Do Old Temperature Graphs Prove that Agencies are Lying?

One of the enduring points of debate on social media has to do with claims that NASA and NOAA (and for that matter, pretty much the entirety of climate science) is fabricating the warming we've been experiencing since the 20th century. The "proof" of this sometimes comes from comparing old graphs of global temperatures with newer graphs. The difference between them is thought to be proof that NASA, NOAA (and the rest of climate science) is lying. Tony Heller is perhaps the most popular purveyor of this kind of conspiratorial thinking, and this week, he posted a couple videos where he recycles the same kind of stuff he's been saying for over a decade. His main conspiratorial argument in these comes from a graph from 1974. There are a couple versions of it that can be found online; one is from Newsweek from April 28, 1975.

The source for this is said to be the National Center for Atmospheric Research. The description of this graph in the main body of the article says, "Murray Mitchell of the National Oceanic and Atmospheric Administration reveals a drop of half a degree in average ground temperatures in the Northern Hemisphere between 1945 and 1968. According to George Kukla of Columbia University, satellite photos indicated a sudden, large increase in Northern Hemisphere snow cover in the winter of 1971-72." This would indicate that this graph is Northern Hemisphere Temperatures. The data points in this graph appear to be spaced about 5 years apart, but it looks hand drawn. There's another, roughly identical version of this same graph.

This graph appeared in the Des Moines Sunday Register (and possibly elsewhere). The units have changed, and earlier data was added to the beginning of the graph. This earlier data is said to be "Northern Hemisphere Only." The title, "A Century of Global Climate Changes" gives the impression that the solid black line is global, and the NH temperatures were added to complete the century. Here we're explicitly told here that we're being given 5-year averages. The data for these graphs appear to be nowhere except in these news stories, so we have to just trust that whatever dataset they had access to, they represented accurately, but the graphs themselves appear to basically hand drawn. The 5-year increments for the datapoints don't appear to be precisely where they belong. This plus the thick line suggests that this graph is not intended to be extremely accurate, and even taking what's here to be accurate to 0.01 C may be asking too much. Now let's compare this to Tony Heller's portrayal of the difference between these graphs and NASA's current time series.

Heller's Scribbles on a Newspaper Graph

Here we're told that "NASA surface temperature graphs have completely erased the post-1940 cooling seen in 1974 NCAR graphs." And since NASA agrees with pretty much all other GMST datasets (see top graph), this is "Overwhelming Evidence Of Collusion." He assumes that the "NCAR 1974" graph is correct, which means and NASA is lying, which means that, since all these datasets agree, "There is no conceivable way this graph could have been produced without lots of collusion between the four 'independent' agencies."

Note: when some dude says that everyone is lying except him, the chances are much higher that the dude making the accusation is either deluded or lying than that he's right and everyone else is actually lying. But let's test this claim. Maybe we can do a little better than scribbling on a graph.

Since the data called "NCAR 1974" doesn't exist anywhere, there's no way to download it. And as we've seen, it's more likely this is Northern Hemisphere temperatures, not global. It's also pretty much certain that this is land-only data, not land and ocean. So already we can see that Heller's claim is based on an apples to oranges comparison. I carefully attempted to duplicate the above graph in 5-year increments. I'm pretty certain I was able to reproduce it with each data point within 0.01 C of what's on that graph. Since I put this in a spreadsheet, these datapoints will be spaced properly 5 year apart. I then downloaded the current data for two of NASA's datasets, both a land-only dataset and a northern hemisphere dataset. Ideally, we'd use a NH land-only dataset, but I haven't found access to that. I set both to the 1881-1910 baseline. Here's what I came up with.

These graphs are clearly not identical, but they don't look nearly as different as Heller would have you believe. The shape of the NCAR graph looks different than the newspapers because of the scale. I set the dimensions of the graph to be more similar to what we're used to seeing. Anyway, the way I lined these up, there's little less cooling following the early 1940s, but there is cooling. But already we can see these imaginary conspiracies with multiple agencies colluding to fabricate warming is a bit far fetched. When we compare what's called "NCAR 1974" to actual datasets, rather than Heller's scribbles, many of the differences diminish considerably. This is due to 1) an imprecisely drawn newspaper graph and 2) uncertainty about what was actually being reported in that graph. It's probably NH land-only temperatures, not global temperatures. But this does not explain all the differences. Can we say more?

Zeke Hausfather actually weighed on this on Twitter. I find Twitter to be a frustrating way to make multi-stepped arguments, but the point he makes is still valid. The most obvious point he makes is, "It shouldn't surprise anyone to learn that our collection of historical observational data has grown by orders of magnitude since the 1970s. Back then (proto-) global temp estimates were based on data from a few hundred primarily mid-latitude N. hemisphere land-only stations." In other words, NASA didn't have as many thermometers around the globe in 1974 as they do now. New observational data has been added that provide more "historical observational data" for the time frame between 1880 and 1970 than was available to the NCAR 1974 dataset. You can see this here.

Notice with the upgrade to v4 over v3, the number of thermometers increased dramatically. So the dataset that NASA (and other agencies) are using is better now even for the years prior to 1970. We've also seen that scientists have been able to identify nonclimatic biases in the data, that when removed, slightly altar global temperature trends. Overall, bias correction decreases the global warming trend, since adjustments warm temperatures prior to 1940. However, much of this warming before 1940 was due to adjustments to sea surface temperatures. Land temperature adjustments generally increase the warming trend, but not by as as much as SST adjustments.

Effect of Bias Correction for Land Temperatures

Effect of Bias Correction for SSTs

In other words, the big picture about Heller's accusation is that the graph he's using is not very well drawn, and it's not even terribly clear what it represents. However, NASA's current dataset is actually better than what is represented in "NCAR 1974." It represents more thermometers and more sophisticated bias correction. The differences are simply not as substantial as Heller wants you to believe. Rather than evidence that agencies are lying, this is evidence that they are improving the accuracy of the historical record.