Preprint links anthropogenic warming to 15–30 extra hot days yearly

A new global dataset estimates how daily temperatures from 1979 through 2020 would have differed without anthropogenic climate forcing. The resulting Global Daily Natural temperature dataset, or GDNat, provides daily average temperatures on a 0.25° × 0.25° grid and preserves the day-to-day structure of the historical record while removing the modeled influence of human-caused warming. In a regional demonstration, the researchers estimated that anthropogenic warming contributed an average of 15–30 additional days per year with average temperatures reaching or exceeding 35°C across North Africa, the Middle East, South Asia and Australia during 2000–2020.

The preprint, authored by Robert Fofrich Navarro, Kelly E. McCusker, Steven B. Malevich and Robert E.

Kopp, is under review in Earth System Science Data. It addresses a recurring challenge in climate attribution: estimating the contribution of human influence to observed meteorological changes. Such studies commonly compare historical observations with global climate model simulations driven solely by natural forcing.

The authors note that these simulations have limited spatial resolution and may be affected by biases in parameterized climate processes and by uninitialized conditions, which can limit their representation of historical meteorological events. GDNat is designed to provide a higher-resolution, daily alternative for assessing temperature extremes and their relationship to anthropogenic warming. To construct the dataset, the researchers applied quantile delta mapping, a bias-adjustment method that preserves trends while adjusting distributions.

They used daily average temperatures from the fifth generation of the European Centre for Medium-Range Weather Forecasts Reanalysis, or ERA5, together with historical and natural-forcing-only simulations from the Coupled Model Intercomparison Project Phase 6, or CMIP6. The procedure removes the modeled influence of anthropogenic warming from the reanalysis temperatures while retaining their detailed temporal structure. The result is not a direct observation of an alternative climate, but a model-based counterfactual estimate intended to represent what temperatures might have been without human-caused climate forcing.

The researchers demonstrated GDNat by examining changes in the annual number of days on which average temperatures reached or exceeded 35°C. Their regional analysis covered North Africa, the Middle East, South Asia and Australia and focused on 2000–2020. Across the four regions, the study attributed an average of 15–30 additional threshold-exceeding days per year to anthropogenic warming over that period.

This is a regional demonstration involving one specified temperature threshold, rather than a comprehensive assessment of all heat extremes or every part of the globe. The result concerns the number of days meeting the daily-average criterion, not a count of discrete heat-wave events. GDNat is intended as a research resource for quantifying human influence on temperatures and studying extremes against an estimate of a naturally forced climate.

Its usefulness depends on the underlying reanalysis, climate-model simulations and bias-correction method, while the counterfactual cannot be treated as a directly observed world. The study is also a preprint that remains under review.

Subject of Research: A model-based global dataset for attributing anthropogenic effects on daily temperature extremes

Article Title: GDNat: a global, daily, high-resolution, natural-forcing only temperature data set for attribution research

Article References: Fofrich Navarro, R., McCusker, K. E., Malevich, S. B., & Kopp, R. E. (2026). GDNat: a global, daily, high-resolution, natural-forcing only temperature data set for attribution research. https://doi.org/10.5194/essd-2026-658

Image Credits: AI Generated

DOI: 10.5194/essd-2026-658

Keywords: climate attribution, anthropogenic warming, daily temperatures, extreme heat, ERA5 reanalysis, CMIP6, quantile delta mapping, global dataset, GDNat, global, daily, high-resolution