Guest Post by Willis Eschenbach
As many of my posts start out, “I got to thinking about …”.
In this case, I got to thinking about the Berkeley Earth global temperature dataset. So I got their gridded file “Monthly Land + Ocean Average Temperature with Air Temperatures at Sea Ice” covering 1850 to 2021 and took a look at it. I started at the first month’s data, January 1850 … and my jaw hit the floor.
Figure 1. Berkeley Earth surface temperature, January 1850. White areas have no data.
What shocked me was the red-orange circle centered north of New Zealand, as well as the half-circle in northern South America.
Clearly, what they are doing is taking one temperature reading at one point, and extrapolating it to a surrounding area. How big an area? Well, the circle north of Kiwiville has a diameter of ~ 1,600 km (~ 1,000 mi). It covers an area of 8,700,000 square km (3,360,000 sq mi). That’s about the area of the continental US … estimated from one temperature reading.
And there’s no island anywhere near the center of that circle, so it would have been a temperature taken from a ship …
Now, if you look carefully you’ll see that the southern part of the circle is more orange, it’s a bit cooler. That makes me think that they’ve used modern measurements of the temperature gradient around the center, and adjusted them to fit the single surface temperature measurement. To check that, let me go take a look at the January temperatures of that region over time … I’m writing this as I’m analyzing the data, so I’ll be back soon.
OK, here’s what I find.
Figure 2. January temperatures from 1850 to 2021 of a vertical (North/South) slice through the middle of the red circle north of New Zealand in Figure 1. The slices run from 16°S to 46°S. Temperatures are expressed as anomalies around the temperature at the center of the circle, at 31° South latitude.
Looks like my guess was not too wild, they’re using some kind of procedure like that.
But is extrapolating the temperature of an area of the ocean the size of the continental US from one single temperature measurement a reasonable procedure?
Having spent a good chunk of my life at sea, I’d have to wonder. I’ve seen areas where the ocean changed temperature by a few degrees or more in a few hundred meters. Where a cold current hits a warm current, there is often a clear dividing line and little mixing across the line.
And over the land the changes are much larger, like say over northern South America in Figure 1.
So … the whole of the US from one thermometer? Where I live, for example, it almost never freezes. But a kilometer (~ a half-mile) away, it freezes a number of times per year. Here’s the freeze warning for tomorrow. I live near the coast north of San Francisco, in the narrow sliver of green near the coast to the left of the “S” in “Santa Rosa” … it probably won’t freeze here. The stretch along the coast in this area on the western side of the first range of hills, from about 600′ to 900′ (180m to 270m) in elevation, is known locally as “The Banana Belt” because it hardly ever freezes. We grow lemons, limes, and avocados on our patch of soil.
Figure 3. Freeze warning for Wednesday, February 23, 2022
So I’ll leave it to the reader to decide if one thermometer is enough to estimate the temperature of the entire continental US … and while you consider that, here’s a video loop of the coverage of the first twenty years (240 months) of the Berkeley Earth global surface temperature data.
Figure 4. Video loop of the first 240 months of the Berkeley Earth global surface temperature.
I find the changing coverage of Australia over time most perplexing.
At the end of the day I got to wondering … just when did Berkeley Earth finally achieve complete global coverage? Here’s the sad answer.
Figure 5. Percent coverage, Berkeley Earth surface temperature, divided by land and ocean.
Interesting. The effects of the wars on the temperature reports from oceanic shipping are quite visible. And even with extrapolating out so that a single thermometer covers an area the size of the continental US, land coverage didn’t exceed 90% until after WWII … and total coverage wasn’t achieved until 1978.
I have no overarching insights from this research, other than that the spotty nature of not just this Berkeley Earth dataset but most climate data is a continual thorn in the side of researchers, and it makes all conclusions about the climate very tentative.
Oh, yeah … about the title of the post, “SWAG”.
A “WAG” is a “Wild Ass Guess“. And no, I didn’t make that up.
And a “SWAG”, on the other hand?
That’s a far superior creature, a “Scientific Wild Ass Guess“ … like say the various estimates of global average temperatures in the 1800s.
My best wishes to all, blessed rain here, what’s not to like?
As Is My Custom: When you comment, I ask that you quote the exact words you’re discussing, so we can all be let in on the secret of just who and what you are on about.