Guest Post by Willis Eschenbach
As with many of my meanderings through scientific landscapes, this one starts with “I got to wondering …”.
In this case, I got to wondering how well the Central England Temperature (“CET”) matches up with the temperature of the planet.
In part, I was wondering because I keep reading that the Little Ice Age, which bottomed out in about 1700AD, was just a European phenomenon. I’ve often wondered if just a part of the world could cool as much as it did in the Little Ice Age without the rest of the world cooling as well.
The CET is one of the longer temperature records. It’s a curious record in that it’s made up of a combination of temperature records of a changing variety of stations in the general area of Central England. It stretches from 1659 to the present. Here’s the more recent part of the CET record (seasonality removed) and the Berkeley Earth global temperature record.
Figure 1. Monthly Central England Temperature (CET) and Berkeley Earth global temperature.
Hmmm … looking at that it seems that there is very little relationship between the two. The R^2 (lower left corner) is a measure of the closeness of the relationship, varying from R^2 = 0 (no relationship) to R^2 = 1 (total agreement). Pondering the question, I realized that the problem is that over a short period of time, months or years rather than decades or centuries, the temperature in a small area of the planet like Central England varies a lot more than the temperature of the globe.
So what I needed to do was to adjust the short-term variance of the CET to match that of the Berkeley record, while leaving the long-term variations intact.
To do that I first took a LOWESS smooth of the CET data. That gave me Figure 2.
Figure 2. Full Central England Temperature record, along with a LOWESS smooth of the CET. You can see the coldest part of the Little Ice Age around 1700AD.
Then I subtracted the LOWESS smooth from the recent CET data (from 1850 to the present to match the period of the Berkeley Earth data). This left me with just the short-term (months to years, not decades or centuries) variations in the CET data.
I also did the same to the Berkeley Earth temperature data, to determine the short-term variations in that data.
Once I had both sets of short-term variations, I adjusted the average size of the CET short-term variations to match the average size of the corresponding Berkeley Earth short-term variations. Finally, I added the LOWESS smooth back in to reconstruct the original CET data, but with much less short-term variations.
I then used a simple linear regression on the CET data to give the best overall fit to the Berkeley Earth data. Figure 3 shows that result.
Figure 3. Central England Temperature, variance adjusted, compared to the Berkeley Earth global temperature.
This was a big surprise to me, and surprises like this are what keep me doing science. I did not expect the temperature of a small part of England to be in such good agreement with the global temperature. The R^2 is 0.67, much larger than the previous R^2 of 0.07 shown in Figure 1. And since the Little Ice Age is clearly visible in the earlier part of the CET record shown in Figure 2, this greatly ups the odds that the Little Ice Age was a global phenomenon.
Now, I’ve also heard the claim about US temperature records, that the US is only ~ 2% of the global area and thus we shouldn’t expect it to be similar to the global record. So I used the same technique to compare the Berkeley Earth US record with the Berkeley Earth global record. Figure 4 shows that result:
Figure 4. US temperature, variance adjusted, compared to the Berkeley Earth global temperature.
Now, the US is much larger than Central England, and thus as we might expect, the agreement with the global temperature is even better than that of the CET. The R^2 is now up to 0.76. Over the last 170 years, the US temperature has been doing very close to what the global temperature has been doing. Who knew? Certainly not me.
Next, here is the correlation of individual 1° latitude x 1° longitude gridcells, variance-adjusted as described above, with the global average temperature.
Another surprise. The land masses generally correlate well with global average temperature, as does much of the ocean … except the North Atlantic, which is negatively correlated with the global mean.
Finally, with all of the above in mind, I’m gonna go out on a limb here and say that the Little Ice Age was most likely a global phenomenon.
And that was my scientific surprise for today … how was your day?
My very best to all,