The Met Office's Temperature Data Shenanigans: Automation, Urban Sprawl, and Inflated Warming Trends, By Richard Miller (Londonistan)
As a meteorologist with decades of experience, Andrew Sibley's October 1, 2025, takedown of the UK Met Office's temperature measurement practices raises red flags that can't be ignored. In the rush to automate for cost-cutting, swapping human-read mercury-in-glass (MiG) and ethanol-in-glass (EiG) thermometers for platinum resistance thermometers (PRTs) and data-loggers, the Met has potentially skewed daily maxima and minima, feeding into global datasets like HadCRUT5. Faster PRT response times capture fleeting peaks during daytime convection, inflating highs, while urban encroachment on stations (now often Class 4 or 5 under WMO standards, with errors up to 5°C) adds nighttime heat island biases. No published long-term studies assess these shifts, and while the Met insists adjustments are robust, critics like Jennifer Marohasy highlight similar issues Down Under, where probes record hotter maxima 41% of the time. This isn't conspiracy, it's sloppy science amplifying warming narratives without transparency. Let's dissect how the Met might be "playing" with data, intentionally or not.
The Automation Switcheroo: PRTs vs. Glass – Faster Isn't Always Fairer
Back in the 1990s-2000s, the Met phased out human observers for PRTs in Stevenson screens, enabling 15-second readings averaged to minutes, a boon for frequency, but a bust for comparability. PRTs, invented in the 19th century, respond quicker than MiG/EiG thermometers, especially in low winds where screens lack forced ventilation. Glass thermometers, read at 2100 UTC (max) and 0900 UTC (min), missed micro-peaks; PRTs snag them, pushing daytime highs higher amid convective variability.
Nighttime? More stable air lets glass catch up, so minima might not drop as sharply with PRTs, but overall, uncorrected, this biases toward warmer averages. Sibley notes rare overrides for consistency, but no multi-year intercomparison like the 1870s Glaisher-to-Stevenson shift (which adjusted for ~0.7°C radiation bias). Globally, similar transitions feed datasets, yet Met guidance admits PRTs' stability but skips climate impact assessments.
Australia's Bureau of Meteorology (BOM) echoes this: Marohasy's FOI battles unearthed Brisbane Airport parallels where probes ran 0.15°C hotter on average post-2019, with extremes up to 0.7°C, statistically significant (p<0.05). Across Aussie stations, hot-day frequency jumped 18.7% post-AWS rollout. If the Met's PRTs do the same, unadjusted records since the 1990s could exaggerate extremes, inflating long-term warming.
Urban Heat Island Creep: Class 4/5 Stations and Nighttime Nightmares
Compounding this: Urban sprawl. Met sites like Cambridge Botanical Gardens have relocated multiple times since the 1950s as buildings encroached, per archive reports. WMO Class 4/5 ratings (common now) flag errors up to 2-5°C from obstacles like structures, which trap daytime heat and radiate it nightly via concrete/steel. Urban heat islands (UHIs) amplify this; cities 5-10°C warmer than rural, worst at night under calm winds.
The Met claims UHIs are mitigated via homogenisation (comparing to neighbours) and siting in parks/airports, but Class 3-4 UK screens still suffer shading biases. Spencer et al.'s recent work ties population density spikes to spurious warming, unaccounted in many records. Night minima, key for averages, likely skew highest here, yet no dedicated UHI corrections in CRUTEM5 land data, which relies on national adjustments.
HadCRUT5: Land Obs Boom, But Quality Dip?
Land-based data's rising role in HadCRUT5, with statistical infilling for sparse spots, boosts estimated warming, but coincides with PRT/urban shifts. Updates like HadSST4 cut SST biases, but CRUTEM5 land series expand without robust UHI/PRT fixes, relying on national tweaks that may miss local errors. Critics argue 19th-century tweaks get scrutiny, but 1990s changes don't, potentially overstating recent trends. NASA/NOAA parallels show adjustments often cool past temps, but Met's non-infilled variant still lags, hinting unresolved biases.
Time for Accountability: Publish the Parallels, Fix the Flaws
Sibley's call echoes Marohasy's BOM battles: Release parallel data, run rigorous intercomparisons. The Met's opacity, no dedicated PRT studies, UHI downplayed – fuels suspicion of "playing" data to fit alarmist narratives. If post-1990s warming owes more to tech tweaks and tarmac than CO2, HadCRUT5's "faster warming" since the 1970s needs scrutiny. Transparency, not tweaks, builds trust. Meteorologists like Sibley deserve answers, before climate policy rides on potentially fiddled figures, which it likely already does.
Comments