New revelations in very old climate data

Exploring early temperature measurements, and a rundown of recent climate and communications data.

An alcohol thermometer built in 1750 by Claude Simon Passemant, at the Conservatoire national des Artes et Métiers. By Rama, via Wikimedia Commons, CC BY-SA 3.0 fr.

I first became fascinated by old weather and climate data, and the potential revelations they contain, back in 2019. I was working on this story, which was part of the Washington Post’s Pulitzer winning “2°C: Beyond the Limit” series. The piece used the history of the collection of weather data as a window into how we know how much the Earth has warmed: Namely, we have enough old observations, diligently collected by prior generations, to compare the past with the present. That story also contained the suggestion that, by looking through more old records, many not yet digitized, scientists may be able to go back far enough in time to document additional warming that has not yet been accounted for.

That’s why I was so excited to write, for CNN last month, about a new dataset that begins to realize this potential. My story starts like this:

Planet-warming pollution rates exploded after the end of World War II. James Watt’s steam engine launched the Industrial Revolution in 1769. Before that, for thousands of years, humans were clearing forested land for farming, releasing carbon from trees and plants into the atmosphere.

The severity of global warming has long depended on your frame of reference — on what temperature you think was normal for the Earth before humans began changing it. But what year should mark that moment?

That’s what makes a groundbreaking new temperature dataset released by a group of scientists based in the United Kingdom so striking. The datasets used to diagnose the modern history of the planet’s climate — and to proclaim that the world is now very near to 1.5 degrees Celsius (2.7 degrees Fahrenheit) of warming — typically begin with the year 1850.

The new one goes all the way back to 1781.

This extended time frame matters because greenhouse gases in the atmosphere increased 2.5 percent between 1750 and 1850, enough to have caused some warming that the data hasn’t accounted for.

For much more detail, my CNN story is here.

Of course, I’m not merely re-blogging my own work here; I want to add some new content and indeed, it has taken me longer than I planned to get this post together as a result.

The coolest part of all of this, at least in my view, is that to do these calculations at all, today’s scientists must synthesize multiple very old temperature records that were diligently maintained by prior generations of researchers. This requires building on the knowledge compiled in the past by organizing and combining it with what we know now. It underscores just how much climate science is an interdisciplinary field that relies on historical understanding, not just models or physics or what have you.

As the CNN piece puts it:

Early modern records go back to the 17th century. Following the path of the scientific revolution generally, measurements began in Europe — and then spread to North America and around the globe. The Central England Temperature series, the longest of its kind, begins in 1659, stitching together the work of many observers. A temperature record in Uppsala, Sweden began in 1722, aided by the work of Anders Celsius, himself.

The CNN story contained a chart showing the entire continuous temperature record kept in Hohenpeissenberg, in today’s Germany, since 1781. Here, I’d like to visualize a few more of these old records, in a format that parallels the Hohenpeissenberg chart.

For instance, here’s the temperature record for Vienna, Austria beginning in 1775, which happens to show a very large amount of warming over some 250 years:

This record was created by merging measurements originally taken at three different sites, according to Alexander Orlik, a scientist with GeoSphere Austria. The Old University in Vienna is where the record began in 1775, according to Orlik, but then came the founding of the Central Institute for Meteorology and Geomagnetism in 1851, and the measurement location shifted to there. And then there was one more move to the current location in 1872, so scientists had to carefully combine the three.

“To ensure comparability of the measurements from the three locations, the original data were homogenized and combined into a single series,” Orlik explained in an email. The paper documenting the methods used in this process can be found here; Orlik is one of its authors.

Let’s also look more closely at Uppsala. The measurements here began in 1722, but actually lapsed from 1732 to 1738. That’s because Anders Celsius embarked on a multi-year voyage, one that included a jaunt with the French mathematician Pierre Louis Moreau de Maupertuis to today’s Swedish and Finnish Lapland, where they sought to determine the precise shape of the Earth.

But temperatures were being recorded at another location in Sweden (Risinge), and Uppsala’s temperatures could be reconstructed on that basis, based on the statistical relationship between measurements at the different sites. And that’s just one way in which the Uppsala record reflects a great deal of scientific work to fill gaps in the record, to deal with different thermometer placements and station moves, to account the effects of urbanization, and so on — all detailed in this study.

Here’s the result:

Overall, the Uppsala record shows about 2C of warming when the last 10 years (which, here, do not include the super-warm 2023 and 2024) are compared with 1750-1849 or 1850-1900.

Was the globe cooler before 1850?

Flywheel of the Boulton-Watt steam engine, 1785, on display in the Powerhouse Museum in Sydney, Australia. Photo by Newtown grafitti from Sydney, Australia - via Wikimedia Commons, CC BY 2.0

The upshot of the CNN story was that the new GloSAT dataset, and a subsequent analysis using it, suggest there’s some extra human-driven warming (around .1 degrees C of it) that occurred prior to 1850 that we are not currently taking account of. What that means for the current climate discussion is complex, but it is hard to just ignore it.

