Analysis methods

Climate Central’s analysis is based on recent data from weather stations, regional-scale outputs from climate projection models, and a common technique for deducing best-guess local climate projections from regional ones. In its essence, this method involves calculating differences between current and future global climate model simulations, and applying them to observed climate data from the same vicinity.

Climate Central first identified weather stations closest to each city on the list, and then found the nearest regional location with model projections available.

For current patterns, Climate Central looked at August daily high temperatures at each weather station for 1981-2000, counting the total days exceeding 90°, 95° or 100° F over those 20 years, dividing by 20, and rounding off. The answers for each city were the average number of days reaching above each temperature in one August during the 1980s or 1990s — roughly today’s climate. (Any given August might have many more or fewer hot days.)

For future projections, Climate Central averaged outputs from 12 major climate models at the available regional location nearest to each city for 2046-2064, and for 1981-2000 (simulated values, not actual ones). We then calculated changes in the 20-year average monthly maximum temperature between these simulated future and current climates, and added the differences to the actual 1981-2000 weather station data from each city (already described).

This last step created new, simulated data for each city for 20 Augusts in the middle of this century. We then applied the same method that we used with actual 1981-2000 temperatures to estimate the average number of days over each temperature threshold in this future scenario.

The resulting projections give long-term averages, not predictions for any individual year; actual outcomes will vary significantly from year to year due to the natural variability of climate. Furthermore, because the modeling and methods used involve uncertainty, the projections should be taken as best guesses within a range of uncertainty. True long-term averages will likely prove somewhat higher or lower than the projections here. However, all twelve models are unanimous in projecting increased hot days from the present by the middle of the century.

All model outputs used were based on a medium-high greenhouse gas emissions scenario called “A1B” by the Intergovernmental Panel on Climate Change. Carbon dioxide and other greenhouse-gas emissions this decade have exceeded the A1B scenario thus far. This makes the projections here conservative, matching a future in which emissions are reduced compared to the current trend.

The techniques and general climate projections used are well established in the scientific literature (for example, see here and here).