today’s exposure at 1.5°C of warming.

At 2°C, drought could increase by almost four times today’s exposure and in some regions, a rare drought may last more than 12 dry months. A drought of the duration and intensity of today’s one-in-20-year event could, on average, become almost a one-in-10-year event in the future. About 4 percent of all areas that experience a change in drought exposure may be no longer exposed as rainfall patterns shift.

Places exposed to drought by developed-economy standards, by warning level

Coastal flooding: Rare events today expected to occur more often across many low-lying regions at 2°C, with some intensity increase

Coastal flooding—defined here as a very rare event causing at least knee-deep inundation—occurs when high tides and storm surges force seawater to cover normally dry coastal land.¹ In 2022, Hurricane Ian pushed seawater inland across southwest Florida, causing dozens of fatalities and extensive damage to homes and infrastructure far from the shoreline.²

Today, such flooding can affect low-lying coastal areas worldwide, including many small island nations, parts of northern Europe, South and Southeast Asia, and the coast of the Gulf of Mexico, which the United States calls the Gulf of America. Some more places would be exposed to such rare events as sea levels further rise with warming.¹ In these places, even modest surges or high tides can trigger inundation. At 1.5°C of warming, coastal land exposed to these very rare floods could expand by about 6 percent, and today’s flood event may occur five times more often on average.

By 2°C at 2050, exposure could increase by roughly 15 percent compared with today. Today’s one-in-100-year event could become a one-in-13-year event by that time. Average depths of the one-in-100-year event could become 10 percent higher on average, though with wide regional variation.

Places exposed to coastal flooding by developed-economy standards, by warning level

Riverine flooding: Present today, with slightly greater extent and some frequency and intensity increase at 2°C

Riverine flooding, defined here as a very rare event bringing at least knee-deep inundation, occurs when rivers and streams overflow onto normally dry land due to excessive rainfall or snowmelt.¹ In July 2021, extreme flooding along the Meuse and Rhine rivers in Europe killed more than 220 people and caused about €46 billion in losses, primarily in infrastructure.²

Today, places exposed to riverine flooding are widespread across many of the world’s major river basins, with flood depths in such rare events often exceeding one meter. The share of land exposed to riverine flooding could increase by about 2 percent at 1.5°C of warming.

At 2°C, the global land exposed could increase by about 4 percent. Extreme riverine floods could become slightly more frequent then, with today’s one-in-100-year event occurring on average once in 80 years. The average flood depth in exposed regions may rise modestly, by about five to ten centimeters, though with wide regional variation. Isolated areas in temperate and polar basins may experience reduced flood risk as rainfall and snowpack patterns shift.

Places exposed to riverine flooding by developed-economy standards, by warning level

Excessive rainfall flooding: Widespread today, expanding in reach at 2°C

Excessive rainfall (pluvial) flooding, defined here as a rare event causing at least ankle-deep inundation, occurs when typically short bursts of heavy rainfall overwhelm the capacity of soil or drainage systems to absorb or channel water, causing water to pool on normally dry land or pavement.¹ In 2023, a record downpour in Hong Kong flooded subway stations and halted stock exchange trading, with damages exceeding $100 million.²

Rainwater flooding is already widespread globally. Many locations experience rare events where flood depths are 15 centimeters or more, enough to inundate streets, damage crops and property, and disrupt urban systems. At 1.5°C, exposed landmass that has a 5 percent chance of flooding from excessive rainfall could expand by nearly 10 percent compared with today, though isolated areas may see decreases.

At 2°C, global land exposure could increase by more than 20 percent. At this level of warming, a one-in-20-year event could become slightly more frequent, roughly one in 17 years, and the global average depth of such a flood would rise only slightly, by two centimeters.

Places exposed to excess-rainfall flooding by developed-economy standards, by warning level

Freezing days: Common today across temperate and polar regions, becoming less widespread with warming at 1.5°C and beyond

As global temperatures rise, the freezing days hazard—at least four weeks of days when the average temperature is below 0°C—could occur in fewer places.¹ This may bring benefits as it could reduce cold-related deaths and crop damage. However, it also brings risks such as shrinking glacial freshwater reserves and thawing permafrost.²

Today, freezing days are common across northern continental interiors, including most of Canada, the northern United States, and northern Europe into Russia. At 1.5°C of warming, about 2 percent of land currently experiencing freezing days is expected not to have them.

At 2°C, the global land exposed could decrease by about 3 percent. Areas with decreasing exposure to freezing days then will also be more densely populated. The biggest decreases may occur in parts of North America, with smaller declines in Central Europe and Northeast Asia. Places that continue to see freezing conditions as the global temperature rises are expected to face them for shorter durations each year.

Places exposed to freezing by developed-economy standards, by warning level

Advancing adaptation: How evolving hazards could shape the agenda