Global Extreme Weather and Climate Change Dashboard
Loss Normalization
A normalization asks how much damage a historical event would cause under contemporary societal conditions. Economic losses are never a good place to detect changes in climate; weather and climate data will always serve that purpose better. Since Chris Landsea and I introduced methodologies of normalization in the 1990s, there have been many dozens of normalization studies for regions and phenomena around the world. This page summarizes that literature.
Top line conclusions:
- Normalized losses are not increasing but overall losses are increasing.
- Storms and floods dominate economic losses.
- North America and Asia see the greatest economic losses.
- Drought, storms, floods, and extreme temperatures account for almost all loss of life.
- Asia, Africa, and Europe see the greatest loss of life.
- A robust peer-reviewed literature presents a strong consensus that societal factors underlie economic impacts, not changes in climate.
Weather/climate disaster deaths per 100,000 people, worldwide
AnnualEM-DAT world weather/climate disaster deaths (same categories as the death pie below) divided by World Bank world population, both by year. Unit: deaths per 100,000 people.
Source: EM-DAT (CRED/UCLouvain) via Our World in Data; World Bank population (SP.POP.TOTL)
Computing trend…
A single mass-casualty event dominates most high years here: 1991 (Cyclone Gorky, Bangladesh, roughly 138,000 deaths), 1998 (Hurricane Mitch and South/East Asian flooding), 2003 and 2010 (European and Russian heat waves), 2008 (Cyclone Nargis, Myanmar). 2022-2024 run elevated relative to the mostly-quiet 2011-2021 stretch, driven substantially by heat-wave mortality (see the Heat Waves page) rather than a single dominant event. Deaths run as a count of people, not a currency, so no inflation adjustment applies; unlike the economic series above, no single literature source is extended here -- this chart comes directly from the two public sources cited, not from a specific published finding.
Where economic losses are greatest
Mohleji, S. & Pielke, R. Jr. (2014), "Reconciliation of Trends in Global and Regional Economic Losses from Weather Events: 1980–2008," Natural Hazards Review 15(4) documented the pattern: storms dominate global weather-related economic losses, and North America dominates regionally. This page extends that pattern through 2025 using EM-DAT data (CRED/UCLouvain, via Our World in Data's public mirror).
By phenomenon
Storms leads at 56% of the total.
- Storms $3,176.1B 55.8%
- Floods $1,696.4B 29.8%
- Drought $414.7B 7.3%
- Wildfires $264B 4.6%
- Extreme temps $114.3B 2.0%
- Landslides $23B 0.4%
By continent
North America leads at 51% of the total.
- North America $2,913.4B 51.2%
- Asia $1,833.3B 32.2%
- Europe $584.4B 10.3%
- South America $162.6B 2.9%
- Oceania $134.5B 2.4%
- Africa $60.1B 1.1%
Where deaths are greatest: a different geography than economic loss
By phenomenon
Drought leads at 33% of the total.
- Drought 584,640 33.1%
- Storms 484,153 27.4%
- Extreme temps 367,414 20.8%
- Floods 289,010 16.3%
- Landslides 39,911 2.3%
- Wildfires 3,582 0.2%
By continent
Asia leads at 37% of the total.
- Asia 663,197 37.5%
- Africa 642,787 36.3%
- Europe 337,192 19.1%
- North America 65,487 3.7%
- South America 56,280 3.2%
- Oceania 3,767 0.2%
Loss of life shows a strikingly different geography than economic loss. North America accounts for over half of economic losses but under 4% of the weather-related death toll over the identical 1980–2025 window — Africa and Asia together account for nearly 74% of global weather-related deaths: drought drives 90% of Africa's toll (the 1983-85 Ethiopian famine and Sahel droughts); storms drive 62% of Asia's (including the 1991 Bangladesh cyclone's ~138,000 deaths).
| Continent | Weather-related deaths, 1980–2025 | % of world | Dominant cause |
|---|---|---|---|
| Asia | 663,197 | 37.5% | Storms (62% -- tropical cyclones, incl. the 1991 Bangladesh cyclone's ~138,000 deaths) |
| Africa | 642,787 | 36.3% | Drought (90% of the region's total -- the 1983-85 Ethiopian famine and Sahel droughts) |
| Europe | 337,192 | 19.1% | Extreme temperatures (97% -- chiefly the 2003 and 2010 European heat waves) |
| North America | 65,487 | 3.7% | Storms (68%) |
| South America | 56,280 | 3.2% | Floods (80%) |
| Oceania | 3,767 | 0.2% | Landslides (34%, of a very small total -- fairly evenly split across causes) |
EM-DAT via Our World in Data.
Normalization adjusts historical economic disaster losses to a common base year — removing the effects of societal change (population growth, wealth, development) from a loss time series to estimate what losses past events would cause under today's conditions. An unbiased normalization should show trends consistent with the actual climatological record of the relevant hazard; where they diverge, the divergence signals a possible bias in the normalization method. This table reproduces the full study list (Table 1) from Pielke, R. (2020), "Economic 'normalisation' of disaster losses 1998–2020: a literature review and assessment," Environmental Hazards, which reviewed every normalization study published 1998 through June 2020.