Note that we do not actually see any indication of this in the Vienna and Uppsala records above. In both of these locations, the era of 1750-1850 is not colder than the currently used “preindustrial” era of 1850 to 1900. But for the globe overall, GloSAT finds the opposite — i.e., cooler temperatures prior to 1850. Of course, that won’t necessarily be reflected in the readings in any single location. Rather, it is the result of merging land station data, like those shown above, with newly analyzed air temperature measurements over the oceans from various late 1700s and early 1800s commercial vessels, such as whalers and the ships of the British East India Company.

For some, this discussion about warming prior to 1850, potentially driven by early industrialization (see the image of a 1785 steam engine above) raises questions about something that has been called the “Little Ice Age.” The GloSAT researchers notably do not use this term in the main text of their paper (it is in one citation), but it is out there. When I shared my story on Facebook, it was all over the comments section. So let’s address it briefly.

On a global basis, the period from 1790-1850 does appear to have been a bit cooler than what came after it, based on the new dataset. The question, then, is perhaps what you call this period, or what you consider to be the causes of its relative coolness (one of them, clearly, was large volcanic eruptions).

Climate scientists tend to have a bit of a strained relationship with the “Little Ice Age” concept. The most recent report from the U.N.’s Intergovernmental Panel on Climate Change opts not to use the term “extensively” due to definitional issues. Indeed, researchers have raised doubts about whether there was any globally consistent “Little Ice Age,” though there’s certainly a marker of it in the northern hemisphere. Even if temperatures were a bit cooler, the phrase “ice age” remains problematic. As one group of researchers puts it: “Compared to the changes in the proper ice ages, the so-called Little Ice Age (LIA) is a very short-lived and puny climate and social perturbation.”

It doesn’t really matter what you call, it, though. The point is that the new record suggests that at least from the late 18^th century up to 1850, the Earth was a little bit cooler than it was from 1850 to 1900, the period we currently call “preindustrial” in climate discussions. The question then becomes, did early industrialization and ongoing agriculture and deforestation contribute anything to the warming between around 1750-1849 and 1850-1900? And here, scientists are currently suggesting a small but non-negligible value around .1C of warming.

Switching gears, a few additional updates:

A not so icy Arctic Ocean in December of 2025

We’ve been in an striking place with Arctic sea ice lately, and I’ve remade one of my more popular charts to better reflect this:

At the far right, Arctic ice extent is close to the lowest on record as of the end of 2025. It’s right up there (or, down there) with 2016, 2017, and 2024.

Indeed, for a significant stretch of days in the Arctic in December of 2025, sea ice was at a record low, according to data from the National Snow and Ice Data Center. It seems possible the entire month of December 2025 will also have the lowest average Arctic sea ice on record for that month. (We’ll know very soon.)

Remember why this matters. As more and more ice forms, we’re heading towards the sea ice “maximum,” the day of the highest daily extent, which generally falls in March. 2025 already saw the lowest maximum on record. Will 2026 be even lower?

It’s looking like another year of newspaper job losses

Meanwhile, after delays due to the government shutdown, we finally are catching up in terms of U.S. jobs data from the Bureau of Labor Statistics. In particular, a recent data release holds sad news for the newspaper industry for September and October:

Estimated jobs in September were 82,800 and in October, preliminarily, 81,800. While there had not been much change earlier in the year, these numbers are considerably lower than 86,000 in December of 2024 and 87,200 in January of 2025.

In short, barring some big late year change, it would seem we’re looking at another year of job losses in the newspaper industry. This losing streak is, as you can see above, basically unbroken going back many years. If there is a “bottom,” we have not yet found it.

For my broader discussion of the importance of newspapers, see here; I expect I’ll be writing a full update for 2025 soon:

Low newspaper coverage of climate change in November 2025, and what it means

Speaking of newspapers, I’ve written extensively about their coverage of climate change, using the MECCO (Media and Climate Change Observatory) dataset, which charts the volume of coverage in 5 major U.S. papers, among other indicators. In particular, I suggested a few months back that coverage levels appear to be in decline after a major rise from roughly 2015-2021.

I advanced this thesis tentatively, with many caveats. But I’m growing more convinced of it. Why?

Well we now have data updated through November, the month of the latest U.N. climate conference in Brazil, which ran from Nov. 10-21. These late-year conferences are often peak moments for climate coverage, driving a seasonal pattern in the data. This November, however, the volume of coverage at the 5 leading U.S. newspapers (362 articles in total) was lower than in any other month of 2025 except for May, and getting closer to the pandemic low value of 297 in June 2020. In other words, there was a climate conference, but no resulting spike in U.S. coverage. That’s pretty off pattern.

Globally, in contrast, there was at least a small November uptick, as you can see in a different MECCO dataset that tracks the world.

What are the causes of this? Well, the Trump administration didn’t send representatives to the conference in Brazil, which may have detracted from the sense of journalistic urgency for some in the U.S., and nullified some possible storylines.

However, I continue to think this is more a zeitgeist thing, where there’s a broader shift away from considering climate change an urgent issue, for reasons too complex to go into here. Nothing major about the science has changed, of course. But I think there is just fatigue, reflecting in part the domestic and global failures to show significant progress on the issue.