84.1 heatwave days per station were recorded during 1930 to 1944, compared to 26.1 during 1965 to 1979, marking the largest sustained concentration of extreme heat events in the 127-year United States record. Findings published in Theoretical and Applied Climatology in March 2026 report a dataset built from 1,211 long-term weather stations covering 97.6% of the contiguous United States, using more than 40 million daily observations without homogenization to track seasonal temperature extremes from 1899 through 2025. The dataset isolates summer maximum temperatures and winter minimum temperatures to capture the physical limits of heat and cold exposure that directly affect human systems and infrastructure thresholds. Each station contributes raw observations mapped onto a 0.5° grid using inverse-distance weighting, where closer stations exert greater influence due to shared air mass behavior. This method preserves localized extremes driven by pressure systems, solar radiation, and terrain effects rather than smoothing them through statistical adjustment, producing a high-resolution record of how extreme temperatures occurred across the United States since 1899.
1936 produced 6.7 record hot days per station compared to an expected 1.2, driven by persistent high-pressure systems that compressed and heated air masses over central and eastern regions. The data show that 1936 alone affected roughly 22% of the contiguous United States with all-time heat records, while subsequent peak years such as 1934 and 1931 produced spatial coverage below 5%. These atmospheric blocking patterns prevent cooler air intrusion and maintain subsiding air, which warms as it descends and compresses toward the surface. Modern extreme heat events recorded after 2000 do not reach the same spatial dominance or frequency when measured across the full station network. Western regions record higher recent frequencies while eastern regions remain below mid-20th century peaks.
Cold extremes follow a different physical mechanism driven by Arctic air mass displacement, where dense cold air moves southward under strong pressure gradients and rapidly lowers surface temperatures. The dataset records 3.7 extreme cold events per station in 1899, tied to a single Arctic outbreak that produced the highest concentration of record low temperatures in the time series. Cold extremes decline sharply after the 1990s, with recent 15-year periods showing roughly half the frequency observed before 1997. Nighttime surface conditions control these extremes, where clear skies and low wind allow cold air to pool in shallow layers near the ground. Urban development stores heat in buildings and pavement, releasing energy at night and raising minimum temperatures, which reduces the occurrence of extreme cold events across populated areas.
Heatwave frequency shows strong multi-decadal variability, with 84.1 heatwave days per station recorded during 1930 to 1944 compared to 26.1 during 1965 to 1979. Heatwaves are defined as at least six consecutive days exceeding the 90th percentile, producing sustained cumulative heating over multi-day periods. Stationary atmospheric circulation patterns trap warm air over a region, allowing heat to build over consecutive days without disruption. Southern regions record between 53 and 69 heatwave days per 15-year window due to more stable summer patterns, while northern regions record between 25 and 37 due to more frequent weather system movement. Western regions show higher recent heatwave counts while eastern regions remain below mid-century levels.
Cold wave frequency declines across most regions, with 15-year running averages falling to 7.9 events per station by 2025 compared to higher values earlier in the record. Cold waves are defined using a 10th percentile threshold over consecutive days, capturing sustained cold exposure events that impact energy systems and agriculture. Stable atmospheric conditions allow cold air to pool near the surface, particularly at night when radiative cooling is strongest. Urbanization increases surface roughness and heat retention, disrupting this process and raising nighttime temperatures. This mechanism reduces the number of extreme cold events recorded at stations near developed areas.
Total extreme temperature days decline from over 120 per 15-year period in the 1930s to approximately 75 after 1960, representing a reduction of around 30%. This shift is driven by the drop in cold extremes rather than a consistent rise in heat extremes. The difference between the hottest and coldest annual extremes narrows by approximately 6 degrees Fahrenheit over the record, indicating a contraction in the range of temperature variability. Minimum temperatures rise more consistently than maximum temperatures, reducing the lower bound of extremes while leaving the upper bound more variable. The net result is fewer total extreme events even as some regions record higher heat-related metrics.
Regional differences appear across the dataset, with the Pacific Southwest, Pacific Northwest, and Four Corners regions recording their highest number of heatwave days in the most recent 15-year period. These areas experience more stationary high-pressure systems and reduced cloud cover, which increase solar heating and prolong heat events. Regions such as the Ohio Valley and Upper Midwest record declines of more than 30 record events compared to mid-20th century peaks. National averages combine these opposing regional patterns, where increases in one area offset decreases in another. Geographic variation reflects differences in atmospheric circulation, land surface conditions, and regional climate drivers.
Threshold-based metrics such as counting days above 95 degrees Fahrenheit produce different results because they emphasize regions that already experience frequent high temperatures. The dataset records an overall decline of 8.3% in these days across the contiguous United States since 1899, despite increases in specific western regions. Small temperature shifts near the threshold produce large changes in counts, amplifying regional signals. Stations in cooler regions contribute fewer events under this metric, reducing their influence on the national average. Percentile-based metrics adjust for local climate conditions and show stronger alignment with historical variability, including the dominance of pre-1940 heat extremes.
Non-climatic influences produce measurable changes in temperature records, particularly for minimum temperatures, where urbanization increases values by more than 5 degrees Fahrenheit in some locations over the past century. Heat stored in buildings and infrastructure is released at night, preventing cooling and raising recorded minimum temperatures. Maximum temperatures are less affected because daytime heating generates vertical mixing, distributing heat through a deeper layer of the atmosphere. This difference makes cold extreme metrics more sensitive to land use changes than heat metrics. Localized influences remain measurable within the dataset despite station comparisons and corrections.
The most recent 15-year period records 21.1 hot extremes per station compared to 35.1 during 1925 to 1939, while cold extremes fall below eight events per station in recent decades. Western regions record higher recent heat extremes while eastern regions remain below early 20th century peaks. Total extreme temperature days remain below historical highs due to the sustained reduction in cold events. Observations extend through 2025 using the same station network and measurement methods.
The dataset extends through continued station observation with more than 1,200 stations maintaining at least 92% data completeness across the record. Data collection continues through the National Centers for Environmental Information and associated archives using consistent daily maximum and minimum temperature measurements. Extreme events are tracked using fixed percentile and threshold definitions applied uniformly across all stations. The record remains accessible for verification and analysis using the same grid mapping and station-based methodology.
Source:
Christy, J.R. (2026). Declines in hot and cold daily temperature extremes in the conterminous US, 1899–2025. Theoretical and Applied Climatology.
https://doi.org/10.1007/s00704-026-06200-3