Across all 69 studies now in this table, 53 find no detectable trend in normalized losses. Of the 16 that do, 10 find decreasing trends and 6 find increasing trends. 3 attribute an increase in losses to human-caused climate change (Grinsted et al. 2019; Willoughby et al. 2024; Willoughby et al.'s 2025 reply, whose claimed increase is not statistically significant) — all 3 are compromised by their use of the fatally flawed ICAT dataset (see the ICAT/BDD column below). Pielke (2020) identifies the first of those, Grinsted et al. 2019, as an outlier at odds with the other U.S. hurricane-loss normalizations among its original 53 studies, and with the independent climatological record of U.S. landfalling hurricanes, which shows no detected trend since 1900. Paprotny et al. 2025 makes a different kind of attribution claim entirely — it attributes a decrease in European flood losses to adaptation, not to climate change driving losses up.
The full study list (69 studies)
The table below updates Pielke (2020) through August 2026 - If I've missed any, let me know. The ICAT/BDD column flags studies built on the fatally flawed ICAT Damage Estimator dataset or improperly combined ICAT/NOAA "Billion Dollar Disasters" dataset to create a Frankenstein dataset — I have called for these flawed studies to be retracted from the literature (see Pielke 2025). Pielke (2024), "Scientific integrity and U.S. 'Billion Dollar Disasters'" (2024, npj Natural Hazards), critically evaluates that NOAA dataset specifically. Click any column header to sort.
| Study | Region | Phenomena | Period | Finding | Attributed? | ICAT/BDD? |
|---|---|---|---|---|---|---|
| Martinez (2020) | United States | Tropical cyclones | 1900–2018 | None | No | N |
| Grinsted et al. (2019) | United States | Tropical cyclones | 1900–2018 | Detected — Increase | Yes | Y |
| Chen et al. (2018) | China | Tropical cyclones | 1983–2015 | None | No | N |
| Ye and Fang (2018) | China | Tropical cyclones | 1985–2010 | Detected — Decrease | No | N |
| Weinkle et al. (2018) | United States | Tropical cyclones | 1900–2017 | None | No | N |
| Klotzbach et al. (2018) | United States | Tropical cyclones | 1900–2016 | None | No | N |
| Fischer et al. (2015) | China | Tropical cyclones | 1984–2013 * | None | No | N |
| Estrada et al. (2015) | United States | Tropical cyclones | 1900–2005 | Detected — Increase | No | N |
| Bouwer and Wouter Botzen (2011) | United States | Tropical cyclones | 1900–2005 | None | No | N |
| Nordhaus (2010) | United States | Tropical cyclones | 1900–2005 | Detected — Increase | No | N |
| Zhang et al. (2009) | China | Tropical cyclones | 1983–2006 * | None | No | N |
| Schmidt et al. (2009) | United States | Tropical cyclones | 1950–2005 | None | No | N |
| Pielke et al. (2008) | United States | Tropical cyclones | 1900–2005 | None | No | N |
| Pielke et al. (2003) | Latin America & Caribbean | Tropical cyclones | 1944–1999 | None | No | N |
| Raghavan and Rajesh (2003) | India | Tropical cyclones | 1977–1998 * | None | No | N |
| Collins and Lowe (2001) | United States | Tropical cyclones | 1900–1999 | None | No | N |
| Pielke and Landsea (1998) | United States | Tropical cyclones | 1926–1995 | None | No | N |
| Du et al. (2019) | China | Floods | 1990–2017 | Detected — Decrease | No | N |
| Paprotny et al. (2018) | Europe | Floods | 1870–2016 | None | No | N |
| Wei et al. (2018) | China | Floods | 2000–2015 * | Detected — Decrease | No | N |
| Fang et al. (2018) | China (Yangtze River) | Floods | 1998–2014 * | Detected — Decrease | No | N |
| Pérez-Morales et al. (2018) | Spain | Floods | 1975–2013 | None | No | N |
| Stevens et al. (2016) | United Kingdom | Floods | 1884–2013 | None | No | N |
| Barredo et al. (2012) | Spain | Floods | 1971–2008 | None | No | N |
| Hilker et al. (2009) | Switzerland | Floods | 1972–2007 | None | No | N |
| Chang et al. (2009) | Korea | Floods | 1971–2005 | Detected — Increase | No | N |
| Barredo (2009) | Europe | Floods | 1970–2006 | None | No | N |
| Downton et al. (2005) | United States | Floods | 1926–2000 | Detected — Decrease | No | N |
| Fengqing et al. (2005) | China | Floods | 1950–2001 | None | No | N |
| Pielke and Downton (2000) | United States | Floods | 1932–1997 | None | No | N |
| Andres and Badoux (2019) | Switzerland | Extratropical storms | 1972–2016 | None | No | N |
| Stucki et al. (2014) | Switzerland | Extratropical storms | 1859–2011 | None | No | N |
