As the long cold winter of 2025-26 in North America drags on, we are predictably told that heat is “shattering” records when not “smashing“ them. The latter being a forecast not a report, complete with map giving the impression the continental United States had been submerged in lava or some such. There’s a lot of ranting and raving. But what there is not, in any of these stories that we could find, is any mention of where and how these temperatures were measured or how far back such records go. The dopiest illustration of this point is Heatmap thundering mathily “A Once-in-a-4,433-Year Heat Wave Is Hitting the Western U.S.” How the devil can they know such a thing? Even ignoring the pseudo-precision (exactly 4,433? Not 4,434 or 4,432?) unless they have solid temperature records going back many multiples of 4,433 years it’s insolent rubbish. They don’t even know what kind of heat wave hit there in 1825, let alone 2,025 BC. The heatwave led to such a predictable wave of instant attribution that we deal with it separately in an item below. But here we want to ask what they think they know and how they think they know it.
Of course when you get that kind of nonsense World Weather Attribution can’t be far behind. And isn’t. According to Heatmap:
“‘The extent and magnitude of this particular heat wave is without comparison to anything that we’ve seen in March,’ John Abatzoglou, a professor of Climatology at the University of California, Merced, who specializes in climate impacts in the West, told me. That’s partially because this heat wave would be ‘virtually impossible for the time of year in a world without human-induced climate change,’ per a report released Friday by scientists from World Weather Attribution.”
No. The reason we’ve never seen anything like it is because we haven’t been looking. WWA, as we explain elsewhere, only looked at data back to 1950. But if we did look further, say to the 1930s, the government changed the data. Not how you do science… outside say the USSR.
Naturally news outlets were all over the supposed all-time record high for the U.S. in March set in Martinez Lake, Arizona. And with the curiosity so famously absent from modern journalism they did not ask questions like how long temperatures have been measured there, since the community itself only dates back to 1958, and how rigorous was the measurement? Well, not. It was something “2 miles west-northwest of Martinez Lake, AZ” that doesn’t seem to be an official weather station that uses established protocols, and as far as we were able to determine its records stretch back only two years. But nobody was very curious when they got to shriek “Record-torching March heat ‘virtually impossible’ without climate change” (climate change being, we say again, a statistical description of changing weather not a causal force) as:
“Multiple all-time March heat records crumpled on Wednesday and Thursday, March 18-19, as one of the most extreme weather events in world history blitzed the Southwest U.S. and far northwest Mexico with unprecedented March heat.”
If you’re wondering what tabloid produced that rant, it was Yale Climate Connections. And no, the story didn’t mention Urban Heat Islands. Or the past. Why cause talk?
Well, Matthew Wielicki wants to. He goes and looks at statewide temperature records, some of which do go back to the 1890s, the blink of an eye geologically but an eternity plus to journalists, and says:
“What these records reveal is not stability followed by sudden disruption. They reveal a system that has always been highly variable.”
And moreover “where large spikes in March temperatures occur well before the recent period of elevated atmospheric carbon dioxide.” And as he insists:
“The most severe heatwaves in United States history still occur in the early twentieth century, particularly during the nineteen thirties. As discussed in my prior work, Heat waves show no correlation with atmospheric GHG concentrations or emissions, the Dust Bowl era remains the benchmark for extreme heat in the observational record. These events occurred when atmospheric carbon dioxide concentrations were far lower than today. This is not a minor detail. It is central to the discussion. Any claim that modern heat events are fundamentally different must reconcile with the fact that more extreme events occurred under very different climatic conditions.”
And that we know very little about such events except in the very recent past. On this point Chris Martz posted:
“Weather records were made to be broken. We have at best 150 years of instrumental data, maybe only 50 with good quality over a large area, on a planet that is 4.5 billion years old.”
Which is not, or should not be, to say that it is impossible to detect trends. Rather, it is possible iff, as the mathematicians say (“if and only if”), you have sufficiently long and precise temperature records that you can pick signal out of noise, and they must be long and precise indeed if you’re working with things are inherently variable.
Like, say, weather. In which, despite those very short and incomplete data sets, though you might not know it from reading most media outlets, records are set all the time. For instance, in complaining about how hot it had been last winter the Washington Post eventually got around to the bit where last January set a bunch in the US:
“In addition to January 2025 having more cold than warm records, it was the first month since May 2023 to reach such a feat. More than 1,500 records for cold were set between Jan. 19 to Jan. 23.”
Of course the Post went right back to how hot it had been. But imagine what they’d have said if over 1,500 records for heat had been set between Jan. 19 and Jan. 23 2026? Assuming they hadn’t run out of inflammatory verbiage.
Also imagine what they’d have said if they’d bothered to see how far back our records went for any of those places where cold or hot records were set. Because if the answer is, say, 1952, those records mean nothing.
Trends are obviously central to the climate change debate. In a variable data set such as is often seen in these debates, there is much debate as to whether the trend in the data is up, down or flat. In my mathematical ignorance, I did an Internet search (I don't use Google) and immediately came across Holt's Linear Trend method. I make no doubt there are others. I haven't done a deep dive into this subject. Nowhere have I seen any mention of just how the trends in the data set have been derived. Eyeballing it? Holt's method? Another one? I presume that the scientific community has an accepted method(s) of deriving these trends. Would it not be reasonable to say, "According to ____________ method of derivation, the trend is ..." In the days I dealt with statistical data, we were strictly enjoined to state what we had removed from it as being bad or outliers, and what methods we used to draw conclusions from it. Are they?