It’s an impressive scientific achievement, and the authors deserve recognition for solving such a long-standing puzzle.

However, I busted out laughing when I read the underlying study in PNAS, entitled The 1831 CE mystery eruption identified as Zavaritskii caldera, Simushir Island (Kurils).

What made me laugh was their claim that the volcano caused a global temperature drop of 1°C (1.8°F, aka “Nearly Two”). I’ve looked at lots of volcanic eruptions. There are links to my research on eruptions and temperature here. I have found that there is no consistent response in the global average surface temperature. I thought “This should be interesting”. I looked at the study to see what evidence they have.

They use the Berkeley Earth land-only temperature data to verify their claim … the problem is that even using their methods, the Berkeley data doesn’t support their claim. The image and the caption in Figure 1 below are from their study.

Figure 1. ORIGINAL CAPTION: Figure S1: Time series of instrumental temperature anomalies spanning 1790–1840 CE. The period 1831–1833 CE is shaded grey. Instrumental temperature records are from Berkeley Earth [www.berkeleyearth.org, (8)] and show estimated global land average temperatures. These represent annual temperature anomalies relative to the 1790–1807 CE mean (a period without major volcanic eruptions).

Squint as I might at Figure 1, I’m not seeing the 1°C temperature drop post-1831. Not even close.

Plus, the Berkeley data is monthly, but they’re using annual averages. This is throwing away a bunch of valuable data.

So I decided to take a proper look at the temperature record of all the major volcanic eruptions for the last two centuries. This eruption information is kept by the Smithsonian Institute in their Volcanoes of the World database. I used only the largest volcanoes, those with a Volcanic Explosivity Index of 6 or more. Here’s that record, compared to the same Berkeley Earth temperature dataset used in their study.

Figure 2. Large volcanic eruptions and the Berkeley Earth global monthly land surface temperature anomaly record.

Here’s the curious part. Of these 7 large volcanoes, there’s only one that is followed by a clear drop in temperature. That one is Pinatubo, the poster child for the claim that eruptions affect global average surface temperatures. All the rest of these big volcanoes show nothing anomalous. Even the two biggest eruptions, Tambora 1815 and Krakatau 1883, were not followed by global cooling in the Berkeley Earth record. Temperatures actually rose after each of those eruptions.

Anyhow … that’s why I laughed when they said they’d found a 1°C (1.8°F) cooling resulting from the 1831 volcano. I didn’t even have to look at the record above to know the chances were small that a large effect would be visible.

And in the event, Figure 2 shows the Zavaritskii volcano occurred halfway through a cooling period, and the cooling post-volcano and pre-volcano are indistinguishable.

I say that this lack of temperature response to eruptions is explained by my hypothesis that a variety of emergent phenomena work to thermoregulate the earth’s temperature. In particular, when temperatures fall due to say eruptions injecting aerosols into the stratosphere, the climate responds inter alia by cumulus cloud fields and thunderstorms forming later in the day. This allows more sunshine to be absorbed by the surface, which opposes the volcanic cooling.

For a different look at this question of temperature changes due to eruptions, I took a look at the UAH MSU satellite-based atmospheric records. Figure 3 shows eruptions and the temperatures at four different atmospheric levels from the top down—stratosphere, tropopause, middle troposphere, and lower troposphere.

Figure 3. UAH MSU satellite-based atmospheric temperature anomalies, in order of altitude from highest (top panel) down to lowest (bottom panel). Red lines show the dates of eruptions with a Volcanic Explosivity Index (VEI) of 5 or more. Numbers following the names show the VEI of the eruption.

Now, this shows the curious nature of the atmosphere—there is very little correlation between stratospheric temperature anomalies (top panel) and the rest of the lower atmosphere (lower three panels). There are clear indications of the 1982 and 1991 volcanoes in the stratosphere … but by the time that works its way down to the lower troposphere, the situation is quite different. There, only the Pinatubo/Cerro Hudson pair shows a temperature drop following the eruption, and even that is not visibly different from a number of other temperature drops during the period.

And by the time the effect of the eruption-injected stratospheric aerosols makes it down to the surface, we get Figure 4.

Figure 4. A closeup of the recent end of the data shown in Figure 2, with additional eruptions. Vertical red lines show VEI 5 and VEI 6 volcanic eruptions during the satellite era along with the Berkeley Earth global monthly land surface temperature record.

If the Mount St. Helens eruption caused the subsequent cooling, did the El Chichon eruption cause the subsequent warming? Once again we see the ~0.5°C temperature drop after Pinatubo, but other than that, there’s no clear indication that eruptions affect the global average surface temperature.

Yes, I know that this goes against the revealed wisdom. That wisdom says:

Eruption –> Stratospheric SO2 Aerosols –> Increased Albedo –> Less Sunlight –> More Cooling

Sounds logical, but it ignores the fact that the climate is a dynamic system that is constantly changing and adapting to a variety of forcings and factors. The revealed wisdom stops too soon. I say it goes like this:

Eruption –> Stratospheric SO2 Aerosols –> Increased Albedo –> Less Sunlight –> Cooling –> Delayed Formation Of Daily Tropical Cumulus Cloud Fields –> Decreased Albedo –> More Sunlight –> Less Cooling

plus

Eruption –> Stratospheric SO2 Aerosols –> Increased Albedo –> Less Sunlight –> Cooling –> Delayed- or Non-Emergence Of Daily Tropical Thunderstorm Formation –> Decreased Thunderstorm Cooling –> Less Cooling

Both of these emergent phenomena, the daily appearance of sunlight-reflecting tropical cumulus cloud fields and cooling thunderstorms, work to oppose the cooling action of the volcanic stratospheric aerosols, and to prevent large swings in the global surface temperature.

I provide interesting evidence for this hypothesis in my post below.