From ClimateRealism

By Anthony Watts

In a recent USA Today story “Hottest summer ever? US breaks longstanding record,” the writer claims that the United States just experienced its hottest summer ever based on the National Oceanic and Atmospheric Administration’s (NOAA) latest temperature analysis. That claim is misleading, based on an incomplete causal analysis. The best available evidence suggests that much of the reported “record” depends on adjusted temperature datasets and weather stations increasingly affected by urban development, while the nation’s highest-quality climate network tells a very different story.

USA Today accurately repeated NOAA’s announcement that the average temperature for June through August 2026 was the warmest in the 132-year record for the contiguous United States, edging out 1936 and 2021 by just 0.4°F.

That sounds straightforward enough, but unfortunately, it isn’t.

The first problem is terminology. USA Today repeatedly calls this the “hottest summer ever.” NOAA itself does not. NOAA consistently describes 2026 as the “warmest summer on record” based on average temperatures. The word “hottest” never appears in NOAA’s announcement. That distinction matters because “warmest average” and “hottest” are not the same thing.

Meteorologist Chris Martz carefully examined NOAA’s data and found that the reported record was driven largely by higher than normal overnight minimum temperatures rather than record daytime highs. Using average daytime maximum temperatures, 2026 ranks behind several earlier summers, including the legendary Dust Bowl summer of 1936.

That fact is hardly a trivial. Most Americans experience summer heat primarily during the daytime, not as an average of daytime highs and nighttime lows. Martz also points out something almost entirely absent from the media coverage. Although the national average narrowly set a record, very few parts of the country actually experienced their hottest summer.

Using NOAA’s own data, he found that only seven of 344 climate districts, barely two percent, recorded their warmest summer on record in 2026. Those districts were concentrated almost entirely in the Desert Southwest, see in Figure 1 below. By contrast, during the infamous summer of 1936, fifty-one climate districts set all-time summer records.

That paints a much different picture than headlines suggesting the entire nation experienced unprecedented heat, when it did not.

Then there is an even larger issue that deserves far more attention.

NOAA’s long-term temperature record is not based solely on raw observations. It relies on a “homogenized” dataset in which station records are adjusted to account for changes in observation times, instrumentation, station moves, and other factors. NOAA openly acknowledges using these adjusted datasets for its national rankings.

Whether those adjustments perfectly capture reality is a legitimate scientific question; Martz argues they may not.

He notes that some adjustments are larger than the tiny 0.4°F margin separating 2026 from 1936, raising reasonable questions about whether the new record is as certain as headlines suggest. He also discusses concerns that NOAA’s pairwise homogenization algorithm may not adequately remove gradual urban heat island effects and could unintentionally blend urban warming into nearby rural stations. Whether one ultimately agrees with every aspect of that critique, these are serious methodological issues worthy of discussion, but the USA Today never opens that discussion.

There is another dataset that deserves attention.

The United States Climate Reference Network, or USCRN, is widely regarded as America’s gold standard surface temperature network. Unlike many older Cooperative Observer stations, USCRN sites were specifically designed to avoid contamination from expanding cities, parking lots, buildings, and other artificial heat sources.

The USCRN Figure 2 graph below tells an interesting story.

The USCRN network data shows one extraordinary warm month during March 2026, with a national temperature anomaly reaching nearly 7.7°F above average. But June, July, and August do not stand out nearly as dramatically. Instead, they fall within the normal range of variability seen across the roughly twenty-year USCRN record.

If America truly experienced an unprecedented hot summer across its highest-quality climate observing network, one might expect the USCRN record to show an equally extraordinary signal. It doesn’t, and that discrepancy deserves investigation, not dismissal.

One plausible explanation is the urban heat island effect (UHI).

Many of NOAA’s legacy stations are located at airports or sites that have experienced decades of surrounding development. Asphalt, concrete, buildings, jet traffic, air conditioning exhaust, and expanding urbanization all contribute to elevated nighttime temperatures. Those are precisely the temperatures that drove much of NOAA’s reported national record. A perfect example of such a situation is the weather station is Reno, NV where an investigation revealed large warm biases in official temperature measurements. A similar problem has been identified by Roy Spencer, Ph.D. with temperatures recorded in and reported for Phoenix.

That possibility becomes even more significant in light of extensive station surveys documenting widespread siting problems throughout the U.S. Historical Climatology Network. The issue is not whether cities are warmer. Everyone agrees they are.

Then there is the problem of “ghost stations,” with NOAA reporting temperature data from stations that no longer exist. NOAA simply makes up temperature measurements for a huge number of closed stations based on averaging from surrounding stations, to maintain record continuity – but made up or assumed temperatures are not actual data.

The issue is whether artificial heat island biases are being fully removed and assumed “data” from ghost stations not counted, before national climate conclusions are announced. Reasonable scientists can debate that question, journalists should at least acknowledge it, but they don’t

Instead, USA Today presented NOAA’s adjusted temperature ranking as settled fact while ignoring both the methodological caveats and the existence of higher-quality observational data that display no record setting warmth. The result is another headline that generates maximum attention while minimizing  scientific complexity and compromising accuracy.

Good climate journalism should do the opposite. It should explain uncertainty, distinguish between adjusted datasets and pristine observations, and recognize that a record based on a few tenths of a degree deserves careful examination rather than unquestioning repetition.

Readers deserve more than headlines declaring “hottest ever.” They deserve the whole complicated story, with all the ifs, ands, buts, and “we just don’t know or can’t be sure,” included.