| Barredo (2010) | Europe | Extratropical storms | 1970–2008 | None | No | N |
| Simmons et al. (2013) | United States | Tornadoes | 1950–2011 | None | No | N |
| Brooks and Doswell (2001) | United States | Tornadoes | 1890–1999 | None | No | N |
| Boruff et al. (2003) | United States | Tornadoes | 1900–2000 | None | No | N |
| Sander et al. (2013) | United States | Convective storms | 1970–2009 | Detected — Increase | No | N |
| Crompton et al. (2010) | Australia | Wildfire | 1925–2009 | None | No | N |
| Choi et al. (2019) | Korea | All weather | 1965–2015 | Detected — Decrease | No | N |
| Reyes and Elias (2019) | United States | Crop loss | 2001–2016 * | Mixed | No | N |
| McAneney et al. (2019) | Australia | All weather | 1966–2017 | None | No | N |
| Paul and Sharif (2018) | Texas, United States | Hydrometeorological | 1960–2016 | None | No | N |
| Bahinipati and Venkatachalam (2016) | India | All weather | 1972–2009 | None | No | N |
| Zhou et al. (2013) | China | Natural disasters | 1990–2011 * | None | No | N |
| Crompton and McAneney (2008) | Australia | All weather | 1967–2006 | None | No | N |
| Choi and Fisher (2003) | United States | All weather | 1951–1997 | None | No | N |
| Pielke (2019) | World | All disasters & weather-only | 1990–2017 * | Detected — Decrease | No | N |
| Watts et al. (2019) | World | All disasters | 1990–2016 * | None | No | N |
| Daniell et al. (2018) | World | Multi-hazard | 1950–2015 | Detected — Decrease | No | N |
| Mohleji and Pielke (2014) | World | All weather-related | 1980–2008 * | None | No | N |
| Neumayer and Barthel (2011) | World | All weather-related | 1980–2008 * | None | No | N |
| Visser et al. (2014) | World | All weather-related | 1980–2010 * | None | No | N |
| Miller et al. (2008) | World | All weather-related | 1950–2005 | None | No | N |
| Alstadt et al. (2022) | United States | Methodology (applied to hurricanes) | n/a (method paper) | n/a | No | N |
| Boccard (2021) | global | All disasters | 1970–2019 | Reported, not confirmed normalized | No | N |
| Das et al. (2022) | Indian states (crops, housing, utilities) | Floods | 1953–2011 | Mixed | No | N |
| Deo et al. (2022) | southwest Pacific Islands | Tropical cyclones | n/a (economic exposure assessment) | n/a | No | N |
| Hadavi et al. (2022) | Quebec & Ontario, Canada | Windstorms | 2008–2021 * | n/a | No | N |
| McAneney et al. (2022) | New Zealand | All natural disasters (incl. earthquakes) | 1968–2019 | Unclear | No | N |
| Ying et al. (2021) | China | Tropical cyclones | 1984–2017 | Unclear | No | N |
| Zhang et al. (2023) | United States | Tornadoes | 1954–2018 | Yes — significant national decline | No | N |
| Willoughby et al. (2024) | United States | Tropical cyclones (Atlantic) | 1900–2022 | Yes — increase | Yes | Y |
| Muller et al. (2025) | United States | Hurricanes | 1900–2022 | None | No | Y |
| Pielke (2025) | United States (ICAT dataset critique) | Hurricanes | — | n/a | No | N |
| Willoughby et al. (2025) | United States (reply to Pielke 2025) | Tropical cyclones (Atlantic) | 1900–2022 | Partial — 0.6%/yr, not significant | Yes | Y |
| Paprotny et al. (2025) | Europe | Floods | 1950–2020 | Detected — Decrease | Yes — to adaptation, reducing losses | N |
| Paprotny et al. (2024) | Europe (HANZE v2.1 database) | Floods | 1870–2020 | Reporting completeness dominates apparent trends | No | N |
| Vessey et al. (2026) | United States | Hurricanes | 1979–2024 | Incidental upward fit, not the study's aim | No | N |
| Botzen et al. (2021) | n/a | Methodology | — | n/a | n/a | N |
* Period under 30 years — the IPCC's typical minimum window for a climate (not weather) trend assessment.
My 2020 paper's bottom-line conclusion remains current
"[This review] finds little evidence to support claims that any part of the overall increase in global economic losses documented on climate time scales is attributable to human-caused changes in climate, reinforcing conclusions of recent assessments of the Intergovernmental Panel on Climate Change." The paper frames this as consistent with, not contrary to, IPCC AR5's conclusion that "loss trends have not been conclusively attributed to anthropogenic climate change" and that "increasing exposure of people and economic assets has been the major cause of long-term increases in economic losses from weather- and climate-related disasters (high confidence)."