The two warmest years in the record of the weather station at Port Angeles, Washington, and the five warmest in the record of the station at Whiteriver, Arizona, were never measured by either station. Both are part of NOAA’s U.S. Historical Climatology Network, 1,218 long-running stations in the 48 contiguous states, and when a station stops reporting, NOAA’s adjusted file for the network continues its record with values estimated from surrounding stations and marked with the flag E. NOAA’s raw file for Port Angeles holds three recorded months in 2008 and nothing after, yet the adjusted file continues the station year by year through 2025, and its estimates for 2015 and 2016 stand about a degree above the warmest year the station recorded in full. The Washington State Climate Office’s trend tool shows Port Angeles as a continuous record from 1896 through 2022 warming 0.11°F per decade, a record whose last 14 years are entirely estimates, and neither the tool’s page nor the University of Washington’s announcement of it mentions estimates or closed stations, the announcement describing the data as federal weather observations. At Whiteriver every month from 2006 on is an estimate, including the months of 2006 through 2008, when the station was still reporting and NOAA’s procedure set its recorded readings aside and replaced them. The five warmest years in its record all fall between 2017 and 2025 and are all estimates, the warmest of them about 1.7 degrees above the warmest year the station recorded in full. In 2026 the U.S. Geological Survey reported a rise of about 2.5°F at the station from 1980 to 2023, in support of tribal members’ accounts of a changing climate, over a period whose last 18 years, 41 percent of it, consist entirely of estimates. The survey’s 58-page report states that estimated data from 2009 to 2023 were used, which omits the three years in which recorded readings were replaced, and its four-page fact sheet, which carries the 2.5-degree figure, does not say that estimated data were used.

It's in a computer model that they put the assumptions into Learn more at co2coalition.org/

I was arrested, interviewed under caution for 2 hours by two sturdy lllllllesbians and my case sent to the @CPSUK for the crime of… ‘causing offence to myself’.

Let's start something! We clearly need to define "AI" for what it truly is. #ArtificialIgnorance

Almost every closed station in the dataset is one that inconveniently shows no long-term warming trend, only cyclicity, despite attempts to manufacture one via adjustments. And most were closed in the early 2010s, right after climategate. Colour me increasingly sceptical.

Lol @ the people claiming that this is a “strongly negative PDO” 😂😂😂

Wow! Latest ECMWF 12z is going all out in the stratosphere, now forecasting an actual Stratospheric Warming event with a wind reversal, making it an official SSW event. It will likely not remain at this, as this would be an unprecedented event, and the model trend map shows quite a spread between the last runs. But even seeing this suggested at this time is really insane. All the maps are available in my Polar Vortex hub (link in the comments) @judah47 I have seen you noticed the 12z ECM, but did you know it also has a proper reversal? 😁

NOAA’s file does not just fill holes, which can be expected when dealing with large datasets. It also overwrites real readings at nearly every station. The replacements lean cool in the past and warm in the present, the same shape as NOAA’s adjustments overall, and in 2025 that includes a full year of readings at 29 working stations, Key West’s airport among them. To be clear, the estimates are not only for closed stations. NOAA’s file also replaces readings that stations actually recorded. This happens when NOAA’s adjustment procedure decides that a stretch of a station’s record is too short to correct reliably, as NOAA explained in 2014; it discards the readings and puts estimates in their place. From 1895 through 2025 there are 93,047 such months, at 1,217 of the 1,218 stations. The top chart answers a simple question: when NOAA replaces a recorded daytime high with an estimate, is the estimate warmer or cooler than the reading it replaced? Each bar is a decade. A bar below zero (blue) means the estimates in that decade were on average cooler than the readings; a bar above zero (orange) means warmer. The comparison is made after NOAA’s “reading-time correction” has already been applied to the readings, so the one adjustment NOAA has documented and tested is not the cause of what you see. The result is that in every decade from the 1890s through the 2000s the estimates are cooler than the readings, by roughly half a degree to a full degree Fahrenheit before 1980 and by two or three tenths of a degree from the 1980s through the 2000s. In the 2010s and the 2020s the estimates are warmer than the readings, by 0.06 and 0.33 degree. The bottom chart shows how many recorded months were replaced in each decade: between 5,000 and 9,000 in every decade through the 2000s, fewer recently. However, for 2025 the replaced readings include every single month at 29 stations that reported a full year of data, among them Key West International Airport, Baton Rouge Metro Airport, Astoria Airport, and the National Weather Service office in Blacksburg, Virginia. The number of such stations has risen from 6 in 2015 to 29 in 2025.