From The ClimateRealism

By Anthony Watts

Phys.org recently published “Global warming amplifies extreme day-to-day temperature swings, study shows,” reporting on a new study in Nature Climate Change which claims that human-caused warming is intensifying rapid day-to-day temperature fluctuations, in the process creating a new category of climate hazard. This is false. The underlying study not only relies almost entirely on notably flawed climate-model assumptions, but astonishingly also ignores well-documented, real-world factors that directly affect short-term temperature variability, such as the urban heat island effect (UHI) and temperature-station siting biases. These omissions fatally undermine the study’s conclusions.

Phys.org’s press release asserts that “extreme day-to-day temperature changes have become more frequent and intense across low- to mid-latitude regions,” with researchers’ “optimal fingerprinting” methods confirming greenhouse gases as the primary cause. It further claims this daily volatility creates “a climate roller coaster” harmful to public health. The underlying Nature Climate Change study echoes this framing, presenting global maps of amplified temperature swings but never addresses the integrity—or contamination—of the surface data employed.

This is the study’s central flaw: Urban Heat Island (UHI) contamination artificially raises nighttime minimum temperatures, causing next-day temperatures to begin from an already elevated baseline. Artificial surfaces—pavement, buildings, vehicles, heat-retaining infrastructure—release stored heat overnight, pushing lows upward and exaggerating the apparent magnitude of day-to-day variations. Yet the terms “urban heat island” or UHI do not appear even once in the Nature Climate Change publication or the press release. Nor does the paper mention temperature-station placement, metadata quality, microsite compliance with quality control, or any discussion of observational uncertainty.

These omissions are inexcusable, because, as the linked Heartland Institute report Corrupted Climate Stations: The Official U.S. Temperature Record Remains Fatally Flawed demonstrates, approximately 96% of NOAA climate stations fail to meet the National Oceanic and Atmospheric Administration’s (NOAA) own siting requirements, and are corrupted by localized heat sources such as asphalt, rooftops, HVAC exhaust, machinery, or reflective surfaces. Heartland’s report shows that compliant stations—those free of artificial heat contamination—exhibit about half of the warming trend found in the corrupted network. See the figure below. Note the difference between the blue line (the uncorrupted stations) and the orange and red lines.

These findings are backed by NOAA’s own siting standards and confirm that station bias, not climate physics, drives much of the exaggerated warming in the U.S. record, and by extension, much of the world.

If the observational foundation is distorted, the derived conclusions—such as “increasing day-to-day temperature volatility”—are also distorted.

Further, the Nature Climate Change authors use global reanalyses and model-based “optimal fingerprinting” to attribute these amplified swings to greenhouse gases. But a model in which biased data serves as a foundation will simply reinforce or exacerbate those biases. None of the physical mechanisms the authors propose—soil moisture changes, pressure variability, drought feedbacks—can be meaningfully evaluated if the input dataset contains systematic nighttime warming caused by UHI and faulty station placement.

This is precisely why the U.S. Climate Reference Network (USCRN), which avoids artificial heat sources by design, shows smaller warming trends than the older, urban-contaminated networks. See the USCRN figure below showing maximum temperatures in the U.S. since 2005. Note the lack of increased temperature extremes (peaks) today compared to 20 years ago, or any obvious upward trend in the data.

Yet the current study never references USCRN data—or any equivalent bias-free baseline.

Even their health-risk findings rely on heat-biased data. If nighttime temperatures are artificially elevated due to urbanization and poor siting, the apparent “roller-coaster swings” reflect local land-use change, not global climate physics. Any claimed health correlations with mortality data would therefore be mixing genuine weather effects with measurement artifacts.

What’s missing, in other words, is climate science’s most basic requirement: separating real climate signal from non-climatic noise.

UHI, land-use change, asphalt expansion, vegetative loss, waste heat emissions, and the explosive growth of built environments all elevate nighttime minimums and increase apparent variability. The study treats all variability as atmospheric in origin, ignoring the man-made thermal reservoirs embedded in modern cities.

A rigorous scientific analysis would have:

• Examined UHI contamination explicitly.
• Segregated rural high-quality stations from urban or microsite-compromised ones.
• Compared modeled projections to USCRN’s pristine, bias-free observational network.
• Quantified the effect of siting violations documented in my 2022 Corrupted Climate Stations study.

Instead, the authors rely on global models and “fingerprinting” techniques that simply assume the data are valid. That assumption is not supported by the evidence, and the report’s conclusions cannot be meaningfully distinguished from the biases embedded in the underlying data. This is not “a new extreme climate hazard”—it is the predictable result of modeling data with errors, then attributing those errors to greenhouse gases.

Phys.org does its readers a disservice by not fact checking the Nature study and examining other possible causes for its conclusions. Instead of a thoughtful scientific examination of the novel claims made in the study it is reporting on, Phys.org’s readers got a fawning, uncritical summary of the flawed Nature study, which exaggerates day-to-day variability based on the well-documented role of UHI and siting flaws. That’s not inquisitive journalism, that’s promotion; in reality no better than a press release.