Extreme event attribution

• 1. Introduction

  ›

  Extreme event attribution asks a question that once seemed unanswerable: when a catastrophic storm, heatwave, or flood strikes, how much of that disaster was shaped by human-caused climate change? For most of the twentieth century, scientists could track the overall warming of the planet and pin that trend on greenhouse gas emissions. But saying that a single hurricane or a single drought bore the fingerprints of industry seemed beyond reach. That view began to shift in the early 2000s, and a discipline now known as attribution science has since grown into a tool used by researchers, courts, and insurance companies alike. The questions it poses are specific and quantifiable: was this heatwave made more likely, more intense, or longer by global warming? By how many degrees? By what factor of probability?

  1356

• 2. A Science Born from Doubt

  ›

  The Intergovernmental Panel on Climate Change acknowledged the problem in its 1990 First Assessment Report, stating that "it is virtually impossible to demonstrate a cause-effect relationship with high confidence from studies of a single variable." The IPCC did note, however, that confidence in attribution grows as climate predictions are borne out by observed data, pointing to a "fingerprint method" that identifies signals unique to human-caused warming.

  By 2001, the IPCC's Synthesis Report had found that detection and attribution studies "consistently find evidence for an anthropogenic signal in the climate record of the last 35 to 50 years." The same report concluded that over the preceding fifty years, the estimated rate and magnitude of warming from greenhouse gases alone was comparable to, or larger than, observed warming. The science was advancing, but it still stopped short of the individual event.

  The turning point came in 2004, when researchers applied a concept called fraction attributable risk to a study of the 2003 European heatwave. That analysis provided the first direct statistical link between human-caused climate change and a single extreme weather event. Greater computing power in the 2000s enabled repeated simulation of weather scenarios, and conceptual breakthroughs in the early to mid 2010s consolidated the field. By 2016, the American Meteorological Society was prepared to state that "the science has now advanced to the point that we can detect the effects of climate change on some events with high confidence."

  78101226

• 3. How Attribution Works

  ›

  German climatologist Friederike Otto framed the central ambition of attribution science plainly: it aims to answer, within a news cycle, whether climate change played a role in a specific extreme event. That two-week deadline is not arbitrary. It means attribution studies must be fast enough to be relevant while public attention is still focused on the disaster.

  The method rests on a comparison between two imagined worlds. In the first, the "factual" world, greenhouse gas emissions are included in climate models as they actually are. In the second, a "counterfactual" world, those anthropogenic emissions are removed. Scientists simulate each world many thousands of times and then measure how frequently the event in question occurs in each. The difference in probability or intensity between the two worlds is attributed to human influence.

  Atmospheric scientist Xubin Zeng described the process in four steps: measuring the magnitude and frequency of a given event using observed data; running computer models to verify those observations; running the same models against a baseline with no climate change; and applying statistics to compare the results. The entire framework rests on peer-reviewed methods, which is what makes it possible to publish findings quickly.

  What counts as an "extreme" event can itself be expressed several ways: as a percentile rank among historical events, as standard deviations from historical norms, as a probability of occurrence in a given year, or as a return period such as a "100-year drought." Results across different studies sometimes diverge, because researchers may define "extreme" differently, analyze different periods of the historical record, or use climate models with different spatial resolutions. A study of peak wind speed and a study of average wind speed for the same storm can legitimately reach different conclusions.

  6413116

• 4. What the Numbers Show

  ›

  A November 2024 review by Climate Central covered more than 600 climate attribution studies spanning nearly 750 extreme weather events and trends. It found that climate change made 74% of those events more likely or more severe. Roughly 9% were made less likely or less severe.

  The specificity of individual findings is striking. The 2020 Siberian heatwave was found to be more than 500 times as likely as similar events in 1900, with a best estimate reaching 600 to 99,000 times more likely. The 2021 Western North America heat wave was judged 150 times more likely. The 2022 United Kingdom heatwave carried a return period of 100 years for two-day average temperatures and 1 in 1000 years for single-day maximum temperatures in one region; the overall event was made at least ten times more likely by climate change.

  For hurricane events, the findings address both frequency and physical character. Hurricane Harvey in August 2017 was found to be three times more likely and 15% more intense than it would have been without anthropogenic warming. Hurricanes Helene and Milton in 2024 each had roughly 44% and 45% of their losses in Florida attributed to climate change, and storm systems like them are up to 20% wetter under current conditions.

  The influence on human life is equally quantified. A review covering the decade from 2004 through 2024 found that climate change intensified the ten deadliest extreme weather events of that period, contributing to at least 570,000 deaths. A separate worldwide review in 2022 found tens of thousands of deaths directly attributable to heat extremes. A review of rainfall events in the North Atlantic basin attributed a combined $500 billion in damages to climate change's influence.

  146362227

• 5. Unequal Heat, Unequal Days

  ›

  Between May 2023 and May 2024, researchers tracked how many days each country experienced extreme heat with global warming compared to what would have occurred in a world without it. Suriname logged 182 days of extreme heat with warming versus 24 without. Ecuador recorded 180 versus 10. Guyana saw 174 versus 33. El Salvador counted 163 versus 15, and Panama 149 versus 12.

  Probability ratios computed for 2023 and 2024 heat events show similarly stark contrasts by region. Events in the Marshall Islands and Micronesia carried a probability ratio of 35, meaning they were 35 times more likely under current climate conditions. Indonesia and the Philippines each showed ratios between 24 and 29. Central America showed a ratio of 13. Spain and Portugal showed 8. Southern South America showed 7.

  For tropical birds, the signal runs across seven decades: a 25-38% reduction in global abundance from 1950 through 2020 has been attributed to human-caused intensification of heat extremes. The 2019-20 Australian bushfire season was found at least 30% more likely, with heat extremes at least twice as likely. The 2017-2018 Tasman Sea marine heatwave was judged to have an overall intensity that was "virtually impossible" without anthropogenic forcing.

  Heatwaves are the most accurately modeled of all extreme event types, because long-term temperature data is more reliable than datasets for other kinds of events. Rainfall attribution is also relatively tractable. Droughts, snowstorms, tropical storms, and wildfires are more complicated to model, partly because they depend on atmospheric dynamics, soil moisture, and surface cover that are harder to simulate.

  2337303129

• 6. From Science to Accountability

  ›

  Attribution science has moved from academic journals into courtrooms. It has been cited in climate litigation against companies for causing climate change and against governments for failing to address it. Researchers applied extreme event attribution to 213 historical heatwaves from 2000 through 2023, finding that "carbon majors" (fossil fuel and cement producers) had enabled between 16 and 53 heatwaves that would have been virtually impossible in a preindustrial climate.

  The Sabin Center for Climate Change Law at Columbia University maintains searchable databases organized under five categories: climate change attribution, extreme event attribution, impact attribution, source attribution, and attribution in the courts. Carbon Brief hosts an interactive map covering more than 600 studies across nearly 750 extreme weather events. The American Meteorological Society publishes annual special collections titled "Explaining Extreme Events of year from a Climate Perspective."

  In late 2025, the Copernicus Climate Change Service announced the formation of a permanently operating attribution office with the goal of delivering two assessments per month, each published within a week of an extreme weather event. Those assessments are designed to inform government policy, assist insurance companies with risk assessment, and provide evidence for climate litigation. The 2026 Chilean wildfires, for instance, were subsequently found to be three times more likely due to climate change, and 2.5 times more likely in Patagonia specifically, an example of the kind of rapid, actionable finding the attribution office is built to produce.

  3825212620